PCB: separate psu gnd from audio gnd ???

[Neeno]

Hi guys...
I'm working on a api 500 series pcb.
As I can see from the specifications the psu and the audio gnd is separated on the edge connector and i know that is a good practice to keep them separated and to join them on the star ground.

I suppose that the star ground point is should be inside the lunchbox.

I would like to use 2 ground planes one for each layer of the board, one for the psu (top layer) and one for the audio gnd (bottom layer) and then connect them to the edge connector separately; pin5=audio gnd, pin 13 for the psu gnd.

Now some questions:
1- Would it work ???
2- Would the groud planes make the board act like a capacitor ?
3- There is a better solution ?
4- If yes, what's the best solution ?

Thank you for your help !

 

[Ian MacGregor]

Neeno,
You might want to double check your 500 series rack and make sure that the audio and power grounds are actually tied together in the power supply. I'm not 100% sure that everyone does the same thing here (anyone know for sure??). There is also a chassis ground and I have been informed that sometimes the chassis ground is connected to audio ground inside the supply, sometimes not (depends on manufacturer). You can always keep the grounds isolated, but inculde internal jumpers just in case.

I'm not sure about the advantage of hacing two separate ground planes instead of one single ground plane. Maybe try having one audio gnd plane (since there will probably be more components connecting to audio gnd) and set up your power gnd traces in a star ground configuration. Again, I'm not so sure which is better, I'm sorta thinking aloud.

Ian

 

[keefaz]

Aren't the power ground and audio ground connected in an API opamp ? (it has only one gnd pin for example)

 

[Samuel Groner]

Using two grounds doesn't make much sense to me (for this application) and I would not consider this this to be good practice. I suggest separate grounds (no need for a plane though) for LEDs and relays, but that's about it. For two ground planes you need a four layer board, which is likely overkill for such simple circuitry, expensive and a pain to (de-)solder.

I think I'd like to have a look at the schematic though--from your writing I have little idea how the two grounds are used.

Would the groud planes make the board act like a capacitor?

Sure--but (one of the) the cool thing about ground planes is that you know that the stray capacity goes to ground and not to anything else. The capacity gets larger, but you know where it is.

Samuel

 

[Neeno]

Hi guys !

Ian MacGregor:
actually i don't own a 500 series rack (i'll get one for Christmas probably) but i think that since they are kept separated on the edge connector there should be a good reason... (well, that's what i think...)

Samuel:
Probably i didn't explained it well, sorry, but my english is still very bad...
The board is a dual layer board, and for ground plane i mean a plane that surrounds the traces on both sides, obiviously it connects all the psu gnd on one side and audio grounds on the other...

Actually i don't really know how to proceed with that... :shock:

 

[JohnRoberts]

Ground is a circuit node but not a single point that can have a fixed voltage. Current flowing in ground wires will contaminate this ground reference quality. PS regulation will only be relative to ground at a single point inside the PS. A separate ground reference wire could eliminate the current errors in ground voltage but the regulated voltages will still be affected by current draw so this is of questionable value.

IMO power ground and chassis ground should be hard connected but audio ground at most should only be connected at one point and differentially refereced between inputs, outputs, and internal nodes.

A second ground wire may be used as that one signal ground connection but IMO with a proper design where that one ground point is connected is not extremely important. If the electronics have dubious PSRR this ground point should be clean relative to supply regulation.

So the extra wire "could" make sense for a given design strategy. Trying to bus a clean audio ground within a chassis is nice in concept but difficult in practice, as that reference can get contaminated by interference and even use. This sounds like a carry over from old school valve design where active elements were expensive and scarce. Please take my word for it, busing a clean ground reference across a 6' wide console doesn't work very well.

If the PS is distant from the circuitry consuming the current, compliance between the reference trace and current carrying trace can cause other problems. Also loop area between the traces may introduce susceptibility to magnetic interference. Grounds should probably be capacitor coupled using HF ceramic disc caps and wiring carefully routed.

JR

 

[Samuel Groner]

Can't we see the schematic where it might get clear what "PSU ground" means?

Samuel

 

[Neeno]

Here we go...
This is the Api 500 Edge connector specification.
There is a separate connection for Chassis Ground, Common (Audio GND) and PSU Common...

EDGE CONNECTOR PIN OUT
Pin 1 CHASSIS GROUND
Pin 2 OUTPUT + (+4 LEVEL)
Pin 3 OUTPUT + (-2 LEVEL)
Pin 4 OUTPUT -
Pin 5 COMMON
Pin 6 525 STEREO LINK
Pin 7 INPUT- (-2 LEVEL)
Pin 8 INPUT- (+4 LEVEL)
Pin 9 INPUT+ (-2 LEVEL)
Pin 10 INPUT+ (+4 LEVEL)
Pin 11 GAIN TRIM RESISTOR
Pin 12 +16VDC
Pin 13 POWER SUPPLY COMMON
Pin 14 -16VDC
Pin 15 +48VDC

I'm working on a simple Api 312 schematic, if necessary i'll post it but actually i don't have protel installed on my laptop...

I think that the TRS (DI input) should be connected to the CHASSIS GND (this is what i normally do in my gear, i tie them togheter and then i connect them to the star ground) with the input trannie shielding and, in this case, the output trannie core.

I considered all the audio grounds everything that is in the signal path and goes to GND on the schematic using the ground plane on the bottom of the board.

The IC, DOA and switches are connected to PSU COMMON, using the top ground plane...

I hope this helps you understand what I'm doing...

 

[sodderboy]

In API speak, from the 550 manual:

Term. 1 is Chassis Ground

Term. 5 is Shield Ground (Power Common)*

Term. 8 is Input Low (Power Ground)*

Term. 15 is +/- Power Supply Common*

*These terminals are internally connected.
[/API speak]

The Technical Earth in the rack would be at a single connection near the IEC ground pin.

If I were going to do anything different than the way API spec's it, I would follow the Neve spec for wiring their classic consoles: with few exceptions, all balanced shields(API's Shield Ground) connect to chassis at the module, and all unbalanced shields(API's Input Low) connect to 0vl or "B-neg" (API's "+/- Power Supply Common) on the backplane connector. Technical Earth would be the same- at a single point 0vl and chassis are connected. On the Neve or SSL for that matter it is near the power input to the console. In a rack it is close to the IEC connector.
So to do that I would connect, in the module, 1 and 5, and also 8 and 15. At the end of the buss, 5 and 15 would wire-off to Technical Earth. If an old 550 was connected that did not have this change, then it would defeat the purpose. If you want to change the backplane, you have to change any old modules that might be plugged.
Mike

PS: I am working on a console who's design follows the exact opposite of the above. 0vl(Audio Common or +/- Power Supply Common) is tied to chassis at EVERY FADER; 64 in all. At least 64 "technical earths"! And it is no surprise that the buss leakage is unacceptable. It was made this way, and I know that when I remove all those improper connections, the buss leakage and noise floor will be greatly reduced, to the level of the next possible design flaw.

 

[Neeno]

The edge connector from api website:

 

[JohnRoberts]

Inside consoles I prefer robust grounding. Connecting faders to chassis ground will be good for shielding but require differential referencing for both signals going to and coming from. Since faders are typically 10k they need to be connected to a real ground to manage crosstalk. Best practice is to ground the fader at the channel strip to keep circulating signal currents local to that strip and perhaps save on one of two differential amps.

FWIW the quality of differential coming from the fader is perhaps the more noticeable one because it will show up in fader kill and crosstalk measurements. Not properly referencing any signal path can lead to signal corruption and degradation.

JR

 

[sodderboy]

I think we are saying the same thing, but the words chassis, ground, etc. get confusing like different names for a sandwich made on a long roll . At least the word "earth" does not appear anywhere in this thread!

I am saying that 0vl (Audio Common, PS Ground) is connected to the fader chassis and the console chassis AND the fader bottom at 64 points in my current patient, which is bad design. My reference is every Neve, API, Trident, SSL, etc. I have fixed or modified. None of those consoles, when installed properly, had channels leaking into the mix buss at -50dB.

My fix in this case will be to disconnect the fader bottom from chassis at every fader and I know that that will solve most of my crosstalk problem. The fader chassis will still be connected to console chassis, and my 0vl, which I you called "real ground" will still be connected to fader bottom. I will control the one and only place where they connect together.
Mike

 

[JohnRoberts]

Ground can never be a single voltage across any distance or combination of currents.

There are two basic design philosophies. "Brute force" where you use 00 gauge welding cable or solid copper bus bar to keep resistive losses low, or live with the sundry different voltages that build up across any real ground and use differential subtraction to reference toward and away from chassis or different local grounds. In practice larger system designs use a combination of the two. Brute force ground for chassis, shielding, and power supply distribution and returns. Clean signal grounds are actually multiple local "brute force" ground references that signals pass to or from differentially.

Since modern consoles are a series of gain stages, by selectively locating where their feedback networks physically connect to ground serves to reference either from or toward a local ground depending on the circuit function.

To get 100dB of fader kill you must reference to the fader itself, some consoles even connect two wires to the fader ground for a Kelvin connection (signal current flows in one, other serves as reference).

Think of every ground as a voltage that must be managed.

JR

 

[sodderboy]

[quote author="JohnRoberts"]
To get 100dB of fader kill you must reference to the fader itself, some consoles even connect two wires to the fader ground for a Kelvin connection (signal current flows in one, other serves as reference).

Think of every ground as a voltage that must be managed.

JR[/quote]

Does your use of "ground" translate into Neve's use of "B neg", which is also their audio common? I really think we are saying the same thing. In your example, does fader bottom ("ground"?) connect to audio common for signal current, and also connect to console chassis for reference?

If you can point me in the direction of a professional mainframe audio console that ties power supply 0 volts to the chassis at multiple places, I would like to study it. I would be an outlier for me. I have also never seen that in small modular racks like the API lunchbox or Prism.
Mike

 

[JohnRoberts]

I don't spend a lot of time reading schematics for other console designs, so I'm not familiar with everybody's nomenclature.

PS ground usually connects to chassis ground at one point, usually near where the PS enters the console, if external supply. There will be some impedance between PS ground to local signal ground to channel input ground, but this will not be a hard short as that would create multiple ground loops. Since input shields are routinely tied to chassis it not uncommon to encounter loops with external line level wiring which the console designer has little control over so all input grounds are properly differentialled to mitigate loop problems.

Speaking in general about faders, if the fader has a metal case that can and should be allowed to mechanically make contact with chassis ground. If that ground is present on a solder-able terminal on the fader, that can be telescoped to shield the wiring between fader and channel PCB but should not connect to ground elsewhere.

Ideally the fader itself needs a 4 wire connection. #1 top of fader, #2 wiper, #3 fader element bottom or fader ground, and #4 fader ground reference. Since #3 is carrying fader signal current, there will be a voltage drop across it. #4 connects to a higher resistance termination so will suffer minimal IR error. Since it is common to follow channel faders with a 10 dB gain stage, this gain stage in either inverting or non-inverting topology is executed with differential input resistors so signal is extracted from the wiper relative to fader ground at the fader. Using this topology the fader kill will be whatever noise and crosstalk is present on the fader ground reduced by the CMRR of the differential amp (1% resistors don't hurt).

In general the fader ground could use the shield of a 3 conductor shielded cable. I am not comfortable with connecting fader ground hard to chassis ground but in principle with a proper differential you could reject any chassis ground noise and crosstalk by 40 dB or more. Another complication from connecting fader ground to chassis is that signal being sent to the fader would also need to be referenced to that particular chassis ground. Since chassis ground is a catch all for noise this would have to be done on a per channel basis.

It is far simpler to connect the fader ground to the local channel signal ground which will be consistent with the audio it is manipulating. There are sundry workable approaches (some better than others) for managing grounds and signal flow within consoles, I offer this as one way I know that works. It doesn't mean all others won't work. YMMV.

JR

Note: if the fader element bottom is electrically connected to the fader case you have little choice about grounding but my recollection is that the resistive element is usually isolated at least on all the faders I've worked with.

 

[sodderboy]

I'll take that as a very wordy affirmation of what I said.

Back to the OT: API designed "audio common" and "power supply common" as the same trace on the board, they just take it to different edge contacts and call it different things. Same trace, different names when split to three edge contacts- Shield Ground, Audio Common, and PS Common.
The plan to have circuit returns changed from the original API design might cause unforseen problems with what John and others have said, in other threads, about PCB "ground" design. And your two planes will have to both connect to the three contacts aforementioned anyway. Will this be an improvement from the original? Probably not, but then I am not a PCB designer.
Mike