Trident Series 65 grounding: theory and practice

[radardoug]

Hi,
    I note you have a noise level of at least -83 all inputs deselected. And it is mainly hum.
Start by disconnecting everything from your console, and then do the noise measurement.
Make sure you do a differential measurement, not connecting your measuring ground to the console.
If you still have hum then your power supply has hum in its output.
Then start connecting your input signals, monitoring the hum as you go.
If the hum level starts to rise then your inputs are dumping noise into your ground.
Have you isolated all pin 1's and run them separately to a.c. ground?
If not, do so.
Record an acoustic guitarist. Once you have got a good signal level, ask him to stop playing.
Measure the signal to noise .
Your -83 will be way below the standing room noise level.
Have a think about why you are obsessing about your console noise.
Go out and have a good evening with your friends.
Much better use of your time.

 

[PRR]

> 60Hz and its harmonics are still showing up.

That didn't get in there by magic.

It pretty much has to be coming from the wall-power. Inadequately filtered power supplies (have you considered two truck batteries?), or wall-power lines run too close to an inadequately closed chassis.

It could come in with the sources.

Or *in your measurement setup*.

Unless you are depending on local filtering to control wall-power cleanliness (a poor plan for a large system), your "signal ground bus" is almost certainly not the place to put your copper.

 

[leigh]

Unless you are depending on local filtering to control wall-power cleanliness (a poor plan for a large system), your "signal ground bus" is almost certainly not the place to put your copper.

Thanks for chiming in, PRR.

When you say "local" filtering, are you meaning the big power filter caps on each channel module? Or "local" as in "somewhere inside the console" vs only being done within the power supply?

 

[JohnRoberts]

I don't presume to speak for PRR but he is suggesting another angle of attack. Namely how is that hum getting in to corrupt the grounds?

One common vector is AC ripple on regulated and/or unregulated power supplies getting pumped*** into grounds through the very PS caps that we use to clean up the PS rails. (ironic isn't that).

Back in the '70s I speculated about powering a console with car batteries but got over that notion in time. 

Once simple test, does the hum level drop between the time you turn off the PS and the rails collapse? If yes, maybe you could clean up the external PS rails more before they get inside the console to corrupt your grounds.

Not exactly finesse, but more of a direct attack, than throwing copper bus bar at the ground. 

JR

***A secondary more subtle vector for PS ripple to get into the audio signal is through op amp PSRR. A decent op amp should have 100 dB of PSRR but a summing amp with 30 inputs has 30dB of noise gain, so that input referred noise is then only down 70 dB. Of course the rails "should" be pretty clean inside a friggin console.

 

[leigh]

Had a big breakthrough today, also a bit of a "Doh!" moment.

As I had mentioned, I've been doing noise tests on the summing amps with all the input channels un-assigned from the master bus. The goal was to isolate just the noise present within the summing amp stages. However, what I had failed to connect the dots about was that the 8 console groups were still "assigned" to the master bus, as they don't have any assignment switches, and are permanently connected to the master bus.

Attached are two graphs of the console's output noise (boosted 36 dB in the DAW). First is with the 8 group strips in place, second is with the 8 group strips pulled out of the console. In both of these, there are no input channels assigned to the master bus.

So, hooray that the 60Hz & harmonics peaks I was seeing don't permanently live in the summing amp section. The summing amps are very quiet by themselves.

However, as soon as I start assigning input channels to the master bus, or re-install some group strips, those 60Hz & friends come right back.

Circling around to what PRR and John are bringing up: the power supply measures very low ripple. While powering the console, and looking at AC voltage on the supply lines, the DMM eventually settles down to 0.1mVAC. That means, for all intents and purposes, no ripple, right?

Which leaves... radiated sources of 60Hz? Or it's being brought in on the inputs. I guess that'll be a next step, pulling every last input and listening then.

 

[JohnRoberts]

The difficulty in this simple analysis is that in use, the bus amps amplify the difference between the local master section 0V reference, and sundry channel 0V references by N+1 for the number of stems assigned. So 16 stems assigned means any difference between 0V gets boosted 17x.

To empirically determine if there is any room for improvement from additional filtering of PS rails consider adding a few ohms of series resistance in PS rails and extra large caps (grounded at PS 0V.). If you measure any change in noise floor than there may be benefit from pursing futher.

Note: This is attacking just PS noise not crosstalk and other issues with sloppy 0V references.

JR

 

[leigh]

The difficulty in this simple analysis is that in use, the bus amps amplify the difference between the local master section 0V reference, and sundry channel 0V references by N+1 for the number of stems assigned. So 16 stems assigned means any difference between 0V gets boosted 17x.

Right... I understand that that's how the noise floor will behave in general. So, I do expect to see overall noise rise as I assign more channels. But, if there's no 60Hz flying around inside the console, then why are 60Hz and its harmonics jumping way above the rest of the noise floor, as soon as I start assigning channels?

I'm starting to experiment with the copper bus bar, will see if that can improve anything.

One measurement I noted today: while the AC ripple on the rails stays consistently low everywhere, the DC level of the rails drops by about 0.2 V from one end of the console to the other. I'm not too concerned about that per se, unless it would be adding to the noise issues somehow. I can't see how it would, but please correct me if I'm wrong.

 

[JohnRoberts]

The difficulty in this simple analysis is that in use, the bus amps amplify the difference between the local master section 0V reference, and sundry channel 0V references by N+1 for the number of stems assigned. So 16 stems assigned means any difference between 0V gets boosted 17x.

Right... I understand that that's how the noise floor will behave in general. So, I do expect to see overall noise rise as I assign more channels. But, if there's no 60Hz flying around inside the console, then why are 60Hz and its harmonics jumping way above the rest of the noise floor, as soon as I start assigning channels?

Because apparently it is in there. The console chassis should be grounded to mains, and that should shunt external magnetic and electrostatic noise to ground. The buds wiring should be inside that chassis shield so clean from external noise sources.

For PS ripple I would expect 120Hz and higher harmonics, not pure 60 Hz. Strong 60 Hz suggests a magnetic field from say a transformer inducing noise. My point is to measure it, you need to see down another 20-30dB in the noise floor and not clearly relative to any single 0V reference..

I'm starting to experiment with the copper bus bar, will see if that can improve anything.

One measurement I noted today: while the AC ripple on the rails stays consistently low everywhere, the DC level of the rails drops by about 0.2 V from one end of the console to the other. I'm not too concerned about that per se, unless it would be adding to the noise issues somehow. I can't see how it would, but please correct me if I'm wrong.

IxR losses probably normal.

Good luck

JR

 

[leigh]

Have you isolated all pin 1's and run them separately to a.c. ground?

Interesting idea. To clarify - you mean, take all input pin 1 (even for line inputs, not just mic), and run them all to a central ground point? Rather than, say, to the ground connector on each input channel card?

 

[JohnRoberts]

Pin 1s go to chassis ground, close by not some distant central ground point.

of course opinions vary

JR

 

[leigh]

Start by disconnecting everything from your console, and then do the noise measurement.

Finally did this tonight, and boom, that's where all the 60Hz harmonics were coming in.

In my defense, I had previously disconnected the far end of the input snake. It connects to the DA converters via DB25s, so I had just pulled the DB25 end of things, since that was easier and quicker to do. I figured those input lines were "dead", having been pulled from any active sources.

But today, being out of any other ideas for how those 60Hz harmonics were getting into a console that is both well-shielded and fed with a ripple-free power supply, I finally pulled all the 1/4" input lines out the back of the console. And that harmonic series dived down 12 to 16dB, into the general noise floor. (60Hz itself only dropped by a few dB.)

Dang. Ooooooops.


My understanding of this is that the shields of the input lines are picking up power line hum, like passive antennae, as they travel the 20 or so feet from converter to console (around the back of two racks of gear and a computer.) And that I should either:

1. Lift pin 1 of all those lines at the console end, or

2. Rework the console's input jacks (as suggested above) so that pin 1 goes right to chassis ground and not to the input channel's audio ground.

Regarding option 1, I've read it's common grounding practice to make the console the central ground of a studio, and if you're going to start lifting pin 1's, you should leave the console end shields intact, and only lift on the outboard end of things. However, if I did that, I wouldn't keep this dirty ground noise out of the input channels.

Which leaves me with option 2 as right way to do it.

 

[PRR]

> Rework the console's input jacks... so that pin 1 goes ... not to the input channel's audio ground.

http://www.google.com/search?q=pin+1+problem

http://www.rane.com/pdf/ranenotes/Pin_1_Revisited.pdf

 

[ricardo]

Good memory! Ciletti did write about some ground mods for the Trident S65. Here is his article:

http://www.mixonline.com/mag/audio_analog_maintenance_upgrade/ (the article's photos unfortunately aren't showing there, but you can see them here

Not sure if Dan Kennedy's stuff is representative of Eddie's grounding philosophy.

In the Eddie grounding revamp article I remember, his 'bus bar' is about 2" long with 2 grounds from each module brought separately to it.  It's more 'star earth' than a bus bar.

So think carefully before you adopt huge brute force Cu bus bars.  If you are putting sewage (eg earth returns from decoupling caps) on a bus bar, it's impossible to make it sufficiently low R & low L

You ALWAYS need to consider what happens if your earths, grounds, chassis, bus bars have finite R & L .. cos they DO. 

It looks as if you have other earthing problems that have nothing to do with bus bars or their lack.

 

[gyraf]

Similar: http://groupdiy.com/index.php?topic=49877.0

 

[JohnRoberts]

Start by disconnecting everything from your console, and then do the noise measurement.

Finally did this tonight, and boom, that's where all the 60Hz harmonics were coming in.

In my defense, I had previously disconnected the far end of the input snake. It connects to the DA converters via DB25s, so I had just pulled the DB25 end of things, since that was easier and quicker to do. I figured those input lines were "dead", having been pulled from any active sources.

But today, being out of any other ideas for how those 60Hz harmonics were getting into a console that is both well-shielded and fed with a ripple-free power supply, I finally pulled all the 1/4" input lines out the back of the console. And that harmonic series dived down 12 to 16dB, into the general noise floor. (60Hz itself only dropped by a few dB.)

Interesting..

My understanding of this is that the shields of the input lines are picking up power line hum, like passive antennae, as they travel the 20 or so feet from converter to console (around the back of two racks of gear and a computer.) And that I should either:

Not sure antenna is the best analogy, but yes shields connected to the console "should" harmlessly dump any noise current into a low impedance chassis ground. This should not cause noise inside the console. Such coincidental noise is the very pin 1 problem that people write about. 

There are other possibilities. While many things are called ground loops, a large area loop formed by shield ground that intercepts a magnetic field will induce a current and voltage in that loop just like a single turn transformer winding.

Another possibility is that some external powered product does not have a grounded chassis, so connecting to the console provided the actual path to ground for that distant chassis too. This should be harmless and good practice if that external product is well designed but if it is dumping noise into the console output ground, and that path to ground is not low impedance the external currents can corrupt the console's internal grounding.

1. Lift pin 1 of all those lines at the console end, or

Lifting pin one is a cheat and not the correct connection. Pin 1 should be bonded to chassis ground right at the jack, to keep HF noise out of the console.

2. Rework the console's input jacks (as suggested above) so that pin 1 goes right to chassis ground and not to the input channel's audio ground.

This is the correct connection, but may not be enough alone to resolve the issue, if the rest of the input circuitry is not well balanced and independent of that ground.

Regarding option 1, I've read it's common grounding practice to make the console the central ground of a studio, and if you're going to start lifting pin 1's, you should leave the console end shields intact, and only lift on the outboard end of things. However, if I did that, I wouldn't keep this dirty ground noise out of the input channels.

I prefer to think of studio grounds like a sewer... to drain away undesirable noise. All of the audio, in transit between studio gear should be carried on audio + and audio - lines with a separate sewer line for the shield noise.

Which leaves me with option 2 as right way to do it.

Time to start experimenting with what you need to do... Maybe play with lead dress to see if where the wires run makes a difference in hum pick-up.

JR

 

[leigh]

http://www.google.com/search?q=pin+1+problem

http://www.rane.com/pdf/ranenotes/Pin_1_Revisited.pdf

Good stuff, I had read those again recently. Muncy coined this "Pin 1 Problem" in 1995, and the Trident Series 65 pre-dates that by over ten years. A little redesign is in the cards, I suppose.

So, is this your preferred "Right" connection, or do you favor connecting the shield via a small cap (so RF still goes to chassis but low freqs do not), ferrite bead, or some other more complicated setup?

 

[leigh]

In the Eddie grounding revamp article I remember, his 'bus bar' is about 2" long with 2 grounds from each module brought separately to it.  It's more 'star earth' than a bus bar.

Yeah, that was Eddie's own Soundcraft 200B grounding mod, which is what is discussed in that thread that Gyraf linked to.

So think carefully before you adopt huge brute force Cu bus bars.  If you are putting sewage (eg earth returns from decoupling caps) on a bus bar, it's impossible to make it sufficiently low R & low L

Oh yeah, no doubt. I've been doing more than thinking, I've been testing and measuring at each stage of the game.

The stock Trident S65 already has a "bus bar" type scheme - it's just that its bus bar is an extra-wide (1/4") PCB trace on the motherboard. If I go the "brute force" route, I would be adding a 1/2" x 1/8" copper bar running in parallel to that existing motherboard trace. The same signals/currents would exist on it as before, it would just be a much lower impedance bar. So, while it may not help, I don't think it's going to hurt. But I'll be testing it first to see if it helps, before going through the work of installing the bar and running 40 jumper wires to it.

 

[leigh]

Another possibility is that some external powered product does not have a grounded chassis, so connecting to the console provided the actual path to ground for that distant chassis too. This should be harmless and good practice if that external product is well designed but if it is dumping noise into the console output ground, and that path to ground is not low impedance the external currents can corrupt the console's internal grounding.

This was a great clue, and got me to pull the very last connections from the back of the console (save the power and the mix L/R outs): the 8 Aux outs.

And therein lay the final culprit: Auxes 1 and 2 had been normalled to a headphone amp. A headphone amp made by a company starting with the letter "B". A headphone amp that had been powered off, but was still plugged into the wall via a grounded outlet. Unplugging either the Aux jacks, or unplugging the headphone amp's power cord, caused the 60Hz noise peak to finally flatten out into the general noise floor.

Thank you for all the help. This has been such a learning experience.

 

[radardoug]

Eddies grounding mod is still not satisfactory. Soundcraft provided 1 ribbon wire for the chassis ground through the ribbon. Eddies joined that to the electronics ground and put a bigger wire on it. Much better to cut the chassis wire off at the connectors, and ground the connector to the aluminium chassis right at the connector. But make sure the chassis ground carries through all metalwork joins. Scrape off screws and use starwashers to ensure good bonding. The chassis ground should have its own wire right back to mains earth in the power supply, not connected to anything else.

 

[JohnRoberts]

Another possibility is that some external powered product does not have a grounded chassis, so connecting to the console provided the actual path to ground for that distant chassis too. This should be harmless and good practice if that external product is well designed but if it is dumping noise into the console output ground, and that path to ground is not low impedance the external currents can corrupt the console's internal grounding.

This was a great clue, and got me to pull the very last connections from the back of the console (save the power and the mix L/R outs): the 8 Aux outs.

And therein lay the final culprit: Auxes 1 and 2 had been normalled to a headphone amp. A headphone amp made by a company starting with the letter "B". A headphone amp that had been powered off, but was still plugged into the wall via a grounded outlet. Unplugging either the Aux jacks, or unplugging the headphone amp's power cord, caused the 60Hz noise peak to finally flatten out into the general noise floor.

Thank you for all the help. This has been such a learning experience.

As tempted as we are to blame the Behringer, in fact just connecting another ground to the console should not cause it to hum and buzz even with an actual ground loop, since the signal should not be on the ground. The external duplicate ground connection appears to be corrupting the console's internal noise floor. The very definition of pin 1 problem.

At this point, making the symptoms go away will probably be easier than properly curing the patient. Maybe use some isolation transformers to float any problematic external grounds.

JR