NOISE! And having a tough time tracking it down.

Greetings, all, and happy holidays!

So I've got a noise issue that's been driving me nuts for the past few wees.  I've finished up another version of the PCB for my take on the classic FET comp design.  I've built a handful of these for myself over the past few years, but never have I encountered a noise problem like this one.  The spectrum is clearly 120Hz and harmonics, but it's LOUD, and almost sounds like a cool synth tone.  I've attached an image of the waveform.  Anyway, my grounding scheme for this version hasn't changed in any way, which is why it's so damn frustrating.  I have a ground plane that terminates on the chassis where the safety ground on the IEC mains outlet terminates, a series RC RF filter on the input XLR (as per Bill Whitlock's Jensen white paper), and the output XLR is left floating.  Also tried chassis-grounding that, just in case, but no change.

An MP3 of the noise can be found here (gained up quite a bit, FYI):

https://soundcloud.com/user-7226859/comp-test_1/s-hTIwO

Thanks, all!

No idea what you are talking about, other than it's a FET compressor.
A schematic would certainly help!
(The MP3 sounds like a power supply problem.)

Agreed. Schem please! Seeing the layout (trace/mask) might help, too.

120Hz is USUALLY noise from the power supply post-rectification, especially if a filter cap goes bad or is simply insufficient. (It's double the wall AC.) If it's loud, it could be that your filter caps are bad completely. You can also audio probe to see where the signal is first audible

PS schems attached.  Nothing out of the ordinary.

- 7824 jacked up to 30V
- zener-regulated -10V
- LM317 adjustable to 15V

All filter caps are brand new Panasonic FC (2200uF @ 50V), and PCB layout for 30V/-10V portion of the PS hasn't changed.  15V supply for relays, stereo link and active side chain filters is the only thing added to the layout - that was previously on a little daughter card elsewhere in the chassis.  I've completed removed that 15V supply from the PCB to see if the problem was somewhere in there, but removal makes no difference.

PCB layout of the PS section attached.  Anything that appears unterminated (filter caps, for example) in the image is terminated to the (hidden) ground plane.

Thanks all!

15V Supply image attached....

PCB layout attached.  Will label and re-upload if that helps!

check all grounds with ohm meter

CJ:

Never more than on Ohm or two between any ground points.  No voltage potential exists on any ground point, either.

midwayfair said:

Agreed. Schem please! Seeing the layout (trace/mask) might help, too.

120Hz is USUALLY noise from the power supply post-rectification, especially if a filter cap goes bad or is simply insufficient. (It's double the wall AC.) If it's loud, it could be that your filter caps are bad completely. You can also audio probe to see where the signal is first audible

Click to expand...

John:

Appreciate the response.  Schems attached in earlier comments.  I guess my primary question has been answered:  this is a power supply problem, as opposed to a grounding issue or a faulty component somewhere in the audio circuit.  With all other grounding/EMF/coupling-related noise issues I've ever had, poking fingers around changes the level of noise.  This is so constant and relentless it's driving me nuts.  The 'scope I use is in the office, locked up until next week.

Could faulty power transformer or rectifier diodes generate this noise, or is it indeed most likely a filter cap problem?  I've got enough parts around that I'll probably just rebuild the PS and see what happens.

Thanks again!

120 Hz noise is rectifier noise most of the time. Look at everything that connects post rect bridge and pre the last few smoothing caps. Also, check up on what your power filter is doing and how it's connected.

Are your transformer AC input voltages normal (compared to previous units) ?

The first image looks like a regulated supply rail that is loosing regulation (eg. the load is too heavy, or the input supply has too little storage capacitance or too little voltage margin above the regulated output voltage).

Can you check using an oscilloscope, or do you just have a multimeter?  Check the DC and AC levels before and after each regulator.  If the regulator is working then there should be no AC voltage on its output.  There should be sufficient voltage margin between regulator input and output.

trobbins said:

Are your transformer AC input voltages normal (compared to previous units) ?

The first image looks like a regulated supply rail that is loosing regulation (eg. the load is too heavy, or the input supply has too little storage capacitance or too little voltage margin above the regulated output voltage).

Can you check using an oscilloscope, or do you just have a multimeter?  Check the DC and AC levels before and after each regulator.  If the regulator is working then there should be no AC voltage on its output.  There should be sufficient voltage margin between regulator input and output.

Click to expand...

Thanks for the response.  Much obliged!

Pre- and post-rectifier voltages are all virtually identical to other units.  AC on transformer secondary is normal.  AC on regulator output measures next to nothing (.2mV) on multiple units.  Regulator is dropping about 7 Volts.  Can't 'scope until I can get into my office over break.

I haven't changed the design of the 30V/-10V.  The filter caps before the regulator are 2200uF, and post is a 6.8uF tantalum. 

What you said about load has piqued my interest.  Maybe I screwed up a collector resistor or two somewhere by a factor of 10.  I've done that before on other (tube) projects.  BUT, the compressor works just fine and sounds like the others, except for the maddening noise, and the regulator is faithfully spitting out just under 30V (29.7V to be exact) at all times...

Could there be something squirrelly going on with a rectifier diode?  How about overtightening the transformer mounting clamp (it's toroidal)?  Would a problematic transistor ever make this type of noise?

Maybe you need to wait for the CRO, and try and determine where the noise enters the signal chain.

trobbins said:

Maybe you need to wait for the CRO, and try and determine where the noise enters the signal chain.

Click to expand...

UPDATE:  Finally had a chance to scope this sucker and here's what I found:

- absolutely no visible noise in  30V or -10V rails, no visible noise in ground
- clearly visible noise waveform all over the signal path.

Will be isolating different areas of the circuit tomorrow and will update then.

trobbins said:

Are your transformer AC input voltages normal (compared to previous units) ?

The first image looks like a regulated supply rail that is loosing regulation (eg. the load is too heavy, or the input supply has too little storage capacitance or too little voltage margin above the regulated output voltage).

Can you check using an oscilloscope, or do you just have a multimeter?  Check the DC and AC levels before and after each regulator.  If the regulator is working then there should be no AC voltage on its output.  There should be sufficient voltage margin between regulator input and output.

Click to expand...

ANOTHER UPDATE:

Noise appears to come from the input amp.  I lifted the cap that couples the input stage to the output stage and there's no noise in the output amp.  Also doesn't seem to come from the

Any ideas?  Getting desperate here.

Noise source found:

It was the 15V power supply for relays, LEDs and op-amp voltage split.  Textbook LM317 adjustable regulator.  Full removal of the ENTIRE circuit eliminated the noise.

Here's the question, though: would a separate ground for that portion of the circuit on the PCB fix the problem?  Then tie that separate ground to the chassis at the star ground point.  What exactly is it about that little power supply sub-circuit that was dumping noise into the rest of the circuit?

This is what I've done on earlier versions, more or less: the 15V supply was  on a separate PCB, which was then tied to star ground.  Those comps are whisper quiet.

For the curious, the circuit is attached earlier in the thread.

sircletus said:

This is what I've done on earlier versions, more or less: the 15V supply was  on a separate PCB, which was then tied to star ground.  Those comps are whisper quiet.

Click to expand...

So you have a Hum-Free proven circuit. Do exactly the same and get the same hum-free results.