The personal website of Scott W Harden
July 31st, 2016

Python Real-time Audio Frequency Monitor

A new project I'm working on requires real-time analysis of soundcard input data, and I made a minimal case example of how to do this in a cross-platform way using python 3, numpy, and PyQt. Previous posts compared performance of the matplotlib widget vs PyQtGraph plotwidget and I've been working with PyQtGraph ever since. For static figures matplotlib is wonderful, but for fast responsive applications I'm leaning toward PyQtGraph. The full source for this project is on a github page, but here's a summary of the project.

I made the UI with QT Designer. The graphs are QGraphicsView widgets promoted to a pyqtgraph PlotWidget. I describe how to do this in my previous post. Here's the content of the primary program:

from PyQt4 import QtGui,QtCore
import sys
import ui_main
import numpy as np
import pyqtgraph
import SWHear

class ExampleApp(QtGui.QMainWindow, ui_main.Ui_MainWindow):
    def __init__(self, parent=None):
        pyqtgraph.setConfigOption('background', 'w') #before loading widget
        super(ExampleApp, self).__init__(parent)
        self.setupUi(self)
        self.grFFT.plotItem.showGrid(True, True, 0.7)
        self.grPCM.plotItem.showGrid(True, True, 0.7)
        self.maxFFT=0
        self.maxPCM=0
        self.ear = SWHear.SWHear()
        self.ear.stream_start()

    def update(self):
        if not self.ear.data is None and not self.ear.fft is None:
            pcmMax=np.max(np.abs(self.ear.data))
            if pcmMax>self.maxPCM:
                self.maxPCM=pcmMax
                self.grPCM.plotItem.setRange(yRange=[-pcmMax,pcmMax])
            if np.max(self.ear.fft)>self.maxFFT:
                self.maxFFT=np.max(np.abs(self.ear.fft))
                self.grFFT.plotItem.setRange(yRange=[0,self.maxFFT])
            self.pbLevel.setValue(1000*pcmMax/self.maxPCM)
            pen=pyqtgraph.mkPen(color='b')
            self.grPCM.plot(self.ear.datax,self.ear.data,
                            pen=pen,clear=True)
            pen=pyqtgraph.mkPen(color='r')
            self.grFFT.plot(self.ear.fftx[:500],self.ear.fft[:500],
                            pen=pen,clear=True)
        QtCore.QTimer.singleShot(1, self.update) # QUICKLY repeat

if __name__=="__main__":
    app = QtGui.QApplication(sys.argv)
    form = ExampleApp()
    form.show()
    form.update() #start with something
    app.exec_()
    print("DONE")

This project uses a gutted version of the SWHEar class which I still haven't released on githib yet. It will likely have its own project folder. For now, take this project with a grain of salt. The primary advantage of this class is that it makes it easy to use PyAudio to automatically detect input sound cards, channels, and sample rates which are likely to succeed without requiring the user to enter any information.

All files used for this project are in a GitHub folder

Audio Visualizer Screenlet

2016-09-05: Okko adapted this project into a screenlet (cross platform) which also includes an installer for Windows: https://github.com/ninlith/audio-visualizer-screenlet

Markdown source code last modified on January 18th, 2021
---
title: Python Real-time Audio Frequency Monitor
date: 2016-07-31 15:21:11
tags: python, old
---

# Python Real-time Audio Frequency Monitor

__A new project I'm working on requires real-time analysis of soundcard input data, and I made a minimal case example of how to do this in a cross-platform way using python 3, numpy, and PyQt.__ Previous posts compared performance of the [matplotlib widget](https://www.swharden.com/wp/2016-07-30-live-data-in-pyqt4-with-matplotlibwidget/) vs [PyQtGraph plotwidget](https://www.swharden.com/wp/2016-07-31-live-data-in-pyqt4-with-plotwidget/) and I've been working with [PyQtGraph](http://www.pyqtgraph.org/) ever since. For static figures matplotlib is wonderful, but for fast responsive applications I'm leaning toward PyQtGraph. The [full source for this project is on a github page](https://github.com/swharden/Python-GUI-examples/tree/master/2016-07-37_qt_audio_monitor), but here's a summary of the project.

<div class="text-center img-border img-small">

![](demo-1.gif)

</div>

![](https://www.youtube.com/embed/lDS9rI0o6mM)

__I made the UI with QT Designer.__ The graphs are _QGraphicsView_ widgets promoted to a pyqtgraph_ PlotWidget_. I describe how to do this [in my previous post](https://www.swharden.com/wp/2016-07-31-live-data-in-pyqt4-with-plotwidget/). Here's the content of the primary program:

```python
from PyQt4 import QtGui,QtCore
import sys
import ui_main
import numpy as np
import pyqtgraph
import SWHear

class ExampleApp(QtGui.QMainWindow, ui_main.Ui_MainWindow):
    def __init__(self, parent=None):
        pyqtgraph.setConfigOption('background', 'w') #before loading widget
        super(ExampleApp, self).__init__(parent)
        self.setupUi(self)
        self.grFFT.plotItem.showGrid(True, True, 0.7)
        self.grPCM.plotItem.showGrid(True, True, 0.7)
        self.maxFFT=0
        self.maxPCM=0
        self.ear = SWHear.SWHear()
        self.ear.stream_start()

    def update(self):
        if not self.ear.data is None and not self.ear.fft is None:
            pcmMax=np.max(np.abs(self.ear.data))
            if pcmMax>self.maxPCM:
                self.maxPCM=pcmMax
                self.grPCM.plotItem.setRange(yRange=[-pcmMax,pcmMax])
            if np.max(self.ear.fft)>self.maxFFT:
                self.maxFFT=np.max(np.abs(self.ear.fft))
                self.grFFT.plotItem.setRange(yRange=[0,self.maxFFT])
            self.pbLevel.setValue(1000*pcmMax/self.maxPCM)
            pen=pyqtgraph.mkPen(color='b')
            self.grPCM.plot(self.ear.datax,self.ear.data,
                            pen=pen,clear=True)
            pen=pyqtgraph.mkPen(color='r')
            self.grFFT.plot(self.ear.fftx[:500],self.ear.fft[:500],
                            pen=pen,clear=True)
        QtCore.QTimer.singleShot(1, self.update) # QUICKLY repeat

if __name__=="__main__":
    app = QtGui.QApplication(sys.argv)
    form = ExampleApp()
    form.show()
    form.update() #start with something
    app.exec_()
    print("DONE")

```

This project uses a gutted version of the SWHEar class which I still haven't released on githib yet. It will likely have its own project folder. For now, take this project with a grain of salt. The primary advantage of this class is that it makes it easy to use PyAudio to automatically detect input sound cards, channels, and sample rates which are likely to succeed without requiring the user to enter any information.

All files used for this project are [in a GitHub folder](https://github.com/swharden/Python-GUI-examples/tree/master/2016-07-37_qt_audio_monitor)

### Audio Visualizer Screenlet

2016-09-05: Okko adapted this project into a screenlet (cross platform) which also includes an installer for Windows: https://github.com/ninlith/audio-visualizer-screenlet

<div class="text-center img-border">

[![](widget_thumb.jpg)](widget.png)

</div>
July 31st, 2016

Live Data in PyQt4 with PlotWidget

After spending a day comparing performance of MatplotlibWidget with PlotWidget, when it comes to speed PlotWidget wins by a mile! Glance over my last post where I describe how to set up the window with QT Designer and convert the .ui file to a .py file. With only a few changes to the code, I swapped the matplotlib _MatplotlibWidget _with the pyqtgraph PlotWidget. I easily got a 20x increase in speed (frame rate), and I'm likely to favor PyQtGraph over matpltolib for python applications involving realtime display of data. Just like the previous example using matplotlib, this one uses the system time to control the phase and color of a sine wave in real time. You can grab the full code from my github folder.

When designing the GUI with QT Designer, add a QGraphicsView widget then assign it to become a PyQtGraph object. To do this, follow the steps found on the pyqtgraph docs page:

  • In Designer, create a QGraphicsView widget
  • Right-click on the QGraphicsView and select “Promote To...”
  • Set “Promoted class name” to “PlotWidget”
  • Under “Header file”, enter “pyqtgraph”
  • Click “Add”, then click “Promote”
  • apparently this only needs to be done once per project

In addition to faster frame rate, the PyQtGraph method is easy to interact with. Clicking and dragging scrolls the data, and right-clicking and dragging zooms on each axis. Here's the program code:

from PyQt4 import QtGui,QtCore
import sys
import ui_main
import numpy as np
import pylab
import time
import pyqtgraph

class ExampleApp(QtGui.QMainWindow, ui_main.Ui_MainWindow):
    def __init__(self, parent=None):
        pyqtgraph.setConfigOption('background', 'w') #before loading widget
        super(ExampleApp, self).__init__(parent)
        self.setupUi(self)
        self.btnAdd.clicked.connect(self.update)
        self.grPlot.plotItem.showGrid(True, True, 0.7)

    def update(self):
        t1=time.clock()
        points=100 #number of data points
        X=np.arange(points)
        Y=np.sin(np.arange(points)/points*3*np.pi+time.time())
        C=pyqtgraph.hsvColor(time.time()/5%1,alpha=.5)
        pen=pyqtgraph.mkPen(color=C,width=10)
        self.grPlot.plot(X,Y,pen=pen,clear=True)
        print("update took %.02f ms"%((time.clock()-t1)*1000))
        if self.chkMore.isChecked():
            QtCore.QTimer.singleShot(1, self.update) # QUICKLY repeat

if __name__=="__main__":
    app = QtGui.QApplication(sys.argv)
    form = ExampleApp()
    form.show()
    form.update() #start with something
    app.exec_()
    print("DONE")

This project is available on GitHub: https://github.com/swharden/Python-GUI-examples/tree/master/2016-07-31_qt_PyQtGraph_sine_scroll

Markdown source code last modified on January 18th, 2021
---
title: Live Data in PyQt4 with PlotWidget
date: 2016-07-31 07:28:28
tags: python, old
---

# Live Data in PyQt4 with PlotWidget

__After spending a day comparing performance of MatplotlibWidget with PlotWidget, when it comes to speed PlotWidget wins by a mile!__ Glance over my [last post](https://www.swharden.com/wp/2016-07-30-live-data-in-pyqt4-with-matplotlibwidget/) where I describe how to set up the window with QT Designer and convert the .ui file to a .py file. With only a few changes to the code, I swapped the matplotlib _MatplotlibWidget _with the pyqtgraph _PlotWidget_. I easily got a 20x increase in speed (frame rate), and I'm likely to favor [PyQtGraph](http://www.pyqtgraph.org/) over matpltolib for python applications involving realtime display of data. Just like the [previous example using matplotlib](https://www.swharden.com/wp/2016-07-30-live-data-in-pyqt4-with-matplotlibwidget/), this one uses the system time to control the phase and color of a sine wave in real time. You can grab the full code from [my github folder](https://github.com/swharden/Python-GUI-examples/tree/master/2016-07-31_qt_PyQtGraph_sine_scroll).

<div class="text-center img-border img-small">

![](demo2.gif)
[![](demo2cmd_thumb.jpg)](demo2cmd.png)

</div>

__When designing the GUI with QT Designer, add a QGraphicsView widget then assign it to become a PyQtGraph object.__ To do this, follow the steps found on the [pyqtgraph docs page](http://www.pyqtgraph.org/documentation/how_to_use.html#embedding-widgets-inside-pyqt-applications):

* In Designer, create a QGraphicsView widget
* Right-click on the QGraphicsView and select “Promote To...”
* Set “Promoted class name” to “PlotWidget”
* Under “Header file”, enter “pyqtgraph”
* Click “Add”, then click “Promote”
* _apparently this only needs to be done once per project_

<div class="text-center img-border">

[![](promoted_thumb.jpg)](promoted.png)

</div>

__In addition to faster frame rate, the PyQtGraph method is easy to interact with.__ Clicking and dragging scrolls the data, and right-clicking and dragging zooms on each axis. Here's the program code:

```python
from PyQt4 import QtGui,QtCore
import sys
import ui_main
import numpy as np
import pylab
import time
import pyqtgraph

class ExampleApp(QtGui.QMainWindow, ui_main.Ui_MainWindow):
    def __init__(self, parent=None):
        pyqtgraph.setConfigOption('background', 'w') #before loading widget
        super(ExampleApp, self).__init__(parent)
        self.setupUi(self)
        self.btnAdd.clicked.connect(self.update)
        self.grPlot.plotItem.showGrid(True, True, 0.7)

    def update(self):
        t1=time.clock()
        points=100 #number of data points
        X=np.arange(points)
        Y=np.sin(np.arange(points)/points*3*np.pi+time.time())
        C=pyqtgraph.hsvColor(time.time()/5%1,alpha=.5)
        pen=pyqtgraph.mkPen(color=C,width=10)
        self.grPlot.plot(X,Y,pen=pen,clear=True)
        print("update took %.02f ms"%((time.clock()-t1)*1000))
        if self.chkMore.isChecked():
            QtCore.QTimer.singleShot(1, self.update) # QUICKLY repeat

if __name__=="__main__":
    app = QtGui.QApplication(sys.argv)
    form = ExampleApp()
    form.show()
    form.update() #start with something
    app.exec_()
    print("DONE")
```

This project is available on GitHub: https://github.com/swharden/Python-GUI-examples/tree/master/2016-07-31_qt_PyQtGraph_sine_scroll
July 30th, 2016

Live Data in PyQt4 with MatplotlibWidget

I keep getting involved in projects which require live data to be graphed in real time. Since most of my back-end is written in Python, it makes sense to have a Pythonic front-end. Cross-platform GUI programming in Python is frustratingly non-trivial, as there multiple window frameworks available (TK, GTK, and QT) and their respective graphical designers (torture, Glade, and QT Designer) and each has its own way of doing things. Add different ways to plot data in the mix (gnuplot, matplotlib, and custom widgets) and it can become a complicated mess. Different framework combinations favor different features (with unique speed / simplicity / elegance), so my goal is to slowly test out a few combinations most likely to work for my needs, and add my findings to a growing github repository. The first stab is using PyQt4 and matplotlib's widget (MatplotlibWidget). Rather than capture data from the sound card (my ultimate goal), I'm going to generate a sine wave whose phase and color is related to the system time. Matplotlib plotting is a bit slow, but the output is beautiful, and their framework is so robust. Here's the output of my first test showing the sine wave generated as well as the console output (showing that each call to the plotting function takes about 40 ms. At this rate, I can expect a maximum update rate of ~25 Hz.

Designing this project was easy, but it was surprisingly hard to figure out how to do this based on examples I found on the internet. This is part of why I wanted to place this example here. The downside of many internet examples is that they did not use Qt Designer to make the window, so their code to create a window and insert the MatplotlibWidget wasn't copy/paste compatible with my goals, and often more complex than I needed. Some internet examples did use Qt Designer to make the window, but left a frame empty which they later manually filled with a widget and attached to a matplotlib canvas. This is fine, but more complex than I need to get started.

First, I designed a GUI with Qt Designer. I dropped a MatplotlibWidget somewhere, and used its default name. I saved this file as ui_main.ui (which is an XML file, ready to be used for multiple programming languages).

Next, I converted the UI file into a .py file with a standalone python script that's an alternative to using pyuic from the command line. The script to do this is ui_convert.py and it calls PyQt4.uic.compileUi():

from PyQt4 import uic
fin = open('ui_main.ui','r')
fout = open('ui_main.py','w')
uic.compileUi(fin,fout,execute=False)
fin.close()
fout.close()

I then created my main program file which populates the matplotlib widget with data. I called it go.py and running it will launch the app. The code explains it all, and there's not much more to say! It produces the output at the top of this post.

from PyQt4 import QtGui,QtCore
import sys
import ui_main
import numpy as np
import pylab
import time

class ExampleApp(QtGui.QMainWindow, ui_main.Ui_MainWindow):
    def __init__(self, parent=None):
        super(ExampleApp, self).__init__(parent)
        self.setupUi(self)
        self.btnAdd.clicked.connect(self.update)
        self.matplotlibwidget.axes.hold(False) #clear on plot()

    def update(self):
        t1=time.time()
        points=100 #number of data points
        X=np.arange(points)
        Y=np.sin(np.arange(points)/points*3*np.pi+time.time())
        C=pylab.cm.jet(time.time()%10/10) # random color
        self.matplotlibwidget.axes.plot(X,Y,ms=100,color=C,lw=10,alpha=.8)
        self.matplotlibwidget.axes.grid()
        self.matplotlibwidget.axes.get_figure().tight_layout() # fill space
        self.matplotlibwidget.draw() # required to update the window
        print("update took %.02f ms"%((time.time()-t1)*1000))
        if self.chkMore.isChecked():
            QtCore.QTimer.singleShot(10, self.update) # QUICKLY repeat

if __name__=="__main__":
    app = QtGui.QApplication(sys.argv)
    form = ExampleApp()
    form.show()
    form.update() #start with something
    app.exec_()
    print("DONE")

This project is on GitHub: https://github.com/swharden/Python-GUI-examples/tree/master/2016-07-30_qt_matplotlib_sine_scroll

Markdown source code last modified on January 18th, 2021
---
title: Live Data in PyQt4 with MatplotlibWidget
date: 2016-07-30 23:15:23
tags: python, old
---

# Live Data in PyQt4 with MatplotlibWidget

__I keep getting involved in projects which require live data to be graphed in real time.__ Since most of my back-end is written in Python, it makes sense to have a Pythonic front-end. Cross-platform GUI programming in Python is frustratingly non-trivial, as there multiple window frameworks available (TK, GTK, and QT) and their respective graphical designers ([torture](https://wiki.python.org/moin/Intro%20to%20programming%20with%20Python%20and%20Tkinter), [Glade](https://glade.gnome.org/), and [QT Designer](http://doc.qt.io/qt-4.8/designer-manual.html)) and each has its own way of doing things. Add different ways to plot data in the mix ([gnuplot](http://www.gnuplot.info/), [matplotlib](http://matplotlib.org/), and [custom widgets](http://qcustomplot.com/)) and it can become a complicated mess. Different framework combinations favor different features (with unique speed / simplicity / elegance), so my goal is to slowly test out a few combinations most likely to work for my needs, and add my findings to a [growing github repository](https://github.com/swharden/Python-GUI-examples). The first stab is using PyQt4 and matplotlib's widget (MatplotlibWidget). Rather than capture data from the sound card (my ultimate goal), I'm going to generate a sine wave whose phase and color is related to the system time. Matplotlib plotting is a bit slow, but the output is beautiful, and their framework is so robust. Here's the output of my first test showing the sine wave generated as well as the console output (showing that each call to the plotting function takes about 40 ms. At this rate, I can expect a maximum update rate of ~25 Hz.

<div class="text-center img-border img-small">

![](demo.gif)
[![](qt4_thumb.jpg)](qt4.png)

</div>

__Designing this project was easy, but it was surprisingly hard to figure out how to do this based on examples I found on the internet.__ This is part of why I wanted to place this example here. The downside of many internet examples is that they did not use Qt Designer to make the window, so their code to create a window and insert the MatplotlibWidget wasn't copy/paste compatible with my goals, and often more complex than I needed. Some internet examples _did_ use Qt Designer to make the window, but left a frame empty which they later manually filled with a widget and attached to a matplotlib canvas. This is fine, but more complex than I need to get started.

__First, I designed a GUI with Qt Designer__. I dropped a MatplotlibWidget somewhere, and used its default name. I saved this file as [ui\_main.ui](https://github.com/swharden/Python-GUI-examples/blob/master/2016-07-30_qt_matplotlib_sine_scroll/ui_main.ui) (which is an XML file, ready to be used for [multiple programming languages](http://doc.qt.io/qt-4.8/designer-using-a-ui-file.html)).

<div class="text-center img-border">

[![](pyqt4-designer_thumb.jpg)](pyqt4-designer.png)

</div>

__Next, I converted the UI file into a .py file__ with a standalone python script that's an alternative to using pyuic from the command line. The script to do this is [ui\_convert.py](https://github.com/swharden/Python-GUI-examples/blob/master/2016-07-30_qt_matplotlib_sine_scroll/ui_convert.py) and it calls PyQt4.uic.compileUi():

```python
from PyQt4 import uic
fin = open('ui_main.ui','r')
fout = open('ui_main.py','w')
uic.compileUi(fin,fout,execute=False)
fin.close()
fout.close()
```

__I then created my main program file which populates the matplotlib widget with data.__ I called it [go.py](https://github.com/swharden/Python-GUI-examples/blob/master/2016-07-30_qt_matplotlib_sine_scroll/go.py) and running it will launch the app. The code explains it all, and there's not much more to say! It produces the output at the top of this post.

```python
from PyQt4 import QtGui,QtCore
import sys
import ui_main
import numpy as np
import pylab
import time

class ExampleApp(QtGui.QMainWindow, ui_main.Ui_MainWindow):
    def __init__(self, parent=None):
        super(ExampleApp, self).__init__(parent)
        self.setupUi(self)
        self.btnAdd.clicked.connect(self.update)
        self.matplotlibwidget.axes.hold(False) #clear on plot()

    def update(self):
        t1=time.time()
        points=100 #number of data points
        X=np.arange(points)
        Y=np.sin(np.arange(points)/points*3*np.pi+time.time())
        C=pylab.cm.jet(time.time()%10/10) # random color
        self.matplotlibwidget.axes.plot(X,Y,ms=100,color=C,lw=10,alpha=.8)
        self.matplotlibwidget.axes.grid()
        self.matplotlibwidget.axes.get_figure().tight_layout() # fill space
        self.matplotlibwidget.draw() # required to update the window
        print("update took %.02f ms"%((time.time()-t1)*1000))
        if self.chkMore.isChecked():
            QtCore.QTimer.singleShot(10, self.update) # QUICKLY repeat

if __name__=="__main__":
    app = QtGui.QApplication(sys.argv)
    form = ExampleApp()
    form.show()
    form.update() #start with something
    app.exec_()
    print("DONE")
```

This project is on GitHub: https://github.com/swharden/Python-GUI-examples/tree/master/2016-07-30_qt_matplotlib_sine_scroll
July 28th, 2016

Opto-Isolated Laser Controller Build

This page documents the design and build of a small device to interface a modern fiber-coupled DPSS laser with old scientific hardware designed to control mechanical relays. This project involves analog and digital circuitry, microcontrollers, and lasers, and it turned out to be a pretty cool build so I'm sharing the design and construction process online in case it will be helpful for others or even my future self.

The existing hardware I must interface is made by Coulbourn Instruments and is essentially just a large multi-channel computer-controlled DAC/ADC. It does its job well (turning lights on and off, recording button presses, etc.), but this new task requires millisecond resolution and modulation patterns which lies outside the specs of this system and software. My goal is to interface a free output line of this old hardware and use it to signal to a new device I build to activate the laser to produce a pulsed pattern. This way there would be no modification to any existing equipment, and no software to install. Further, since this hardware isn't mine, I don't like the idea of permanently modifying it (or even risking breaking it by designing something which could damage it by connecting to it). The specific goal is to allow the existing software to cause the laser to fire 20 ms pulses at 15 Hz for a few dozen cycles of 5s on, 5s off. It's also important to have some flexibility to reprogram the high speed laser stimulation pattern in the future. Experiments are already underway and I need this device to be complete within a couple of days! As much as I'd love to go to the internet and order the perfect cheap components, make a proper PCB, and have a beautiful build completed in a month, my goal is to build this over a weekend using only using parts I already have at my home.

Probing the existing hardware revealed many surprises. After inspecting the existing hardware setup I found an auxiliary output which could be controlled by software. This AUX port has a frustratingly rare connector 1mm dual keyhole touchproof connector which I couldn't buy in bulk on eBay or Amazon, and couldn't figure out the part numbers of on Mouser or Digikey. The manual even says "you may find it convenient to fit them with CI-type connectors" which makes me wonder why it wasn't designed this way in the first place! Luckily the laboratory had an old (broken) device with that connector on it they said I could use for this build. After plugging in the connector, I used a volt meter to measure the output. To my surprise, it wasn't a TTL signal! I expected to see my volt meter read 5V, but it read 28V! After consulting the manual I found mention of this: "Graphic State Notation software is designed for use with our Habitest animal-behavior-analysis environments or any other animal-behavior-testing apparatus that operates on the industry-standard 28-Volt control voltage." I was surprised that 28V signals is a standard for any industry.

Control voltages are negative! Elsewhere in the manual I found the phrase "The power base is capable of delivering 8 Amps of -28 VDC" which made me question the voltage reading I took earlier. The voltmeter showed 28V, but that's the difference between one keyhole connector output and the other. It became apparent that it really may be 0V (GND) and -28mV (an even more curious "industry standard"). I wondered if connecting the negative terminal to ground would destroy the unit (think about how easy this would be to do! If it were a TTL signal, the first thing you would do is connect the negative terminal to ground and start sampling the positive terminal). There was even talk of me interfacing with a different output port (which I hadn't probed, so I didn't know the voltage). Moving forward, I realized I had to tread very carefully. Doing something like connecting two grounds together could permanently damage this system!

Optical isolation should be used as a caution. Not really knowing if I should design to expect a TTL signal, a +28V signal, or a -28V signal, I decided to design a circuit to accommodate all of the above, while achieving total electrical discontinuity from whatever circuit I develop. I'm going to accomplish this using an opto-isolator on the input. I drew the schematic using KiCad as I built the board manually using through-hole construction. I considered laying-out a PCB (I have most of these components in SMT form factors too) but I knew I wouldn't manage a one-weekend turnaround if I went that direction.

Design Notes

  • The input should be able to accomodate any signal (TTL, CMOS, 28V, etc)
  • The input is totally isolated electrically, so this should be very safe on the hardware
  • The microcontroller is a socketed ATTiny85 which I programmed with a Bus Pirate.
  • I decided to rely on a crystal rather than the internal RC clock to improve temporal precision of the output signal. A 11.0592 MHz crystal was chosen because I already had an abundance of them (they're perfect for serial communication at all common baud rates). Any crystal could be used, as long as its frequency is defined in software.
  • Capacitors were added more to ensure oscillation initiates than to bring down the oscillation frequency. (I'm told that omitting them may cause a case where the crystal doesn't resonate as well, but I've never found this in my personal experience.) A good note on microcontroller clocks is in a Microchip PIC application note.
  • I included a "test" button (momentary switch) to simulate having an input signal.
  • Note that R1 must be able to handle the current applied to it. It was mistakenly designed as 1k, and later replaced with 10k. See the bodge note at the bottom of this post for details.

Potential Improvements

  • Forward protection diodes on the input could protect accidental reverse polarity
  • Adding an ICSP header would prevent de-socketing of the MCU if reprogramming is desired
  • The BNC output is directly from a MCU pin. A transistor-buffered output design could be considered to deliver higher current for more confident control of the laser input.

Because there is a possibility that a different output (laser control) pattern may be desired in the future, I considered whether or not I should make the output pattern user-configurable. Adding buttons, a display, and designing a menu system in software would be a lot of work and it's unclear if a protocol change will ever actually be required, so I concluded that I'm going to build this device to the specific task at hand and extended it later if/when needed. If the end user eventually wants the ability to modulate the pattern on their own, the device they ask for would be a very different one than the one I was tasked to create. Since the current pattern is burned into a microchip, a compromise is that I could have new patterns burned into new microchips, and the end-user could change the chip (as long as it's an infrequent occurrence).

EDIT: About a year later a modification was indeed required (something like changing 15 Hz to 40 Hz), and the solution was to burn these two patterns into two microchips. Since they're socketed, they're swappable (albiet a limited number of times). This design worked-out well.

Is a microcontroller/crystal overkill for a system which could be accomplished using an analog circuit? Generating 20 ms pulses at 15 Hz sounds like an easy task for a 555 timer without the need for digital circuitry. I considered this for a while, but concluded that the advantage of the MCU (crystal-disciplined time precision) outweighed the convenience of a purely analog circuit. A 555 timer in astable / multi-vibrator configuration would mostly get the job done, but you would either (1) only allow one output pattern and rely on precision passive components (which I don't have on hand), or (2) allow the end-user to adjust duty/frequency with potentiometers (which would require the output to be quantitatively monitored on an oscilloscope). There's also the issue where RC oscillators can be highly temperature sensitive, so the microcontroller/crystal design seemed like a more robust solution for reliable and defined behavior.

I started the build by measuring/marking drill points. I used a Dremel drill press to make the holes, then smoothed them with a deburring tool. I also drilled holes in the base of the enclosure.

I had an enclosure ready to go. I always buy enclosures in bulk, and even though nice ones tend to be expensive, having them on hand encourages me to build devices as I think of them, rather than making flaky hardware which I have a history of doing which sometimes borders on ridiculousness. I usually stock unfinished Hammond diecast aluminum enclosures for making quick RF projects, and boxes with feet and side vents for fancier projects, but for this task I decided to enclose everything inside a typical (but a little more costly) aluminum enclosure ordered in bulk from eBay. I love using low current LEDs, and I started going with frosted instead of clear LEDs because they're easier on the eyes. Also, I switched to mostly 3mm LEDs instead of 5MM because they look a little nicer in these small enclosures with small black bezels.

I used nicer perfboard with plated holes to build this circuit. Normally I use cheap perfboard with little copper rings glued to one side because it's faster to solder (the copper is so thin it heats quickly), but it's not always a good long-term solution because the copper pads have a tendency to un-stick. I rarely use this nicer perfboard (it is more expensive), but it's nice to have for more reliable builds.

I marked areas of optical isolation with a black marker. This makes it obvious where the potentially dangerous, potentially high-voltage (well, higher than TTL), potentially negative input comes in. No wires or connections should invade this space on the board. The special connector which will connect this device to the scientific hardware is on-site, and I'll have to solder it at the time of delivery/installation. I left an extra hole in the back to accommodate this wire. I didn't have any rubber grommets, but I suspect this build may have benefitted from one.

Once it was all together, the device seemed to perform well. The test button on the back made it easy to inspect the output. I build so many RF circuits that I instinctively reached for a 50-ohm terminator, but the square wave quickly transformed into shark fins (RC curves) reminding me that and I realized 50 ohms is far too low impedance. If it's a TTL signal, let's assume it's virtually infinite impedance. I was uncertain whether or not I should drive the output directly with a microcontroller pin. There may be a need for a buffered output. The microcontroller's datasheet suggests limiting its current to 20 mA per pin (requiring termination of no less than 250 Ohms at 5V), and I'm going to move forward assuming the laser TTL input doesn't sink much current.

How should TTL timing be controlled in software? I want this device to perform identically over long periods of time, favoring reliability and consistency over microsecond time precision. To achieve 15 Hz of 20ms pulses I need 20ms on and 46.666667 ms off. I could probably get pretty close if I wanted to, but I rounded it to 20 ms on and 46 ms off. This gives time for the instruction cycles toggling the output pins to occur (although it's on an order of magnitude faster time scale), which slightly biases the time in the right direction. I considered adding a _delay_us(666) after the _delay_ms(46) but I'm satisfied with it this knowing it's within 1% accuracy of 15 Hz and that precision is locked to that of the crystal (around 10 ppm, or 0.001%).

Hard-coded _delay_ms() is somewhat inelegant. The use of AVR timers should probably be considered as an alternative strategy. Here's an awesome guide on the topic, and here's another. Timers would be preferred if I wanted the program code of the microcontroller to be free to do other things like drive menus or multiplex a display. Since this task is simply required generation of a TTL output pattern (and nothing more), the hard-coded delay method met this need, but I found it useful to consider the asynchronous strategies:

  • Timers: Set the timer to overflow every 1 ms. On overflow, a counting variable would be incremented and a function would be called to determine what to do. At pre-programmed time points (with respect to the counting variable), the output pin would be toggled, or the counting variable would be reset.

  • Output compare registers: Utilize the built-in OCR (output compare register) to turn the output signal on and off. Set the timer to overflow at 15 Hz, turning the output on. Set the OCR (to the fractional point between 0 and the maximum timer value) such that when it is passed, the output is turned off. This way 15 Hz, 20 ms pulses would be continuously running without any code being executed. Input sensing could simply enable and disable the timer.

  • Input interrupts: Why stop at timers? Polling the input pin for a TTL signal puts the chip in an infinite loop. Relying on the AVR's external (pin change) hardware interrupts could eliminate this as well. I always rely heavily on the datasheet when setting these interrupts.

These alternative implementations will be useful in the future if a more accurate time source is desired, an advanced display is added, or menus are implemented which would benefit from letting the pulsing output operate in the background while accepting user input. For now, I'm happy with the blocking delay strategy.

After I was satisfied with construction, I started labeling the enclosure. I want to tip my hat to Onno Hoekstra on this one, as his webpage and some email correspondence helped me realize how good clear labels look when outlined and applied to aluminum enclosures. I'm using a DYMO LetraTag LT-100T Plus label maker and clear tape. It's important to enable the black outline around the text, then I cut carefully slightly outside the outline with regular scissors, and apply the labels with a hobby knife or razor blade.

The morning I delivered the product and added the final proprietary connector which I didn't have at home. It's an inelegant knot-retained configuration, but I think it'll get the job done! It was a surprisingly rare, fully shielded, keyhole-shaped touchproof connector apparently used only in medical applications. At this point, I'm thinking this connector was chosen to (A) protect the user from accidentally shorting a 28V 8A power source (that's over 200 watts!), (B) to prevent you from damaging the equipment by plugging in something that doesn't belong (could you imagine what would happen if this -28V high current source had a BNC connector and you plugged this into something expecting a 5V TTL input?), and (C) prevent you from plugging in anything that wasn't made by this company. The last option is most likely a well-intentioned attempt by the manufacturer to prevent customers from damaging their product rather than the company trying to maintain its status as a sole distributor of accessories, but it makes me wonder. I would have preferred power pole sockets, molded power connectors like those on motherboards, or even barrel connectors! Surely there's a more standard touchproof connector for moderate voltage/currents than this bizarre keyhole connector.

I plugged the device in to the computer, attached the laser, and it worked immediately! I wasn't really surprised that it worked (I tested it extensively at home), but it still felt good to watch the blue laser trigger as it was supposed to. Another interesting one-off project is complete, and have some interesting photos and notes about the build to share on this website. I hope this little device continues to do its job well for many years in its new laboratory home.

Microcontroller Code

#define F_CPU 11059200UL
#include <avr/io.h>
#include <util/delay.h>

int main (void){
    DDRB=(1<<PB0); // set port B pin 0 as an output
    PORTB=0;       // pull all pins low
    while(1){
        while((PINB&(1<<PB2))==0){} // do nothing while the input is low
        PORTB=(1<<PB0);   // TTL ON
        _delay_ms(20);    // stay on for 20ms
        PORTB&=~(1<<PB0); // TTL OFF
        _delay_ms(46);    // stay off for 46ms
    }
}

Here's the batch script I used to compile and load the code onto the microcontroller. I compiled the code with AVR-GCC and copied it onto the microcontroller with a Bus Pirate. Note also that I'm setting the fuses to respect an external oscillator.

@echo off
del *.elf
del *.hex
avr-gcc -mmcu=attiny85 -Wall -Os -o main.elf main.c
avr-objcopy -j .text -j .data -O ihex main.elf main.hex
avrdude -c buspirate -p attiny85 -P com3 -e -U flash:w:main.hex
avrdude -c buspirate -p attiny85 -P com3 -U lfuse:w:0xff:m -U hfuse:w:0xdf:m -U efuse:w:0xff:m
pause

Final Design

Update: Improve Current Handling

After a few days I got an email from someone concerned about the current handling capability of the front-end of the circuit. It was noted that a standard 1/4 watt resistor may not be suitable for R1, as a 28V potential would stress it beyond its specs. With 28V applied, R1 (a quarter-watt resistor) would experience P=IE=28mA*28V=784mW of current! It might last (especially if pulsed), but it also might fail with time. The advantage of the R1/D1/R2 system is that the output current will be identical across a wide range of input voltages. The disadvantage is that it's hard to predict how much current R1 needs to be expected to tolerate. I could have placed five 4.7k resistors in parallel to replace R1 (this would let me handle over 1 watt of input power), but I instead simply upped it from 1kOhm to 10kOhm. This further reduced the current that the opto-isolator sees (now only about 0.2 mA) but it seems to work still. I'm satisfied with this modification, but a little disappointed I didn't catch it sooner. Note that the new input resistor (a 10k R1) should now only have to dissipate about 80mW, well within its specs.

Update: H11B1 minimum current and AC noise

What is the minimum current required to confidently activate the optoisolator? I considered that a 10K input resistor on 28V would only allow 2.8 mA to pass, and I was unsure if this would reliably activate the optocoupler. Considering only 3.3V will persist after the zener (a ~11.7% current retaining ratio, if that's valid), I figured that a best 330µA were passing through the opto-isolator. That seems outside of the specs of the device, because their datasheet graphs always start at 1mA. I decided to run some tests at my home to see how this device performed with lower input currents. I determined that a 10k resistor still works with 5V (500 µA into the device), but checking the output on the oscilloscope I realized that the device operates only partially, and slowly at that low voltage/current. The darlington transistor configuration is very high gain, which is the only reason this works at all, but such low currents are sensitive to parasitic capacitance and infiltrating RF currents. Because of this it seems the chip takes a few ms to activate and deactivate. Since this application only uses 5s on and 5s off inputs, it's fine for this application, but I wouldn't expect high speed pulsing of the input signal to work well. Furthermore, in my breadboard I realized I was getting funny output currents oscillating around 60Hz, which made me suspicious that the device was picking up AC somehow. I realized it was from pin 6 (the exposed darlington base). Normally the LED is so strong is blasts the device fully on or off, but hovering on the edge like this, that pin is picking up signals that influence the open state of the transistor inside. Since it's not connected to anything anyway, I cut the pin off as close to the microchip as I could, and noticed an instant improvement in 60Hz rejection. In conclusion, I wouldn't try to reliably drive an opto-isolator with a complex pattern using less than 1 mA, but it seems to work well for simple on/off control.

This is the output of the unmodified H11B1:

This is the output of the H11B1 with the base pin removed:

Conclusion: This was an interesting build! I'm satisfied with the use of optical isolation here to adapt two incompatible systems with unknown/unspecified input/output properties, and the lack of electrical connections between the inputs and outputs gave me high confidence to experiment along the way.

Markdown source code last modified on May 4th, 2021
---
title: Opto-Isolated Laser Controller Build
date: 2016-07-28 02:33:12
tags: microcontroller, circuit, old
---

# Opto-Isolated Laser Controller Build

**This page documents the design and build of a small device to interface a [modern fiber-coupled DPSS laser](http://www.lasercentury.com/product.asp?id=612) with [old scientific hardware](http://www.coulbourn.com/product_p/h02-08.htm) designed to control mechanical relays.** This project involves analog and digital circuitry, microcontrollers, and lasers, and it turned out to be a pretty cool build so I'm sharing the design and construction process online in case it will be helpful for others or even my future self.

<div class="text-center img-border">

[![](IMG_7304_thumb.jpg)](IMG_7304.jpg)

</div>

**The existing hardware I must interface is made by [Coulbourn Instruments](http://www.coulbourn.com/) and is essentially just a large multi-channel computer-controlled DAC/ADC**. It does its job well (turning lights on and off, recording button presses, etc.), but this new task requires millisecond resolution and modulation patterns which lies outside the specs of this system and software. **My goal is to interface a free output line of this old hardware and use it to signal to a new device I build to activate the laser to produce a pulsed pattern.** This way there would be no modification to any existing equipment, and no software to install. Further, since this hardware isn't mine, I don't like the idea of permanently modifying it (or even risking breaking it by designing something which could damage it by connecting to it). The specific goal is to allow the existing software to cause the laser to fire 20 ms pulses at 15 Hz for a few dozen cycles of 5s on, 5s off. **It's also important to have some flexibility to reprogram the high speed laser stimulation pattern in the future.** Experiments are already underway and I need this device to be complete within a couple of days! As much as I'd love to go to the internet and order the perfect cheap components, make a proper PCB, and have a beautiful build completed in a month, my goal is to build this over a weekend using only using parts I already have at my home.

<div class="text-center img-micro">

[![](hardware-stack_thumb.jpg)](hardware-stack.jpg)
![](hardware-stack2.jpg)
[![](hardware-pci_thumb.jpg)](hardware-pci.jpg)

</div>

**Probing the existing hardware revealed many surprises.** After inspecting the existing hardware setup I found an auxiliary output which could be controlled by software. This AUX port has a [frustratingly rare connector 1mm dual keyhole touchproof connector](http://shop.cephalon.eu/Braebon,-Keyhole-to-two-touchproof-(1,5mm)-connect/ItemDetails.aspx?9=GB&5=0574&11=1531) which I couldn't buy in bulk on eBay or Amazon, and couldn't figure out the part numbers of on Mouser or Digikey. The manual even says "_you may find it convenient to fit them with CI-type connectors_" which makes me wonder why it wasn't designed this way in the first place! Luckily the laboratory had an old (broken) device with that connector on it they said I could use for this build. After plugging in the connector, I used a volt meter to measure the output. To my surprise, it wasn't a [TTL signal](https://en.wikipedia.org/wiki/Transistor%E2%80%93transistor_logic)! I expected to see my volt meter read 5V, but it read 28V! [After consulting the manual](http://www.coulbourn.com/v/vspfiles/assets/manuals/GraphicStateNotation2-101SoftwareUserGuide.pdf) I found mention of this: "_Graphic State Notation software is designed for use with our Habitest animal-behavior-analysis environments or any other animal-behavior-testing apparatus that operates on the industry-standard 28-Volt control voltage._" I was surprised that 28V signals is a standard for any industry. 

**Control voltages are negative!** Elsewhere in the manual I found the phrase "_The power base is capable of delivering 8 Amps of -28 VDC_" which made me question the voltage reading I took earlier. The voltmeter showed 28V, but that's the difference between one keyhole connector output and the other. It became apparent that it really may be 0V (GND) and -28mV (an even more curious "industry standard"). I wondered if connecting the negative terminal to ground would destroy the unit (think about how easy this would be to do! If it were a TTL signal, the first thing you would do is connect the negative terminal to ground and start sampling the positive terminal). There was even talk of me interfacing with a different output port (which I hadn't probed, so I didn't know the voltage). _Moving forward, I realized I had to tread very carefully. Doing something like connecting two grounds together could permanently damage this system!_ 

**Optical isolation should be used as a caution.** Not really knowing if I should design to expect a TTL signal, a +28V signal, or a -28V signal, I decided to design a circuit to accommodate all of the above, while achieving _total electrical discontinuity_ from whatever circuit I develop. I'm going to accomplish this using an [opto-isolator on the input](https://en.wikipedia.org/wiki/Opto-isolator). I drew the schematic using [KiCad](http://kicad-pcb.org/) as I built the board manually using through-hole construction. I considered laying-out a PCB (I have most of these components in SMT form factors too) but I knew I wouldn't manage a one-weekend turnaround if I went that direction.

<div class="text-center">

[![](schematic-4_thumb.jpg)](schematic-4.png)

</div>

### Design Notes

*   The input should be able to accomodate any signal (TTL, CMOS, 28V, etc)
*   The input is _totally_ isolated electrically, so this should be very safe on the hardware
*   The microcontroller is a socketed [ATTiny85](http://www.atmel.com/images/atmel-2586-avr-8-bit-microcontroller-attiny25-attiny45-attiny85_datasheet.pdf) which I programmed with a [Bus Pirate](https://www.swharden.com/wp/2016-07-14-controlling-bus-pirate-with-python/).
*   I decided to rely on a [crystal rather than the internal RC clock](http://www.electroschematics.com/9481/avr-clock-source-fuse-bits/) to improve temporal precision of the output signal. A 11.0592 MHz crystal was chosen because I already had an abundance of them (they're [perfect for serial communication at all common baud rates](http://wormfood.net/avrbaudcalc.php)). Any crystal could be used, as long as its frequency is defined in software.
*   Capacitors were added more to ensure oscillation initiates than to bring down the oscillation frequency. (I'm told that omitting them may cause a case where the crystal doesn't resonate as well, but I've never found this in my personal experience.) A good note on microcontroller clocks is in a [Microchip PIC application note](http://ww1.microchip.com/downloads/en/appnotes/00826a.pdf).
*   I included a "test" button (momentary switch) to simulate having an input signal.
*   Note that R1 must be able to handle the current applied to it. It was mistakenly designed as 1k, and later replaced with 10k. See the bodge note at the bottom of this post for details.

### Potential Improvements

*   Forward protection diodes on the input could protect accidental reverse polarity
*   Adding an [ICSP](http://www.ladyada.net/learn/avr/programming.html) header would prevent de-socketing of the MCU if reprogramming is desired
*   The BNC output is directly from a MCU pin. A [transistor-buffered](https://en.wikipedia.org/wiki/Buffer_amplifier#Single-transistor_circuits) output design could be considered to deliver higher current for more confident control of the laser input.

**Because there is a possibility that a different output (laser control) pattern may be desired in the future, I considered whether or not I should make the output pattern user-configurable.** Adding buttons, a display, and designing a menu system in software would be a lot of work and it's unclear if a protocol change will ever actually be required, so I concluded that I'm going to build this device to the specific task at hand and extended it later if/when needed. If the end user eventually wants the ability to modulate the pattern on their own, the device they ask for would be a very different one than the one I was tasked to create. Since the current pattern is burned into a microchip, a compromise is that I could have new patterns burned into new microchips, and the end-user could change the chip (as long as it's an infrequent occurrence).

_EDIT: About a year later a modification was indeed required (something like changing 15 Hz to 40 Hz), and the solution was to burn these two patterns into two microchips. Since they're socketed, they're swappable (albiet a limited number of times). This design worked-out well._

**Is a microcontroller/crystal overkill for a system which could be accomplished using an analog circuit?** Generating 20 ms pulses at 15 Hz sounds like an easy task for a [555 timer](https://en.wikipedia.org/wiki/555_timer_IC) without the need for digital circuitry. I considered this for a while, but concluded that the advantage of the MCU (crystal-disciplined time precision) outweighed the convenience of  a purely analog circuit. A 555 timer in [astable / multi-vibrator configuration](http://www.circuitbasics.com/555-timer-basics-astable-mode/) would mostly get the job done, but you would either (1) only allow one output pattern and rely on precision passive components (which I don't have on hand), or (2) allow the end-user to adjust duty/frequency with potentiometers (which would require the output to be quantitatively monitored on an oscilloscope). There's also the issue where RC oscillators can be highly temperature sensitive, so the microcontroller/crystal design seemed like a more robust solution for reliable and defined behavior.

**I started the build** by measuring/marking drill points. I used a Dremel drill press to make the holes, then smoothed them with a deburring tool. I also drilled holes in the base of the enclosure.

<div class="text-center img-small img-border">

[![](IMG_7186_thumb.jpg)](IMG_7186.jpg)
[![](IMG_7188_thumb.jpg)](IMG_7188.jpg)
[![](IMG_7191_thumb.jpg)](IMG_7191.jpg)
[![](IMG_7192_thumb.jpg)](IMG_7192.jpg)
[![](IMG_7203_thumb.jpg)](IMG_7203.jpg)

</div>

**I had an enclosure ready to go.** I always buy enclosures in bulk, and even though nice ones tend to be expensive, having them on hand encourages me to build devices as I think of them, rather than making flaky hardware which I [have a history of doing](https://www.swharden.com/wp/2011-01-16-first-homebrew-qso-ever/) which sometimes [borders on ridiculousness](https://www.swharden.com/wp/2010-06-03-aj4vd-qrss-vd-mept/). I usually stock [unfinished Hammond diecast aluminum enclosures](http://www.alliedelec.com/hammond-manufacturing-1590g/70165610/) for making quick RF projects, and [boxes with feet and side vents](https://www.amazon.com/s/ref=nb_sb_noss?url=search-alias%3Daps&field-keywords=aluminum+electronic+project+enclosure+box) for fancier projects, but for this task I decided to enclose everything inside a typical (but a little more costly) aluminum enclosure ordered in bulk from eBay. I love using low current LEDs, and I started going with frosted instead of clear LEDs because they're easier on the eyes. Also, I switched to mostly 3mm LEDs instead of 5MM because they look a little nicer in these small enclosures with small black bezels.

<div class="text-center img-small img-border">

[![](IMG_7191_thumb.jpg)](IMG_7191.jpg)
[![](IMG_7206_thumb.jpg)](IMG_7206.jpg)
[![](IMG_7204_thumb.jpg)](IMG_7204.jpg)
[![](IMG_7208_thumb.jpg)](IMG_7208.jpg)

</div>

**I used nicer perfboard with plated holes to build this circuit.** Normally I use [cheap perfboard](https://www.circuitspecialists.com/solderable-perf-boards) with little copper rings glued to one side because it's faster to solder (the copper is so thin it heats quickly), but it's not always a good long-term solution because the copper pads have a tendency to un-stick. I rarely use this nicer perfboard (it is more expensive), but it's nice to have for more reliable builds.

<div class="text-center img-small img-border">

[![](IMG_7258_thumb.jpg)](IMG_7258.jpg)
[![](IMG_7215_thumb.jpg)](IMG_7215.jpg)
[![](IMG_7216_thumb.jpg)](IMG_7216.jpg)
[![](IMG_7252_thumb.jpg)](IMG_7252.jpg)
[![](IMG_7259_thumb.jpg)](IMG_7259.jpg)

</div>

**I marked areas of optical isolation with a black marker.** This makes it obvious where the potentially dangerous, potentially high-voltage (well, higher than TTL), potentially negative input comes in. No wires or connections should invade this space on the board. The special connector which will connect this device to the scientific hardware is on-site, and I'll have to solder it at the time of delivery/installation. I left an extra hole in the back to accommodate this wire. I didn't have any [rubber grommets](http://www.ebay.com/sch/i.html?_from=R40&_trksid=p2050601.m570.l1313.TR0.TRC0.H0.Xrubber+grommets.TRS0&_nkw=rubber+grommets&_sacat=0), but I suspect this build may have benefitted from one.

<div class="text-center img-border">

[![](IMG_7232_thumb.jpg)](IMG_7232.jpg)

</div>

**Once it was all together, the device seemed to perform well.** The test button on the back made it easy to inspect the output. I build so many RF circuits that I instinctively reached for a 50-ohm terminator, but the square wave quickly transformed into shark fins (RC curves) reminding me that and I realized 50 ohms is far too low impedance. If it's a TTL signal, let's assume it's virtually infinite impedance. I was uncertain whether or not I should drive the output directly with a microcontroller pin. There may be a need for a buffered output. The microcontroller's datasheet suggests limiting its current to 20 mA per pin (requiring termination of no less than 250 Ohms at 5V), and I'm going to move forward assuming the laser TTL input doesn't sink much current.

<div class="text-center img-border img-medium">

[![](IMG_7284_thumb.jpg)](IMG_7284.jpg)
[![](scope_thumb.jpg)](scope.jpg)

</div>

**How should TTL timing be controlled in software?** I want this device to perform identically over long periods of time, favoring reliability and consistency over microsecond time precision. To achieve 15 Hz of 20ms pulses I need 20ms on and 46.666667 ms off. I could probably get pretty close if I wanted to, but I rounded it to 20 ms on and 46 ms off. This gives time for the instruction cycles toggling the output pins to occur (although it's on an order of magnitude faster time scale), which slightly biases the time in the right direction. I considered adding a _delay_us(666) after the _delay_ms(46) but I'm satisfied with it this knowing it's within 1% accuracy of 15 Hz and that precision is locked to that of the crystal (around 10 ppm, or 0.001%).

**Hard-coded \_delay\_ms() is somewhat inelegant.** The use of [AVR timers](http://www.atmel.com/Images/Atmel-2505-Setup-and-Use-of-AVR-Timers_ApplicationNote_AVR130.pdf) should probably be considered as an alternative strategy. Here's an [awesome guide on the topic](http://maxembedded.com/2011/06/introduction-to-avr-timers/), and [here's another](http://www.avrfreaks.net/forum/tut-c-newbies-guide-avr-timers?page=all). Timers would be preferred if I wanted the program code of the microcontroller to be free to do other things like drive menus or multiplex a display. Since this task is simply required generation of a TTL output pattern (and nothing more), the hard-coded delay method met this need, but I found it useful to consider the asynchronous strategies:

*   **Timers**: [Set the timer](http://eleccelerator.com/avr-timer-calculator/) to overflow every 1 ms. On overflow, a counting variable would be incremented and a function would be called to determine what to do. At pre-programmed time points (with respect to the counting variable), the output pin would be toggled, or the counting variable would be reset.

*   **Output compare registers:** Utilize the built-in OCR (output compare register) to turn the output signal on and off. [Set the timer](http://eleccelerator.com/avr-timer-calculator/) to overflow at 15 Hz, turning the output on. Set the OCR (to the fractional point between 0 and the maximum timer value) such that when it is passed, the output is turned off. This way 15 Hz, 20 ms pulses _would be continuously running without any code being executed_. Input sensing could simply enable and disable the timer.

*   **Input interrupts:** Why stop at timers? Polling the input pin for a TTL signal puts the chip in an infinite loop. Relying on the AVR's external (pin change) hardware interrupts could eliminate this as well. I always [rely heavily on the datasheet](http://www.atmel.com/images/atmel-2586-avr-8-bit-microcontroller-attiny25-attiny45-attiny85_datasheet.pdf) when setting these interrupts.

**These alternative implementations will be useful in the future** if a more accurate time source is desired, an advanced display is added, or menus are implemented which would benefit from letting the pulsing output operate in the background while accepting user input. For now, I'm happy with the blocking delay strategy.

<div class="text-center img-border img-small">

[![](IMG_7265_thumb.jpg)](IMG_7265.jpg)
[![](IMG_7262_thumb.jpg)](IMG_7262.jpg)
[![](IMG_7301_thumb.jpg)](IMG_7301.jpg)
[![](IMG_7300_thumb.jpg)](IMG_7300.jpg)

</div>

**After I was satisfied with construction, I started labeling the enclosure.** I want to tip my hat to [Onno Hoekstra](http://www.qsl.net/pa2ohh/) on this one, as [his webpage](http://www.qsl.net/pa2ohh/tlabels.htm) and some email correspondence helped me realize how good clear labels look when outlined and applied to aluminum enclosures. I'm using a [DYMO LetraTag LT-100T Plus](https://www.amazon.com/DYMO-LetraTag-Personal-Hand-Held-1733013/dp/B001B1FIW2/) label maker and [clear tape](https://www.amazon.com/DYMO-Labeling-LetraTag-Labelers-Black/dp/B00006B8FA). It's important to enable the black outline around the text, then I cut carefully slightly outside the outline with regular scissors, and apply the labels with a hobby knife or razor blade.

<div class="text-center img-border img-small">

[![](IMG_7310_thumb.jpg)](IMG_7310.jpg)
[![](IMG_7308_thumb.jpg)](IMG_7308.jpg)
[![](IMG_7307_thumb.jpg)](IMG_7307.jpg)
[![](IMG_7305_thumb.jpg)](IMG_7305.jpg)

</div>

**The morning I delivered the product** and added the final proprietary connector which I didn't have at home. It's an inelegant knot-retained configuration, but I think it'll get the job done! It was a [surprisingly rare, fully shielded, keyhole-shaped touchproof connector](http://shop.cephalon.eu/Braebon,-Keyhole-to-two-touchproof-(1,5mm)-connect/ItemDetails.aspx?9=GB&5=0574&11=1531) apparently used only in medical applications. At this point, I'm thinking this connector was chosen to (A) protect the user from accidentally shorting a 28V 8A power source (that's over 200 watts!), (B) to prevent you from damaging the equipment by plugging in something that doesn't belong (could you imagine what would happen if this -28V high current source had a BNC connector and you plugged this into something expecting a 5V TTL input?), and (C) prevent you from plugging in anything that wasn't made by this company. The last option is most likely a well-intentioned attempt by the manufacturer to prevent customers from damaging their product rather than the company trying to maintain its status as a sole distributor of accessories, but it makes me wonder. I would have preferred power pole sockets, molded power connectors like those on motherboards, or even barrel connectors! Surely there's a more standard touchproof connector for moderate voltage/currents than this bizarre keyhole connector.

**I plugged the device in to the computer, attached the laser, and it worked immediately!** I wasn't really surprised that it worked (I tested it extensively at home), but it still felt good to watch the blue laser trigger as it was supposed to. Another interesting one-off project is complete, and have some interesting photos and notes about the build to share on this website. I hope this little device continues to do its job well for many years in its new laboratory home.

<div class="text-center img-border">

[![](IMG_7316_thumb.jpg)](IMG_7316.jpg)

</div>

### Microcontroller Code

```c
#define F_CPU 11059200UL
#include <avr/io.h>
#include <util/delay.h>

int main (void){
    DDRB=(1<<PB0); // set port B pin 0 as an output
    PORTB=0;       // pull all pins low
    while(1){
        while((PINB&(1<<PB2))==0){} // do nothing while the input is low
        PORTB=(1<<PB0);   // TTL ON
        _delay_ms(20);    // stay on for 20ms
        PORTB&=~(1<<PB0); // TTL OFF
        _delay_ms(46);    // stay off for 46ms
    }
}
```

**Here's the batch script I used to compile and load the code onto the microcontroller.** I compiled the code with AVR-GCC and copied it onto the microcontroller with a [Bus Pirate](https://www.swharden.com/wp/2016-07-14-controlling-bus-pirate-with-python/). Note also that I'm [setting the fuses to respect an external oscillator](http://www.engbedded.com/fusecalc/).

```bash
@echo off
del *.elf
del *.hex
avr-gcc -mmcu=attiny85 -Wall -Os -o main.elf main.c
avr-objcopy -j .text -j .data -O ihex main.elf main.hex
avrdude -c buspirate -p attiny85 -P com3 -e -U flash:w:main.hex
avrdude -c buspirate -p attiny85 -P com3 -U lfuse:w:0xff:m -U hfuse:w:0xdf:m -U efuse:w:0xff:m
pause
```

### Final Design

<div class="text-center img-border">

[![](IMG_7304_thumb.jpg)](IMG_7304.jpg)

</div>

### Update: Improve Current Handling

**After a few days I got an email from someone concerned about the current handling capability of the front-end of the circuit.** It was noted that a standard 1/4 watt resistor may not be suitable for R1, as a 28V potential would stress it beyond its specs. With 28V applied, R1 (a quarter-watt resistor) would experience P=IE=28mA*28V=784mW of current! It might last (especially if pulsed), but it also might fail with time. The advantage of the R1/D1/R2 system is that the output current will be identical across a wide range of input voltages. The disadvantage is that it's hard to predict how much current R1 needs to be expected to tolerate. I could have placed five 4.7k resistors in parallel to replace R1 (this would let me handle over 1 watt of input power), but I instead simply upped it from 1kOhm to 10kOhm. This further reduced the current that the opto-isolator sees (now only about 0.2 mA) but it seems to work still. I'm satisfied with this modification, but a little disappointed I didn't catch it sooner. Note that the new input resistor (a 10k R1) should now only have to dissipate about 80mW, well within its specs.

<div class="text-center img-border">

[![](IMG_7368_thumb.jpg)](IMG_7368.jpg)

</div>

### Update: H11B1 minimum current and AC noise

**What is the minimum current required to confidently activate the optoisolator?** I considered that a 10K input resistor on 28V would only allow 2.8 mA to pass, and I was unsure if this would reliably activate the optocoupler. Considering only 3.3V will persist after the zener (a ~11.7% current retaining ratio, if that's valid), I figured that a best 330µA were passing through the opto-isolator. That seems outside of the specs of the device, because their datasheet graphs always start at 1mA. I decided to run some tests at my home to see how this device performed with lower input currents. I determined that a 10k resistor still works with 5V (500 µA into the device), but checking the output on the oscilloscope I realized that the device operates only partially, and slowly at that low voltage/current. The darlington transistor configuration is very high gain, which is the only reason this works at all, but such low currents are sensitive to parasitic capacitance and infiltrating RF currents. Because of this it seems the chip takes a few ms to activate and deactivate. Since this application only uses 5s on and 5s off inputs, it's fine for this application, but I wouldn't expect high speed pulsing of the input signal to work well. Furthermore, in my breadboard I realized I was getting funny output currents oscillating around 60Hz, which made me suspicious that the device was picking up AC somehow. I realized it was from pin 6 (the exposed darlington base). Normally the LED is so strong is blasts the device fully on or off, but hovering on the edge like this, that pin is picking up signals that influence the open state of the transistor inside. Since it's not connected to anything anyway, I cut the pin off as close to the microchip as I could, and noticed an instant improvement in 60Hz rejection. In conclusion, I wouldn't try to reliably drive an opto-isolator with a complex pattern using less than 1 mA, but it seems to work well for simple on/off control.

This is the output of the unmodified H11B1:

<div class="text-center img-border img-small">

[![](IMG_7495_thumb.jpg)](IMG_7495.jpg)
![](SDS00008.bmp)

</div>

This is the output of the H11B1 with the base pin removed:

<div class="text-center img-border img-small">

[![](IMG_7497_thumb.jpg)](IMG_7497.jpg)
![](SDS00010.bmp)

</div>

**Conclusion:** This was an interesting build! I'm satisfied with the use of optical isolation here to adapt two incompatible systems with unknown/unspecified input/output properties, and the lack of electrical connections between the inputs and outputs gave me high confidence to experiment along the way.
July 19th, 2016

Realtime Audio Visualization in Python

Python's "batteries included" nature makes it easy to interact with just about anything... except speakers and a microphone! As of this moment, there still are not standard libraries which which allow cross-platform interfacing with audio devices. There are some pretty convenient third-party modules, but I hope in the future a standard solution will be distributed with python. I appreciate the differences of Linux architectures such as ALSA and OSS, but toss in Windows and MacOS in the mix and it gets to be a huge mess. For Linux, would I even need anything fancy? I can run "cat file.wav > /dev/dsp" from a command prompt to play audio. There are some standard libraries for operating system specific sound (i.e., winsound), but I want something more versatile. The official audio wiki page on the subject lists a small collection of third-party platform-independent libraries. After excluding those which don't support microphone access (the ultimate goal of all my poking around in this subject), I dove a little deeper into sounddevice and PyAudio. Both of these I installed with pip (i.e., pip install pyaudio)

I really like the structure and documentation of sounddevice, but I decided to keep developing with PyAudio for now. Sounddevice seemed to take more system resources than PyAudio (in my limited test conditions: Windows 10 with very fast and modern hardware, Python 3), and would audibly "glitch" music as it was being played every time it attached or detached from the microphone stream. I tried streaming, but after about an hour I couldn't get clean live access to the microphone without glitching audio playback. Furthermore, every few times I ran this script it crashed my python kernel! I very rarely see this happening. iPython complained: "It seems the kernel died unexpectedly. Use 'Restart kernel' to continue using this console" and I eventually moved back to PyAudio. For a less "realtime" application, sounddevice might be a great solution. Here's the minimal case sounddevice script I tested with (that crashed sometimes). If you have a better one to do live high-speed audio capture, let me know!

import sounddevice #pip install sounddevice

for i in range(30): #30 updates in 1 second
    rec = sounddevice.rec(44100/30)
    sounddevice.wait()
    print(rec.shape)

Here's a simple demo to show how I get realtime microphone audio into numpy arrays using PyAudio. This isn't really that special. It's a good starting point though. Note that rather than have the user define a microphone source in the python script (I had a fancy menu system handling this for a while), I allow PyAudio to just look at the operating system's default input device. This seems like a realistic expectation, and saves time as long as you don't expect your user to be recording from two different devices at the same time. This script gets some audio from the microphone and shows the values in the console (ten times).

import pyaudio
import numpy as np

CHUNK = 4096 # number of data points to read at a time
RATE = 44100 # time resolution of the recording device (Hz)

p=pyaudio.PyAudio() # start the PyAudio class
stream=p.open(format=pyaudio.paInt16,channels=1,rate=RATE,input=True,
              frames_per_buffer=CHUNK) #uses default input device

# create a numpy array holding a single read of audio data
for i in range(10): #to it a few times just to see
    data = np.fromstring(stream.read(CHUNK),dtype=np.int16)
    print(data)

# close the stream gracefully
stream.stop_stream()
stream.close()
p.terminate()

I tried to push the limit a little bit and see how much useful data I could get from this console window. It turns out that it's pretty responsive! Here's a slight modification of the code, made to turn the console window into an impromptu VU meter.

import pyaudio
import numpy as np

CHUNK = 2**11
RATE = 44100

p=pyaudio.PyAudio()
stream=p.open(format=pyaudio.paInt16,channels=1,rate=RATE,input=True,
              frames_per_buffer=CHUNK)

for i in range(int(10*44100/1024)): #go for a few seconds
    data = np.fromstring(stream.read(CHUNK),dtype=np.int16)
    peak=np.average(np.abs(data))*2
    bars="#"*int(50*peak/2**16)
    print("%04d %05d %s"%(i,peak,bars))

stream.stop_stream()
stream.close()
p.terminate()

Result

The results are pretty good! The advantage here is that no libraries are required except PyAudio. For people interested in doing simple math (peak detection, frequency detection, etc.) this is a perfect starting point. Here's a quick cellphone video:

I've made realtime audio visualization (realtime FFT) scripts with Python before, but 80% of that code was creating a GUI. I want to see data in real time while I'm developing this code, but I really don't want to mess with GUI programming. I then had a crazy idea. Everyone has a web browser, which is a pretty good GUI... with a Python script to analyze audio and save graphs (a lot of them, quickly) and some JavaScript running in a browser to keep refreshing those graphs, I could get an idea of what the audio stream is doing in something kind of like real time. It was intended to be a hack, but I never expected it to work so well! Check this out...

Here's the python script to listen to the microphone and generate graphs:

import pyaudio
import numpy as np
import pylab
import time

RATE = 44100
CHUNK = int(RATE/20) # RATE / number of updates per second

def soundplot(stream):
    t1=time.time()
    data = np.fromstring(stream.read(CHUNK),dtype=np.int16)
    pylab.plot(data)
    pylab.title(i)
    pylab.grid()
    pylab.axis([0,len(data),-2**16/2,2**16/2])
    pylab.savefig("03.png",dpi=50)
    pylab.close('all')
    print("took %.02f ms"%((time.time()-t1)*1000))

if __name__=="__main__":
    p=pyaudio.PyAudio()
    stream=p.open(format=pyaudio.paInt16,channels=1,rate=RATE,input=True,
                  frames_per_buffer=CHUNK)
    for i in range(int(20*RATE/CHUNK)): #do this for 10 seconds
        soundplot(stream)
    stream.stop_stream()
    stream.close()
    p.terminate()

Here's the HTML file with JavaScript to keep reloading the image...

<html>
<script language="javascript">
function RefreshImage(){
document.pic0.src="03.png?a=" + String(Math.random()*99999999);
setTimeout('RefreshImage()',50);
}
</script>
<body onload="RefreshImage()">
<img name="pic0" src="03.png">
</body>
</html>

Operation

I couldn't believe my eyes. It's not elegant, but it's kind of functional!

Why stop there? I went ahead and wrote a microphone listening and processing class which makes this stuff easier. My ultimate goal hasn't been revealed yet, but I'm sure it'll be clear in a few weeks. Let's just say there's a lot of use in me visualizing streams of continuous data. Anyway, this class is the truly terrible attempt at a word pun by merging the words "SWH", "ear", and "Hear", into the official title "SWHear" which seems to be unique on Google. This class is minimal case, but can be easily modified to implement threaded recording (which won't cause the rest of the functions to hang) as well as mathematical manipulation of data, such as FFT. With the same HTML file as used above, here's the new python script and some video of the output:

import pyaudio
import time
import pylab
import numpy as np

class SWHear(object):
    """
    The SWHear class is made to provide access to continuously recorded
    (and mathematically processed) microphone data.
    """

    def __init__(self,device=None,startStreaming=True):
        """fire up the SWHear class."""
        print(" -- initializing SWHear")

        self.chunk = 4096 # number of data points to read at a time
        self.rate = 44100 # time resolution of the recording device (Hz)

        # for tape recording (continuous "tape" of recent audio)
        self.tapeLength=2 #seconds
        self.tape=np.empty(self.rate*self.tapeLength)*np.nan

        self.p=pyaudio.PyAudio() # start the PyAudio class
        if startStreaming:
            self.stream_start()

    ### LOWEST LEVEL AUDIO ACCESS
    # pure access to microphone and stream operations
    # keep math, plotting, FFT, etc out of here.

    def stream_read(self):
        """return values for a single chunk"""
        data = np.fromstring(self.stream.read(self.chunk),dtype=np.int16)
        #print(data)
        return data

    def stream_start(self):
        """connect to the audio device and start a stream"""
        print(" -- stream started")
        self.stream=self.p.open(format=pyaudio.paInt16,channels=1,
                                rate=self.rate,input=True,
                                frames_per_buffer=self.chunk)

    def stream_stop(self):
        """close the stream but keep the PyAudio instance alive."""
        if 'stream' in locals():
            self.stream.stop_stream()
            self.stream.close()
        print(" -- stream CLOSED")

    def close(self):
        """gently detach from things."""
        self.stream_stop()
        self.p.terminate()

    ### TAPE METHODS
    # tape is like a circular magnetic ribbon of tape that's continously
    # recorded and recorded over in a loop. self.tape contains this data.
    # the newest data is always at the end. Don't modify data on the type,
    # but rather do math on it (like FFT) as you read from it.

    def tape_add(self):
        """add a single chunk to the tape."""
        self.tape[:-self.chunk]=self.tape[self.chunk:]
        self.tape[-self.chunk:]=self.stream_read()

    def tape_flush(self):
        """completely fill tape with new data."""
        readsInTape=int(self.rate*self.tapeLength/self.chunk)
        print(" -- flushing %d s tape with %dx%.2f ms reads"%\
                  (self.tapeLength,readsInTape,self.chunk/self.rate))
        for i in range(readsInTape):
            self.tape_add()

    def tape_forever(self,plotSec=.25):
        t1=0
        try:
            while True:
                self.tape_add()
                if (time.time()-t1)>plotSec:
                    t1=time.time()
                    self.tape_plot()
        except:
            print(" ~~ exception (keyboard?)")
            return

    def tape_plot(self,saveAs="03.png"):
        """plot what's in the tape."""
        pylab.plot(np.arange(len(self.tape))/self.rate,self.tape)
        pylab.axis([0,self.tapeLength,-2**16/2,2**16/2])
        if saveAs:
            t1=time.time()
            pylab.savefig(saveAs,dpi=50)
            print("plotting saving took %.02f ms"%((time.time()-t1)*1000))
        else:
            pylab.show()
            print() #good for IPython
        pylab.close('all')

if __name__=="__main__":
    ear=SWHear()
    ear.tape_forever()
    ear.close()
    print("DONE")

I don't really intend anyone to actually do this, but it's a cool alternative to recording a small portion of audio, plotting it in a pop-up matplotlib window, and waiting for the user to close it to record a new fraction. I had a lot more text in here demonstrating real-time FFT, but I'd rather consolidate everything FFT related into a single post. For now, I'm happy pursuing microphone-related python projects with PyAudio.

Display a single frequency

Use Numpy's FFT() and FFTFREQ() to turn the linear data into frequency. Set that target and grab the FFT value corresponding to that frequency. I haven't tested this to be sure it's working, but it should at least be close...

import pyaudio
import numpy as np
np.set_printoptions(suppress=True) # don't use scientific notation

CHUNK = 4096 # number of data points to read at a time
RATE = 44100 # time resolution of the recording device (Hz)
TARGET = 2100 # show only this one frequency

p=pyaudio.PyAudio() # start the PyAudio class
stream=p.open(format=pyaudio.paInt16,channels=1,rate=RATE,input=True,
              frames_per_buffer=CHUNK) #uses default input device

# create a numpy array holding a single read of audio data
for i in range(10): #to it a few times just to see
    data = np.fromstring(stream.read(CHUNK),dtype=np.int16)
    fft = abs(np.fft.fft(data).real)
    fft = fft[:int(len(fft)/2)] # keep only first half
    freq = np.fft.fftfreq(CHUNK,1.0/RATE)
    freq = freq[:int(len(freq)/2)] # keep only first half
    assert freq[-1]>TARGET, "ERROR: increase chunk size"
    val = fft[np.where(freq>TARGET)[0][0]]
    print(val)

# close the stream gracefully
stream.stop_stream()
stream.close()
p.terminate()

Display Peak Frequency

If your goal is to determine which frequency is producing the loudest tone, use this function. I also added a few lines to graph the output in case you want to observe how it operates. I recommend testing this script with a tone generator, or a YouTube video containing tones of a range of frequencies like this one.

import pyaudio
import numpy as np
import matplotlib.pyplot as plt

np.set_printoptions(suppress=True) # don't use scientific notation

CHUNK = 4096 # number of data points to read at a time
RATE = 44100 # time resolution of the recording device (Hz)

p=pyaudio.PyAudio() # start the PyAudio class
stream=p.open(format=pyaudio.paInt16,channels=1,rate=RATE,input=True,
              frames_per_buffer=CHUNK) #uses default input device

# create a numpy array holding a single read of audio data
for i in range(10): #to it a few times just to see
    data = np.fromstring(stream.read(CHUNK),dtype=np.int16)
    data = data * np.hanning(len(data)) # smooth the FFT by windowing data
    fft = abs(np.fft.fft(data).real)
    fft = fft[:int(len(fft)/2)] # keep only first half
    freq = np.fft.fftfreq(CHUNK,1.0/RATE)
    freq = freq[:int(len(freq)/2)] # keep only first half
    freqPeak = freq[np.where(fft==np.max(fft))[0][0]]+1
    print("peak frequency: %d Hz"%freqPeak)

    # uncomment this if you want to see what the freq vs FFT looks like
    #plt.plot(freq,fft)
    #plt.axis([0,4000,None,None])
    #plt.show()
    #plt.close()

# close the stream gracefully
stream.stop_stream()
stream.close()
p.terminate()

Display Left and Right Levels

import pyaudio
import numpy as np

maxValue = 2**16
p=pyaudio.PyAudio()
stream=p.open(format=pyaudio.paInt16,channels=2,rate=44100,
              input=True, frames_per_buffer=1024)
while True:
    data = np.fromstring(stream.read(1024),dtype=np.int16)
    dataL = data[0::2]
    dataR = data[1::2]
    peakL = np.abs(np.max(dataL)-np.min(dataL))/maxValue
    peakR = np.abs(np.max(dataR)-np.min(dataR))/maxValue
    print("L:%00.02f R:%00.02f"%(peakL*100, peakR*100))

Output

L:47.26 R:45.17
L:47.55 R:45.63
L:49.44 R:45.98
L:45.27 R:49.80
L:44.39 R:45.75
L:47.50 R:46.96
L:41.49 R:42.64
L:42.95 R:41.39
L:49.56 R:49.62
L:48.29 R:48.80
L:45.03 R:47.62
L:47.99 R:49.35
L:41.58 R:49.21

Or with a tweak...

import pyaudio
import numpy as np

maxValue = 2**16
bars = 35
p=pyaudio.PyAudio()
stream=p.open(format=pyaudio.paInt16,channels=2,rate=44100,
              input=True, frames_per_buffer=1024)
while True:
    data = np.fromstring(stream.read(1024),dtype=np.int16)
    dataL = data[0::2]
    dataR = data[1::2]
    peakL = np.abs(np.max(dataL)-np.min(dataL))/maxValue
    peakR = np.abs(np.max(dataR)-np.min(dataR))/maxValue
    lString = "#"*int(peakL*bars)+"-"*int(bars-peakL*bars)
    rString = "#"*int(peakR*bars)+"-"*int(bars-peakR*bars)
    print("L=[%s]\tR=[%s]"%(lString, rString))

Graphical Output

Markdown source code last modified on January 18th, 2021
---
title: Realtime Audio Visualization in Python
date: 2016-07-19 04:44:48
tags: python, old
---

# Realtime Audio Visualization in Python

__Python's "batteries included" nature makes it easy to interact with just about anything... except speakers and a microphone!__ As of this moment, there still are not standard libraries which which allow cross-platform interfacing with audio devices. There are some pretty convenient third-party modules, but I hope in the future a standard solution will be distributed with python. I appreciate the differences of Linux architectures such as [ALSA](http://www.alsa-project.org/) and [OSS](https://en.wikipedia.org/wiki/Open_Sound_System), but toss in Windows and MacOS in the mix and it gets to be a huge mess. For Linux, would I even need anything fancy? I can run "`` cat file.wav > /dev/dsp ``" from a command prompt to play audio. There are some standard libraries for operating system specific sound (i.e., [winsound](https://docs.python.org/2/library/winsound.html)), but I want something more versatile. The [official audio wiki page on the subject](https://wiki.python.org/moin/Audio/) lists a small collection of third-party platform-independent libraries. After excluding those which don't support microphone access (the ultimate goal of all my poking around in this subject), I dove a little deeper into [sounddevice](http://python-sounddevice.readthedocs.io/en/0.3.3/) and [PyAudio](http://people.csail.mit.edu/hubert/pyaudio/). Both of these I installed with pip (i.e., `` pip install pyaudio ``)

__I really like the structure and documentation of sounddevice, but I decided to keep developing with PyAudio for now.__ Sounddevice seemed to take more system resources than PyAudio (in my limited test conditions: Windows 10 with very fast and modern hardware, Python 3), and would audibly "glitch" music as it was being played every time it attached or detached from the microphone stream. I tried streaming, but after about an hour I couldn't get clean live access to the microphone without glitching audio playback. Furthermore, every few times I ran this script it crashed my python kernel! I very rarely see this happening. iPython complained: "_It seems the kernel died unexpectedly. Use 'Restart kernel' to continue using this console_" and I eventually moved back to PyAudio. For a less "realtime" application, sounddevice might be a great solution. Here's the minimal case sounddevice script I tested with (that crashed sometimes). If you have a better one to do live high-speed audio capture, let me know!

```python
import sounddevice #pip install sounddevice

for i in range(30): #30 updates in 1 second
    rec = sounddevice.rec(44100/30)
    sounddevice.wait()
    print(rec.shape)
```

__Here's a simple demo to show how I get realtime microphone audio into numpy arrays using PyAudio.__ This isn't really that special. It's a good starting point though. Note that rather than have the user define a microphone source in the python script (I had a fancy menu system handling this for a while), I allow PyAudio to just look at the operating system's default input device. This seems like a realistic expectation, and saves time as long as you don't expect your user to be recording from two different devices at the same time. This script gets some audio from the microphone and shows the values in the console (ten times).

```python
import pyaudio
import numpy as np

CHUNK = 4096 # number of data points to read at a time
RATE = 44100 # time resolution of the recording device (Hz)

p=pyaudio.PyAudio() # start the PyAudio class
stream=p.open(format=pyaudio.paInt16,channels=1,rate=RATE,input=True,
              frames_per_buffer=CHUNK) #uses default input device

# create a numpy array holding a single read of audio data
for i in range(10): #to it a few times just to see
    data = np.fromstring(stream.read(CHUNK),dtype=np.int16)
    print(data)

# close the stream gracefully
stream.stop_stream()
stream.close()
p.terminate()
```

<div class="text-center img-border">

[![](01_thumb.jpg)](01.png)

</div>

I tried to push the limit a little bit and see how much useful data I could get from this console window. It turns out that it's pretty responsive! Here's a slight modification of the code, made to turn the console window into an impromptu [VU meter](https://en.wikipedia.org/wiki/VU_meter).

```python
import pyaudio
import numpy as np

CHUNK = 2**11
RATE = 44100

p=pyaudio.PyAudio()
stream=p.open(format=pyaudio.paInt16,channels=1,rate=RATE,input=True,
              frames_per_buffer=CHUNK)

for i in range(int(10*44100/1024)): #go for a few seconds
    data = np.fromstring(stream.read(CHUNK),dtype=np.int16)
    peak=np.average(np.abs(data))*2
    bars="#"*int(50*peak/2**16)
    print("%04d %05d %s"%(i,peak,bars))

stream.stop_stream()
stream.close()
p.terminate()
```

### Result

The results are pretty good! The advantage here is that _no_ libraries are required except PyAudio. For people interested in doing simple math (peak detection, frequency detection, etc.) this is a perfect starting point. Here's a quick cellphone video:

![](https://www.youtube.com/embed/xUzyDPsesK8)

I've made [realtime audio visualization (realtime FFT) scripts with Python before](https://www.swharden.com/wp/2013-05-09-realtime-fft-audio-visualization-with-python/), but 80% of that code was creating a GUI. I want to see data in real time while I'm developing this code, but I _really_ don't want to mess with GUI programming. I then had a crazy idea. Everyone has a web browser, which is a pretty good GUI... with a Python script to analyze audio and save graphs (a lot of them, quickly) and some JavaScript running in a browser to keep refreshing those graphs, I could get an idea of what the audio stream is doing in something kind of like real time. It was intended to be a hack, but I never expected it to work so well! Check this out...

__Here's the python script to listen to the microphone and generate graphs:__

```python
import pyaudio
import numpy as np
import pylab
import time

RATE = 44100
CHUNK = int(RATE/20) # RATE / number of updates per second

def soundplot(stream):
    t1=time.time()
    data = np.fromstring(stream.read(CHUNK),dtype=np.int16)
    pylab.plot(data)
    pylab.title(i)
    pylab.grid()
    pylab.axis([0,len(data),-2**16/2,2**16/2])
    pylab.savefig("03.png",dpi=50)
    pylab.close('all')
    print("took %.02f ms"%((time.time()-t1)*1000))

if __name__=="__main__":
    p=pyaudio.PyAudio()
    stream=p.open(format=pyaudio.paInt16,channels=1,rate=RATE,input=True,
                  frames_per_buffer=CHUNK)
    for i in range(int(20*RATE/CHUNK)): #do this for 10 seconds
        soundplot(stream)
    stream.stop_stream()
    stream.close()
    p.terminate()
```

__Here's the HTML file with JavaScript to keep reloading the image...__

```html
<html>
<script language="javascript">
function RefreshImage(){
document.pic0.src="03.png?a=" + String(Math.random()*99999999);
setTimeout('RefreshImage()',50);
}
</script>
<body onload="RefreshImage()">
<img name="pic0" src="03.png">
</body>
</html>
```

### Operation

I couldn't believe my eyes. It's not elegant, but it's kind of functional!

![](https://www.youtube.com/embed/80lAehBMUbE)

__Why stop there?__ I went ahead and wrote a microphone listening and processing class which makes this stuff easier. My ultimate goal hasn't been revealed yet, but I'm sure it'll be clear in a few weeks. Let's just say there's a lot of use in me visualizing streams of continuous data. Anyway, this class is the truly _terrible_ attempt at a word pun by merging the words "SWH", "ear", and "Hear", into the official title "SWHear" which seems to be [unique on Google](https://www.google.com/search?q=%2Bpython+%2Bswhear). This class is minimal case, but can be easily modified to implement threaded recording (which won't cause the rest of the functions to hang) as well as mathematical manipulation of data, such as FFT. With the same HTML file as used above, here's the new python script and some video of the output:

```python
import pyaudio
import time
import pylab
import numpy as np

class SWHear(object):
    """
    The SWHear class is made to provide access to continuously recorded
    (and mathematically processed) microphone data.
    """

    def __init__(self,device=None,startStreaming=True):
        """fire up the SWHear class."""
        print(" -- initializing SWHear")

        self.chunk = 4096 # number of data points to read at a time
        self.rate = 44100 # time resolution of the recording device (Hz)

        # for tape recording (continuous "tape" of recent audio)
        self.tapeLength=2 #seconds
        self.tape=np.empty(self.rate*self.tapeLength)*np.nan

        self.p=pyaudio.PyAudio() # start the PyAudio class
        if startStreaming:
            self.stream_start()

    ### LOWEST LEVEL AUDIO ACCESS
    # pure access to microphone and stream operations
    # keep math, plotting, FFT, etc out of here.

    def stream_read(self):
        """return values for a single chunk"""
        data = np.fromstring(self.stream.read(self.chunk),dtype=np.int16)
        #print(data)
        return data

    def stream_start(self):
        """connect to the audio device and start a stream"""
        print(" -- stream started")
        self.stream=self.p.open(format=pyaudio.paInt16,channels=1,
                                rate=self.rate,input=True,
                                frames_per_buffer=self.chunk)

    def stream_stop(self):
        """close the stream but keep the PyAudio instance alive."""
        if 'stream' in locals():
            self.stream.stop_stream()
            self.stream.close()
        print(" -- stream CLOSED")

    def close(self):
        """gently detach from things."""
        self.stream_stop()
        self.p.terminate()

    ### TAPE METHODS
    # tape is like a circular magnetic ribbon of tape that's continously
    # recorded and recorded over in a loop. self.tape contains this data.
    # the newest data is always at the end. Don't modify data on the type,
    # but rather do math on it (like FFT) as you read from it.

    def tape_add(self):
        """add a single chunk to the tape."""
        self.tape[:-self.chunk]=self.tape[self.chunk:]
        self.tape[-self.chunk:]=self.stream_read()

    def tape_flush(self):
        """completely fill tape with new data."""
        readsInTape=int(self.rate*self.tapeLength/self.chunk)
        print(" -- flushing %d s tape with %dx%.2f ms reads"%\
                  (self.tapeLength,readsInTape,self.chunk/self.rate))
        for i in range(readsInTape):
            self.tape_add()

    def tape_forever(self,plotSec=.25):
        t1=0
        try:
            while True:
                self.tape_add()
                if (time.time()-t1)>plotSec:
                    t1=time.time()
                    self.tape_plot()
        except:
            print(" ~~ exception (keyboard?)")
            return

    def tape_plot(self,saveAs="03.png"):
        """plot what's in the tape."""
        pylab.plot(np.arange(len(self.tape))/self.rate,self.tape)
        pylab.axis([0,self.tapeLength,-2**16/2,2**16/2])
        if saveAs:
            t1=time.time()
            pylab.savefig(saveAs,dpi=50)
            print("plotting saving took %.02f ms"%((time.time()-t1)*1000))
        else:
            pylab.show()
            print() #good for IPython
        pylab.close('all')

if __name__=="__main__":
    ear=SWHear()
    ear.tape_forever()
    ear.close()
    print("DONE")
```

![](https://www.youtube.com/embed/E94MuHtdg6Y)

__I don't really intend anyone to actually do this,__ but it's a cool alternative to recording a small portion of audio, plotting it in a pop-up matplotlib window, and waiting for the user to close it to record a new fraction. I had a lot more text in here demonstrating real-time FFT, but I'd rather consolidate everything FFT related into a single post. For now, I'm happy pursuing microphone-related python projects with PyAudio.

### Display a single frequency

Use [Numpy's FFT() and FFTFREQ()](https://docs.scipy.org/doc/numpy/reference/routines.fft.html) to turn the linear data into frequency. Set that target and grab the FFT value corresponding to that frequency. I haven't tested this to be sure it's working, but it should at least be close...

```python
import pyaudio
import numpy as np
np.set_printoptions(suppress=True) # don't use scientific notation

CHUNK = 4096 # number of data points to read at a time
RATE = 44100 # time resolution of the recording device (Hz)
TARGET = 2100 # show only this one frequency

p=pyaudio.PyAudio() # start the PyAudio class
stream=p.open(format=pyaudio.paInt16,channels=1,rate=RATE,input=True,
              frames_per_buffer=CHUNK) #uses default input device

# create a numpy array holding a single read of audio data
for i in range(10): #to it a few times just to see
    data = np.fromstring(stream.read(CHUNK),dtype=np.int16)
    fft = abs(np.fft.fft(data).real)
    fft = fft[:int(len(fft)/2)] # keep only first half
    freq = np.fft.fftfreq(CHUNK,1.0/RATE)
    freq = freq[:int(len(freq)/2)] # keep only first half
    assert freq[-1]>TARGET, "ERROR: increase chunk size"
    val = fft[np.where(freq>TARGET)[0][0]]
    print(val)

# close the stream gracefully
stream.stop_stream()
stream.close()
p.terminate()

```

### Display Peak Frequency

If your goal is to determine which frequency is producing the loudest tone, use this function. I also added a few lines to graph the output in case you want to observe how it operates. I recommend testing this script with a tone generator, or a YouTube video containing tones of a range of frequencies [like this one](https://www.youtube.com/watch?v=WfbMNJj2C4I).

```python
import pyaudio
import numpy as np
import matplotlib.pyplot as plt

np.set_printoptions(suppress=True) # don't use scientific notation

CHUNK = 4096 # number of data points to read at a time
RATE = 44100 # time resolution of the recording device (Hz)

p=pyaudio.PyAudio() # start the PyAudio class
stream=p.open(format=pyaudio.paInt16,channels=1,rate=RATE,input=True,
              frames_per_buffer=CHUNK) #uses default input device

# create a numpy array holding a single read of audio data
for i in range(10): #to it a few times just to see
    data = np.fromstring(stream.read(CHUNK),dtype=np.int16)
    data = data * np.hanning(len(data)) # smooth the FFT by windowing data
    fft = abs(np.fft.fft(data).real)
    fft = fft[:int(len(fft)/2)] # keep only first half
    freq = np.fft.fftfreq(CHUNK,1.0/RATE)
    freq = freq[:int(len(freq)/2)] # keep only first half
    freqPeak = freq[np.where(fft==np.max(fft))[0][0]]+1
    print("peak frequency: %d Hz"%freqPeak)

    # uncomment this if you want to see what the freq vs FFT looks like
    #plt.plot(freq,fft)
    #plt.axis([0,4000,None,None])
    #plt.show()
    #plt.close()

# close the stream gracefully
stream.stop_stream()
stream.close()
p.terminate()
```

### Display Left and Right Levels

```python
import pyaudio
import numpy as np

maxValue = 2**16
p=pyaudio.PyAudio()
stream=p.open(format=pyaudio.paInt16,channels=2,rate=44100,
              input=True, frames_per_buffer=1024)
while True:
    data = np.fromstring(stream.read(1024),dtype=np.int16)
    dataL = data[0::2]
    dataR = data[1::2]
    peakL = np.abs(np.max(dataL)-np.min(dataL))/maxValue
    peakR = np.abs(np.max(dataR)-np.min(dataR))/maxValue
    print("L:%00.02f R:%00.02f"%(peakL*100, peakR*100))
```

__Output__

```
L:47.26 R:45.17
L:47.55 R:45.63
L:49.44 R:45.98
L:45.27 R:49.80
L:44.39 R:45.75
L:47.50 R:46.96
L:41.49 R:42.64
L:42.95 R:41.39
L:49.56 R:49.62
L:48.29 R:48.80
L:45.03 R:47.62
L:47.99 R:49.35
L:41.58 R:49.21

```

Or with a tweak...

```python
import pyaudio
import numpy as np

maxValue = 2**16
bars = 35
p=pyaudio.PyAudio()
stream=p.open(format=pyaudio.paInt16,channels=2,rate=44100,
              input=True, frames_per_buffer=1024)
while True:
    data = np.fromstring(stream.read(1024),dtype=np.int16)
    dataL = data[0::2]
    dataR = data[1::2]
    peakL = np.abs(np.max(dataL)-np.min(dataL))/maxValue
    peakR = np.abs(np.max(dataR)-np.min(dataR))/maxValue
    lString = "#"*int(peakL*bars)+"-"*int(bars-peakL*bars)
    rString = "#"*int(peakR*bars)+"-"*int(bars-peakR*bars)
    print("L=[%s]\tR=[%s]"%(lString, rString))
```

### Graphical Output

<div class="text-center img-border">

[![](audio-graphical_thumb.jpg)](audio-graphical.png)

</div>
Pages