Ground loops within a single chassis?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'm planning to build an enclosure with the following things in it:

- Power entry module (PEM) with fusing and internal EMI filter
- Ground loop breaker (GLB) consisting of bridge rectifier, resistor(s), capacitor
- Connex SMPS800RS with +48 VDC and 0 VDC outputs (the 0 VDC output appears to be isolated from input power ground although I'm not 100% sure)
- Some sort of +7.5 VDC SMPS whose 0 VDC output is fully isolated from ground
- 2x TPA3255EVM and 2x some other TPA3255 amp boards for 8 amp channels altogether, which have plated mounting holes connected to their incoming power ground/common
- freeDSP Aurora which has plated mounting holes connected to its incoming power ground/common

Diagram to follow, but essentially the way it will be connected is as follows:

- Mains power from the PEM is connected to both power supplies. The mains ground is tied to chassis using a grounding bolt (the metal body of the PEM is also contacting the chassis directly)
- The GLB is fed from the grounding bolt and has "output" tied to the SMPS800RS input power ground and USB connector/cable ground
- The amp boards are fed from the SMPS800RS outputs.
- The power for the Aurora is from the +7.5 VDC SMPS outputs.
- The audio input for the Aurora is via USB connector from a PC.
- The Aurora will transmit audio out via balanced outputs to all 8 amp channels, probably (small chance of single-ended connections)
- The mounting holes of all amp boards and Aurora are isolated from chassis (i.e. not connected to ground).
- The only conductors electrically bonded to chassis are:
-- Incoming mains ground
-- Grounding bolt for safety - which feeds the input of the GLB
- All wiring is contained within the enclosure, with the exception of mains power cable, USB cable, and speaker cables

Some questions:

1) Will I have a ground loop between the two power supplies, or their downstream devices, because neither of their outputs are referenced to ground? Again, not 100% sure that the SMPS800RS output is fully isolated from ground but I get what seems to be an open connection when I try to check resistance between 0 VDC output (marked "GND", surprisingly, on the PCB) and the input ground terminal.

2) From a ground loop standpoint, does it matter if the audio output from the Aurora is single-ended or balanced to the amp boards?

3) Should I bond the mounting hole(s), or some other connection/terminal/reference intended to be grounded, of the amp boards to the output of the GLB so that the amps don't float?

4) Should I bond the mounting hole(s), or some other connection/terminal/reference intended to be grounded, of the Aurora to the output of the GLB? The USB connection prevents the Aurora from floating, I think.

5) What else should I do to mitigate ground loops in this fully-self-enclosed system?
 
Last edited:
Layout attached. I didn't show the amp boards (out of laziness) but I don't think they have anything special; they have +48 VDC and 0 VDC ("ground") power inputs, and SE as well as balanced analog signal inputs. Their mounting holes are tied to 0 VDC ("ground") via PCB. If needed, I'll draw them up as well.
 

Attachments

  • layout-Model 20201122r0.pdf
    23.4 KB · Views: 103
Some cursory comments that may assist:

Best to add in a single amp board and show that the Aurora analog ground will connect to and amp board ground, which connects back to the Connex 0V.

Do either power supply board have EMI related caps from the primary side circuitry to: the AC mains ground (for Connex); the pcb mounting holes; the secondary side circuitry?

Do you have a capacitance meter that may be able to measure a stray capacitance value from primary to secondary side circuitry on the power supplies? Any measured capacitance may not be directly relevant, but there can be parasitic capacitance coupling of noise from primary to secondary side circuitry.

Imho it is also worthwhile trying to prepare a block diagram showing ground connections that can easily indicate loops. If you google keywords like ground, noise, emi, smps, loop and look around the images you should see a few examples where a schematic is trying to portray loop paths for noise or ground currents. You may luck upon a schematic that is close to your situation.

One external loop will include the incoming USB equipment, unless it is battery powered and otherwise isolated.
 
Some cursory comments that may assist:

Best to add in a single amp board and show that the Aurora analog ground will connect to and amp board ground, which connects back to the Connex 0V.

Do either power supply board have EMI related caps from the primary side circuitry to: the AC mains ground (for Connex); the pcb mounting holes; the secondary side circuitry?

Do you have a capacitance meter that may be able to measure a stray capacitance value from primary to secondary side circuitry on the power supplies? Any measured capacitance may not be directly relevant, but there can be parasitic capacitance coupling of noise from primary to secondary side circuitry.

Imho it is also worthwhile trying to prepare a block diagram showing ground connections that can easily indicate loops. If you google keywords like ground, noise, emi, smps, loop and look around the images you should see a few examples where a schematic is trying to portray loop paths for noise or ground currents. You may luck upon a schematic that is close to your situation.

One external loop will include the incoming USB equipment, unless it is battery powered and otherwise isolated.
Thanks, I was just getting tired last night and couldn't get myself to do the full layout. Will do soon.

I do have a capacitance meter but don't know if it would be sensitive enough to pick up stray capacitance. It can go down to picoF in theory.

Will check on PS input bypass caps.
 
The main way to fight ground loops is to keep all flow and returns close coupled, if you do this you shouldn't have any major problems. Granted, it's easier said than done at times.
I think I can, I'm just mainly worried about what would happen because the Aurora and the amp boards are floating (but "grounded" to each other).

Actually, I think the Aurora and all amp boards will use the USB source ground as signal common... might not be a problem after all.
 
Proponent of star topology. Both ground and power as there is no difference between them. Without getting exotic or trying to measure things that your fingers on the probe are greater than, it will get you into the -120 dB or better range without difficulty.

People forget. Ground is an arbitrary point to use as a reference. Current flows where there is a potential and a path. Electrons are immune from symbols on a piece of paper. We should be talking about "parasitic" or " spurious" currents rather than the limited case of our chosen "ground".
 
Scott,

What we call power and what we call ground are just made up notation so we can talk the same language. Electrons move from negative to positive, however they can get there. As an example, my MG was built with "positive ground". So the chassis was the sink for electrons, not the source as is most cars. It makes no difference. You could put a ground symbol on the positive rail and label your mid point - 12, and you most negative -24. Electrons would not know the difference. So, the very same unintended currents that may flow in what we call ground can flow in what we call power. Th only difference is outside the chassis where we are connecting safety ( mains referenced) as we are given an external reference.
 
@454Casull
Your SMPS input ground should be connected direct to PE of your mains socket, not through the ground lifter. Ground lifter should be connected on output side of your SMPS if you like it to get proper grounding to the metal case of your equipment.
I think that you know what are you doing. If SMPS have 0V already grounded to the metal frame of your SMPS, ground lifter will not have any effect.
People told you that mains input of the SMPS have probably some filter inside and current which flow through the filter have to be earthed with PE wire in your mains input cable. You put the ground lifter on that path and noise will spread through your audio ground.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.