Whine when playing video

[linusfargo]

If you set the brightness of the screen to the maximum, the noise disappear. This means that it is the PWM used to set the brightness of the screen which is causing this.
We will have to see if it is possible to add some shield around the components which are setting the brightness.

Awesome find on the brightness setting. I concur setting screen brightness to maximum does alleviate the noise.

[Firestorm]

If you set the brightness of the screen to the maximum, the noise disappear. This means that it is the PWM used to set the brightness of the screen which is causing this.
We will have to see if it is possible to add some shield around the components which are setting the brightness.

Good job on identifying which of the PWMs is causing this. I also confirm the whining noise disappears in brightness above 90% or below 30% (rough estimates).
So listen to music in darkness or on sunny days only ;-)

Ok, seriously I hope we can figure out how to tweak the hardware.

As a side notice: in general the audio output has quite some white noise. But I do not complain about this, it is normal for cheap devices and does not disturb me as much as the whining PWM noise. After all the whole pinebook costs the same as just the audio interface of my desktop PC.

[phuzy]

If you set the brightness of the screen to the maximum, the noise disappear. This means that it is the PWM used to set the brightness of the screen which is causing this.
We will have to see if it is possible to add some shield around the components which are setting the brightness.

As other's have also confirmed. Setting the brightness to 100% makes the whine go away. Will take a closer look at the board to see if the components can be identified. Thanks for sharing!

[pfeerick]

Sounds like an inductor related to the LCD backlight PWM driver circuitry needs a visit from Mr Epoxy

[Samarium151]

There are rc filters on input output lines to speakers. (P14 of Pinebook pro schematic). I’m not as sure about headphone jack but if I’m reading it right filter components have different values for head phone . Given this you might not have a whine via headphone. A change to the filter compost values might be a fix at the cost of a narrower frequency range. The cheap home brew solution might be to wrap the speaker leads around a ferrite donut to act as inductor and perhaps more of a low pass filter. This will probably not work for a very high pitched noise but if it is the pulse width modulator for brightness is should be low enough to work. Waiting on my pbp so I can’t try unfortunately.

[m42uko]

On my PBP, I can also hear a fairly loud noise (whine) through the speakers under some conditions.

The noise in the circuit seems to be coming from the 5V regulator for USB + Periphery (VCC5V0_USB). Probably, it's operating under non-ideal conditions (too low load), which causes some ripple at "high" frequency, which then couples into the output power amps (U19/U20), and thus out the speakers. I haven't gotten around to do any real measurements yet, but here's my initial analysis.

The symptoms are:
* PBP idle, with nothing connected: No noise
* PBP idle, with low-power USB device connected (like a mouse):  quiet noise
* PBP playing audio (but muted), with nothing connected: fairly loud noise
* PBP playing audio (but muted), with higher power USB device connected (USB drive): no noise
* Backlight brightness has some effect, but compared to the USB load changes, it's negligible.
* Headphone jack has no noise (it's directly driven from the DAC, and not through the power amp ICs)

Now about possible solutions. There is a filter in the power supply of the audio power amp (L11, C307, C308), which, however, is pretty much ineffective as the series inductor (L11) has been replaced with a 0-Ohm resistor (at least on my board). It might help replacing this part with either a properly dimensioned inductor or a low-value resistor (maybe around 10 Ohms?). Adding a 100 nF-ish capacitor in parallel with C307 might also help.

If we're unlucky, the noise doesn't couple through the power supply lines, but through data lines or even through the air. In that case, the solution would be a lot more tricky.

I don't have the components to test this right now, so I can't say for sure whether this will help or not.

Has anybody else tried anything like this already? Also, does anybody have the datasheet to the power amp (XA9108). This would help to estimate the proper values for the filter.

[zaius]

Also, does anybody have the datasheet to the power amp (XA9108). This would help to estimate the proper values for the filter.

I looked, but all I found is this webpage, which I can't read.

http://www.oriic.com/product/4922.htm

[linusfargo]

There are rc filters on input output lines to speakers. (P14 of Pinebook pro schematic).  I’m not as sure about headphone jack but if I’m reading it right filter components have different values for head phone .  Given this you might not have a whine via headphone.   A change to the filter compost values might be a fix at the cost of a narrower frequency range.  The cheap home brew solution might be to wrap the speaker leads around a ferrite donut to act as inductor and perhaps more of a low pass filter.  This will probably not work for a very high pitched noise but if it is the pulse width modulator for brightness is should be low enough to work.  Waiting on my pbp so I can’t try unfortunately.

After reading your analysis of the audio circuit, found that disabling the sound card in sound preferences seems to eliminate the noise. If you are using the stock distro, you can right click on the speaker icon in the menu bar, select sound preferences. Then click on the hardware tab, select es8316-spk-sound. Finally change the profile in the drop down menu to off.

[m42uko]

So I just took some time to mess around with the audio circuitry on my PBP. I desoldered L11 (0 Ohm resistor) and connected some wires so that I can easily try different components. Long wires aren't ideal, especially when troubleshooting EMI issues, but as you'll see from my findings below, my setup shouldn't be a problem here.

Okay, so here's what I tried:

Replace L11 with a resistor.
75 Ohms: Noise goes down noticeably (but not gone), but the audio amp cuts out (audio gone for half a second, then back for half a second) already at low volumes.
23 Ohms: Noise goes down only a bit, and still cuts out at high volumes

Replace L11 with an inductor:
Random one I had lying around: No effect whatsoever

Add a 100 nF in parallel with the PA supply:
No effect.

Power the PA from my external lab power supply:
No noise, full volume achievable. (That also sorta validates my test setup.)

Just because it might be interesting, here's the power draw of both power amps:
disabled: 0 mA
idle: 20 mA
quiet listening: ~30 mA
full volume: up to 300 mA

That's a lot more than I expected, and it explains why the RC filter above caused the amp to cut out, but it's what we have to work with. Also, I was hoping I could maybe run the PA off of 3.3 V, which would've allowed to maybe add a linear regulator, but I tried it with my lab supply and it started to cut out at higher volumes already at 4.5 V.

So, in summary, I don't have a solution yet. At least we know for certain now where the noise is coming from (the 5V supply).

Any ideas on what else I could try? I guess taking a look at the source (the 5V regulator) might help (and that would be the proper way to do it), but it's also the one that's a lot harder to fix now that the boards are finished. Not to mention it's way easier to fry the board by playing around with these components.

Thanks,
Markus

PS: Notebooks with random wires coming out of the side sure look fun^^

[m42uko]

Another day, another attempt. This time, I added the following circuit instead of L11:

And while it does not solve the problem entirely, it significantly reduces the noise down to an acceptable level. There is enough space on the PBP to fit this circuit, but it's still three components for one. Yet, I don't really have any other idea at the moment. Well, apart from messing with the SMPS chip, but I don't feel like risking to fry the entire board when I only have one to play with.