According to here the Librem 5 LCD backlight is configured to use PWM at 200Hz. Now 200Hz is low enough to cause eyestrain, headaches and other neurological problems in people with damaged eyes/nervous systems (such as myself).
a) Why was 200Hz chosen?
b) Do you know if there are any negative effects from raising the PWM frequency to avoid eyestrain?
Keep up the good work
I was hoping it would be higher than 200Hz.
Me too, but it looks easy to adjust (in theory) if you can recompile the kernel.
I’m curious to know the upper frequency limit because if it is too low it may make the screen unusable for me.
This is about the dev kit panel which is most likly not the same as in the final phone, if i remember it corectly from other discousions.
So i think the final display is not chosen yet.
Positive aspect of the phone being pushed back is they still have time to integrate feedback like this into the final product, if they so choose. I hope Purism will raise that at least a little.
Fair enough. Sorry to jump to conclusions not based on a final version of the phone.
No need to be sorry. I think its the right time to raise concern for this. If the final display isn’t choosen yet this might be counted in. And i wasn’t aware that PWM is a configurable thing for diaplays. I allways thought it is kinda fixed in hardware. So thanks for this discussion.
So would be nice if some one from Purism could comment if this i on the list for the finale chooise. Maybe @dcz?
200Hz was basically guesswork, I determined it should be “okay” based on how much I notice flicker on my laptop. If someone has some actual data on what LED duty cycle is good for humans, that would help a lot.
The PWM in the display will have some upper frequency bound though. I’m not aware of any downsides of increasing frequencies yet, but I’ll ask around.
It‘s not the might not be the best or most academic source, but as i like to read theire review because they mesure such things like pwm, notebookcheck.com considers pwm over 1000hz high and not noticable for most people.
Within the year of 2019 here is to recognize that: “52% of all tested devices do not use PWM to dim the display”. And, according to TFTCentral PWM issue is a game about to win or lose some of potential buyers … trust me, please, my eyes have over 50 years of what quality … life experience. PWM is Health Issue as well.
I have to correct my self. Here they call 363Hz high. But they also link to studies which indicates that up to 2500/3000hHz there are people who are effected. So if the hardware could do this high it might be a goal to achive this. And maybe also for the other librem devices. The librem 13 review sadly has no PWM mesurement.
Our L5 screen should be like the one from Razer Phone 2: 5.72" IGZO IPS Ultramotion LCD, 1440x2560px resolution (16:9); wide color gamut, 120Hz refresh rate; 580nit max brightness or similar … with support for both 720p50 and 720p60 vertical frequencies. According to Notebookcheck.net “the display backlight flickers at 2451 Hz (Likely utilizing PWM) Flickering detected at a brightness setting of 20 % and below. There should be no flickering or PWM above this brightness setting”. And, as known, OLED displays always use PWM of around 250 Hz and therefore are for me no-go!
Good, flicker free performance at minimal brightness levels is important.
Yet, there is already professional software solution called IRIS for all Librem users out here not already pleased with their screen. Why not to invite Mr. Daniel Georgiev on board and see what he can do for Librem 5 during development phase?
The current plan is to use this to drive the backlight, which means no PWM, and therefore no flicker.
I read a study from a lighting industry publication that claimed anything above 10Khz is safe, but this hasn’t really been studied to the extent it’s needed.
Apple for most MacBooks since 2010 (and probably iPhones) have used 100Khz PWM below 50% brightness. So good even cheap oscilloscopes were “fooled” into thinking it was PWM free.
A bit ambitious, but maybe for the Librem 5 2.0 or whatever next year
No PWM would be better, but I have damaged eyes and they were fine with the LG V20 which has similar ~2.4Khz PWM below a certain brightness (37%)
A while ago I tested some different backlight frequencies on my laptop. I found that:
- 200Hz flicker was fairly obvious and noticeable
- 1000Hz was less apparent in normal use, but obvious when specifically looking for flicker
- 2000Hz was on the verge of being impossible to discern.
- 4000Hz was more or less imperceptible.
I don’t really like the idea of PWM for backlights. Just because it’s not perceptible doesn’t mean it’s not searing my retinas with super bright light 200 times a second! But I’m not someone who finds it much of a problem in practice. Still, I’m glad the phone won’t have it.
Well, it looks like you’ve got the purpose of my intention. But first and to be correct, with LG V20 measurable flicker was not detected at brightness levels of 90% and 100%. And it was 5.7 inches IPS with only 16M colors, with same (mentioned) resolution: 1440×2560 pixels. Therefore here (relativly outside of the PWM topic), let me clear up why I left 720p number(s): Vivante GC7000Lite GPU should support 720p60 directly over the MIPI DSI to the capacitive touch screen meaning (simplifying here) 720×2=1440 … resulting in much better 720p picture (and its color) reproduction (in short). To my understanding 1080p and higher resolutions will be projected over USB-C (without usage of Mini-HDMI). Is it 1440 pixels 5.7 IPS screen nowadays more expensive than 720 pixels 5.7 IPS screen? Resolution close to 720×1440 Rocktech JH057N00900 is 1440×2960 with almost identical 18.5:9 aspect ratio (instead of 2:1).