Delay chip with the longest clean time?

[Taylor]

Pretty much all there in the title: what delay chip has the longest time without underclocking? Ideally without extra ram, but not necessarily.

What's being used in these digital delays like the Digitech Digidelay, which has 4 seconds? Couldn't find the answer to that in searching.

[R.G.]

The simple "delay chips" are limited in time to about one second.

The longer delays are not simple delay chips. They have multichip digital setups with A/D converters, D/A converters, microcontrollers, possibly gate array chips or FPGAs, possibly DSP chips and more RAM.

There are not any really simple solutions. Even the delay chips with external ram are limited in memory address lines, so putting on more RAM does nothing to extend the delay time - the chip cannot address the bigger RAM.

[Taylor]

What chips go up to 1 second? I've only found the PT2395 which goes up to 800ms with extra RAM.

[R.G.]

"about one second" - 800ms is about one second.

Yes, it needs an alternate RAM Chip. However, you cannot add even more RAM because it cannot address the extra RAM, being address bit limited. 800ms is obtained with an obsolete and difficult to find 256k bit ram chip. You can put an easier to find 1Mb chip on it, but it can still only address 256k of the 1Mb. Even if you add external logic to get the extra address bits, the internal logic can't control the addresses for some of the nicer functions of the chip and changing timing can produce some very weird results.

How do I know this?

[Auke Haarsma]

It also depends on what you perceive as 'clean'.

My PT2395 goes well over 1 second, even over 2 and is still 'pretty' clean.

Maybe try one of the loopers like Tonegods Payback and convert it to a delay. At short settings it can stay 'pretty' clean. At 20+ seconds it is lofi, at lower settings, it stays more clean.

[Mark Hammer]

Ah, the age old question of what constitutes "clean".  I've had a good many non-productive debates about that very concept with my kids.

There is "clean" that is accompanied by relatively full bandwidth for the original signal.  In the case of guitar or bass, let's say this goes out to maybe 8-10khz.  There is "clean" that does not provide full bandwidth, but at least does not suffer from any aliasing or similar artifacts resulting from undersampling etc.  There is "clean" that provides full bandwidth with minimal or negligible artifact, but reduces the delay capacity.  I imagine there are a bunch of other "cleans" too.

I've asked the question here a couple of times regarding what the major "inflection points" are with respect to delay time.  In other words, at what point in the continuum from zero to infinite delay does one start to think about the use of the delay in different terms.  So, for example does a delay time of 400msec involve using the pedal in any manner that is qualitatively different from, say, 200 or 300msec?  Does the nature of the use start to change as we hit 600 or 700msec?  Is 1000msec that much different in terms of one's strategic thinking from 800msec?  And so on.

Finally, although the amount of on-board RAM in those standalone delay chips like the PT2399 is fairly modest, I remember only too well the "digital delay project" in my old 1980 issue of Polyphony (the orange one with Gary Numan on the cover, for those folks here of a certain vintage).  It had 8-bit A-D/D-A, requiring a handful of chips to accomplish, and 16k of DRAM, requiring 16 DRAM chips and another fistful for addressing, and yielded a few hundred milliseconds at best, with very modest bandwidth (And I remember the days when 16k of DRAM for an S-100 system would set you back a couple of hundred bucks!).  The board would have been big enough to demand a rack chassis, and the power supply would have had to have been 1 amp capacity at least.

When you consider that the PT2399 essentially packs everything above into a single chip that can run off a 9v battery and comes with MORE memory capacity on board, I think we're doing okay.  When Danelectro can turn out the FAB Echo for the price of a movie ticket and medium popcorn, and still turn a profit, I think we're doing REALLY okay.

The point is not to smack you on the wrist and scold you for being spoiled.  The point it that the rate of miniaturization is progressing at a rate that we can probably expect to see chips to replace the PT2395, et al, WITH sufficient RAM to support 1.6sec delay (or more) long before any of us here have lost interest in guitar and delay pedals.  Be patient.  It'll come soon enough.