Odd speaker behavior

NoSloppy · October 13, 2021, 3:44pm

This looks like DC offset to me. Wiring here is straight speaker pads to a mono audio jack interface to the speaker.
Anyone see any thing like this before? I have no idea why the speaker would polarize like that when audio plays, unless it’s simply a broken amp. It is not the audio itself, as the speaker reacts the same to spoken talkie from the board.
Thoughts?

profezzorn · October 13, 2021, 4:45pm

Do you have highpass filtering on?
Does it happen with 5.9?
Does it change behavior if you modify this line and change 30 to something like 30000?

github.com

profezzorn/ProffieOS/blob/14b3b8ff582c63de361abfae8f1b0a2159f9ea9e/sound/amplifier.h#L31

    
      
          
          
  bool Active() {
          //    if (saber_synth.on_) return true;
          //    if (audio_splicer.isPlaying()) return true;
              if (beeper.isPlaying()) return true;
              if (talkie.isPlaying()) return true;
              for (size_t i = 0; i < NELEM(wav_players); i++)
                if (wav_players[i].isPlaying())
                  return true;
              uint32_t t = millis() - last_enabled_;
              if (t < 30) return true;
              return false;
            }
          
          
  void Enable() {
              dac.begin();
              last_enabled_ = millis();
          #ifdef AUDIO_CONTROL_SGTL5000
              if (!sgtl5000_enabled) {
                sgtl5000_1.enable();
                sgtl5000_1.volume(0.7);

What if you change the 10 on this line to something bigger? (put the 30 from above back and try one thing at a time…)

github.com

profezzorn/ProffieOS/blob/14b3b8ff582c63de361abfae8f1b0a2159f9ea9e/common/booster.h#L51

    
      
              pinMode(boosterPin, OUTPUT);
              digitalWrite(boosterPin, LOW); // turn the booster off
              last_enabled_ = millis();
            }
          
          
  void Loop() override {
              STATE_MACHINE_BEGIN();
              while (true) {
                while (Active()) YIELD();
                // 10 * 0.1 s = 1 second
                for (i_ = 0; i_ < 10 && !Active(); i_++)
          	SLEEP(100);
                if (Active()) continue;
                STDOUT.println("Booster off.");
                digitalWrite(boosterPin, LOW); // turn the booster off
                // pinMode(amplifierPin, INPUT_ANALOG); // Let the pull-down do the work
                on_ = false;
                while (!Active()) YIELD();
              }
              STATE_MACHINE_END();
            }

What does your config file look like?

profezzorn · October 13, 2021, 4:46pm

Basically, I doubt that it’s DC offset, because there is no analog amplifier in a proffieboard.
It could be an equivalent problem in the digital pipeline, however, I think it’s more likely that there are remnants of the previous sound stuck somewhere in the pipeline.

NoSloppy · October 13, 2021, 5:11pm

OS 5.9, no highpass stuff.
This is not my personal situation, asking for a friend.
Tests as per above will be done tonight. I’ll be back. Thanks.

profezzorn · October 13, 2021, 5:30pm

It almost looks like the amplifier is driving the speaker to 100% positive or negative when we stop sending data to it. It’s not something I remember seeing before. Maybe you’re right, maybe it’s something wrong with the amplifier…

NoSloppy · October 13, 2021, 5:46pm

Oh. It looks to me it’s at rest when no audio, but then “sucked in” when audio plays.
Either way, IMO it’s not a code issue since this is the only reported instance (that I know of anyway), more likely hardware…but ALSO the only time I’ve ever seen it.
The proper test would be to remove the audio jack interface snd try speaker direct to pads as designed.

profezzorn · October 13, 2021, 5:50pm

Maybe one of the wires is shorted to 5 volts?
That would possibly create this sort of bias, and it would go away when the 5v booster gets turned off. Although I’m not sure how that would sound…

NoSloppy · October 14, 2021, 5:58am

The first edit results: the popping went away for 30 seconds.
However there was a hiss introduced during that time. Then a in loud pop at 30 seconds and then silence.

profezzorn · October 14, 2021, 6:36am

The first edit makes the amplifier (and the 5v power supply) stay on for 30 seconds. The second edit makes only the 5v power supply stay on for 30 seconds. I’ll be curious to see how the second edit turns out.

NoSloppy · October 14, 2021, 6:37pm

Editing booster.h only had no apparent effect. Results same as the original video.

profezzorn · October 14, 2021, 8:14pm

So, assuming those edits work the way I intended, this would mean that the weird speaker behavior is related to turning the amplifier on/off, and not related to turning the 5v booster on/off.
Has this board always behaved like this, or did this suddenly start happening?

Sebastianl · October 24, 2021, 9:53pm

Sorry for the late add in! I haven’t had a spare moment until now. I installed this saber, the popping may have been present from the onset. I can’t verify that as it was a using headphones with low volume to test the wireless set up. I had an over site and did not test the set up with a standard speaker either. After putting it on my bench again I haven’t seen any bridging from the 5v to the speaker or power pads. I then removed the 1/4 in jack and tested it with a speaker. The popping has persisted. I plan to just swap this board out for another one as I’m thinking something is wrong with it. If I can answer any questions let me know. I tried to upload the video but it doesn’t like my iPhones format.

NoSloppy · October 25, 2021, 12:08am

Here’s the vid you couldn’t upload.

profezzorn · October 25, 2021, 2:09am

If swapping the board out is an option, please do that. If that ends up not helping, then we should dig deeper into what is actually causing this.