DOD 680 delay sound issue

[Oneoff1]

Hi everyone,
I have 2 DOD 680 delays, both are SAD4096 based, both pedals are working and both sad chips are fine (Swapped between the two pedals).
One pedal has TL081 as the master clock and the other LF351
The problem - one pedal repeats sounds normal but on the other I get repeats that sounds 'metallic' and a little thin. It's like the repeats are passing something Like a high pass filter Combined with a ring modulator...
The clock on the pedal was checked and compared with the other one and both looks stable and clean on my scope. The timing of both clocks going into the BBD are perfect and the phasing between them looks exactly like they should on the Sad4096 tech notes. the voltages are all right, no noise, all pots are clean, bias is calibrated,
The 571 ic was swapped between the pedals and it doesn't change the issue.
I started checking the resistors values, the clock timing capacitors value,
But I'm starting to get out of ideas here...
I'm pretty sure it's something around the filtering  or the 571 cmpander ic but I'm not sure what to look for.  Should I swap these old electrolytic caps? Not sure it's gonna help...
can the difference in the clock opamp type make any difference?

Many thanks for your input!

http://www.dirk-hendrik.com/temp/dod680.gif

[Scruffie]

Reading that schematic hurts my eyes this late at night but it looks like the feedback is fed directly from the BBD output to the BBD input so it doesn't get any filtering at all.

Just to clarify, when you say repeats, do you mean this is affecting the first delay and all subsequent echoes or is it just on the echoes?

If it's just the echoes, are you sure there's no differences between the units? An extra filtering cap anywhere?

If it's the first delay too, probably worth audio probing and comparing between the units to see if the filters are off for some reason, could just be cap tolerance or part drift.

There's no clock cancellation capability but you say the chips work fine in one another so seems that wouldn't be too bigger issue.

[willienillie]

Should I swap these old electrolytic caps?

I would for sure.  Might not be your issue, but can't hurt.

[ElectricDruid]

Could one of the two outputs not be present? If you've swapped the chip, it can't be the chip, but it could be a corroded socket. Or even a broken or cracked track between pins 8 and 9.

Just a thought - the "ring mod" comment made me think of what the waveform from a single output looks like. It's chopped on-and-off at the clock frequency, exactly like a ring mod.

Personally I wouldn't replace the caps unless you've got reason to believe that's where the fault is. Doing it "just because" is a fishing expedition, not fault finding. And it's not true to say it can't hurt - there's always the risk of lifting a track or damaging something on these old boards.

[willienillie]

"Just because" they're LONG past their expected life span, and this thing plugs into the wall.  And probably not the best quality caps when they were new.  I wouldn't replace electros on a vintage battery-powered unit that sounded good.  I don't know this circuit and I'm an idiot anyway, but the "ring mod" aspect made it sound to me like possibly some noise wasn't being filtered out properly, which is usually the job of electros.  Clock noise, rectified AC noise, I don't know.  But yeah be careful.  New caps won't hurt, but bad technique sure can.

[Oneoff1]

Thanks for the replies guys!
I forgot to mention that it's most noticeable with long delays. When the delay knob is on its half and up it sounds (sort of) ok.
Scruffie:
you're right but if I'm not wrong there's a split in the outputs that goes down trough a bunch of capacitors and many more parts to the 571 (which by itself must act like a 'sound changing' device among its role as a compressor), then to the outputs.
The problem is just on the echo, and the boards are identical except that master clock opamp.

ElectricDruid :
that's actually making a lot of sense although the outs hooked together and It's hard to test.
But physical damage? Maybe

I thought about another option- since the delay pot is actually controlling the master clock, it's probably impossible to hear cracks and noise if the pot is dirty inside because it's not
On the audio line. Maybe this pot introduces noise to the timing section of this unit?

I also ran a sine wave to both units and it looks similar until I'm moving the delay pot on the bad unit,
Then the same sine wave gets a little additional bump on the side and it's changing and moving up and down when the knob is turned.
Not sure what to make of it...

[Scruffie]

The 571 compresses the signal on the way in and then restores the dynamics on the way out, they're not 100% perfect so they do alter the dynamics a little but it's not doing any filtering. If its caps were bad it could be unbalanced or struggling to perform its duties properly but your issue sounds more filter based.

The delay control is just an op amp 'relaxtion' oscillator, it produces a variable frequency square wave and then the 4013 divides this in to two anti phase clock signals for the BBD. You say you've probed the clocks and they're square so, doesn't sound like this is an area of concern.

I really think an audio probe is the way to go on this one.

[ElectricDruid]

The delay control is just an op amp 'relaxtion' oscillator, it produces a variable frequency square wave and then the 4013 divides this in to two anti phase clock signals for the BBD. You say you've probed the clocks and they're square so, doesn't sound like this is an area of concern.

+1 agree. If the clocks looks good on the scope, then they are good. If the frequency is stable and the shape is good, they're fine.

"Just because" they're LONG past their expected life span, and this thing plugs into the wall.  And probably not the best quality caps when they were new.

True. That would suggest replacement of the 470u and 10u around the regulator if you had problems with mains hum. But that doesn't seem to be the issue here. Also worth pointing out that heat is mostly what kills electrolytic caps, so it's a far worse problem in tube amps than anywhere else.

Oneoff, have you tried o'scoping the power rails? If there were a lot of AC leakage, willienillie could be right. I'd be surprised you didn't hear it all the time though, but it's worth a check just to eliminate it.

[Oneoff1]

Yes I did scoped power rails, on the DC side and on the AC after the transformer, also no hum or noise of any kind on this section.

I'm pretty sure that the problem is maybe on the filtering section as the good unit sounds more
Like it has a low pass filter to filter these high harsh
Sounds.
Does anyone knows where I should look for any type of filter?

Scruffie: yes I did probe before and after the 571 while feeding a sine wave to both pedals.
Other then a little amplitude difference and the 'bump' I described before , I don't see
Any other changes.
The 'bump' is an addition to the sine and it appears in different levels while turning
The delay pot.  I'm sure it's some kind of distortion but not sure where's the source. For sure it's after the BBD and not before.

[Mark Hammer]

It's not always the solution, but sometimes it is.  Check the bias on the BBD in the problematic one.  There is a region in the range of BBD bias settings where the BBD passes signal that is almost clean, but not quite.  Happily, the 680 allows one to listen to the delay-only signal, which will facilitate identifying - by sound - whether the bias setting is okay.  I'd suggest setting the delay time for shortest possible, and cranking the Repeats to max.  That will make any bias-setting errors stick out like a sore thumb.

Again, I'm not saying that IS the nature of the problem, but it might be, and in any event is worth at least crossing off the list so any subsequent roubleshooting can be more productive.

[Oneoff1]

Thanks mark, I did that while listening (I guess listening is the only way anyway because bias setting range in the reticon manual is quite large).
I'm sure I came to the best delay sound with minimum distortion. To my ears anyway.

[Mark Hammer]

So I guess we can cross that off the list.

[Scruffie]

See on the schematic the op amp after the 571 with 3 x 100k resistors and some caps and the two op amps at the bottom, also each with 3 x 100k resistors + caps? Those are the BBD input and output filters, try having a listen to the output of the input filter, then the BBD output and then each of those output op amp sections.

As you have two units you should be able to compare them to see if the working unit matches the metallic unit.

[Oneoff1]

Sounds like a good idea! Thanks for the info about the filters.
when you say 'listen', do you mean probe it with a scope? If not then what do you think will be the best way to listen to a certain point on the circuit?

[Scruffie]

https://www.diystompboxes.com/pedals/debug.html

Audio Probe.

[Oneoff1]

Hey guys!
Do you think it'll be ok to replace the 0.47uF electrolytic capacitors on leg 1 and 16
Of the 571 with Mylar non polarized caps?
I'm not familiar with what this IC does. These
Legs named "react" In the datasheet.
Thanks!

[ElectricDruid]

Do you think it'll be ok to replace the 0.47uF electrolytic capacitors on leg 1 and 16
Of the 571 with Mylar non polarized caps?

Yes, it'll be fine. 470n is a fairly typical value for those caps, and it won't care polarised or non-polarised.

I'm not familiar with what this IC does. These
Legs named "react" In the datasheet.

The chip is a "compander" e.g. compressor+expander, used for noise reduction. A little more about it here:

https://electricdruid.net/noise-reduction-with-companders/

The two caps act as smoothing for the envelope following/level detecting part of the chip that controls the VCA. If one of those was bad, you might get the output amplitude-modulated by a rectified version of the input, so it's a good thought.

[j_flanders]

I guess listening is the only way anyway because bias setting range in the reticon manual is quite large
I'm sure I came to the best delay sound with minimum distortion. To my ears anyway.

while feeding a sine wave to both pedals.

While feeding the sine wave (let's say 400Hz), probe the output of the bbd chip, connect to your DAW (or oscilloscope), and use a 'live frequency response (FFT)' setting/plugin.
In the ideal situation you'll see a single vertical line or spike at the 400Hz point.

In reality you'll see:

-spikes at 50/60Hz (and their multiples/harmonics) : hum

-spike at 10kHz : fixed clock whine,  the clock frequency, around 10kHz for longest delay time, around 100kHz for shortest delay time and anything in between depending on the delay knob setting. There are two out of phase 10kHz clock signals at the bbd output. Ideally they cancel each other out. In reality their level might differ a bit so cancellation isn't 100% and it shows up. Some circuits have a balance trimpot or two resistors to make the two out of phase clock signal levels equal so that cancellation is near 100%.

-spike at 400Hz : the input sine wave

-spikes at the harmonics of 400Hz : distortion Frequencies 800Hz, 1200Hz etc. These are the result of the distortion caused by a less than ideal bbd bias setting. Adjust the bbd bias until you see the lowest amount of harmonics (=the least amount of distortion) for an optimal bbd bias setting. When you have a live FFT/frequency plot view you can see the amount and output level of those harmonics being reduced while turning the bias trimpot towards its ideal setting.
Note:If the input signal is too strong they will not go away or even reduce much at any bias setting. If too low they might not all show up.
I set repeats to min and delay time towards max.

-spikes at the sum and difference of the clock and input frequency : input signal dependent clock whine A metallic sound, ring mod, aliasing, frequency modulation, heterodyning, sum and difference harmonics. (or whatever the correct term is). With a 400Hz input and a 10kHz clock you'll see spikes at 9,6KH, 9,2kHz, 8,8kHz etc. (difference harmonics) and also 10,4kHz, 10,8kHz, 11,2kHz etc. (sum harmonics).
These are musically unrelated frequencies and give that metallic sound. The higher order the harmonic the lower its output level.
With a 400Hz input signal and a 10kHz clock signal it will take more than 10 harmonics (10kHz - 4000Hz) to come down to a somewhat audible range. By then the (tenth) harmonics is also low in output.
But with a 4000Hz input signal already the first and second hamonic (10kHz-4kHz=6kHz and 10kHz-2*4kHz=2kHz) come down to the audible range.

This explains why you do not have the metallic sound at short delay times. At short delay times the clock frequency is around 100kHz instead of 10kHz and it will take an awful lot of harmonics to come down from 100kHz to the audible range than from 10kHz.

How to solve the problem:
-increase the lowest clock frequency: but this reduces the maximum delay time
-increase the lowest clock frequency and add another bbd chip: but this is expensive
-decrease the maximum allowed input frequency: but this makes the repeats dark sounding (this is what that multiple order low pass filter does)

Most of the time it's a compromise: somewhat limit the lowest clock freq and thus max delay time, somewhat limit maximum input frequency and thus brightness of repeats, somewhat accept clock whine at higher delay times (lower clock freq)


The way to methodically bias your bdd as suggested works well but for your specific aliasing distortion/clock whine issue I don't immediately have a solution. But at least now you know what the mechanics behind this problem are.

I don't know how bad the problem is but if the 500k delay time pots differ in actual value (one 450k and the other 550k) that would lead to different maximum delay times and different lowest clock frequency and thus more clock whine in one pedal than the other at long delay times.

PS:
Another methodical way to bias a bbd is to feed it a too loud input sine wave so that the bbd (slightly) distorts it.
Then probe the bbd output and while looking at the live wave form, adjust the bias so that top and bottom of the wave form are symmetrically clipped. In a less than ideal bias the top or bottom is clipped more than the other. But I personally prefer the previously suggested method.

[Oneoff1]

Thanks so much for this information! Very interesting and useful. This is definitely a 'cut and save' material for me

I have some questions :

Why the 'sum and difference' problem happens? Is it parts tolerances? Capacitors getting old? If so- changing some electrolytic caps or any other 'critical' parts might also help?

Can the filtering section of the circuit have anything to do with my problem?

'Decreasing the max allowed frequency'- Do you mean changing the master oscillator frequency? It is quite doable because it also has 100k resistor on the PCB connected across the delay pot, so changing this value up and down might do.

How can I increase the lowest clock frequency only? Practically, what is the difference between this and the option above?

BTW- I didn't measure the maximum master clock frequency of both of my pedals, but the master oscillator frequency I measured (when delay pot was at minimum) was 13.9 KHz VS.
16.3 kHz (in the metallic sounding pedal)
Doesn't look like a huge difference, bur maybe it is...

Thanks again for taking the time answering!

[j_flanders]

Why the 'sum and difference' problem happens? Is it parts tolerances? Capacitors getting old? If so- changing some electrolytic caps or any other 'critical' parts might also help?

https://en.wikipedia.org/wiki/Aliasing
https://en.wikipedia.org/wiki/Anti-aliasing_filter

Can the filtering section of the circuit have anything to do with my problem?

If it does not cut enough highs, it could. You'll have brighter repeats but also more aliasing artifacts. See the example above of 400hz vs 4000Hz input signal.

'Decreasing the max allowed frequency'- Do you mean changing the master oscillator frequency? It is quite doable because it also has 100k resistor on the PCB connected across the delay pot, so changing this value up and down might do.

How can I increase the lowest clock frequency only? Practically, what is the difference between this and the option above?

Those were more like general design considerations for bbd delays in general. Also pointing out there isn't a real solution without some consequences.
It's like saying: if you don't like the clock whine at longer delay times, then don't turn up the delay time knob too far...

BTW- I didn't measure the maximum master clock frequency of both of my pedals, but the master oscillator frequency I measured (when delay pot was at minimum) was 13.9 KHz VS.
16.3 kHz (in the metallic sounding pedal)
Doesn't look like a huge difference, bur maybe it is...

They're around 10kHz as in my examples, thus far ok.
By logic, the 13,9kHz should be more metallic sounding. The lower the clock frequency the faster the artifacts will enter the audible range.

Maybe you can post a short clip to hear how bad it really is.

If it's really, really bad, ElectricDruid posted a clever insight:

Just a thought - the "ring mod" comment made me think of what the waveform from a single output looks like. It's chopped on-and-off at the clock frequency, exactly like a ring mod.