Help needed PT2399 Tap Tempo Delay

[Mr. Lime]

Here's one of my first echo/delay designs I made a PCB for.
It's a PT2399 Delay feeding a BTDR-3 Belton Brick. Similar to the EQD Ghost Echo but with some additions like the StompLFO to modulate delay time, a dry/wet mix control and an active BMP tone control.

Voltages seem fine, the clean sound can be blended in and the StompLFO seems to modulate something.
My problem is, that there is no guitar signal in the delay path. It's like the PT2399 get's no input signal.
I turned the feedback trim to 100k, I can't hear output on pin 12 or pin 15.
The PT2399 has already been replaced once..
Am I missing something essential around R3, C2, the feedback path, R5 and R6?




Thank you!

[vigilante397]

First things first, can we see your PCB layout? Secondly can we see your populated board? Lastly can we see some voltages?

[Mr. Lime]

Sure, here are some pictures and the measured voltages:

PT2399:
1: 4.9V
2: 2.45
3: 0
4: 0
5: 2.9V
6: 2.5V
7: ~3.5V
8: ~0.9V
9: 2.5V
10: 2.5V
11: ~3V
12: ~1.2V
13: 2.5V
14: ~3.6V
15: 2.5V
16: 2.5V

IC5:
1: 3.4V
2: 3.4V
3: 3.2V
4: 0V
5: 3.4V
6: 3.4V
7: 6.3V
8: 7.3V

IC3:
1: 3.4V
2: 3.4V
3: 3.4V
4: 0V
5: 3.4V
6: 3.4V
7: 3.4V
8: 7.3V

StompLFO:
1: 5V
2: 5V
3: 2.5V
4: 5V
5: 5V
6: 0V
7: 5V
8: 0V

R25 (47R bridged because of high voltage drop)

[anotherjim]

IC5b has a DC offset. In the Schematic, R20 is looking at 0v instead of Vb on the gain trimmer.

[Mr. Lime]

IC5b has a DC offset. In the Schematic, R20 is looking at 0v instead of Vb on the gain trimmer.

Ah I see, so I have to rewire the gaintrim to Vb instead of GND when I'm home.
I'm afraid this isn't responsible for the missing delayed input signal..?

[anotherjim]

On the PT2399, all pins from 10 to 16 should be about 2.5v. It appears you have caps in all the right places there so there will be an error or fault responsible for those wrong voltages.

[ElectricDruid]

The StompLFO should be ok with those voltages.

Pin 4's an input and has an internal pull-up, so I'd expect that to read high, but I doubt that's a deal-breaker. It may be that I'm wrong, I'm not sure I've ever tested it like that.

[Mr. Lime]

I soldered the gain trim to vb instead of GND but as expected this didn't change the situation.
There is still alternating voltage on the PT2399 pins 11, 12, 14 but 9 and 10 are okay what's weird.
Otherwise I would have suspected the 5V regulator as it's the only common conntection between them according to the block diagram..



Removed the StompLFO but that didn't help either so I don't think the LFO is the problem..

I assume the circuit on the schematic is fine and shouldn't cause failure on it's own?

[ElectricDruid]

Checking the circuit, is there a cap missing on that input filter? From R5/R6 down to ground?

[anotherjim]

pins 11,12 are the last part of the digital to analogue conversion. That Pin11 has an offset to 3v might suggest the chip is locked up. Take the control transistor out and fit a 10k resistor between E and C and see if the delay has signal then. With 10k, the delay should be really obvious.

The input side of the PT2399 isn't how I'd like it - you don't have virtual ground mixing between the dry input and the feedback, so they can fight each other. The resistors on pins 15,16 look like leftovers from a multipole filter. If you short R5 across with a wire, you should get an improved input. The Feedback trimmer must never go to 0ohm a 10k in series with it would be a good idea.

I would also take the PT2399 out of its socket and check the socket pins for any stray voltage. With the chip out,  only pin1 should have any volts.

[Mr. Lime]

Checking the circuit, is there a cap missing on that input filter? From R5/R6 down to groun

I've seen some designs without that cap and it forms just a lowpass which has not to be used, isn't it?

pins 11,12 are the last part of the digital to analogue conversion. That Pin11 has an offset to 3v might suggest the chip is locked up. Take the control transistor out and fit a 10k resistor between E and C and see if the delay has signal then. With 10k, the delay should be really obvious.

The input side of the PT2399 isn't how I'd like it - you don't have virtual ground mixing between the dry input and the feedback, so they can fight each other. The resistors on pins 15,16 look like leftovers from a multipole filter. If you short R5 across with a wire, you should get an improved input. The Feedback trimmer must never go to 0ohm a 10k in series with it would be a good idea.

I would also take the PT2399 out of its socket and check the socket pins for any stray voltage. With the chip out,  only pin1 should have any volts.

I jumped R5. The control transistor went out and a 10k resistor took it's place.
Still no difference beside the LFO has no impact any more - of course.
The voltage on the pins are still alternating.

..so the Dwell pot should be tied to vb instead of GND?

Removed the PT2399 and there's voltage on the socket:
1: 5V
2: 0
3: 0
4: 0
5: 0.12 - 0.22V alternating
6: 0
7: 0
8: 0
9: 0.12 - 0.22V alternating
10:0.12 - 0.22V alternating
11: 0
12: 0
13: 0
14: 0
15: 0
16: 0

[ElectricDruid]

Checking the circuit, is there a cap missing on that input filter? From R5/R6 down to ground

I've seen some designs without that cap and it forms just a lowpass which has not to be used, isn't it?

Have you got any links to some examples?

To me, it looks like a multiple-feedback filter (MFB) with some parts missing. Doing a basic simulation of it gives results that don't prove anything much, but it doesn't seem to be filtering usefully.

[anotherjim]

I've seen other PT2399 designs where the MFB input filter from the datasheet has had the extra cap stripped out. Once someone does it in a successful published design, it gets copied.  It won't not-work, just isn't optimal. For guitar circuits, only a single feedback resistor and cap is adequate, just as IC6a.

Pins 5 and 9/10 happen not to connect to any other part of the circuit. The small voltage reading is probably just stray radiated noise picked up because those points are "floating" and nothing to worry about.

At this point, I would try another PT2399 (this is one chip I never buy in less than 5 units at a time). But first, confirm that with the chip out, you can audio probe a signal at the socket pin16.

[Mr. Lime]

It's already the 4th or 5th chip I'm trying.


Strange, if I take the signal from pin 15 to the amp, there's a clean and loud guitar signal to hear, similar to pin 2 of IC5.
On pin 16 is a kind of oscillation going on. Pin 14 the same - just louder. After R3 the signal is very weak and sounds distorted in the whole network till pin 15.
I don't get how pin 15 can sound so clean when pin 16 sounds like that?!
Tried to unbridge R5 again and add a cap for a lowpass but that didn't help either..

Here's a sound sample of pin 16:
https://soundcloud.com/user-958053490/pt2399-fail

Can a damaged resistor (R3, R5) or cap (C2, C3) lead to such noise?
But then again, how can pin 15 sound clean?

[anotherjim]

As long as pin2 reference is stable, then the internal opamps should be too.
Do you see on the schematic that if the Dwell pot is minimum and the feedback trimmer is also minimum, then the input signal sees an AC short to ground through the 100nF C20? I would take the trimmer out or disconnect the Dwell pot wiper so you can rule out that path as a problem.
If you get good audio on pin15 but nothing on pin11, then the PT2399 is not running.

Those PT2399 can lock up by an internally stored static charge which can linger for some time. It has been found that it can be cured by shorting all the chip pins together for a few minutes, easily done by pushing it into a piece of foil.

[Mr. Lime]

Checking the circuit, is there a cap missing on that input filter? From R5/R6 down to ground

I've seen some designs without that cap and it forms just a lowpass which has not to be used, isn't it?

Have you got any links to some examples?

To me, it looks like a multiple-feedback filter (MFB) with some parts missing. Doing a basic simulation of it gives results that don't prove anything much, but it doesn't seem to be filtering usefully.

Sorry Tom, I seem to miss your question, here's a schematic without that cap:
https://www.pedalpcb.com/docs/PitchWitch.pdf
https://www.pedalpcb.com/docs/SpiritBox.pdf
https://www.pedalpcb.com/docs/MagnetronDelay.pdf
I think I have seen some more but can't find them at the moment, has to be an EQD delay I guess.

[ElectricDruid]

Checking the circuit, is there a cap missing on that input filter? From R5/R6 down to ground

I've seen some designs without that cap and it forms just a lowpass which has not to be used, isn't it?

Have you got any links to some examples?

To me, it looks like a multiple-feedback filter (MFB) with some parts missing. Doing a basic simulation of it gives results that don't prove anything much, but it doesn't seem to be filtering usefully.

Sorry Tom, I seem to miss your question, here's a schematic without that cap:
https://www.pedalpcb.com/docs/PitchWitch.pdf
https://www.pedalpcb.com/docs/SpiritBox.pdf
https://www.pedalpcb.com/docs/MagnetronDelay.pdf
I think I have seen some more but can't find them at the moment, has to be an EQD delay I guess.

Thanks very much, that's more than enough for starters. I guess the question is "Is that just a mistake that's been copied, or is there some other reason for doing it that way?". Perhaps I can find time to do a few simulations and have a look.

[Mr. Lime]

Thanks very much, that's more than enough for starters. I guess the question is "Is that just a mistake that's been copied, or is there some other reason for doing it that way?". Perhaps I can find time to do a few simulations and have a look.

You're welcome!
When I drew my schematic I just combined some parts from here and there which seem to work for most delays as a starting point..
I assumed that my circuit won't work properly at the first try but this thing really drives me mad! 

As long as pin2 reference is stable, then the internal opamps should be too.
Do you see on the schematic that if the Dwell pot is minimum and the feedback trimmer is also minimum, then the input signal sees an AC short to ground through the 100nF C20? I would take the trimmer out or disconnect the Dwell pot wiper so you can rule out that path as a problem.
If you get good audio on pin15 but nothing on pin11, then the PT2399 is not running.

Those PT2399 can lock up by an internally stored static charge which can linger for some time. It has been found that it can be cured by shorting all the chip pins together for a few minutes, easily done by pushing it into a piece of foil.

I removed the connection from the wiper but still no difference, so I guess the feedback path was okay.
Pin 11 puts out a similar sound to pin 16, just a little louder.

I've updated the schematic here:


Even a new PT2399 doesn't help. I can't imagine that they are all faulty, there must be something else causing the problems.

[anotherjim]

You should visit the Little Angel thread. There you will find examples of builders that didn't have working PT2399's but in the frustrating position of not being able to prove it one way or another. When I first started using PT2399's, I bought x10 (from a reputable source) of which I killed x3.

Can you get a scope on pin5 and see if you have the VCO clock working?

BTW, the input is just an opamp. You should pick up negligible signal on the input pin16, since opamp feedback action should maintain that pin at the pin2 DC reference.