DOD FX-90 troubleshooting

[CheapPedalCollector]

I'm still assuming the pedal worked correctly before, if it has issues even with a new chip it shouldn't be chip's functional blocks fault

If the pedal worked correctly when I bought it, I wouldn't have done a thing to it.

[Rob Strand]

No I meant C35 or 36 can't read it... The cap at 571's output pins.
If it leaks DC, it may affect output signal level.
I'm still assuming the pedal worked correctly before, if it has issues even with a new chip it shouldn't be chip's functional blocks fault

If C35 leaked it could screw up the circuit *after* C35 but I would expect the signal *on* pin 10 (output) to be largely unaffected as the opamp should drive a leaky load no problems.

When you get tricky problems like this it's a good idea to check *everything*, even the functional blocks.

One quirk about this circuit is pin 14 and pin 15 are DC coupled.  This is a very common practice on the NE570/571 but not entirely recommended by the datasheet, as they use separate coupling caps on each pin.   Normally it's OK since the whole chip runs of the same internal Vref.   However, maybe some replacement chips don't work so well with the pins connected.   Having said that, since the problem was there with the original chip and with the new chip it seems unlikely this is the cause.

Another angle I had a while back was the cause isn't the NE570 but the power supply voltage.  Since the NE570 has it's own Vref the bias voltage on pin 10 *is not* set by the supply voltage.    The bias voltage on pin 10 is set by the 10k (R38) on pin 12.   On many DOD pedals the power supply voltage is set with the simple regulator (Q3, Q10).   If the supply voltage is too low for the bias set on U2B pin 10 the U2B's output could saturate.   The reason I didn't go down this path before is because you would expect to hear some distortion on U2B's output, and we would also expect the waveforms to look bad on the oscilloscope.  U2A doesn't have a problem but it's connected slightly different.   Nonetheless look at the previous measurements:  the supply is Vcc=8.1V and pin 10 is biased at 6.3V.  It's not biased optimally.   You could lower the bias voltage by increasing R38 (10k on pin 12) to 12k or 15k.   

My gut feeling is there's a broken track or bad solder joint but there's nothing supporting that either.  DC voltages all good.  So that's only leaves the track on pin 14 and that has been checked!

There's so little circuit there yet the problem isn't clear.   That's why a few posts back I recommended lifting C35 and C30 and characterize that circuit isolation.   Check the DC voltage on pin 16 agrees with the AC input level.

You could build that part of the circuit on a breadboard and try different NE570 chips, see what happens.

[CheapPedalCollector]

I think you might be on to something there Rob, lemme mess with it some more tomorrow when I'm finished with these other things I have to do.

[bean]

One quirk about this circuit is pin 14 and pin 15 are DC coupled.  This is a very common practice on the NE570/571 but not entirely recommended by the datasheet, as they use separate coupling caps on each pin.   Normally it's OK since the whole chip runs of the same internal Vref.   However, maybe some replacement chips don't work so well with the pins connected.

I picked up on this tip after reading "Making Music with the NE570 Compander".

In all the delays I've messed with those pins were always DC coupled, but I started using separate caps for each pin as suggested.

[CheapPedalCollector]

Thank you MadBean, hope I can find hard copies of those books.

[allesz]

Is everything grounded properly. I remember old dod's being only chassis grounded, a loose nut my cause Power flutters.

[CheapPedalCollector]

Is everything grounded properly. I remember old dod's being only chassis grounded, a loose nut my cause Power flutters.

Yes, that's basic stuff I always check when repairing things.

[CheapPedalCollector]

Rob, can I buy you a coffee? You're a genius. Thank you so much, I would have never figured that out.

I hooked up my new signal generator and did some measurements, and everything looked good on pin 16.

I swapped the 10k resistor for 15k and now there's ~5.9v on the pins with 8.1 on the supply. and I have echo again. What's the optimal voltage so I can hook up a pot and find the value I need to put in there? Is it around 1/2 the supply on pin 13 ?

[CheapPedalCollector]

OK yes 1/2 supply is correct, I found 39K was perfect and set the voltages to 4.1 volts with 8.2 on the supply. Enjoying my FX90.

[Rob Strand]

Rob, can I buy you a coffee? You're a genius. Thank you so much, I would have never figured that out.

I hooked up my new signal generator and did some measurements, and everything looked good on pin 16.

I swapped the 10k resistor for 15k and now there's ~5.9v on the pins with 8.1 on the supply. and I have echo again. What's the optimal voltage so I can hook up a pot and find the value I need to put in there? Is it around 1/2 the supply on pin 13 ?

Very cool indeed.  It's amazing it turned out to be that - it's like the last throw of the grappling hook to escape the bad guys.   All those FX90 pedals biased so close to the wind.

OK yes 1/2 supply is correct, I found 39K was perfect and set the voltages to 4.1 volts with 8.2 on the supply. Enjoying my FX90.

It should be *very* close to optimal.  The default resistors inside the NE570/NE571 set-up the output bias point at 3V which is half-rail on a 6V supply (the minimum supply rail).  You might be able to split hairs scraping an 0.2V extra swing using the oscilloscope only to find out it doesn't match-up with standard resistor value.

FWIW, the compressor half of the NE570/NE571 on that pedal is biased at around mid supply.

[ElectricDruid]

Nice work both! 

[CheapPedalCollector]

Yeah I'm quite happy that was it. Hopefully someone else will find this thread useful in the future to fix theirs.

[Rob Strand]

Yeah I'm quite happy that was it. Hopefully someone else will find this thread useful in the future to fix theirs.

I have a nagging feeling in my head this has come up before.

[CheapPedalCollector]

I have a nagging feeling in my head this has come up before.

I did search before, but couldn't find anything about this particular issue, only the clock drift issue.

[Rob Strand]

I have a nagging feeling in my head this has come up before.

I did search before, but couldn't find anything about this particular issue, only the clock drift issue.

Maybe this one:
https://www.diystompboxes.com/smfforum/index.php?topic=73461

[CheapPedalCollector]

Yeah nothing of use in that thread, seems a design flaw with the unit.

[Rob Strand]

It occurred to me maybe the problem was on another pedal!

It might depend on the specifics of the power supply circuit, since they
don't all use the same divider network on the base of the PSU transistor.
The problem might only be on the 9V pedals.

Possible non-optimal bias on NE570/NE571
Expander stage in these DOD pedals,and probably more:

DOD FX-90 Delay
DOD R-820 Flanger   ? ? ?

Symptom:
- No or weak delay signal

Problem:
- Vout biased too high on NE570/NE571 expander stage

Solution:
- rebias Vout to mid supply

Fix:
- Increase 10k on NE570/NE571 INV input of expander stage to 39k

18V models, should be OK
DOD 585 Analog Delay
DOD 575 Flanger

Edited

[CheapPedalCollector]

I think this is reasonable to assume. It's very odd they worked at all originally like this, puzzling.

DOD 565 Chorus should be fine as well. The R5106 also works in these pedals where SAD512D is used.

I'm not entirely sure which other pedals in the 9V use the NE570/571, but I would think later models would have this corrected.

[Rob Strand]

I think this is reasonable to assume. It's very odd they worked at all originally like this, puzzling.

DOD 565 Chorus should be fine as well. The R5106 also works in these pedals where SAD512D is used.

I'm not entirely sure which other pedals in the 9V use the NE570/571, but I would think later models would have this corrected.

What I think happened is someone forgot that the NE570/571 bias point depends on the resistor values on the NE570/571 and not on the supply.  They took a 12V/18V design and translated it to the common 9V format.   The NE570/571 just shaved by on the prototypes and the error wasn't picked up.

I started to do some searches on-line but it's too haphazard.  That's when I remembered DOD had 18V pedals.   I've got a stash of DOD schematic and it wouldn't take long to go through those to find the 9V units with NE570/571's.   Now I think about it they probably only used that combination on the 9V Analog delays.  The 9V flangers and choruses I saw dropped the NE570/571.  I guess that greatly reduces the extent of problem - maybe only the FX-90 and FX-96.  The FX-96 schematic on-line looks wrong so I'd have to check out my stash.