How to "Sshhhut" up a PT2399 once and for all!

[merlinb]

Just now, on the bench, I have been unable to measure any change in noise performance with the delay resistor connected to digital ground rather than analog ground. I am wondering if something more subtle is going on in Earthscum's examples?

[WhenBoredomPeaks]

Got ya.. So on the Rebot 2.5 we cut the trace between Pin 3 and 4 lift the ground side of the 47uf from pin 2 and send that to system ground

I have made this horrible cut between those traces, my knife is balls deep in the pcb but they (pin 3 and 4) are still connected according to my dmm.



Should i go deeper? (inception)

[Hides-His-Eyes]

Well, first you should see if there's 10R between them or 0. Both will cause a "beep" on a continuity test.

[Vince_b]

I have made this horrible cut between those traces, my knife is balls deep in the pcb but they (pin 3 and 4) are still connected according to my dmm.

Even if you cut all the way through your pcb you will still have continuity. It looks like pins 3 and 4 are internally connected. I tested a pt2399 out of a circuit and I read 9 ohm between pins 3 and 4.
If you read 0 ohm it looks like you haven't cut deep enough on the trace, but if you read a very small resistance it's normal.

[slacker]

Unless you've soldered the PT2399 in, pull it and then measure between the traces.

[WhenBoredomPeaks]

You guys were right, when i pulled out the chip, there were no connection between the legs, when i measured it in the socket, there were about 15 ohms resistance between those legs.

Now i lift that 47 uf cap and connect it to analog ground and will update the thread with the results.

Just to make sure, this should "eliminate" the noise from the repeats?

[Earthscum]

Just to make sure, this should "eliminate" the noise from the repeats?

It SHOULD, and has so far for me. What I have been able to eliminate with it is the digital hiss that gates in while you play, and suddenly disappears when you mute your strings. That hiss has been plaguing me with these things for awhile now, and thus far has only been able to be hushed with extensive filtering on the output amp and integrators.

Playing with the LA, I am still running 10nF in the integrators, but bumped up from 1nF to 10nF in the amp stages. I still get a bit of bleedthrough at longer delays... to be expected, bandwidth considerations and all. I also did the bias upgrade, and added a 150pF cap across the 470k in the buffer amp. Now it's REALLY quiet... even with the smallish integrator caps. Only issue I had was that I stupidly put in one of the newer chips, so I had a bit of a fuss firing it up (since I used a pot to adjust delay, I just cranked that up before I plugged in).

As for the Sewer Pipes, that thing was inspired (and originally started on the board) as anchovie's Noise Ensemble. That one was most noticable. I still get all the nice reverber-ish sounds, but no more digital hiss gated in during the playing. I'm gonna try his NE out again when I get home tonight and see if it works with that noise monster. Should make that circuit alot more fun to play with.

This fix hasn't gotten rid of any of the normal hiss, just cuts out a major portion of the digital hissing that comes through during playing. I don't think this will affect the op amps' noise figures, just isolation of the digi from the analog so the chip can compensate for clock noise internally.

[culturejam]

Nice developments, lads! 

[frequencycentral]

I'm in the middle of doing a layout for a tube amp with a Belton brick. The Belton (which is said to contain multiple PT2399) also has two grounds, signal ground and power ground. I was just looking at the GGG D-Verb schematic, it has the 7805 common hooked up to power ground, with no connection to signal ground apart from however it's connected within the Belton. May be something, may be nothing, but I'm wondering if we do the same with PT2399 - connect the common of the 7805 to digital ground - we might well shed some more unwanted audio artefacts?

http://www.generalguitargadgets.com/pdf/ggg_verb_sc.pdf?phpMyAdmin=78482479fd7e7fc3768044a841b3e85a

[Earthscum]

----
but I'm wondering if we do the same with PT2399 - connect the common of the 7805 to digital ground - we might well shed some more unwanted audio artefacts?

Hmmm... That would put the reg ground ~10 ohms above analog ground. Would that 10R make a difference in the performance of the 7805, or just limit current (which the chip does anyways, right?). 5V/10=500mA, or am I thinking from the wrong end of the voltage? lol. Good eye. I'm gonna try and motivate on a SP that is complete tonight (gotta buffer some things, and some diode limiting), so I'll probably give it a shot as well.

If we can get a good method of stability down, I can forsee a whole new slew of PT2399 abuse and torture in the near future. Especially if it leads to wider useable bandwidths at longer delays.

[nordine]

Breadboarded a Rebote 2.5 and tried both, delay pot to analog ground AND digital ground
there's no change in the digital hissing

btw, i assume, you're talking about the digital hiss on the repetitions? that distinction hasn't been made clear yet

cause there's three hiss sources on Rebote Delay:
power source hiss (easily solved by big caps to ground)
opamp hiss (102-103 caps in the feedback loop)
repeats hiss (unable yet to find a cure)<---is this the one? 

[Fp-www.Tonepad.com]

....Tried it. No noticeable change. Your experience with this 'mod' might be different, though.
Fp

[merlinb]

It SHOULD, and has so far for me. What I have been able to eliminate with it is the digital hiss that gates in while you play, and suddenly disappears when you mute your strings.

I had the same problem once, which turned out to be a bad connection. I discovered that if I hit slapped the strings in 'just the right way' the PT would shut up, but I could make it hiss again by stroking the strings in 'just the right way'. Reflowing the joints fixed the problem.

I don't think you will ever get good bandwidth with long delay times because it is fundamentally limited by quantization noise. The slower the clock, the worse the digitisation gets, and no amount of effort can change the inside of the chip! More's the pity...

[anchovie]

I just had a look at the datasheet for the HT8972, which is Holtek's pin-for-pin replacement. Their application circuit has the caps from pins 7 & 8 going to digital ground, as well as the delay time pot.

[WhenBoredomPeaks]

Ok i did the mod on Francisco's board. I tried to find out what happened through monitoring it with earphones through an interface.
The repeats are dirty (that is the chip's fault) and my guitar's noise levels are bigger than the other parts of the circuit but i think it sounded the same the last time i heard it.

[egasimus]

I don't think you will ever get good bandwidth with long delay times because it is fundamentally limited by quantization noise. The slower the clock, the worse the digitisation gets, and no amount of effort can change the inside of the chip! More's the pity...

Well, the solution is 2x PT2399. Or more...

[~arph]

It SHOULD, and has so far for me. What I have been able to eliminate with it is the digital hiss that gates in while you play, and suddenly disappears when you mute your strings.

I had the same problem once, which turned out to be a bad connection. I discovered that if I hit slapped the strings in 'just the right way' the PT would shut up, but I could make it hiss again by stroking the strings in 'just the right way'. Reflowing the joints fixed the problem.

I don't think you will ever get good bandwidth with long delay times because it is fundamentally limited by quantization noise. The slower the clock, the worse the digitisation gets, and no amount of effort can change the inside of the chip! More's the pity...

I'm having the exact same hiss problem Earthscum describes. But instead of a bad joint it appears it can be traced to the stompswitch. Sometimes after engaging the pedal the hiss it there, sometimes it is not. I figured it might have to do with some static discharge going into the circuit too as it is very heavy on the feedback. ( listen here: http://www.youtube.com/watch?v=8KauS8rvR3o ) (that is without the hiss, straight from the breadboard)


I'll try the digital ground sugggestions here and see if it makes a diffference in my circuit. The pin 7,8 caps to digital ground do seem to make sense too. Seperating the 78L05 GND to digital ground is interesting too.

[slacker]

Just now, on the bench, I have been unable to measure any change in noise performance with the delay resistor connected to digital ground rather than analog ground.

I couldn't hear any difference either. I can't tell any difference just by disconnecting pin 4 from analogue ground either. In my pedal there's already quite a lot of filtering of the audio so maybe that's masking any improvement gained doing this. Perhaps designs with less filtering will benefit from this.

The only change I noticed was with pin 4 disconnected, the pedal latched up and gave no delay with the delay resistance at minimum at power up,  this is less than the now recommended 1k, I don't recall it ever doing this before. This might just have been a one off thing though, I haven't tried to repeat it.

[~arph]

I've been tinkering with the PT2399 yesterday and found that reducing the pin 7 and 8 capacitors made a big difference in my circuit. 
I'm running it with a 560 ohms resistor to digital ground and caps for pin 7, 8 to digital ground too. With this short delay set up I was experiencing noise but only when sound was running through. When the input signal is gone there is no noise. It's not a constant hiss, but more of a crackling sound. This was with the pin 7, 8 caps at .1u at default value. It seems that the crackling noise can be shifted up in ffrequency by lowering the cap values there. I found 15nF to be optimal.


Also I tried swithing Dgnd to Agnd and noticed quite a difference. With Agdn connected to Dgnd I had more volume and a more bassy/hi-fi sound. With it disconnected, a little less volume and some bass loss. It preferred the separated Agnd, Dgnd sound. Also it seemed that the delay time was a little less stable with Agnd and Dgnd disconnected. It seemed to move around a (tiny little) bit, which added character.

So all in all, I did experience a difference, but not in noise. It took other tinkering to get the noise down.

I think we're not finished exploring this IC.

[Mugshot]

why dont we just email the manufacturer and ask info about the pins and ground (A/D) connections? i mean, we cannot rely on their datasheet alone, but working blindly on this chip could take so much time.