Help with my delay design. Can´t get only one repeat!

[Morocotopo]

Here´s the story. I´m designing a four PT2399 delay, very large, sort of a multi tap delay functionally. I have it pretty much finished, but there´s something I can´t correct. I want to get only one repeat from each delay chip with the repeats knob at minimum, but I can´t! I always get a second repeat, about 20-25dB lower, but still audible

I tried many ways to solve this, even removed the repeat pot and R, so effectively removing any feedback path, but I STILL get a second low volume repeat. So the problem seems to be feedback in the delay lines themselves. I honestly can´t see how that can happen, so I publish the schem to see if anyone can see how could this happen. I wasn´t going to publish it till it was ready, but I can´t solve this last snag.

First, a sound clip of the problem (this thing has four delay chips in series, so you´ll hear first the original note, then four repeats and then the unwanted, low volume extra four repeats):

https://soundcloud.com/truemostro/delay-bleed

The schem (it´s on two pages for more clarity)





Some thoughts: It´s on breadboard, quite a mess:



Maybe I´m getting crosstalk because of all the flying wiring and exposed parts leads? The rotary switch on the left of the picture, could that cause such bad crosstalk? I mean, there´s no gain in any place, except the in opamp that has pre-emphasis, so about 15 dB gain over 2K more or less. Maybe the PT2399´s opamps have really bad crosstalk specs?

Well, I´m stumped. If anyone is brave enough to look at the schem, and give me some hints, I´ll be grateful. Hope I explained myself clearly.

If anyone wants an explanation of the main circuit blocks, I´ll do it.

[Fender3D]

Do the extra repeats occur on dry out also?
It may easily be a crosstalk, since you're on breadboard, I think the rotary's not guilty 
You may use U2a to feed reference voltage, and see if it decouples better...

I know I'm a cheap ass when dealing with components number... why don't toss Q9 and make U2b non inverting with 1M input impedance? Pre-emphasis will go to inverting input.

Again, you may avoid Q1 and Q8 FETs and use 2 NPN shorting respectively U10's pins 1 and 16 (check PDS 1000), using the same switching logic

[Morocotopo]

Federico, thanks. Dry out is not hooked up, I will hook it and test.

God idea about making 1/2V with an opamp, I´ll try it. I also thought that there could be crosstalk through 1/2V or even ground, or 9V...

If I make U2b non inverting the signal path would be inverting overall, not that in a non true bypass pedal it matters... also, I was trying to avid an opamp as first in the signal path to avoid big value resistors to minimize noise, such as the ones I should use to get high input impedance . But I´ll investigate the suggestion.

About the switching, you mean shorting the 1 and 16 pins to ground? Hmmm.

More details: when I turn the level pot to minimum, the wet signal dissapears completely, no problem there, no "ghost" repeats. I tried tapping the signal at pin 14 of the fourth delay chip, there I got the delayed signal only, after it travelled through the four chips, and then the "ghost" repeats... can´t figure that one out. The fourth chip feeding itself signal from output back to input?

[Fender3D]

About the switching, you mean shorting the 1 and 16 pins to ground? Hmmm.

Yes, sorry I missed the "to GND" statement...

I was talking about this with another member on the other forum, during the blackout, also...

[Morocotopo]

Oh, Ok. i tried putting a 100K in parallell with the compressor rect cap in pin 1 and got a nice "swell" effect in the repeats. Might add a switch for it.

Couldn´t try your suggestions yet, I´ll have some time at the end of the week. In the mean time, anyone has any more suggestions?

[slacker]

Impressive work, I can't see anything on the schematic that would cause the problem you're having the only way for the delayed signal to get fed back seems to be through the repeats pot. Have you tried completely removing the repeats pot, just pull it off the board, that should stop any repeats.
I would try what the others have suggested and try disconnecting parts of the circuit to see if you can pinpoint where the problem is. For example you could disconnect all the output stuff leaving just the input section and the PT2399 circuits.

[anotherjim]

Very impressive.
The only way I can think of how a PT2399 could feedback to itself without a deliberate path, is coupling via the internal V/2 connection between amps - but you have good decoupling to supply with 47uF on pin 2 so that shouldn't happen. So, is the pin 2 decoupling working? I can see the caps in the photo are there so it's a long shot.

[Hatredman]

If it makes you feel better, I have a DD3 with the same behaviour - a ghost repeat om min setting.

[Morocotopo]

Impressive work, I can't see anything on the schematic that would cause the problem you're having the only way for the delayed signal to get fed back seems to be through the repeats pot. Have you tried completely removing the repeats pot, just pull it off the board, that should stop any repeats.
I would try what the others have suggested and try disconnecting parts of the circuit to see if you can pinpoint where the problem is. For example you could disconnect all the output stuff leaving just the input section and the PT2399 circuits.

Slacker, thanks. The thing is really nice in operation, noise is pretty much under control, I reallly want to get the best performance I can out of it. I removed the repeats pot and the following R, totally breaking the signal path, and it still had ghost repeats... Voodoo!!!

Anotherjim, as soon as I have some time i´ll try modding the V/2, using an opamp rather than a resistor divider, to see if it improves things, and /or tweaking it. A 100n ceramic cap in parallell with the electrolitic perhaps?

Hatredman, interesting that your DD3 does the same... so I´m not alone in my problem!

[anotherjim]

No, I mean the internal V/2 of the PT2399 - it's op-amps are using that (though I agree with earlier comment that you are hammering the plain resistor divider making it Vref for the everything else).

[samhay]

If the PT2399's internal Vref (pin 2) were oscillating, you would probably hear it as a pitch warbling of the repeats - that's how the 'Little Angel' works anyway...

[Morocotopo]

No, I mean the internal V/2 of the PT2399 - it's op-amps are using that (though I agree with earlier comment that you are hammering the plain resistor divider making it Vref for the everything else).

If the PT2399's internal Vref (pin 2) were oscillating, you would probably hear it as a pitch warbling of the repeats - that's how the 'Little Angel' works anyway...

Oh. I have the Merlinb notes on the chip, says that the internal V/2 is made with two 5.6K resistors. Could I use an opamp connected to pin 2, to get a stiffer V/2 for the internal opamps? Might be worth a try...

Samhay,I haven´t heard an obvious pitch warble, but I have an LFO to modulate the thing´s delay time. Might disconnect that and see if there´s warble without it.

Guys,you´ve given me a few things to try. Will do on the weekend.

[anotherjim]

Shouldn't need to mess with the internal V/2. A working smoothing cap on pin 2 should keep it stable. It's only supplying the chips own reference needs. Just check your 47uF on the pin2's is a good connection.
You could probably get away with having 2 resistive V/2 networks for everything else. One for the input side circuits, one for the output side.

[StephenGiles]

I love your breadboard, mine is leaning towards that look with my potential rackmount triggered filter ramblings!!

[Morocotopo]

Well, tried all the suggestions and a few more ideas and nothing. Still has ghost repeats... maybe an exorcist will do?



The only thing left to try is using an opamp for V/2, since I have a unused one I´ll try it, but really I don´t believe that will solve the ghosting. I´m betting the problem is on the breadboard / wires everywhere / grounding a mess / crosstalk / exposed parts leads / non shielding, or all of them (!!). So, I think I´ll call it a finished schem and start designing the PCB, hoping that will solve the issue. Too optimistic? Hmmm. Have made a few PT2399 delays before, so I know how to design a PCB for this type of circuit.

So, on to designing the PCB. This thing will not beat me!! Well, maybe...