2

Trying to add a MIDI synthesizer & speakers to a keyboard here. Here's the schematic:

schematic

simulate this circuit – Schematic created using CircuitLab

But I was getting annoying crackling and popping noise. I did some debugging and found my problem is very similar to the one in this question:

Ground loop? Switch mode power supply noise

The problem is the amplifier board connects the negative inputs to the power ground. These are then connected to the headphone ground, which is really 1.5V*, causing the worst possible ground loop! I guess there's another coupling capacitor inside the computer to protect the headphone ground from being shorted. It seems in theory if the headphone ground stays perfectly constant, then there will be no ground loop current and therefore, no noise. But in reality, there will be current flowing from the headphone ground.

So, I cut the traces on the amplifier board that connected the negative inputs to ground. The noise went away finally.

But I'm still utterly at a loss at why would an amplifier connect the negative inputs to ground? That would only work if you plan on powering it from another isolated power supply.

*I measured HPCOM @ 1.5V without load. The Allwinner A10 datasheet (22.1) says, "Support to automatic change from Virtual Ground to True Ground to protect headphone amplifier"

Yale Zhang
  • 577
  • 6
  • 17
  • 1
    *That would only work if you plan on powering it from another isolated power supply.* Which isn't an unreasonable expectation to make? – Marcus Müller Aug 13 '17 at 12:40
  • Because a differential signal with the negative connected to ground behaves like a single-ended signal. Your amplifier wants differential signals, your computer outpts single-ended signals. Most audio systems use single-ended systems, referenced to ground (3.5mm jacks, 1/4 inch jacks, cinch connectors, etc.) I think your computer is doing stranger things here (putting your headphone ground at 1.5V to actual ground?). Having signals referenced to ground is the standard, hence the amplifier connects negative input to ground. – Joren Vaes Aug 13 '17 at 12:43
  • 2
    @JorenVaes possibly the computer has an intentionally simple audio output, and no dual supply, so that it needs a virtual ground in the middle between supply (probably 3V) and GND to be able to operate its own output amps. – Marcus Müller Aug 13 '17 at 12:44
  • @MarcusMüller I guess that would make sense, but at the same time, shouldn't audio signals like that be AC-coupled internally anyways? It just seems to me that this is going to give you issues was soon as you connect that output to a mains-referenced amplifier that has chassis (and thus cinch ground iirc) referenced to earth... but practical audio circuits is not my expertise so I don't know what the common practices are in a system like a computer headphone output. – Joren Vaes Aug 13 '17 at 12:47
  • 1
    @JorenVaes yeah, not my expertise, either. And yes, should be AC-Coupled, but then again, that output seems to be labeled "headphones out", so maybe the left and right signals are AC coupled, but ground isn't. – Marcus Müller Aug 13 '17 at 13:04
  • @MarcusMüller Focus on Marcus's key point: **headphone out** is meant for a floating load. Anything downstream had better have good common-mode rejection. Count yourself lucky that capacitive-coupling worked out. – glen_geek Aug 13 '17 at 15:00
  • @MarcusMüller. isolated power supply means a bulky transformer, which wouldn't go well with today's mobile devices (I'm going to put the SBC & amp inside the keyboard). The headphone ground is a virtual ground according to the datasheet (see footnote I added). – Yale Zhang Aug 14 '17 at 00:53
  • The amplifier in question is a Sanwu SW-HF82. I also have a Sure electronics AA-AB32231 board and it also ties in- to ground. When I first discovered this, I was shocked - I thought the designer had to be drunk. Then I saw the reference design in the TPA3110 datasheet which does the same thing, so I figured it had to be for a reason. – Yale Zhang Aug 14 '17 at 01:05
  • Ground is just a cruel joke. Learn about star grounds. And examine how long paths into and from the "star" cause useful isolation. Read the final example in this answer of mine https://electronics.stackexchange.com/questions/306862/should-i-really-divide-the-ground-plane-into-analog-and-digital-parts/306896#306896 – analogsystemsrf Aug 14 '17 at 05:39
  • Nothing wrong with the amp design but why is your HP com is not grounded. You may need to open up your PC to take a lool. – Jason Han Aug 14 '17 at 07:19

0 Answers0