GND Scheme questions

[pdf64]

Apologies for my previous ’pin 1’ comment, on refreshing myself regarding the topic, I realise having not tinkered with balanced systems for a good few years, I’d forgotten a lotta stuff and was under a misapprehension

[dai h.]

hey, no worries. Hard to tell what was going on from the first diagram anyway, plus a bit of a complicated subject.

Anyway (while trying to watch some of the WC games), found a couple more chassis pics:







re: electronically balanced circuits: I don't have a very good understanding and I only sort of vaguely recognize the Dimension D input circuit which looks a bit weird (with different values on each leg (usually they seem to be similar). The Elliott Sound Products article doesn't seem to think this is particularly good balancing scheme, and that may be true but if it only sees higher level (line level) circuits (looks like that is what it's meant to see) and the cables are not particularly long or not in a difficult noise environment, maybe this lesser scheme will still work okay. Sort of like not necessarily needing super duper low noise and high CMR(common mode rejection: expression of how good the balanced is at some frequency, example (-)60dB@1 khz = the line has the ability to suppress noise affecting the balanced lines (pins 2&3) to 1/1000th) for a mic pre all the time. If you are using a relatively higher output condenser mic, plus close micing a loud source, it might be totally fine to use a so-so mic pre (I think less gain needed = less distortion if more gain is produced using less feedback in the circuit, plus a hotter signal means less susceptibility to noise so aspects like the layout are maybe less critical). Foley work (sound effects for TV, movies, etc.) seems to need the super duper though, with preamps boasting really high CMRR. Interesting reading some of the literature on one site where they talk about difficulties sourcing a relay for phantom power switching (they found one satisfactory but ultimately found it too expensive so for their mic pre you have to plug into another input and use a short cable to use a P48 mic). (My understanding) you want things to be tight (loop areas small), (physically) symmetrical, matched impedances (tight parts matching). Also imbalances matter less the bigger the difference is between output and input, so if the Dimension D balanced input is mediocre but the input signal is relatively large line level and a very low impedance, maybe another factor adding to a satisfactory result.

[Davebassman]

I'm totally confused! Haha

Something that I thought it was simple, it's very complex.

This unit have two trimmers at the balanced input to adjust CMR. Not necessary for unbalanced.

[dai h.]

re: the fat wire that connects to the power supply common: it appears to be the circuit ground to chassis connection.

re: the CMR trimmer, my understanding was it wasn't desirable in terms of upsetting physical symmetry (my general understanding: try to keep the two balanced lines the same physically so the cancellation is as complete as possible--something like twisted tight, things close together, maybe use modern surface mount parts). But I don't quite get how that works exactly--if it means one thing is improved but not another?

[Mick Bailey]

Davebassman - I replied to your PM - did you see this?

Looking through AES48 raises a question in relation to this particular build, in that the standard does not consider the mix of unbalanced 1/4" connections in conjunction with balanced connectors. One of reasons for terminating an XLR screen to chassis at the entry point and not connecting it to the internal circuitry is to prevent EMI from radiating within the enclosure. However, as soon as 1/4" unbalanced/isolated sockets are used, the screen connection is carried into the enclosure to the signal ground and defeats efforts to fix the 'pin 1 problem' with the XLR inputs.

If 1/4" non-isolated sockets are used and are connected to chassis ground, then this fixes that problem but introduces another in that return current paths have to flow between different points in the chassis which may give rise to hum. It also means that star grounding cannot be used.

My thought would be to accept the compromise of using isolated 1/4" sockets, but in this respect the grounding is not really much of an improvement on the original Roland design.

[Davebassman]

Davebassman - I replied to your PM - did you see this?

Looking through AES48 raises a question in relation to this particular build, in that the standard does not consider the mix of unbalanced 1/4" connections in conjunction with balanced connectors. One of reasons for terminating an XLR screen to chassis at the entry point and not connecting it to the internal circuitry is to prevent EMI from radiating within the enclosure. However, as soon as 1/4" unbalanced/isolated sockets are used, the screen connection is carried into the enclosure to the signal ground and defeats efforts to fix the 'pin 1 problem' with the XLR inputs.

If 1/4" non-isolated sockets are used and are connected to chassis ground, then this fixes that problem but introduces another in that return current paths have to flow between different points in the chassis which may give rise to hum. It also means that star grounding cannot be used.

My thought would be to accept the compromise of using isolated 1/4" sockets, but in this respect the grounding is not really much of an improvement on the original Roland design.

I didn't receive any notification, sorry for that!!

And thanks for your reply, at last someone who bring me some light.

I now that the problem is the coexistent of unbalanced and balanced connections. And it's very confusing, because if I use an unbalanced connections I will have ground loop in some instance, although I fixed the problem of pin 1. So, my thought is:

1.- I will always use XLR connections, only in a few times I will use unbalanced and I assume the risk of ground loop, noise, etc
2.- Also I would mount a Ground Break Looper.

I attach some schematics with a GBL. What do you think? What of those two It's better? I think that the correct would be the gnd scheme 6.

Many thanks!

P.D.: Isolated jacks, of course.

[Mick Bailey]

The second drawing better complies with AES48 and is preferable in that when the balanced connections are used there's little possibility of radiated EMI into the enclosure. You may also want to screen the signal leads internally to reduce hum pickup from the transformer. I'm not entirely sure that the PCB ground to the loop breaker is in the best place - my thought is to connect this to the PSU 0v at the -15v/0v/+15v point on the PCB. If space is an issue you could use a pair of 6A4 diodes back-to-back instead of the bridge for the loop breaker, as per Fender's '63 reverb reissue..

At this stage you can only assume that the ground scheme of the PCB track is optimised. The currents are quite small and a bonus is that someone has already designed and built this so you would expect it to be OK. My only other though is possibly to connect all of the unbalanced grounds together and connect a single lead to a point on the PCB and treat that as a local star connection for those sockets.

[Davebassman]

The second drawing better complies with AES48 and is preferable in that when the balanced connections are used there's little possibility of radiated EMI into the enclosure. You may also want to screen the signal leads internally to reduce hum pickup from the transformer. I'm not entirely sure that the PCB ground to the loop breaker is in the best place - my thought is to connect this to the PSU 0v at the -15v/0v/+15v point on the PCB. If space is an issue you could use a pair of 6A4 diodes back-to-back instead of the bridge for the loop breaker, as per Fender's '63 reverb reissue..

At this stage you can only assume that the ground scheme of the PCB track is optimised. The currents are quite small and a bonus is that someone has already designed and built this so you would expect it to be OK. My only other though is possibly to connect all of the unbalanced grounds together and connect a single lead to a point on the PCB and treat that as a local star connection for those sockets.

Yes, I will use shield cable for the internal audio connection.

Thanks for the advice's. About the 0V place, this 0V is the same in all the GND plane no? that it matters where it connect? I have enough space for loop breaker, but I like the back to back diode. It's mandatory the use of 6A4 diodes? and for the resistor, between 10 a 47r would be fine. And how many watts?

About the XLR ground. I've seen a lot of different wiring about this. The people from Analog Classics, say:

"The chassis should be connected to 0V at – and only at – the ground (pin1) of one of the input XLR’s. In case you choose a IEC power inlet connector with a power ground, you should connect it to this point also."

I understand why the chassis should connect to 0V. But necessary to pin1? why? And a lot of documentation about grounding, says that the earth ground need to be near the power socket, with a short wire. I think this is contradictory.

I have a bolt on chassis near the IEC socket, and I connect the earth ground and the 0V signal GND from pcb to this bolt. It's not the same? Or I missing something?

It seems that doesn't exist a rule or normalization standard for the XLR ground. Only it's my opinion.

A lot of confusion everywhere...

[Mick Bailey]

0v and signal ground are all connected through the traces on the PCB. The 0v connection at the PSU is the lowest impedance point on the board, though this is not necessarily the optimum position for hum reduction, depending on the PCB layout. With a new build I use a clip lead to establish the best location to chassis. If this was a guitar amp then I'd be first looking at connecting the PCB ground at or close to the input socket, but you have 4 inputs on this unit. If you're really hung up over this, build a growler and test the grounding once the unit is operational.

The diodes in the loop breaker need to conduct the full current to chassis ground in the event of a serious mains fault. The 6A4 is rated at 6A,400V. If the current is greater than this the diode failure mode is to short out. A bridge is even better, but takes up more space. Either way, consider the current involved in a short and the diodes have to be capable of handling the current. Signal diodes or low current rectifiers are not suitable. You may not need a loop breaker - try the unit with a direct wired connection first. If you get ground loop hum, experimentally insert a test breaker consisting of a pair of 1N1007 diodes (if you don't have the correct values), the resistor and cap. If that works, replace it with the correct components. The resistor should be a FP type, 1/4W or 1/2W.

As a repeat of post #9, the mains earth from the receptacle should be positioned as close to it as possible and on its own dedicated bolt. This is how every commercial item of equipment is done, though on older items you sometimes see a shared bolt with the PCB ground.

Whatever anyone here suggests, you'll always find a counter-argument, difference of opinion or contradiction on another site.

[Davebassman]

0v and signal ground are all connected through the traces on the PCB. The 0v connection at the PSU is the lowest impedance point on the board, though this is not necessarily the optimum position for hum reduction, depending on the PCB layout. With a new build I use a clip lead to establish the best location to chassis. If this was a guitar amp then I'd be first looking at connecting the PCB ground at or close to the input socket, but you have 4 inputs on this unit. If you're really hung up over this, build a growler and test the grounding once the unit is operational.

The diodes in the loop breaker need to conduct the full current to chassis ground in the event of a serious mains fault. The 6A4 is rated at 6A,400V. If the current is greater than this the diode failure mode is to short out. A bridge is even better, but takes up more space. Either way, consider the current involved in a short and the diodes have to be capable of handling the current. Signal diodes or low current rectifiers are not suitable. You may not need a loop breaker - try the unit with a direct wired connection first. If you get ground loop hum, experimentally insert a test breaker consisting of a pair of 1N1007 diodes (if you don't have the correct values), the resistor and cap. If that works, replace it with the correct components. The resistor should be a FP type, 1/4W or 1/2W.

As a repeat of post #9, the mains earth from the receptacle should be positioned as close to it as possible and on its own dedicated bolt. This is how every commercial item of equipment is done, though on older items you sometimes see a shared bolt with the PCB ground.

Whatever anyone here suggests, you'll always find a counter-argument, difference of opinion or contradiction on another site.

Here some progress:

- Isolated jacks.
- Pin 1 of every XLR to chassis.
- Signal GND to first input JACK, and from there to the other jack grounds.
- I will get 0v from filter caps and goto chassis/safety ground.
- Every in / out cable has shilded, only from the signal ground.

I'm going the right way?

[Mick Bailey]

I think so, though you never really know how a new build is going to work out until you plug it in and use it. It could be that you need to insert a loop breaker from the PCB to chassis, but best to leave doing this until the unit is operational and only do this if necessary.

Because the panel is anodized you need to check for ground continuity between each XLR pin 1 and chassis, and between the panel and the rest of the enclosure. Anodizing is a really good insulator.

[Davebassman]

I think so, though you never really know how a new build is going to work out until you plug it in and use it. It could be that you need to insert a loop breaker from the PCB to chassis, but best to leave doing this until the unit is operational and only do this if necessary.

Because the panel is anodized you need to check for ground continuity between each XLR pin 1 and chassis, and between the panel and the rest of the enclosure. Anodizing is a really good insulator.

Finish!

I needed to make some connections between the chassis panels. Now I have continuity on all chassis parts. Test with jack with 0 noise or hum. Great!

[Mick Bailey]

That's a really neat build, and no small effort to populate those boards. Does it live up to expectations as an effect?

[Davebassman]

That's a really neat build, and no small effort to populate those boards. Does it live up to expectations as an effect?

Yes! I like it very much. It's a subtle chorus effect and will open up the stereo field. I think it's the hardest I've ever built. I'll upload something for you to hear.