666 hz whine in the Holy Grail

Started by j_flanders, June 02, 2021, 06:04:49 AM

Previous topic - Next topic

Rob Strand

QuoteThere are three kinds of tantalum capacitors you can run into
There's also modern-day "fusible" tantalums which attempt to address the shorting  tantalum issue.  IIRC these are targeted at automotive (and maybe medical) where you don't want hazards after a component failure.
Send:     . .- .-. - .... / - --- / --. --- .-. -

j_flanders

#41
Thanks for the valuable information concerning those tantalum caps.
Apart from the ethical aspects I didn't any of that.
It got me confused and somewhat weary to try them though, even for a very short test.

Here's what I found in the mean time:

There's a loud whine on all the +V lines.
From loudest to least loud:
- input of the 7805 (around 0.650 mV rms measured with my DMM)
- output of the 7805 +5 Digital line
- output of the 7805 +5V Analog line

I cannot hear it on any of the + 2.3V bias lines
Depending on whether the sleeve of the DC-jack is connected to sleeve of the output jack (through the chassis) I hear a faint whine on ground.

The whine on the input of the 7805 is a lot louder with the EHX unregulated adapter than with a regulated OneSpot.
Between those two I don't hear a difference in whine at the output of the 7805 though.
With the unregulated psu (around 12V unloaded) the 7805 runs around 90° C, while with the regulated Onespot around 70°C.

Since there's a 4,7Ohm voltage dropping resistor before the regulator I could try a cap there to filter out the ripple.
I tried a 1000µF cap and it drastically reduced the whine at the input of the 7805.
It did however make very little difference to the whine at the output. At least through headphones I could not tell if it was quieter.

I checked all the e-caps in this particular pedal. I thought I had done it before but apparently not, because I found that C12 (1µF) reads 100nF (being the C7 0.1µF cap parallel next to it).
All the other filter e-caps measure within spec.
I replaced the cap but it made no difference.

Trying to rule out transformer/ switching PSU related issues I tried an all battery setup:
Zoom G3 amp modeler powered by 4 AA cells + headphones.
Holy grail powered by 6 D cells.
It made zero difference. There's still a huge whine (666Hz and all its multiples going to 20kHz and beyond) on the 7805 input, output and finally on the output of the pedal.

Although I tried various filter caps in the past, I used them with leads that were clearly too long. The additional cap + long leads either didn't help or made things worse.
This time I held caps straight to the pcb.
What has helped is adding a 470µF cap at the output of the 7805 (from +out to gnd).
It doesn't need to be directly on the output pins. Anywhere there's access to a +5V Analog or Digital point, I can hold that cap, going to a ground spot nearby and it makes the whine at the output of the pedal inaudible.
But that's all it does. There's still the same whine on the 7805 input and on all the +5V A and D lines and on ground.

So far, to remove the whine on the output of the pedal I can:
- add a small cap in the NFB loop of the output opamp
- add a 1 to 2 k resistor at the output of the output opamp
- add a big e-cap from +5V A or +5V D to ground anywhere on the pcb.

I haven't isolated the sleeve of the DC-jack from the chassis yet.
Only when I crank a JCM800 type amp I can hear a very faint high pitch whine among the deafening white noise. And only when I hold my head/ears in a particular position towards the cab so the tiny whine can beam straight into my ear.
However, that whine is only there in Spring mode and on Flerb. It's gone when set to Hall mode.
Hall mode happens to be the only mode where none of the PIO (0,1,2) pins are connected to ground.
I would have to further investigate if isolating the DC-jack sleeve from chassis would help or if the wire from the mode-switch to ground is better off to another ground spot on the pcb.

Can the 666Hz be caused by ripple current? 'Pulsed drawn' by the dsp chip at 666Hz or other related frequency? Just guessing here as I cannot think of anything else, now that I've tried the all battery setup.
Would it explain why a big e-cap on the input of the 7805 doesn't not help because the source of the whine is after the 7805? But then why is it louder at the input of the 7805 than at its output?

I also wonder if the huge amounts of white noise in this pedal is related to the whine. I'll start another thread to cover cures for the white noise.



anotherjim

I suppose the CPU in that is surface mount? Those flat packages can become moisture traps, but also hide flux residue. There could be spots of low resistance causing higher current pulses.

Was there a version with program keys and a display? That might explain the unused pins that are not grounded and they could be handing out digital hash to nowhere.

Anyway, I'd give it a good soaking with a solvent cleaner in the hope of washing out any crap.

Rob Strand

Something that popped into my head a while back.  Is there is any correlation between the power supply revision (size of caps and resistors) and the units that oscillate (if that's the problem) ?
Send:     . .- .-. - .... / - --- / --. --- .-. -

j_flanders

Quote from: anotherjim on September 14, 2021, 11:02:43 AM
I suppose the CPU in that is surface mount? Those flat packages can become moisture traps, but also hide flux residue. There could be spots of low resistance causing higher current pulses.

Was there a version with program keys and a display? That might explain the unused pins that are not grounded and they could be handing out digital hash to nowhere.

Anyway, I'd give it a good soaking with a solvent cleaner in the hope of washing out any crap.
It's surface mount.
There's no version with program keys and a display that I'm aware of. The more elaborate versions (Holier Grail, Holiest Grail) don't have any of that.
They're following the datasheet/ app notes as far as NC and GND pins. As well as all the other implementations aspects. The only thing I cannot judge is their implementation of power supply distribution and separation of digital and analog ground. Some important information wrt that is underneath the big chip.
By solvent cleaner you mean IPA?

j_flanders

#45
Quote from: Rob Strand on September 14, 2021, 06:01:46 PM
Something that popped into my head a while back.  Is there is any correlation between the power supply revision (size of caps and resistors) and the units that oscillate (if that's the problem) ?
I think I have to come to the conclusion that all of them have whine. If an affordable one comes along I'll buy a 4th and/or 5th pedal just to confirm.

The ones I have where it was actually a problem, those had bad e-caps which were either dried out or had leaked.
I can only guess what caused one or more e-caps to fail in every pedal. Age? Heat from that 90° 7805 regulator inside an unvented box? The oscillation/ripple itself?
Anyway, after replacing those dead caps the whine was still there but no longer objectionable, kinda hard to hear over the massive amounts of white noise this pedal has.

After I added the 470µF cap It became inaudible.

During my probing I was listening through headphones and no audio signal input.

I was just curious to find out what caused it in the first place as people have been whining about this whine for the past 20 years and not a single solution or explanation has been found.

As said before it also depends on what comes after the pedal. It could explain why some people don't have/hear the whine.
I have been using one for years without noticing the issue because I either had a Boss pedal after it or because I used a mixer with the Holy grail set for 100% wet and mixing in the dry signal. And since the wet signal has no whine I never heard it.




Rob Strand

QuoteI think I have to come to the conclusion that all of them have whine. If an affordable one comes along I'll buy a 4th and/or 5th pedal just to confirm.

So it's only a matter of degree.

Quote
The ones I have where it was actually a problem, those had bad e-caps which were either dried out or had leaked.
I can only guess what caused one or more e-caps to fail in every pedal. Age? Heat from that 90° 7805 regulator inside an unvented box? The oscillation/ripple itself?
Anyway, after replacing those dead caps the whine was still there but no longer objectionable, kinda hard to hear over the massive amounts of white noise this pedal has.

After I added the 470µF cap It became inaudible.

The interesting thing is there is a degree of whine which depends on the size of the cap.   I'm assuming the cap at the input of the regulator.    Worst case is stuffed caps which don't work at all.

Oscillation tends to be present then when you put on the means to stop it it stops entirely.      The "dialing down" type problems usually mean current pulses along tracks and the bigger caps buffer the current pulse essential preventing the pulse going down certain track.    The tracks of concern are usually ground traces.

So maybe the processor is producing the current pulses and the layout has a problem.

QuoteAs said before it also depends on what comes after the pedal. It could explain why some people don't have/hear the whine.
I have been using one for years without noticing the issue because I either had a Boss pedal after it or because I used a mixer with the Holy grail set for 100% wet and mixing in the dry signal. And since the wet signal has no whine I never heard it.

I don't get how following a effects pedal or changing the output resistors can help a grounding issue.  If the Boss pedal and the HG shared the same power supply then, yes, you can get cancellations or reductions on noisy grounds.    With separate power supplies on each very doubtful it would fix a ground issue.

So at the end of the day I still don't understand what's going on!

QuoteI was just curious to find out what caused it in the first place as people have been whining about this whine for the past 20 years and not a single solution or explanation has been found.
No doubt it's a tricky problem.  One obscure detail setting it all off.

The reason for the fizz on the Boss SD-1 was never really identified.  That went around for years.   The solutions avoid the issue instead of identifying the cause but they do it an good way.

I think you made a good effort - best so far by anyone.
Send:     . .- .-. - .... / - --- / --. --- .-. -

j_flanders

#47
For the moment I've given up. Maybe some day I'll remove all the wiring and try to find out where or how the ground connections have an impact.

On all of the pedals I have managed to bring down the whine to an inaudible level with either (or combined):
- 470 µF cap across the output of the 7805, or anywhere on the pcb from +5V analog or digital to ground.
- 2k2 resistor at the output, soldered on the 3PDT switch, between pcb output wire and the 3PDT switch lug
- smallish cap (100-200pF) in the NFB loop of the output op amp
- isolating the DC-jack sleeve/ground from the chassis. With its shoulder, this jack is already set up for this, just increase the hole and use an insulating washer on the outside.

In the mean time I wrote EHX about the problem. This was their reply:
QuoteWe don't have a solution yet on the Holy Grail but we did find a Holy Grail making the noise plus a Holier Grail and Holy Grail Plus that do not make the noise. We will study all three when we can and hopefully find a solution. This is not a high priority for us at the moment so we're not sure when we'll actually get around to doing this study.

Next up is finding a good solution for the rather large amounts of white noise/hiss these pedals have.
I have a crude solution (boosting the input, attenuating the output) that works really well to increase the signal to noise ratio, but I have to incorporate it into the circuit.
I'll start another thread for that.  :)

Big thanks to all who have tried to help and give input, feedback and suggestions.

Rob Strand

QuoteOn all of the pedals I have managed to bring down the whine to an inaudible level with either (or combined):
That's pretty good going.  A repeatable recipe as well.

QuoteNext up is finding a good solution for the rather large amounts of white noise/hiss these pedals have.
Another can of worms.

Start at the input stage.  All bad: inverting stage, large resistor values, noisy opamp (most low power opamps are).  However, that goes against the fact you can remove the noise by boosting the input gain.  So what do I know!

Who knows, perhaps the way the whine is getting in is also letting in other more random junk from the power supply.

I often disconnect input to stages and connect them to Vref  or ground (assuming DC is OK).  Sometime you need to add an input cap.  If the noise goes away it's a sign something before that point is causing the noise.   Can be messy to do on a build board.  Not to mention an SMD build.
Send:     . .- .-. - .... / - --- / --. --- .-. -

j_flanders

#49
Sorry to bring this topic back up.
Along with my quest for less white noise I have also been hunting down that 666hz whine again for the past few months.
Although I had previously found solutions to prevent it from finding a way to my amp, internally it's still there.

Quote from: Rob Strand on September 30, 2021, 05:49:33 AM
QuoteNext up is finding a good solution for the rather large amounts of white noise/hiss these pedals have.
Another can of worms.

Start at the input stage.  All bad: inverting stage, large resistor values, noisy opamp (most low power opamps are).  However, that goes against the fact you can remove the noise by boosting the input gain. 
So what do I know!
You know a lot. :) Just not the reasoning behind this idea.
To avoid the white noise accompanying the dry signal I was using a Boss LS-2 as a splitter + mixer back then.
Although not exactly free from white noise itself the Boss LS-2 is much quieter than the dry signal path in the Holy grail.
So what I did was: mix dry signal through Boss LS-2 with 100% wet signal from Holy Grail.
The problem with the 100% wet signal is that it has a ton of white noise from the DAC outputs.
The 'good' thing about it is that the white noise is constant in amplitude regardless of the input level.
So by boosting the input and attenuating the output by the same amount, I got the same level of reverberated guitar signal out but with attenuated hiss from the DAC outputs. The key being that the DAC noise does not increase by boosting the input signal.
A poor man's compander...

Back to the whine...
Although reluctant to revive a multipage, years old thread I still think it's best to have all info in one thread.
(Forum software should simply alert users they're about to enter an old thread. I get a warning for replying to an old thread. Users should get that warning when opening a revived multipage year old thread. ;))

Maybe years on, someone comes across this thread and finds it useful to have it all in one place.

As for the high pitched whining noise. Here's an update.
I've bought even more of these Holy Grails. :)
I've also bought other versions of this pedal that use the same DSP chip:
A couple of Holier Grails (using CS4811) a Holy Grail Plus (CS4812) and some V1 Holy Grails (CS4811)

Internally they all have the high pitched whining noise!
It's loud on the 5V lines (analog and digital) and less loud on ground and some ground parts have none.

During one of my tests where I was adding additional caps to ground I accidentally put a charged cap with the wrong voltage on CMFILT. I assumed Everything was 5V or 2.3V (bias) but CMFILT has 3.64V.
Doing this 'stopped' the DSP chip and the whine was gone. After a reset/repower it was back there.

So my conclusion is that it's the DSP chip (or some clock or switching inside it) or the way it draws power that is causing the whining noise.
The way ground paths are set up seems to determine whether the whine makes it into the audio path.

After all these years I think I'm willing to let this issue rest.
I haven't found a way to prevent the whining. I don't think there is one.
The solutions previously posted to prevent it from being heard through an amp still work:
Quote from: myself
- 470 µF cap across the output of the 7805, or anywhere on the pcb from +5V analog or digital to ground.
- 2k2 resistor at the output, soldered on the 3PDT switch, between pcb output wire and the 3PDT switch lug
- smallish cap (100-200pF) in the NFB loop of the output op amp
- isolating the DC-jack sleeve/ground from the chassis. With its shoulder, this jack is already set up for this, just increase the hole and use an insulating washer on the outside.
The two most important ones being:
-replacing dried out electrolytic caps (C8, C12, C15, C20)
-isolating the DC-jack sleeve from the chassis.

I have also traced the PCB, front and back.
*This is still a work in progress. I'll update the pictures the coming days with component numbering and some small changes to the digital parts around the pull up resistors and the off board wiring, and some 'heat maps' where most of the whine is located.
But for now:
Hi-res originals:
https://i.postimg.cc/kJPSY7HF/hg-traced1.jpg
https://i.postimg.cc/pWv9Yq9s/hg-traced2.jpg







Rob Strand

QuoteYou know a lot. :) Just not the reasoning behind this idea.
To avoid the white noise accompanying the dry signal I was using a Boss LS-2 as a splitter + mixer back then.
Although not exactly free from white noise itself the Boss LS-2 is much quieter than the dry signal path in the Holy grail.
So what I did was: mix dry signal through Boss LS-2 with 100% wet signal from Holy Grail.
The problem with the 100% wet signal is that it has a ton of white noise from the DAC outputs.
The 'good' thing about it is that the white noise is constant in amplitude regardless of the input level.
So by boosting the input and attenuating the output by the same amount, I got the same level of reverberated guitar signal out but with attenuated hiss from the DAC outputs. The key being that the DAC noise does not increase by boosting the input signal.
A poor man's compander...
Sometimes you just have to weed through all the possible sources until you get a hit.

You debugging trick with the external mixer was a really cool way to isolate the digital audio stuff.

Like you suspect already, the whine and some noise could well be layout.  So that's getting outside of what you can fix (without doing silly things like finding the noise source cancelling it at the output mixer!).

You can get extra hiss from poor anti-aliasing filters at the input of the ADC.  The current filter C6 and R5 are pretty weak.   You need at least a second order filter to keep the noise down.  The cut-off for the second filter R6 and C9 are very high.   I can't remember if you already tried playing with the filters early on in the thread.

You could only check noise from DAC's, with the ADC noise as well, if you can block the input inside the DSP chip and only output zeros on the DAC.  That's so easy to do when you don't have control of the chip/code.
Send:     . .- .-. - .... / - --- / --. --- .-. -