Grounding in a multi-PCB multi-PSU environment

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

living sounds

Well-known member
Joined
Jul 26, 2006
Messages
3,931
Location
Cologne, Germany
I'm wondering about the best way to approach groundingin the following 2-channel DAC scenario:

Switch-mode PSU +/-15VDC powers everything:
- powers directly (with added LCR filters) analog buffer/balancing output board (2 NE5532 in inverting configuration drive XLR outputs, pin 1 is only connected to chassis)
- powers DAC board via an additional linear PSU that puts out +/-7VDC, DAC board again has additional regulators on board to create +3.3VDC for digital and +/-4VDC for resistor ladder; AES/EBU digital input connection is transformer balanced with ground either floating or shield connected to chassis at pin 1 entry
- both DAC board and analog output board have thick PSU ground connections going to the linear PSUs output ground (DAC board) and the +/-15VDC ground (analog buffer board), these PSU grounds are closely connected via low impedance lines
- chassis ground is currently only connected to the board grounds at the point after the +/-VDC switch mode PSU (bolted down directly below the PSU)
- the unbalanced analog output of the DAC board goes to the buffer/balancing board via two microphone cables, the shield joins the DAC boards analog output ground to the buffer board ground

Please see my truly awful (MSPaint) drawing attached. It shows only the relevant grounding.

Here's the problem:

The DAC boards analog ground and the buffer boards audio ground should be connected directly to allow signal flow between the two; otherwise, with grounds connected through the PSU return lines, the return paths cross through dirty PSU ground;

But, the direct connection between the DAC board and the balancing board (with violet spray around in the drawing) also produces a ground loop and potential undesireable return paths for PSU decoupling signals.

There's no measureable increase in hum with the grounds connected, but the low end sounds better without the connection.

Would it make any sense to join the two audio grounds just with a small cap? Or is there another way to optimize things here?

Thanks!
 

Attachments

  • DAC_grounding.jpg
    DAC_grounding.jpg
    210.2 KB · Views: 57
Thanks abbey!

This made a huge difference. I have yet to make roundtrip recordings, but the high end clarity of the output is much improved. Low end seems a little attenuated but is dynamically better than before.

It looks like the attached drawing now. Is that what you had in mind? Chassis connection is now only made to the switchmode PSU ground.
 

Attachments

  • DAC_grounding_abbey_mod.jpg
    DAC_grounding_abbey_mod.jpg
    208.1 KB · Views: 50
BTW, there are also LC filters before the linear PSU. I have connected the capacitors after series inductors to the switchmode PSU ground and not the linear PSU ground.
 
living sounds said:
This made a huge difference. I have yet to make roundtrip recordings, but the high end clarity of the output is much improved. Low end seems a little attenuated but is dynamically better than before.
Honest I cannot predict any sound signature from my recommendations. I just know it's what makes sense in terms of keeping nasty currents where they belong, and experience tells me the consequence is just better sound integrity.

It looks like the attached drawing now. Is that what you had in mind?
Indeed.

Chassis connection is now only made to the switchmode PSU ground.
[/quote] Correct.
 
Thanks! So if I rewire it as in the attached drawing the fact that there is a ground loop doesn't matter?
 

Attachments

  • DAC_grounding_one_PSU.jpg
    DAC_grounding_one_PSU.jpg
    201.7 KB · Views: 43
Thanks... I guess I haven't grasped the concept of grounding quite yet...

If I disconnect 0V at the DAC board's input in the drawing, wouldn't that mean that all the junk from the DAC board's rectification, regulation, decoupling - from the analog and the digital side of the board - travels close to the DAC board's audio ground and then also the buffer board's audio ground - and potentially injects some of it back into the signal? The DAC board has all the rectification, regulation and the digital stuff on the left, but the ladder, the audio outs and of course the audio ground connector on the right.

Also, would it really be OK to have no connection between the switch mode PSU ground and the linear PSU at all?
 
living sounds said:
So I guess this was a stupid question?

Can someone recommend a good book detailing the rules and science behind current return paths?
The last book I read about grounding and shielding was back in the 70s "Noise reduction techniques in electronic systems" c.1976, by Henry Ott.

I don't know that it precisely anticipates your specific application but it covers the basics.

JR
 
living sounds said:
So I guess this was a stupid question?

Can someone recommend a good book detailing the rules and science behind current return paths?
I always liked Douglas Self's Small Signal Audio Design. There's a lot of good explaination about grounding and ground loops and why ground loops are a problem.

The ideal grounding scenario is to have all of your grounds and filter caps converge at one point. But it is also entirely acceptable and probably more practical to have more of a branches of a tree scenario. However, in the later case you have discovered an issue which is that if you want to deliver signal from one leaf to another, that creates a dilemma because if you connect the grounds you get a ground loop but if you don't your signal ground follows a long path around and generally breaks the return follows signal rule. But the ground loop is going to be more of a problem than the ground reference so by disconnecting the ground of DAC regulator (which incidentally is just a regulator, it's not a "linear PSU" and I don't understand why there would be rectification if the source is DC) and leaving only the ground connection between the leaves of the tree, you eliminate the ground loop. It is not ideal that the supply takes a different path because the general rule is that return wires should be physically close to supply wires (again there's the return follows "signal" rule) so that current going out matches current coming back and the two magnetic fields cancel each other and you're not radiating noise. But in this case again you're not using enough current for that to be more of a problem than the ground loop. Ground loops are where most of the noise and hum is going to come from. The essence of a ground loop is that if you have a loop of wire it's like a transformer turn and the whole apparatus is swimming in magnetic fields from mains wires in the walls and such so it causes currents in the wire. Those currents will drop across the impedance of inputs and thus be voltage sources. So by disconnecting the ground of the DAC regulator you break that transformer loop AND you leave the critical signal connection between the DAC output and analog buffer input.
 
Thanks! I mostly tend to think about grounding in terms of keeping "dirt" from being injected into the audio path, so I usually tried to primarily trace/wire ground in such a way that the "dirty" return path is physically closer to the PSU...

Why is there rectification? Well, my linear PSU already had it and there's plenty of voltage to drop anyway, so the less for the regulator to do the better. The DAC board is a ready-made unit that can accept AC power, so rectification is in place.
 
I always though "ground loops" were over blamed as the go to bogey man causing ground related signal integrity issues.  A ground loop is a specific wiring flaw, where a one turn magnetic winding is formed by multiple ground paths and area circumscribed by the loop. In the presence of a significant magnetic field a voltage will be developed across the one turn winding. This is a real design concern inside large old technology power amps using huge old school power transformers. Also an issue inside powered mixers again with a large power transformer in close proximity to signal wiring.

Outside the chassis IMO the more common ground issue from stray unintentional current flowing in common ground paths. Current is like water seeking out the lowest level. Current seeks out lowest resistance path. The whole pin 1 problem design brief is about using differentials to ignore corrupted chassis grounds from common ground currents.

Good luck, this is easier with HF switchers that make smaller magnetic fields compared to old school heavy iron.

JR
 
There wasn't anything in the response that I could see or measure (hum, buzz, noise, distortion), but changing the ground connection in the counterintuitive (to me) way abbey suggested made a major difference wrt to sound quality I can hear.

I still don't understand why it should be wired this way, and why with the way it's connected (only audio ground of DAC pcb goes to PSU ground instead of DAC PSU ground connector) didn't audibly pollute the signal. But it evidently is the right way...
 
living sounds said:
There wasn't anything in the response that I could see or measure (hum, buzz, noise, distortion), but changing the ground connection in the counterintuitive (to me) way abbey suggested made a major difference wrt to sound quality I can hear.

I still don't understand why it should be wired this way, and why with the way it's connected (only audio ground of DAC pcb goes to PSU ground instead of DAC PSU ground connector) didn't audibly pollute the signal. But it evidently is the right way...
Grounding issues will only affect noise and mostly hum at mains frequencies and not "sound quality". If you're noticing a loss in bass or some other change in frequency response, something else would have to be responsible for that.
 
squarewave said:
Grounding issues will only affect noise and mostly hum at mains frequencies and not "sound quality". If you're noticing a loss in bass or some other change in frequency response, something else would have to be responsible for that.

Maybe not grounding, but what about current return issues? I would guess that inductance, resistance, capacitance between audio signal, power and ground lines could impact, for instance, the effectiveness of decoupling caps.

My playing around with grounding in various circuits leads me to believe that it makes a big audible difference beyond hum and noise. Again, probably because the circuit isn't working as it theoretically should.

The "loss of bass" appears to be clocking/jitter related and became more noticeable, probably because the high end was clearer now in comparison.

The discrete DAC with my buffer output stage and grounding fixed denotes the first time that I have heard digital audio competing with analog in terms of the tangible, 3d quality and dynamic integrity (in a converter connected to my PC, a few old CD players did that, too).
 

Latest posts

Back
Top