Hum only with signal present (BBD chorus)

[living sounds]

I've build a Haible Suble Chorus a few years ago. It's a clone of the classic Roland Dimension D unit. Full documentation here:

http://www.jhaible.com/subtle_chorus/jh_subtle_chorus.html


It's working well in principle, but there's one "problem" with it:
Despite being virtually hum free with no signal present, the unit produces quite a lot of hum and other noise components that rise (it seems) in a linear way with input amplitude. I've attached analyzer images with no signal as well as a 1k sinewave.

Since it is hum the PSU was the obvious thing to check. With my voltmeter going down to the milivolt range I couldn't detect any ripple, with or without a signal present on the audio inputs. I next tried a bigger 18V AC transformer instead of the 15V AC one. No difference.

What could be the source of this problem, and is there a reasonable way to fix it?

PSU schematics are attached as well. Thanks!

 

[living sounds]

Graph with signal present.

 

[living sounds]

PSU.

 

[living sounds]

I added an off-board RC filter, and this considerably lowered the 50 hz hum and it's harmonics. But I don't think the problem should exist in the first place, Jürgen Haible knew his stuff...

 

[JohnRoberts]

I've build a Haible Suble Chorus a few years ago. It's a clone of the classic Roland Dimension D unit. Full documentation here:

http://www.jhaible.com/subtle_chorus/jh_subtle_chorus.html


It's working well in principle, but there's one "problem" with it:
Despite being virtually hum free with no signal present, the unit produces quite a lot of hum and other noise components that rise (it seems) in a linear way with input amplitude. I've attached analyzer images with no signal as well as a 1k sinewave.

Since it is hum the PSU was the obvious thing to check. With my voltmeter going down to the milivolt range I couldn't detect any ripple, with or without a signal present on the audio inputs. I next tried a bigger 18V AC transformer instead of the 15V AC one. No difference.

What could be the source of this problem, and is there a reasonable way to fix it?

PSU schematics are attached as well. Thanks!

Well the first thing to realize is that the BBD uses a companding noise reduction wrapped around it. So the dry audio gets compressed before the BBD, then expanded back out after it. This improves the perceived signal to noise since quiet parts get boosted up relative to the BBD noise floor. and then both get expanded down after. However, the louder the signal the less S/N improvement you get. Depending upon where the 0dB inflection point is for the companding , with a loud enough signal you get the full noise floor of the BBD.

So this explains wide band noise that tracks with or is modulated by the signal level. The BBD should not have unusually large amounts of hum, so that could be another different problem. I would look at the BBD audio path quality specifically for hum in it's noise floor. 

I did a lot of work with BBD delays back in the '70s. I not only used companding NR, but used HF pre/de emphasis, and complicated anti-alias, anti-imaging filters around the delay path.
 
It should be easier to find the hum when you look in the right place...

Check the DC bias string feeding the BBD input for hum., but it could also be in any one of the pre/post filters inside the companded loop.  ;D

JR

 

[joaquins]

PSU.

What I can suggest now without much research is that the PS could be quieter with a cap in parallel with R30 and 31 (and another with R32 and 33), about 10µF should be good and a diode in parallel with R29 and 34 (in the way current shouldn't get through it in normal operation) to protect the regulator against this cap charge in shout down and short circuits. That's what most data sheets of those regulators suggest.

JS

 

[living sounds]

What I can suggest now without much research is that the PS could be quieter with a cap in parallel with R30 and 31 (and another with R32 and 33), about 10µF should be good and a diode in parallel with R29 and 34 (in the way current shouldn't get through it in normal operation) to protect the regulator against this cap charge in shout down and short circuits. That's what most data sheets of those regulators suggest.

JS

Already got it reasonably quit with a diode  bridge and a few RC filters between transformer and PSU, but thanks anyway.

And thanks to you, too, John, of course.

 

[joaquins]

What I can suggest now without much research is that the PS could be quieter with a cap in parallel with R30 and 31 (and another with R32 and 33), about 10µF should be good and a diode in parallel with R29 and 34 (in the way current shouldn't get through it in normal operation) to protect the regulator against this cap charge in shout down and short circuits. That's what most data sheets of those regulators suggest.

JS

Already got it reasonably quit with a diode  bridge and a few RC filters between transformer and PSU, but thanks anyway.

And thanks to you, too, John, of course.

You didn't have a bridge? HW rectifier? we should started at that! 

JS

 

[living sounds]

You didn't have a bridge? HW rectifier? we should started at that! 

JS

No, the PSU was as it is shown in the attached schematic. I put an ADDITIONAL diode bridge + RC filters before the PSU.  8)

 

[abbey road d enfer]

I put an ADDITIONAL diode bridge + RC filters before the PSU.

??? Additional bridge? So the signal is rectified twice!... utterly useless, except if the original bridge is VERY faulty. You should have started by replacing it.
RC filters before the PSU? Would indicate very faulty regulators...ummmmh
The LM317/337 have much more regulation than any battery of RC filters, except if the pre-regulator voltage is too low or has too much ripple.
You should check the voltages before and after regulation, with the original setup.
With these additional bridge and RC filters, there would be too much voltage loss.

 

[living sounds]

??? Additional bridge? So the signal is rectified twice!... utterly useless, except if the original bridge is VERY faulty. You should have started by replacing it.
RC filters before the PSU? Would indicate very faulty regulators...ummmmh
The LM317/337 have much more regulation than any battery of RC filters, except if the pre-regulator voltage is too low or has too much ripple.
You should check the voltages before and after regulation, with the original setup.
With these additional bridge and RC filters, there would be too much voltage loss.

I couldn't detect any ripple on the +/- lines with my multimeter (with or without audio signal present) before I added the filter and cannot detect any now. But the hum stays low now with even with a signal present. And the trafo puts out sufficient voltage to account for the drop by the filter.

One of our members here sells the exact same filter for use in the GSSL to lower hum, it works fine.

There may be another reason for the symptoms somewhere on these boards, but the filter cured the symptoms for now.

 

[abbey road d enfer]

I couldn't detect any ripple on the +/- lines with my multimeter (with or without audio signal present) before I added the filter

I don't doubt it. Your multimeter probably requires about 1V to get any sensible indication. You need an oscilloscope or an AC millivoltmeter.

and cannot detect any now. But the hum stays low now with even with a signal present. And the trafo puts out sufficient voltage to account for the drop by the filter.

Please be specific, what voltages are you measuring pre/post regulators.

One of our members

mentioning who would help.

here sells the exact same filter for use in the GSSL to lower hum, it works fine.

There may be another reason for the symptoms somewhere on these boards, but the filter cured the symptoms for now.

Has it really. You say the hum stays low, but I would be satisfied only if the hum completely disappeared.