Homemade oscilloscope to detect PWM DIY guide

GregAtkinson I would not recommend the LiFli for displays. The reason is that it needs a pretty high brightness to operate. When the brightness is too low, the scale would just blink at 2%, indicating a lack of light. It would still output to an oscilloscope, but the output is very noisy, which is a problem since this hides low brightness fluctuations. The device works well for measuring room lighting and lamps, which I believe it is made for. But not for low brightness light sources like displays. On displays I need to show a white screen on a high brightness setting to get enough light.

For display measurements, you are much better off buying some BPW34 off eBay and build the oscilloscope circuit. It works even without soldering as you can see in the latest picture.

If you want to see even tiny fluctuations (1%) in low brightness modes, I can recommend the PicoScope 2204a. Its lowest range of +/-50 mV is a true 8 bit hardware mode. Plus it can do oversampling to simulate additional bits of vertical resolution, meaning it can filter out some noise. Up to 10 bits are useful. They told me the option which covers everything most accurate would be their true 12 bit oscilloscope for $800+, which no doubt would be great but seems overkill. Unless even fluctuations of far less than 1% are responsible for our eye strain, which at this point I have a hard time to believe.

The Hantek 6022BE has some accuracy problems in low voltage modes. Those modes are not real hardware modes but zoomed in from like the +/-100 or +/-200 mV mode. You will still see PWM, but maybe not tiny fluctuations (say below 5% voltage ripple) at low screen brightness. But thanks to the Android software HScope, the 6022BE might be the cheapest option to build a portable solution. We just need to make the photodiode probe circuit more robust so it can survive a trip in a bag easily. Then, with a powerbank and a compatible Android device, everything should be portable.

martin Orientation does not matter without the battery. If the signal is negative, below zero, just switch the probes. I will upload a photo in some minutes.

Viktor Yes, I tried to do that. But the battery had no impact at all. Maybe my soldering was too bad. So I took my second OPT101 and used it without the attached board, effectively overheating it within seconds.
However, I think the OPT101 may be too slow due to the official bandwidth specification of 14 kHz.

Edit: Viktor I realized I soldered the OPT101 to its board the opposite way. Would this make any difference? I had to switch the probe tips to get a positive voltage.

My recent LED measurements (see above) which showed no perfect square waves bothered me a little and I began to wonder whether the LED or the photodiode was too slow for the PWM signal. So I wanted to know how much impact the 9V battery really has, and as you can see in this screenshot, it speeds up the rise time a little.

Blue=9V battery, Red=no battery; Frequency: 5000 Hz, 100K resistors

However, reducing the resistance from 100K down to 10K speeds up the diode almost proportionally by the factor 10:

Blue=10K resistor, Red=100K resistor; No batteries

The downside is that 10K is not only 10 times faster but also delivers a 10 times smaller voltage, so I believe for now 100K are a good trade-off. If anyone has ideas how to make this faster or can measure his OPT101 rise times, let me know. With the Raspberry Pi How-To you can create your own PWM-controlled LED circuit and measure the speed of your oscilloscope setup: https://ledstrain.org/d/375-howto-create-pwm-led-flicker-with-a-raspberry-pi

KM There is a small notch on the photodiode to help with the orientation, this should be the correct layout:

Edit:
Test with a Nokia 6 at <17% brightness:

result from notebookcheck.com:

Edit2:
I tried to create a pwm signal with the arduino status led:

Code:
void setup()
{
pinMode(13, OUTPUT);
}

void loop()
{
digitalWrite(13, HIGH);
delayMicroseconds(100);
digitalWrite(13, LOW);
delayMicroseconds(100);
}

The opt101 can also be connected without soldering:

Hanzebenger I think you should even see 10 MHz PWM - as long as it's full on/off PWM. As you can see in my pictures in which I compared battery vs non-battery waves, a few posts above: https://ledstrain.org/d/312-homemade-oscilloscope-to-detect-pwm-diy-guide/52
The wave will not look like rectangular PWM anymore, and it will float above zero, but you can see spikes that have the correct frequency.

When the flicker is just a small ripple riding on a DC signal though, as seen in "flicker-free" displays, it might not be displayed. But such small flicker may matter: FRC/temporal dithering, unresolved software eye strain - all those have just a very tiny flicker. So I imagine if a backlight has a similar small flicker, it might trigger eye strain, too.
You can see such small waves a little easier when you switch your PicoScope oscilloscope to "AC" mode. The Hantek 6022BE doesn't have such a mode.
In fact I think there is a higher frequency riding on top of your waves (where there is no thin line but distortion) which you could make visible by adjusting the time axis.

I believe the largest part of your rise times is related to the slow CCFL response. You should easily detect 1000 Hz PWM.