PT2399 latch-up issues

[slashandburn]

Hey all.  I'm having a bit of a mare with a PT2399 latching up and I can't seem to solve this. Weirdly if I remove the power for a minute and plug it back in the effect works for a second or two before the latch-up occurs again.

I've been digging though some other threads of similar issues, pin 3 and 4 are both definitely grounded, I've tried swapping in different pt2399 chips and replaced the 50k Delay Time pot. I've also knifed all the traces around pin 6 to be sure there's no short to ground. I'm stumped.

Its a Keeley Magnetic Echo clone based on the schematic here: https://www.pedalpcb.com/docs/MagnetronDelay.pdf

I've built one before using the same layout so im sure the layout isn't the culprit, but might be worth mentioning this one boxed up with a Tremolo (Shoot the Moon) which might affect things. The tremolo side is fine though.

I was considering swapping R24 (in series with the Delay time pot between pin 6 and ground) for a higher value as its only 1k and I've read ideally it should be at least 2k to prevent latch-up, but I'd assume so long as the pot is set to more than 1k that shouldn't be an issue.

Any ideas? I can post voltages, layout and photos if necessary.
Cheers

[anotherjim]

Never had a problem with a 1k minimum pin6 resistor myself.
What is pin 2 doing? Doesn't go anywhere on the schematic. Should have a decoupling cap to 0v on it - not fussy, I fit 10uF.

[slashandburn]

Thanks Jim. Pin two has a 47uf cap to ground.
Probably should have shared in the first post this is the layout I've used:
https://4.bp.blogspot.com/-N2LvxPQl9qc/XcXIgI0gTcI/AAAAAAAAG4I/MdEF646jKiQah8ggN_1iEzuaTfpyvkfgQCK4BGAYYCw/s1600/Keeley%2BMag%2BEcho.png

I'd imagine I've just got a small solder bridge somewhere but can't for the life of me find it and have reflowed everything and scraped between every trace I can find.

Aside from a decoupling cap on pin two to ground, putting pins 3+4 to ground and making sure there's at least 1k to ground off pin 6, is there much else that could be causing the lockup?
And is the "works for a second before locking up" after killing the power to be expected or does that sound weird?
Cheers

[anotherjim]

I've found excessive LFO voltage swing can lock the chip up. I see the LFO runs off 9v and feeds via a cap. This can be dangerous to MOS  devices as together with any internal diodes it can form a charge pump voltage multiplier. I've used LFO's in my own designs running off the same 5v as the PT2399 and DC coupled via resistors to either pin6 or pin2. This cannot force excessive voltage anywhere.
Maybe you can measure the LFO waveform voltage and if it's over 5v peak-peak lower R21 to reduce it.

[duck_arse]

me too - I was thinking lift one leg of that 4u7 from the lfo, see if you get latches still.

[slashandburn]

That's all a bit over my head Jim!  By "measure the LFO waveform voltage" do you just mean get a voltage reading on the wiper of the depth pot? Or something else entirely?

And thanks for the suggestion duckarse! Gave that a go but no joy. I'll have another poke about and swear at it a few more times and if that fails I'll be back with voltages and photos.

[anotherjim]

You'd need a scope to measure the LFO. The voltage at the output of the second opamp goes up and down. The difference between the maximum up voltage and minimum down voltage is the peak to peak voltage swing. Generally, a multimeter on AC volts will give an averaged RMS voltage reading which is lower than the peaks, although if you multiply the RMS reading by 1.4 it will probably be close to a peak reading.

It could be you have a temperamental PT2399 that doesn't like that minimum 1k time resistor -  and the LFO at full depth can periodically add a 4k7 in parallel. Since it's not meant to make delays short enough for chorus, it probably won't harm to try raising the 1k to 2k.