PWM whine near earpiece in low signal phone call and loud sound sent to the other end

[vortex]

So, I forced the modem into 2G mode (weaker 2G reception where I am at than 4G), took a landline at the same location, setup the call, and recorded the sound with the Sound Recorder app installed by default on Mobian. To make the signal worse and increase the transmit power, I held the phone in a metal enclosure with the door slightly ajar so I could hold it and see it (the kind of metal enclosure circuit breakers are put in).

If possible, it would be interesting to repeat the same experiment once more, now with a wired headset connected to the phone.  That way, the built-in microphone would be left unused, so it would be possible to determine whether the issue stems from the built-in microphone or from the routing of the audio signals.  Maybe it's even up to the A/D conversion performed inside the A64 SoC?

Sadly, I don't have a wired headset and the store that would have one is hard to go to with the pandemic unless something changed in the last week (not going to try to get a headset online given how hard it is to get headphones that fit and operate well when buying in person).

One thing to note, it seems the issue is worst on 2G and 2.5G. This could be correlation (poor signal and thus drop to lower G) or it could be causation (e.g. for the same transmit power, 2G is worse than higher G).

I'll try to reproduce the issue, by placing my PinePhone in a metal enclosure, etc.  By the way, back at the time when the modem calls initially started to work on the PinePhone, the dialer application didn't perform the audio routing.  As a result, there was a loud noise during the call, resemblling some sort of weird, metallic, pulsating noise that made it impossible to hear the other end of the call, or the other end to hear me.  After executing the audio setup utility, during the call, this noise disappeared and the audio quality was rather good afterwards.  I wonder if this is somehow related?

It is related to that sound. When I first heard the recorder, I thought it was something very different. But listening to my more recent recording, I think they are related. The pulsating sound has the same spectral characteristics as the more continuous sound, so I think they are the same. Since my Nexus 5 makes similar sounds but doesn't have the issue, I am inclined to think it isn't direct audio coupling from the sound to the microphone. This is why I think it is electronic. I could of course be wrong on this.

So, I think what you just brought up is worth investigating further.

The resulting spectrogram, which is where the bad news starts, is

Nice work on creating the analysis and plotting utility!  The first thing that crossed my mind once I saw these spectrograms was "damn! the noise covers nearly the entire audible spectrum!" and it sadly turned out to be true after reading your explanation.

And then there is an important question. How widespread is this phenomena among PinePhones in low cellular signal conditions? For those that have it, is the frequency the same? This latter one is important for implementing any filter in a widespread fashion. If each phone with the problem has a different frequency, each phone will require tuning of the filter. If it is the same, then the parameters can be found very accurately using one or a handful of phones and used for all with the problem.

I'll try to reproduce the issue on my PinePhone, and hopefully more people would do the same.  That's the only way to know how widespread the issue is, and whether it manifests in the same way on all affected phones.

Yes. With luck, I will be the only one or nearly the only one. Would be really bad if it was pervasive.

Edit: I somehow managed to forget mentioning that the proprietary modem firmware can be updated, which you're probably already aware of, so you should check that as well. You can find the instructions, the latest firmware, and the release notes in this repository.  Further information, including the source code of the update utility, can be found in this forum thread

I am familiar with it and looking forward to where the firmware will eventually go, but I am not quite ready to dive down that rabbit hole yet. A lot of things I have to get setup before I can go there.

That all said, that firmware might be able to have some effect on the problem (e.g. use settings that reduce it).

Edit #2: Here's a very good insight from Biktorgj, who is the primary author of the open-source firmware for the PinePhone's Quectel modem. The quotation below comes from an IRC session on #pinedev, in which I asked Biktorgj to have a look at this forum thread.

Difficult to diagnose without hearing it and finding something to replicate.

I've always heard some noise from the modem during calls that seems to filter to the analog amp in the pinephone but it's never been a static whine, you can hear the "data" being moved between the bts and the modem like when you put a phone near an amplifier, but I'm always on 4G with VoLtE.

That said, the modem does have some gain compensation and white noise generation through the afe loop which can be making the problem worse. First I would try to adjust the output level to the modem, then I'd try without the AFE calibration and that stuff (my firmware doesn't have it).

The modem doesn't have any internal dac/adc for voice (it can be optionally attached to an external codec but the pinephone doesnt have one, so everything that the modem gets is digital).

The other thing that could be tried is setting audio to 16KHz if running stock firmware or to 48k if using the custom one, just in case the resampling is getting artifacts into the audio, but maybe that's useless.

As I mentioned earlier, I misinterpreted the sound at first as a static whine from what the recorder was able to get. The sound is closer to the sound one hears when "data" is moving now that I have had a chance to hear a decent recording.

If the CPU is using a fixed gain and the modem is doing an adjustable gain that adjusts to match the sound level, there are some possible mitigations. One would be to have the gain from the modem decrease more rapidly than it can increase.

Also, if a comb filter needs to be implemented, it could be run from the CPU or the modem, whichever is easier and requires less power.

I don't know if the sampling frequency in the modem would have much effect given that the frequency comb is seen at the CPU. But, there is a chance it might due to interactions between the different close components in the CPU's sound unit due to their close proximity electrically.