Studio Grounding Question

[abbey road d enfer]

It is arranged as in the attached image - where the secondary windings are joined, it is tied to the chassis. Is that correct?

No. It's the junction of C35/C38 that must be tied to chassis.

 

[Gold]

In my defense I did say "0V/chassis". I meant connect to audio common then to chassis. Like in the schematic.

 

[plumsolly]

It's a bit tough to explain how I have arranged it in words and I will add a picture if I am unable to get it across this way, but here goes: I am using a PCB for the output section, and that PCB also has on it the regulator and filtering with the exception of the two 470μF caps (C33/36) and the rectifier diodes (this is because it is a module from a console and it is expecting regulated +/-22V). On the pcb, 0v audio and the ground plane are connected right at the card edge connector. What I have done is run a wire from the card edge connector 0V/ground plane pin to the chassis where it is joined by the junction of C33 & C36 and the two secondary windings.

Reading that back, I do not have high hopes that it will be clear to the reader what is happening - I'll try and get a picture up.

 

[plumsolly]

I didn't learn anything much new in my experiments today. I tried lifting the safety ground (don't worry - it was just for science and I put it right back) and it had no effect. Safety grounds in the mains and power strips appear to be intact and connected to neutral at the breaker as it should be.

I have noticed that the noise is inconsistent - it's not there for a while and then it pops up. The only way I have found to make it go away that works consistently is to shut off my console power supply.

I am attaching a picture of the spectrum, but I don't think it's too revealing.

Thanks again to everyone that has been trying to help - I really appreciate it.

 

[scott2000]

Reading that back, I do have high hopes that it will be clear to the reader what is happening - I'll try and get a picture up.

A picture would be good. Sounds like it would reveal something....

No mention of grounding chassis at caps abbey mentioned yet?

Confusing about the 22v console supply mention tbh...

Transformer struggling??

I can't open schematic yet even though it says download pdf when I copy/paste...prob phone issue....

 

[Gold]

It's a bit tough to explain how I have arranged it in words and I will add a picture if I am unable to get it across this way, but here goes: I am using a PCB for the output section, and that PCB also has on it the regulator and filtering with the exception of the two 470μF caps (C33/36) and the rectifier diodes (this is because it is a module from a console and it is expecting regulated +/-22V). On the pcb, 0v audio and the ground plane are connected right at the card edge connector. What I have done is run a wire from the card edge connector 0V/ground plane pin to the chassis where it is joined by the junction of C33 & C36 and the two secondary windings.

Reading that back, I do have high hopes that it will be clear to the reader what is happening - I'll try and get a picture up.

I guess I'm having trouble picturing it. Are you saying that the Input section is on a separate PCB than the Output section? How does the Input section connect to 0V?

As an experiment I'd use a known good bipolar 15VDC PSU, connect 0V of the Input section to the Output 0V at the edge connector. Then connect that to PSU 0V then PSU 0V to chassis. That's a straightforward way of doing it.

 

[plumsolly]

A picture would be good. Sounds like it would reveal something....

Meant to write "do not have high hopes" I will plan on getting a pic up.

No mention of grounding chassis at caps abbey mentioned yet?

The caps Abbey is referring to are on the PCB and connected to the ground plane which is then taken to the chassis from the edge connector.

Confusing about the 22v console supply mention tbh...

The output section is on a PCB that is from a console module. The way that console was set up, the power supply was regulated +/- 22V, and then each module had its own regulation to bring that down to +/- 18v or so. So the pcb doesn't have its own rectification or large filter caps cause it is expecting regulated DC, so I had to add that in on a terminal strip. Not sure if that's any clearer  :-[

Transformer struggling??

That's a good question - I don't think so, but I should measure the current draw.

Thanks for using your brain power on this.

Ben

 

[plumsolly]

I guess I'm having trouble picturing it.

I figured that wouldn't get it across - I'll get a pic up.

Are you saying that the Input section is on a separate PCB than the Output section?

Yes.

How does the Input section connect to 0V?

Just a molex connector to 0V on the pcb and then a separate connector and wire for the ground plane.

As an experiment I'd use a known good bipolar 15VDC PSU, connect 0V of the Input section to the Output 0V at the edge connector. Then connect that to PSU 0V then PSU 0V to chassis. That's a straightforward way of doing it.

That's a good thought - I have a feeling, though, that the problem is not going to want to develop on the bench - I think it needs to be connected to the same mains power as my console power supply for the noise to develop. I'll see if I can get it to act up though.

Thanks again for your help.

Ben

 

[Bo Deadly]

If you really want to see what's going on you need a high res spectrum. If you can post a wav file I have a script that can generate one like I did here:

  https://groupdiy.com/index.php?topic=76066.msg968004#msg968004

Specifically, ground the input (or use an impedance appropriate for a phono input - not sure what that is) and then record 10 seconds of silence at 96kHz / 24 bit. Pan that track hard right and then export to a 32bit float stereo wav file. Post the wav somewhere where I can get to it, I'll generate the spectrum and post an image of it here.

Tracing noise issues like this is difficult because it can be many things. But a really good high resolution spectrum can give you some clues. The spectrum tools of DAWs are usually pretty crude because there are limits to realtime processing. You can see that in the highly interpolated low frequency data in your pic.

 

[plumsolly]

If you really want to see what's going on you need a high res spectrum. If you can post a wav file I have a script that can generate one like I did here:

  https://groupdiy.com/index.php?topic=76066.msg968004#msg968004

Specifically, ground the input (or use an impedance appropriate for a phono input - not sure what that is) and then record 10 seconds of silence at 96kHz / 24 bit. Pan that track hard right and then export to a 32bit float stereo wav file. Post the wav somewhere where I can get to it, I'll generate the spectrum and post an image of it here.

Tracing noise issues like this is difficult because it can be many things. But a really good high resolution spectrum can give you some clues. The spectrum tools of DAWs are usually pretty crude because there are limits to realtime processing. You can see that in the highly interpolated low frequency data in your pic.

Thanks so much for the offer!

Here is a file https://drive.google.com/file/d/177DaUGqfULEDXxl1WJ1LFVA-v1iWI_0r/view?usp=sharing - It does have 24db of gain applied to it, digitally.

 

[Bo Deadly]

It does have 24db of gain applied to it, digitally.

Here ya go:

This is clearly very bad to the point where something is fundamentally wrong. I would say check continuity between various ground points. Start with pin 1 on the XLR our (or whatever shield connection) and the earth ground pin on your mains plug. It should be near zero. I suspect it's not. Your ground is floating or your missing a connection entirely.

 

[plumsolly]

Ok, here's an annotated picture that hopefully makes my grounding scheme more clear.

 

[plumsolly]

Here ya go:

This is clearly very bad to the point where something is fundamentally wrong. I would say check continuity between various ground points. Start with pin 1 on the XLR our (or whatever shield connection) and the earth ground pin on your mains plug. It should be near zero. I suspect it's not. Your ground is floating or your missing a connection entirely.

Thanks for doing that. Pin one connects are solid in the unit (and lifted at the patch bay on the receiving end) 0V/safety ground/chassis/pcb ground plane are all connected and at one point only.

 

[Gold]

I think that the 0V reference for the input board should be taken from the point where the white wire is connected to the edge connector.

The shields from the output of the Input PCB shouldn’t be connected to 0V. Those should go to chassis on both ends.

 

[plumsolly]

As an experiment I'd use a known good bipolar 15VDC PSU, connect 0V of the Input section to the Output 0V at the edge connector. Then connect that to PSU 0V then PSU 0V to chassis. That's a straightforward way of doing it.

As I feared, it can't get it to act up on the bench - quiet as a mouse. If I don't figure anything else out, maybe I'll bring the bench supply over to the rack.

 

[plumsolly]

I think that the 0V reference for the input board should be taken from the point where the white wire is connected to the edge connector.

I took that 0V from the where input stage was originally on that PCB (it was 990s as summing amps) so I think it should be ok.

The shields from the output of the Input PCB shouldn’t be connected to 0V. Those should go to chassis on both ends.

That point on the edge connector isn't just 0V, it's where the PCB ground plane and 0V meet. I needed to get a ground plane connection over to input PCB for the 990's ground plane connection and thought I could kill birds with one stone - but maybe not?

On a side note - is the conventional wisdom to connect internal shields at both ends? I'm so used to lifting the shield at one end when connecting balanced gear in order to prevent ground loops, that that is what I do inside a chassis usually, too.

That connection probably doesn't even need a shield - it's at line level already by that point.

 

[Bo Deadly]

As I feared, it can't get it to act up on the bench - quiet as a mouse.

That's important. The noise in the spectrum would be highly audible. There's main at >40dB above the noise floor. So it's not EMI from the power transformer or the junction particulars at which grounds are connected together, it's something more fundamental. Something is wired incorrectly.

Again, what does your meter read between pin 1 of the XLR out and the earth ground of your mains connector?

Check to see if there's any current between pin 1 of the XLR out and pin 1 of the gear your connecting to. What does the meter read precisely?

 

[plumsolly]

Here's the noise on the scope. just in case that's useful.

 

[plumsolly]

That's important. The noise in the spectrum would be highly audible. There's main at >40dB above the noise floor.

Oh, I can definitely hear it when it is connected in the rack and exhibiting the noise- it doesn't always. I just meant it is not exhibiting the noise on the bench. I believe it will only exhibit the noise if it is sharing a mains circuit with my console power supply and the console power supply is pulling current.

So it's not EMI from the power transformer

100% agree.

Again, what does your meter read between pin 1 of the XLR out and the earth ground of your mains connector?

< 0.2Ω

Check to see if there's any current between pin 1 of the XLR out and pin 1 of the gear your connecting to. What does the meter read precisely?

1.5-2µA

 

[abbey road d enfer]

I have noticed that the noise is inconsistent - it's not there for a while and then it pops up. The only way I have found to make it go away that works consistently is to shut off my console power supply.

I think it's a very important clue. Surely, the mains witch of the console PSU does not alter the earth circuit, so any ground loop is unchanged by the position on or off of the switch.
It means that turning on the PSU switch introduces significant leakage into the console ground. I would check for a defective Y capacitor. Y capacitors are between Line and Earth and between Neutral and Earth.