Getting rid of console hum via the PSU?

[user 37518]

Are you sure you dont have any unbalanced or sneaky peripherial device causing a ground loop?

 

[elektroivo]

I had similar problems on Soundcraft TS-24. Hum was building up with channels mounted in desk. Try to put only master section in desk, and measure hum, then disconnect ALL voltages coming from PSU except AUDIO power (digital PS, Phantom, etc. Then connect them one by one again, and check for hum (connect Phantom last, master section doesn't need that). if your problem is in PSU (or distribution) you should be able in this way to pinpoint which voltage(es) makes problems.... on one other console I had problem with Phantom.... there was a lot of hum in aux masters when phantom was conected (bad regulation in PSU)

regards, Ivica

 

[Michael Tibes]

...As far as the spectrum is concerned 50 Hz is the most prominent peak, but 15o Hz is almost as much, 250 Hz is the third highest peak. 100 Hz is a lot lower than even higher frequency harmonics.

If I'm not missing something, then your hum should be 100 Hz - shouldn't it? Unless you have an unusual problem: I once worked on an SSL and couldn't get rid of some hum and it turned out to be ONE broken diode in the rectifier bridge. So the cap was charged with 50Hz istead of 100Hz and the voltage fell too low between cycles causing 50Hz hum. It took me days to find that because it didn't come to my mind that the rectifier might be partially broken. Finally checking the voltages after the rectifer with a scope made it obvious.

I don't know how much current you need, but you might be able to find a new dual voltage lab psu on ebay for relatively little money. I run my console on one and it served me well for several years already.

If your psu is working correctly, the solution might be in Eddies article...

Michael

 

[sodderboy]

Definitely all tests must be made with nothing connected except a scope or a single monitor.  All it takes is one strand of a shield in an installation to create noise on a mix bus.  Also, do not have any electronic dimmers in use when listening to the noise floor.

I am not familiar with the mod, but anything from Eddie C. can be trusted as long as it is implemented exactly as he sez.  That being said, changing the grounding scheme from manufacturer's spec is a leap of faith, because sometimes a dubious scheme is there for a reason.  I remember trying to tame noise floors in mini-consoles like Lake or Seck and the cheezy audio common wiring schemes in them were tuned somehow.  Using beefy wire or changing to a scheme used in larger consoles would make the noise floor worse.  Going back to the 22 GA wires routing all over the PCB's would yield the lowest floor.
Mike

 

[living sounds]

Finally went back to the problem. I unsoldered the connections between the grounds on the PCBs - no difference in hum amount. The 150 Hz hum on the channels gets worse the further away from the master bus they are. The closest one adds only 1 db to the hum from the master bus alone, but the furthest adds a whopping 16 db. For the 50Hz hum it doesn't matter, a single channel added to the master bus alone adds 18db of 50 hz hum. This is with the faders at zero, the hum increases in a linear fashion with fader position. I'm not sure if the latter was the case when the grounds were connected.

Can anything be deduced from this?

Thanks!

 

[living sounds]

Anyone?

 

[radiance]

Are these cables soldered to signal ground all the same length?
Is the chassis ground on the input modules connected to the actual chassis via steel parts of pots & or switches?

The idea is to disconnect all grounds (signal, chassis etc. etc.) from each other and see what configuration has the lowest hum.

For example, insert the channel with the most hum, and start fiddling around with ground wires. This way you might get more info on what's actually going on.

 

[JohnRoberts]

Can anything be deduced from this?

Thanks!

Yes.. perhaps.

Finally went back to the problem. I unsoldered the connections between the grounds on the PCBs - no difference in hum amount.

The 150 Hz hum on the channels gets worse the further away from the master bus they are. The closest one adds only 1 db to the hum from the master bus alone, but the furthest adds a whopping 16 db. For the 50Hz hum it doesn't matter, a single channel added to the master bus alone adds 18db of 50 hz hum. This is with the faders at zero, the hum increases in a linear fashion with fader position. I'm not sure if the latter was the case when the grounds were connected.

OK, grounds are apparently not connected... what grounds?  I don't expect many products to work well without grounds connected.

The increase in hum with distance from the master suggests there is no forward referencing between the channels and master section. So logically, input modules that are further away from the master will have progressively more difference in local ground voltage. 

There are several techniques to forward reference the local channel grounds up to the master. The simplest for a typical virtual earth inverting sum amp, is to generate a bus for the grounds too. Using a low value R (say 10-20 ohms) from each input strip to this synthesized "average bus ground" we will collect an average of all the channel ground voltages, and a first order cancellation of the difference in the bus master output due to differential connection (ground noise to + inpiut, signal + ground noise to - input, so ground noise cancels and only signal is left).

The math is straight forward..  imagine a horrible case of a channel with 1v of ground difference between local channel ground and master section ground.  With 10 inputs assigned,  this ground noise is also added into the plus input but divided or averaged by the number of inputs (plus the R to ground in the master) so 1/11th of a volt. But the noise gain,or gain wrt the + input of the sum amp is N+1 or 11x so you get +1V, that cancels out the -1V from that 1V of ground difference being superimposed on channel signal feeding the bus.

As you can imagine.. this differential works best when the number of R's switched into the average bus ground is exactly equal to the number of channels assigned. Some console designs to save cost and complexity leave the ground resistors always connected to the ground average bus assigned counting on the relationship that you get best cancellation when all inputs are assigned which is also the worst case for noise, and when less inputs are assigned the imperfect cancellation will be less of a problem because of the lower noise gain.

I am not familiar with your console schematic, but i will speculate that it does not already have a differential or balanced sum bus since extra buses, resistors, switch poles, etc cost money, and some design engineers prefer to trust brute force (low Z common ground) over science. 

JR

 

[living sounds]

Are these cables soldered to signal ground all the same length?
Is the chassis ground on the input modules connected to the actual chassis via steel parts of pots & or switches?

The idea is to disconnect all grounds (signal, chassis etc. etc.) from each other and see what configuration has the lowest hum.

For example, insert the channel with the most hum, and start fiddling around with ground wires. This way you might get more info on what's actually going on.

The stock configuration has the channels and busses connected in a chain via a ribbon cable, using three strands for common ground and one strand for chassis ground.

http://www.tangible-technology.com/articles/200b.html

I then perfomed the mod linked above, cutting the four 4 ground wires from each module and connecting them via cables of equal lenght to a bus bar. Thinking about it some more, unsoldering the connection of common and chassis ground on the modules again just left the chassis ground unconnected...  (where is that face palm emoticon when you need it?)
But there's still the same amount of hum either way, it must originate somewhere else.

I'd love to test if the stock configuration actually worked better, but that would require making a ribbon cable.

Here are the schematics:

http://www.cem3374.com/docs/Manuals/Soundcraft/200B_Schema.pdf

But I guess I'll have to do more tests first anyway, take out most of the modules and try to isolate the problem... 

 

[living sounds]

Can anything be deduced from this?

Thanks!

Yes.. perhaps.

Finally went back to the problem. I unsoldered the connections between the grounds on the PCBs - no difference in hum amount.

The 150 Hz hum on the channels gets worse the further away from the master bus they are. The closest one adds only 1 db to the hum from the master bus alone, but the furthest adds a whopping 16 db. For the 50Hz hum it doesn't matter, a single channel added to the master bus alone adds 18db of 50 hz hum. This is with the faders at zero, the hum increases in a linear fashion with fader position. I'm not sure if the latter was the case when the grounds were connected.

OK, grounds are apparently not connected... what grounds?  I don't expect many products to work well without grounds connected.

The increase in hum with distance from the master suggests there is no forward referencing between the channels and master section. So logically, input modules that are further away from the master will have progressively more difference in local ground voltage. 

There are several techniques to forward reference the local channel grounds up to the master. The simplest for a typical virtual earth inverting sum amp, is to generate a bus for the grounds too. Using a low value R (say 10-20 ohms) from each input strip to this synthesized "average bus ground" we will collect an average of all the channel ground voltages, and a first order cancellation of the difference in the bus master output due to differential connection (ground noise to + inpiut, signal + ground noise to - input, so ground noise cancels and only signal is left).

The math is straight forward..  imagine a horrible case of a channel with 1v of ground difference between local channel ground and master section ground.  With 10 inputs assigned,  this ground noise is also added into the plus input but divided or averaged by the number of inputs (plus the R to ground in the master) so 1/11th of a volt. But the noise gain,or gain wrt the + input of the sum amp is N+1 or 11x so you get +1V, that cancels out the -1V from that 1V of ground difference being superimposed on channel signal feeding the bus.

As you can imagine.. this differential works best when the number of R's switched into the average bus ground is exactly equal to the number of channels assigned. Some console designs to save cost and complexity leave the ground resistors always connected to the ground average bus assigned counting on the relationship that you get best cancellation when all inputs are assigned which is also the worst case for noise, and when less inputs are assigned the imperfect cancellation will be less of a problem because of the lower noise gain.

I am not familiar with your console schematic, but i will speculate that it does not already have a differential or balanced sum bus since extra buses, resistors, switch poles, etc cost money, and some design engineers prefer to trust brute force (low Z common ground) over science. 

JR

I guess the increase in hum on further away channels can be blamed on me unsoldering common and chassis ground on the channels. But either way there is no significant change in total hum. The schematics can be found here:

http://www.cem3374.com/docs/Manuals/Soundcraft/200B_Schema.pdf

Thanks!

 

[living sounds]

So I went back to this. Fixed a broken trace in the PSU that connects a smaller ripple-reducing electrolytic to ground. Replaced the 100nf polyester caps in the PSU with NGO ceramic of the same value. Didn't make a difference.
Then I went back to the original grounding scheme (made a new flat cable unsoldered common and chassis ground on all modules). This is far worse than the star ground mod (by 10 db or so).

So the hum is lower the other way, but it's still quite a bit above the noise. From my calculations the PSU is taxed far below its maximum capacity.

I've looked for ripple with my (pretty limited) scope and the voltmeter and could not detect significant amounts.

I've also tried (just to test it) and disconnected the primary ground on the PSU, no difference in hum. It must originate in the console.

How low can you realisitically get with hum in a console? With the rest of my gear using balanced connections I have managed to get hum wel below the noise. I this too much to ask in a console?

 

[JohnRoberts]

The answer this time is pretty much the same as the last time...  If it was simple to fix it with an easy mod, it would have been easy to do it right in the first place.

If you have already done the popular mods, you may be as good as it gets. 

At the risk of being repetitious I will ASSume you are laking about noise on a sum bus with a bunch of inputs assigned. The general cause of this is very small differences between the local channel grounds and the master section bus ground. When a lot of inputs are assigned the effective gain of the sum bus amplifies this ground voltage difference,

I suspect some of your mods are a brute force tweak to make the local grounds more like the master ground but this can never be perfect even with a 1" solid copper bus bar.

In some cases to re-engineer a sum bus is like building a new console from scratch on top of or inside your old one,  generally not a good use of time and resources.

JR

 

[living sounds]

Thanks John! There is still hum above the noise on the master bus alone. It does not get worse with all channels added at unity gain. Ultimately it's not excessive, if I run the console fairly hot (there's plenty of headroom) and pull down the master fader 6-8db all the hum/noise is at less than -90db peak on my converter's input. Guess I should call it a day.

The only other thing still bothering me is that the aux sends put out quite a lot of hum, while the bus outputs somehow don't. I wouldn't mind reducing the aux hum since this does worsen the SNR, especially with things like reverb...

 

[user 37518]

Im perhaps a little bit late, but what voltages/current consumption do you need?, depending on that you may consider just dumping the PSU and use a Power One/Condor instead, they are dirty cheap and they are great! hell you can even get a used Acopian on e-bay for scraps. Just saying...

I once saw an Amek 9098 that Brian Roth was installing, Amek chose International/condor PSUs for the entire monster desk, that was the a good call if you ask me, not only do they work great but they are off the shelf so if one gets burnt you can just buy a new one from any electronics dealer. Bri has this theory that most console hot shot designers were "too good" to design something as trivial as a power supply, so they had the summer interns do it  ;D ;D so its better to leave the power supplies to the companies devoted to them like Acopian, Condor... i agree.

 

[Andy Peters]

Bri has this theory that most console hot shot designers were "too good" to design something as trivial as a power supply, so they had the summer interns do it.

I kinda doubt that, because the hot-shot console designers are well aware of the fact that the console essentially modulates the power supply ...

-a

 

[user 37518]

Bri has this theory that most console hot shot designers were "too good" to design something as trivial as a power supply, so they had the summer interns do it.

I kinda doubt that, because the hot-shot console designers are well aware of the fact that the console essentially modulates the power supply ...

-a

Take a pill, its a joke...

 

[JohnRoberts]

Bri has this theory that most console hot shot designers were "too good" to design something as trivial as a power supply, so they had the summer interns do it.

I kinda doubt that, because the hot-shot console designers are well aware of the fact that the console essentially modulates the power supply ...

-a

One observation from inside the tent, most hot shot console designers are better low voltage signal handling guys than significant power designers. I have seen examples of power supply over design and gross under design. Some consoles had such a well deserved reputation for poorly designed PS that spares were routinely purchased and carried to events. 

I would say console designers in general are better about managing PS noise than PS reliability, while even a few decades ago, proper use of differentials throughout was not well understood by all.

========
Re: master bus hum floor... Does this track with master fader, or remain constant? if constant it may be interface between console and external gear.

re: noisy aux send- My suspicion is that everything stays assigned to the aux bus so exhibits worst case noise gain in sum amp.  In principle you could generate a sum of all grounds to feed the sum amp ground pin to create a first order cancellation. Even if you could generate this average ground voltage bus, the cancellation would be imperfect since the source impedance feeding the bus is probably from a pot wiper so will change with different send settings.  If this was dialed in for best null at full off, it would probably be OK in typical use.

For today's off the wall idea, if you have a spare aux send, and they all have the same hum in their noise floor, you could subtract the hum from an unused reference aux, from then others.  Wide band noise would increase, but hum should correlate, and cancel.

JR

 

[living sounds]

Re: master bus hum floor... Does this track with master fader, or remain constant? if constant it may be interface between console and external gear.

A lot of it is still present with the fader down, I need to go over ca. -5db on the master fader to exceed the hum from the master module alone. Could lifting the ground on the console outputs make a difference? It would require some soldering on the chassis connections. There would be other ground connections to the patch bays that hold the console/converters/outboard I/O , but I only interface the master outs with balanced gear, so it's at least possible.

re: noisy aux send- My suspicion is that everything stays assigned to the aux bus so exhibits worst case noise gain in sum amp.  In principle you could generate a sum of all grounds to feed the sum amp ground pin to create a first order cancellation. Even if you could generate this average ground voltage bus, the cancellation would be imperfect since the source impedance feeding the bus is probably from a pot wiper so will change with different send settings.  If this was dialed in for best null at full off, it would probably be OK in typical use.

For today's off the wall idea, if you have a spare aux send, and they all have the same hum in their noise floor, you could subtract the hum from an unused reference aux, from then others.  Wide band noise would increase, but hum should correlate, and cancel.

JR

There's no spare aux send unfortunately. Again, could lifting the ground on the aux outputs make a difference here?

Thanks!

 

[JohnRoberts]

Lifting grounds and hoping it gets better is not a scientific approach.

It appears you have more than one problem, but the best way to fix them is one at a time.

JR

 

[living sounds]

All my gear is connected to these patchbays, and there is no significant hum with the rest of the balanced equipment. Also no hum with the console inputs and returns coming from the same patchbay. Could the problem originate somewhere between the chassis and audio grounds within the console?