Weird power problem?

So I've had my WTPA for a while and it's been working dandy but I've noticed a weird problem as of late. I have mine built with a little mixer in front of it (in the same enclosure) as can be seen here:

Now sometimes it does this weird power sagging kind of thing where if I have something running into input1 that's kind of loud, then I do something on input2 that's also kind of loud, both get quieter and kind of duck each other out. I assumed something funky was happening with my mixer (which is possible as I had a helluva time troubleshooting it in the first place), but today I noticed the same happens when I have stuff coming OUT of the wtpa (which is after the mixer in my wiring).

So something going into wtpa, I record it and start playing it back. Then something different going into the circuit(mixer and/or wtpa) and the output of the party gets ducked out.

I'm running both of a 12v 500ma supply (might be 9v now that I think about it, but it's definitely 500ma).
Is that not enough?
Do these symptoms point to something else?
Am I smashing the amp input? (I generally run into a guitar amp)


  • 500mA should be more than enough, mine works well with 200mA, without the mixer, which is unlikely to draw that much.
    btw nice WTPA!
  • Yeah the mixer is low power consumption too, but I can't think what else could be effecting both circuits (though I guess it could all be happening in the WTPA (the input and output stages).
  • how are the boards interconnected?
    do they work fine separately?
  • They're both pulling power straight from the supply or maybe it's coming after the power filtering for the wtpa? It's pretty messy inside my enclosure. It's also impossible to test them separately due to the aforementioned messiness (they're hardwired, and I believe the connection between them is covered in copper foil, and tapped to a copper foil 'ground plane' on the side of the enclosure etc..).
  • Remember, if the voltage is getting low, you should see the LEDs dim, and the processor will reboot at some point. If you aren't seeing that, it doesn't mean that isn't the problem, but you may want to look at other causes as well.

    Symptoms like that can also be indicative of too-big RC time constants in the bias sections of amplifiers, like C13 and R12 on the input of WTPA or C17 and R18 on the output.
    If those are overly large, a big signal swing can pull the DC bias point to one rail or another and you'll hear exactly what you're talking about.

    While WTPA has plenty of audio foibles, I'm inclined to not suspect WTPA assuming it's operating correctly -- it's designed to have reasonable filter points (1.6Hz input, 0.3Hz f3db on input and output resp) and I haven't heard any of mine do that.
    An easy test is to try your WTPA/mixer with a different supply, with different inputs, and plugging your wtpa/mixer into some other amp on the output side and see which if anything affects your problem.

    Also, might want to post the schematic for your mixer here so we can look at it.
  • Yeah it wouldn't surprise me at all if its the mixer or the 'in between' stuff that is messed up as I did plenty of messing it up when putting it together.

    I'm only speculating on this point as I need to actually sit down and test in a controlled environment (not a gig or rehearsal), but I also want to say that the problem starts after a certain amount of time. Like that at first everything works fine, but after a bit it goes wonky.

    Here is the mixer schematic (from the second of my 2 epic troubleshooting threads):

    I'm going to sit down and test a different power supply and inputs and outputs to see if I can narrow down what's happening.
  • That midrail reference and the output RC are both pretty squishy. Midrail shouldn't matter if your supply is stiff, but try strapping a 10k resistor across the output and see if your problems go away.
  • By output do you mean WTPA output (audio), mixer output (audio), or powersupply output (going to mixer and/or WTPA)?
  • Ah, sorry. I meant the mixer audio out. WTPA should have a 59k to ground (or so) already in place.
  • Does it have to be right on the board (of the mixer) or can it be nearer the input of the wtpa? My mixer board is cocooned in copper foil...
  • Doesn't really matter where for this test.
  • Looks like the cable going from the mixer to the wtpa was only grounded on the wtpa side (there probably was a reason for this). I've soldered a 10k resistor across the input of the wtpa and will do some extensive testing tomorrow (when I can be loud).
  • Ok, so I put a 10k resistor across the input of the wtpa and I think I misunderstood what you're saying to try.

    I literally connected a 10k resistor from the + of the input to the - of the input, so all but 10k of the signal is going to ground now?

    It's suuuuuuper quiet now (the mixer output). The click from the wtpa is significantly louder than anything else I can record into it, and quietish things are smeared into noise (in a not so cool way).

    I'm starting to wonder if you were in fact talking about power instead of signal, or something?

    As a point of reference, I just took a 10k resistor and connected it from one of the pads of the wtpa audio input, to the one right next to it.
  • >>>>Looks like the cable going from the mixer to the wtpa was only grounded on the wtpa side (there probably was a reason for this)

    if you're talking about the outer copper of a shielded wire that would be correct in order to avoid ground loops.
    but to make sure both circuits have the same GND as reference there must be 1 GND connection between them.
  • Yeah that seems like a reason why I would've done that. I'm pretty sure I wired everything to a common ground too when building it.
  • Generally, connect all your grounds unless you have a really good reason not to.
    Ground loop stuff can make problems, but they're kind of like capacitor related problems in that audio people seem to be drawn to them. If all your stuff inside that box is referenced to the same wall wart and wired right you shouldn't worry much about ground loops. Having a weird signal path is a bigger concern.

    I just noticed there are two mixer schematics on the page you referenced. The 10k resistor I suggested as a test across the output of the first circuit (the one with the 1M output). It won't matter on the second one.

    I don't know how your mixer is wired w/r/t the WTPA, so I can't say if your diagram above is correct. Basically, the impedance into WTPA is high and the impedance out of that first mixer is as well. 10k shouldn't attenuate your level in that case. If the RC is different on the mixer (it looks like it is on the second one).

    The question is just one of RC time constants.
    At this point try and get either the WTPA or the mixer to do the behavior alone, then fix it. If neither does it, hook them together and try to get the failure to happen. If it only happens when they're connected then the connection is suspect in some way.
    Make sense?
  • Mine is the second mixer (the buffered one), but with some value tweaks (as per your suggestions in the thread) and bypass caps and other stuff like that (also from that thread).

    I'm going to undo the 10k resistor and do some thorough testing to see what exactly is happening and when.

    So the 'sagging' I'm getting is related to the output RC thing? (4.7uf/10k) If I can recreate the problem easily enough I'll record what's happening too.

    I really want to avoid, if at all possible, decoupling the circuits as it's a terrible mess in there and I just know I'll cause more problems than I'll fix, and I have a few gigs coming up that are WTPA based (only one input, so it's not a problem then).
  • Ah, OK. That RC is actually pretty good, and shouldn't cause trouble really. The one on the first mixer (the 1M) could, though.

    What you're hearing I've heard when a DC bias point starts to float around.
    If it didn't used to do that, and now it does, it's worth checking for a broken connection too.

    And, sure, if you can try a different wall wart (if you haven't) because that's easy. I suppose it could be PSU related too, but I wonder.

    I guess I'd also isolate the two different devices and try them out to see who's the culprit. I'll leave it to you to get creative and figure a way to do that with minimal disassembly.

    And yep, take out the 10k.

  • After taking the 10k resistor out I found I had no audio going through at all...

    After some troubleshooting (including testing just the mixer out) I realized I had turned the input gain on the wtpa all the way down by mistake. The way I have mine built, the input gain is a hidden knob inside the enclosure, so it must've gotten bumped.
    After recalibrating it to guitar level input I tested it for a bit and couldn't get any sag sounding problems.

    Gonna have to get back in there and test it for a longer period time (in case that has anything to do with it).
  • Sounds like progress to me. If you never want to adjust the input gain, and you always use it with a guitar, you could always just replace the pot with some resistors.
  • Why do that when you can just shove a piece of blue-tak where the shaft meets the pot....

    (not that I did that)

    (I totally did that)
  • Personally I find that plugging my inputs into outputs also greatly reduces noise levels.
  • As an aside, I tracked down the problem here. I don't doubt that I have some funky stuff happening in my mixer/wtpa setup, but it turns out my main sampler (Roland SPD-S) is acting a fool with certain samples. Playing it back all ducked and distorted. Quite unusual.
  • In my experience that's usually a bit rate
  • It appears to be 16bit 44.1 just like the rest of the stuff.

    I'm wondering if it might be some weird watermarking thing happening, but I can't see why it would freak out my hardware sampler.

    Even when they would play back normally if I recorded it with the WTPA there would be some crazy high pitched squeal.

    Actually, I just loaded it into an audio editor.
    What on earth do you make of this:
  • If I run it through a high-pass filter (@1hz) it scoots the waveform up and it looks kind of normal (except clearly offset) until the ending bit where it looks completely normal.

    I'm guessing the DC offset, or whatever it is that's actually happening in this waveform is freaking my sampler out.
  • That squeal is probably a function of high frequency noise on the sample's output which is interacting with the aliasing of the WTPA. That sample has something clearly fucked above zero side; if I saw that on some of my gear I'd imagine I was screwing up some signed/unsigned conversion somewhere. Looks like a data error to me.
  • I managed to track down versions of the samples that aren't all jacked up like that. It does now mean I need to pick/convert/export/etc.. all the samples that look/sound like this.

    It's good to have that narrowed down/resolved (and knowing it's not a WTPA/mixer problem)
Sign In or Register to comment.