Multiple PT2399 in series for mega long delays - what have folks done?

[frequencycentral]

I'm considering stringing a few together. Possibly two, maybe four. Optimally one would want to be able to control the delay time of all of them for one knob, so two in series would mean a dual gang, four might mean some LDR jiggerypokery.

So.....I'm askin' - what have people done previously? Does two in series give a decent 1 second of delay? Is excess noise an issue? Heterodyning? Any other ideas on how to implement say four PT2399 in series as far as controlling the delay times goes? Current mirror is a term which rings a bell...........

[CynicalMan]

Couldn't you control delay time with a rotary switch and a pot? One of the PT2399s would be controlled by a pot, and then you'd have a rotary switch to switch in each of the other PT2399, which would all be set at maximum delay. So position 1 (only 1 PT2399 on) would be 0-300ms, position 2 would switch in another for 300-600ms, position 3 would be 600-900ms, and position 4 would be 900-1200ms. I think many people are used to this kind of control anyway from the DD-3 and similar pedals. It would also allow finer control of the delay time.

Still, if you're building such a complex delay, you might as well throw in tap tempo too. 

[Govmnt_Lacky]

Haven't I read somewhere that this is how the Belton Reverb brick is designed? Multiple PTs 

[Suicufnoc]

Still, if you're building such a complex delay, you might as well throw in tap tempo too. 

I'm thinking you could use a Taptation with two digital pots in parallel and the tempo division always set to double to pull this off.

[frequencycentral]

Couldn't you control delay time with a rotary switch and a pot? One of the PT2399s would be controlled by a pot, and then you'd have a rotary switch to switch in each of the other PT2399, which would all be set at maximum delay. So position 1 (only 1 PT2399 on) would be 0-300ms, position 2 would switch in another for 300-600ms, position 3 would be 600-900ms, and position 4 would be 900-1200ms. I think many people are used to this kind of control anyway from the DD-3 and similar pedals. It would also allow finer control of the delay time.

I just started cooking dinner after I posted the initial post. While chopping the jalapenos I came up with the identical solution, and for all the same reasons, finer control etc - great minds again?? Great idea anyway

[slacker]

Here's how I did it for the "double time" mod for the echo base. The short/long switch sets the second one to either the minimum delay time, so it's basically off or to the maximum time, with the first being controlled by a pot as normal.

http://www.diystompboxes.com/smfforum/index.php?topic=60662.msg533527#msg533527

For more than two it might get noisy because the signal is still going through all the PT2399s even when they're "off". It also makes the minimum time equal to the combined minimum time of both chips which might become an issue with more chips.

Another way to do it would be to have them all connected in series with the time of the first controlled by a pot and the others set to a fixed time and simply take the output and feedback from which ever one you want. That might be easier than switching them in and out. Also means you could do weird stuff like taking the output and feedback from different stages.

Sounds like a great idea however you do it.

[Fender3D]

PT2399's clock is from a VCO.
You might use the same voltage, maybe trough a "unlatching" trick so to be safe..., to control the overall delay time.
Your minimum delay time will be chip_min_del x 4 though...

[frequencycentral]

Another way to do it would be to have them all connected in series with the time of the first controlled by a pot and the others set to a fixed time and simply take the output and feedback from which ever one you want.

That's an even better idea, thanks Ian!

Also means you could do weird stuff like taking the output and feedback from different stages.

I'm gonna have to think that one through......

[Suicufnoc]

Also means you could do weird stuff like taking the output and feedback from different stages.

That's quite an interesting thought.  My first though was that you would pull feedback from an earlier stage than the output, giving you a longer initial delay and then closer feedback repeats.  But you could pull the feedback out of a later stage giving you a quick first delay and slower repeats.

[Thomeeque]

I'm considering stringing a few together. Possibly two, maybe four. Optimally one would want to be able to control the delay time of all of them for one knob, so two in series would mean a dual gang, four might mean some LDR jiggerypokery.

So.....I'm askin' - what have people done previously? Does two in series give a decent 1 second of delay? Is excess noise an issue? Heterodyning? Any other ideas on how to implement say four PT2399 in series as far as controlling the delay times goes? Current mirror is a term which rings a bell...........

Check Yerasov DM-5, it answers basically all your questions, it's delay time control is IMO brilliant and could be easily expanded to control four PT2399s. Even modulation could be simply achieved btw. (by modulating those 5V at P5), I plan to try it once. Heterodyning is solved by using slightly different values for R15 and R16.

IMO 1sec from two PT2399s gives already very dirty delay (as does 500ms from one), but it is very subjective. To my taste already 600ms is not perfect but yet acceptable (hear my demo in that thread).

Good luck, T.

[merlinb]

So.....I'm askin' - what have people done previously? Does two in series give a decent 1 second of delay? Is excess noise an issue? Heterodyning? Any other ideas on how to implement say four PT2399 in series as far as controlling the delay times goes? Current mirror is a term which rings a bell...........

This is something I have been planning for a while. Easiest way to control all four delay times would be to connect every delay pin (via individual 1k resistors) to a single pot. Current mirrors would be the sensible choice if you want to get different delay times from each PT, for reverb effects say.

For reduction of noise you want the input signal to be as big as possible, and the PT can handle about 3Vp-p reliably over most of its bandwidth. I amplify by x1.5 at the input of my Equinox, giving me about 3dB quasi companding; depends how much headroom you can sacrifice.
Input filter can be a data-sheet LP filter, since this is not very critical. Then each PT is cascaded with little filtering in between (simple 10kHz LP is fine). The output filtering is the imnportant bit. I was thinking of data sheet 3rd order LP filter, but perhaps with the feedback taken from the 2nd order node (or maybe before any filtering at all), so your feedback single isn't being filtered as heavily every single time it goes around, so the repeats retain brightness.

[Thomeeque]

Then each PT is cascaded with little filtering in between (simple 10kHz LP is fine).

Have you tried it for longer delays? Proper filtering in between stages seems to me essential for cascading.

T.

[R.G.]

I think I must have exercised every possible scheme for ganging up PT2399s, including a few that haven't been suggested. The conclusion I came to is that if you want PT2399s in series, you have the obvious issues: quantization noise at delays over about 400mS, heterodyning and quantizing+heterodyning noise in series, loss of treble if you filter to get rid of the noise with a high pass cutoff. Princeton Technologies was not interested in changing the innards or even telling me more about the insides for less than a 100K unit order.

I did solve the heterodyning issue by locking the oscillators up. That still left quantization noise and its other artifacts for longer delays. I did get a 2 second delay working, but in my judgement, the necessary standard of audio quality was not practically reachable.

My conclusion is that the PT2399 is usable up to 400ms, full stop. Beyond that, you're accepting steadily declining audio quality.

[frequencycentral]

Funnily enough, I just found a dual PT2399 based delay over at the other forum (designed/posted by cpm) which uses 2 x BS170 as a current mirror to control the delay time of the second PT. No link available of course, but do a search for 'Board index » Circuit Analysis » DIY Stompbox designs » 'my new delay, over the top PT2399's'. Maybe then for 4 x PT2399, one delay time pot and 3 x current mirrors.

[frequencycentral]

Easiest way to control all four delay times would be to connect every delay pin (via individual 1k resistors) to a single pot.

I got really bad noise issues last time I tried that. Enough to convince me that the respective pin 6's need to be isolated from each other in some way.

[merlinb]

Easiest way to control all four delay times would be to connect every delay pin (via individual 1k resistors) to a single pot.

I got really bad noise issues last time I tried that. Enough to convince me that the respective pin 6's need to be isolated from each other in some way.

Well, they're current sources so they need some individual resistance to work against, and for four PTs the shared resistance would be one fourth of the value you would use for a single PT. Not that I've tried this topology.

[R.G.]

The real bottom line for delays is that above about 100-500mS, you either need to use what amounts to tape (or electrostatic) recording and playback, or go digital. Other analog techniques have practical issues with noise, fidelity and losses after that.

Of course, the PT2399 is digital. It just has a limited internal memory and method for clocking it.  So it inherits issues with quantization noise, heterodyning, and all the rest, without letting you have access to the digital side of things to ameliorate them.

The simple thing to do is to use one of the clock control inputs as a digital input and drive an external clock into it to force the internal clocks into lockstep. Unfortunately, that does not work for the PT2399, as the clock circuitry is hidden behind that analog current source clock control pin. Overdriving the clock/VCO control pin does not work (ask me how I know that  ) until you precisely damage the input with the perfect amount of electronic damage. I managed this on about one out of ten chips. The other pins don't do what you want either.

Neither does talking to Princeton Technologies. The current source input is the input to a complex current mirror that makes the actual clock, and is not amenable to digital inputs. Nor are the other pins.

The PT2399 is good for what it does, but not very good fodder for growing beyond that. There are places for every technology where you have exploited all the easy gains to get, and you need to step to a different technology to get more results.

[Taylor]

You can make a 2-second, very nice-sounding delay with the FV1. Take the stock delay code from Spin's site, clock the chip at 16k. It's super easy to add in as much grime as you want if you want it to sound like a PT2399.

Just sayin'.   I know that there is an allure in doing with the 2399 that which seems impossible, of making a device do things for which it was not designed. I've even stacked a bunch of 2399s in series just as you're thinking. But in this case I agree with RG that this is a limit of the tech. Of course, I wouldn't be shocked if you make it work and sound good despite us!

[R.G.]

Yep. You can even get a several second delay by extending that further; some BBD circuits filter the delayed signal down as low as 3kHz, so you could clock it down at a bit over 6kHz and get  five second delay. It's the classic tradeoff that Nyquist noted: if you limit the input signal bandwidth (and also output clock filtering) to half the sample rate, you can get alias-free sampled signal reconstruction.

You pay for that with the recovered signal bandwidth. 3kHz filtering makes it sound pretty dull. But some people would be OK with it.

I'm always extremely cautious with running clocks down in the audio midrange. It's very easy to mess up clock isolation and have an intolerable whine. Not completely unavoidable, but very easy to fall into.

[CynicalMan]

3kHz filtering makes it sound pretty dull warm.

Fixed.