- Edited
Fauser LiFli (red) vs. BPW34 @ 100K 9V (blue), 50 kHz PWM:
Homemade oscilloscope to detect PWM DIY guide
13 days later
- Edited
Could you please advise me about oscope results? I have fashioned the BPW34 with the rezistor as posted by @KM .
I am measuring my camera display which is a source of pain. In the first image, the brightness is set on +3 out of a range -5 to +5. It clearly shows PWM on the graph, but it doesnt show the frequency.
On the second image, the brightness is +2, there it shows PWM and freq as well (20 Khz).
Am I doing something wrong? The ***** symbol instead of the freq number also shows with brightness below -3, so -4 and -5 show no freq. Is it because the display then is too dim and the diode cant pick it up? It obviously must have PWM at lower levels as well, if it has PWM at -3. Could it apply for the too bright setting as well?
I am confused as then I do not know how well to trust the measurements I make with this. When I measure with the soundcard oscope, it shows PWM at all brightness levels of 15Khz.
Also there is a red switch on the probe, either to 1x or 10x. When its set on 10x it doesnt show any readings, so I leave it at 1x.
- Edited
martin I think their implementation of their frequency software counter only recognizes waves that have a high peak-to-peak difference. OpenHantek suffers from the same problem. But what you could do is try and move the rulers to manually mark a single wave. Then the frequency might appear somewhere. As seen here in the OpenHantek screenshot: https://ledstrain.org/d/312-homemade-oscilloscope-to-detect-pwm-diy-guide/39 the rulers are labeled "1" and "2" and can be moved manually. Hopefully Hantek's software has such rulers, too.
martin When its set on 10x it doesnt show any readings, so I leave it at 1x.
That is normal behavior, we need to keep it at "1x".
By the way, even PicoScope's frequency counter is not very accurate, so I always use the bars there, too. The counter is useful only for quick evaluation but I rarely use it. Stopping and then manually marking the wave is always more accurate.
martin I am confused as then I do not know how well to trust the measurements I make with this. When I measure with the soundcard oscope, it shows PWM at all brightness levels of 15Khz.
I would trust the oscilloscope, as it is the more precise tool. If you want to, you can create pretty precise PWM with a Raspberry Pi and low budget parts (the breadboard isn't even needed) and verify if the frequencies are correct: https://ledstrain.org/d/375-howto-create-pwm-led-flicker-with-a-raspberry-pi
2 months later
First, I'd like to thank Martin and everyone else who contributed to this. I finally got around to mimicking what Martin came up with, namely:
- BPW34,
- PicoScope 2204A,
- 100,000 Ohm resistor,
- 9V battery.
When I started, I had never used an oscilloscope, but it was pretty easy to figure out. I am so happy that I did this. Now I don't have to guess (so much) at what's causing my nausea, headaches and tinnitus.
I do have a question that is perhaps off topic. I've configured everything I know about:
- Turning the brightness all the way up,
- Going into Intel Graphics Settings and disabling the Intel power saving features (laptop only),
- Going into advanced power options and making sure that the max brightness is 100% (laptop only),
- Going into the bios and making sure that the max brightness is 100% (PC/laptop only),
- Disabling the adaptive brightness/ambient light sensor (if applicable).
But there are still times when the screen flickers with noticeable PWM under certain scenarios. For instance, on my daughter's laptop, when it's on the facial recognition screen when you're trying to log in to Windows, the screen will exhibit PWM (and therefore cause nausea, headaches and tinnitus). The screen does not flicker at full brightness when you're entering a PIN or your password - it's only on the facial recognition screen. I suppose it's trying to reduce glare in the camera.
So here's my question: does anyone know of other scenarios where we should be testing for PWM? The obvious ones are on battery vs. plugged in, and when the battery is low (say, less than 20% remaining).
My problem is that I often get nausea, headaches and tinnitus presumably caused by PWM, but I'm unable to pin down the exact source because our environment is so full of screens and the onset of my symptoms often isn't immediate. If my phone or laptop is occasionally using PWM, it might take me months to pinpoint under exactly what scenarios it happens.
tl,dr; Does anyone know of scenarios, other than what's listed above, where we should be checking for PWM?
GregAtkinson Small power/status LEDs often flicker. Especially pulsating ones (e.g. smartphone notifications). They are small, but could trigger symptoms.
Displays of mobile devices could flicker while the devices are booting and not yet using the user-set "safe" brightness value but some default value that is lower. They could also flicker when entering the lock screen or waking up, displaying a fade animation that in fact could be PWM-controlled in some cases.
Last but not least, it is possible that some other cause triggers your symptoms. Temporal dithering, for example. Or even tiny flicker that is not PWM but just some tiny ripple riding on a DC signal, the former which you could only see by holding the sensor very near to the light source (in my case LED bulbs) and then switching the oscilloscope to AC mode rather than DC. Screens are usually not bright enough to do this reliably. Since I still get eye strain despite ruling out PWM and blue light, every now and then I'm considering buying the Thorlabs sensor which Notebookcheck uses in hopes to get a better, amplified signal, but it is expensive and costs over 300 €. It won't prove much though, since I realized probably all LED bulbs and displays connected to AC power have a small ripple.
Nice thread. Thanks for the info, martin.
5 months later
@martin: Do you think you could build such a headphone jack connector?
https://rmania.net/aksesoari-photodiode-for-measuring-light-pulsations-and-pwm-2358.html
See the pictures in the link. I tried to connect a photodiode with a regular audio cable to my smartphone in the past, but it didn't work. I think because the connector is a little different for phones, given that both output and input are combined. It seems microphone connectors have 3 dividers instead of 2.
It seems they use a free oscilloscope app, probably the one available on F-Droid (at least it looks similar. If we could get this to work, we'd have an ultra low-cost portable PWM detector for frequencies in the audible range.
3 years later
I ordered:
- PicoScope 2204A
- BPW34
- 100k Ohm resistor
Looking forward to finally having my own setup.
I look forward to seeing your data!
4 months later
Someone managed to record the differing PWM and variable display refresh rates on the Pixel 6 Pro. What's interesting is he used an FFT chart which plots the intensity over frequency spectrum instead of a single waveform like "classic" frequency measurements. Now I believe Intel in their inconclusive tests tried something similar with whatever equipment they were using but I didn't look further than that.
Any opinions on this oscilloscope?
https://www.aliexpress.com/item/32259059985.html
An appealing fact about it is that it's portable
Alyosha2001 I researched something similar a number of years ago, but some people claim there are issues with the trigger function and some disparity over what frequencies it can capture in the first place so I ended up passing on it.
Caveat emptor.
2 years later
- Edited
I splurged on a Thorlabs PDA100A2 photodetector for about $450. And actually, I'm glad I did.
I'll be forever grateful to @KM and others on this thread that helped me create a DIY photodetector, but there were a few screens that seemed to be flicker free but still gave me headaches. Here's one of those screens (a supposedly flicker-free Benq). The DIY photodetector in red and the Thorlabs one in blue. Note that the oscilloscope is set to 50 microseconds per div, so my DIY photodetector just doesn't seem to be responsive enough to pick up the flicker.
Here's a couple more comparisons with a flashlight that flickers. Same colors (DIY in red and Thorlabs in blue), The Thorlabs response time is well under 1 microsecond whereas my DIY response time is roughly 10 microseconds.
If anyone wants to borrow my setup I'm sure we could work something out.
Would you be able to share some tips on how to use the Thorlabs product? I have been testing several devices for PWM and temporal dithering at the link below with a cheap photodiode and microscope. Don't think they are truly "flicker free."
https://ledstrain.org/d/2589-testing-for-non-pwm-and-temporal-dithering-displays-laptops-etc
photon78s Here's a video that I created that walks you through the whole process: https://www.youtube.com/watch?v=AoJ2jFgpBik
That should give you a pretty good idea of what to do.
I wonder if we (as a community) could do more to get people started down the path of using a photodetector and oscilloscope. Things like:
- Mass produce a bunch of the DIY detectors and sell them on eBay for like $25 (not trying to make a profit; just cover the costs).
- Produce some how-to content to help get people started.
- Establish a rental program where you could rent the equipment for a nominal fee and try it out for a few weeks.
- Engage other people on this forum to encourage them to give it a shot.
Although there are likely a lot of people on this forum that aren't sensitive to flicker, I suspect there are a lot of people who are sensitive to flicker but they just don't know how to measure it.
Very helpful video! Thanks so much.
Agreed. And also easy ways to measure temporal dithering as well with the cheap microscopes and phones many people already have or can afford.
a month later
can i just use the sound card of the laptop?
8 days later
- Edited
Looking for small handheld portable scope with spectrum analyser mode because just flicker waveform is kind of stupid when radex lupin gives you spectrum view. Any made in china marvels? For decent price?