PT2399 not to spec

[Tightpants]

I have been trying to get a reliable method of controlling the delay time on this bloody chip. The data sheet gives a nice linear relationship between resistance on pin 6 and delay time. I bought some chips a few years ago and they were pretty much exact to the datasheet. This (in my mind) opened up the possibility of controlling the resistance/current by digital means and getting somewhat accurate results.
More recently however, the PT2399 chips have been much more variable with some deviating from the datasheet values by as much as 15%. I tested them by using a fixed resistance and measuring the delay time and clock frequency of different chips. I don't want to implicate any particular suppliers as this has happened now with chips from three different sources so it seems to be pretty widespread. I don't think they are fakes, they sound just as good/bad as the older ones, maybe they are out of spec rejects bought up by a wholesaler or maybe standards have dropped at Princeton. I should add that the relationship between delay time and clock frequency is rock solid as you might expect. This still leaves the possibility of calibrating each individual chip or using feedback of clock frequency to get accurate delay times but this does complicate things considerably 
Anybody else experienced anything like this?

[anotherjim]

Only used some last year. Delay time nothing like the data sheet chart. It also seems that recent ones don't work unless digital ground is tied to analogue ground, despite that you can measure 30ohm between pin 3 & 4 which appears to have always been the case. It seems they always manage to produce 1/2 vcc at the reference pin too, so it they aren't completely dud chips.
I suspect the good ones are in the smd packages, since that's what the major customers would be wanting nowadays.
UK retailer DrTweak sells batch tested PT2399, but I haven't ordered any from that source.

[smallbearelec]

If any of those PT2399 came from me, please drop a line to the smallbearelec address with the lot code. I only buy from Princeton's authorized distributors, so I need to know if there is a QC issue.

[Tightpants]

Seems like it's not just me then. I didn't get any from Small Bear but I'd be happy to check some for you if you want?

[R.G.]

I have authentic, genuine PTC PT2399's available, if needed:

What else in the way of help to people learning to build pedals other than just advertising chips for sale do you have?

You have four posts, and they're all involved with advertising ICs for sale.

[merlinb]

If you're controlling the delay time by digital means then surely you are reading the PT2399's clock anyway, to complete a feedback loop that locks the PT2399 to your chosen delay time (i.e. chosen clock freq). I don't see the problem?

[samhay]

^More recently however, the PT2399 chips have been much more variable with some deviating from the datasheet values by as much as 15%.

What deviates, the delay for given resistance? In any case, is the f(clock) to delay (in ms) conversion factor still about 684 in these chips?

[Govmnt_Lacky]

I have authentic, genuine PTC PT2399's available, if needed:

What else in the way of help to people learning to build pedals other than just advertising chips for sale do you have?

You have four posts, and they're all involved with advertising ICs for sale.

I guess Aron must have heard you... 

[Tightpants]

If you're controlling the delay time by digital means then surely you are reading the PT2399's clock anyway, to complete a feedback loop that locks the PT2399 to your chosen delay time (i.e. chosen clock freq). I don't see the problem?

Yes you are right, it's not a problem in terms of getting the job done. However clock frequency needs to be divided in hardware, feedback systems are clunky in software (at least when I do it) and self calibration/lookup tables take up a lot of memory space.
For my project I wanted to simplify things as much as possible and I had a simple system that seemed to work really well, one small PIC one transistor and a couple of passive components. On the software side, no calibration or feedback required, just substitution of a couple of simple equations and bingo...until I tried some of the more recent chips 

^More recently however, the PT2399 chips have been much more variable with some deviating from the datasheet values by as much as 15%.

What deviates, the delay for given resistance? In any case, is the f(clock) to delay (in ms) conversion factor still about 684 in these chips?

Yes it's the delay time vs resistance. It seems that the response of the oscillator for a given resistance is causing the deviations. The clock frequency / delay time conversion factor seems to be OK (still about 684).   

[samhay]

^Yes it's the delay time vs resistance.

I have this problem too. My chips need significantly less resistance to obtain the same delay.

If you can quickly read the clock and change the delay, then you can use a control loop to set the delay -
if clock != clock +/- tolerance {
if clock < desired clock, delay = --delay, delay = ++delay}

[Tightpants]

^Yes it's the delay time vs resistance.

I have this problem too. My chips need significantly less resistance to obtain the same delay.

If you can quickly read the clock and change the delay, then you can use a control loop to set the delay -
if clock != clock +/- tolerance {
if clock < desired clock, delay = --delay, delay = ++delay}

I know, I tried all that, it worked but I wasn't too happy with it, problems with hysteresis and wobble. Probably my crap software, have you got it working well/got any programming tips???

[samhay]

Afraid I am mostly still in the conceptual stage - bits work, but I keep asking myself how much polishing the turd of a PT2399 really needs and wander off to find other things to do instead.
I haven't tried a control loop yet, but have just enough experience with PID control to know how much of a PITA it can be to get stable.

Good luck.

[Tightpants]

Haha I like that, if I manage it I'm going to call it the "Burnished turd", or if I can get somewhere near, the "Roll it in glitter" delay!

[samhay]

Haha I like that, if I manage it I'm going to call it the "Burnished turd", or if I can get somewhere near, the "Roll it in glitter" delay!

Both most excellent choices.

[Cjuried]

You might also consider using a comparator for phase and frequency detection along with VCO control input, with filtering. The VCO output feeds a simple down-counter which feeds the first phase detector input. The second phase detector input is feed with a stable reference frequency (Quartz crystal or maybe a rubidium-disciplined crystal oscillator). Simply a quality Phase Locked Loop.

If you're controlling the delay time by digital means then surely you are reading the PT2399's clock anyway, to complete a feedback loop that locks the PT2399 to your chosen delay time (i.e. chosen clock freq). I don't see the problem?

[Cjuried]

2nd message posted in error.

[Tightpants]

You might also consider using a comparator for phase and frequency detection along with VCO control input, with filtering. The VCO output feeds a simple down-counter which feeds the first phase detector input. The second phase detector input is feed with a stable reference frequency (Quartz crystal or maybe a rubidium-disciplined crystal oscillator). Simply a quality Phase Locked Loop.

If you're controlling the delay time by digital means then surely you are reading the PT2399's clock anyway, to complete a feedback loop that locks the PT2399 to your chosen delay time (i.e. chosen clock freq). I don't see the problem?

Yes, thanks for that. I was put onto this idea earlier by ~arph (Arnoud) another member here. He posted this link on another thread:http://www.newtone-online.nl/forum/index.php/topic,7603.0.html
Looks very interesting but I haven't had time to have a proper look at it yet (another member was also interested in giving it a go but I don't know if he got it working).

[Cjuried]

Fantastic. Keep me posted if you build it. Would be great to hear the results.