PT2399 delay

[DWBH]

I've read somewhere that the max delay time for a PT2399 delay is around 400-600ms.
However, delays like the Danelectro PB&J, which are based on the PT2399 as well, have a maximum delay time of 1000ms (1 sec).
How is this possible?
And the same with the Danelectro Fab Echo, once modded it allows us to change delay time, but only up to 600ms?

[Mark Hammer]

In theory, you can push most delay chips ut to much longer delays than you'd expect.  The key question is whether one retains the bandwidth and sonic fidelity one is aiming for at those longer delay times.  In most instances, the answer is a resounding No.

That being said, if you apply some serious lowpass filtering to the wet signal so that any grittiness and artifact arising from aliasing, and undersampling, or from cap leakage in the case of BBDs (remember, those caps can't hang onto the signal sample forever!) is essentially smudged over, you can push the boundaries further.

For instance, Mike Irwin designed this delay unit for MODCAN: http://www.modcan.com/modhtml/delay.html
The unit uses 8192 stages of analog delay but is spec'd for 8 seconds max delay time.   Mike told me that if one cranked the regen and delay time up full and came back a while later (I forget how many minutes that was), that the audio signal was still "recognizable".  But of course, "recognizable" is not the same as "usable", and "usable" means different things to a pure electronic musician and to a guitar player working in the format of songs with verses and choruses and a beat.

How does Mike do it?  As the web-page states, two 6-pole tracking filters.  That's 12 'poles' of filtering, about 4-6 more than many delays have.  The tracking part means that when the delay is shorter and can tolerate more bandwidth, that bandwidth is provided, but when the delay is long the bandwidth is seriously restricted (e.g., rolloff at 1khz or maybe even lower).

So, yeah, you CAN squeeze ridiculous amounts of delay time out of a PT2399 if you want, but as a guitar player you probably won't like what you hear after the first 400ms or so.

[DWBH]

So... how does the PB&J stays that clean?

[Mark Hammer]

You can be fabulously "clean" if you are willing to sacrifice bandwidth.  I have no idea what the filters are set to in the PB&J, with respect to corner frequency, nor what sort of audio grime crops up at the longest delay times.

If memory serves, too, doesn't the PB&J use a PT2395 rather than a 2399?  Like the older HT8955, the 2395 can use an external 256k DRAM for up to 800msec of reasonable-fidelity delay.  Big difference compared to the 2399 which has a much smaller buffer space internal to the chip.

[DWBH]

Sorry for my mistake. It is indeed the PT2395, not the PT2399. Sorry for the confusion, and thanks Mark