Flash drive delay pedal

[liddokun]

I've been reading up on theories of using old computer hard drives for delay pedals, recording stuff onto the hard drive then playing back...etc.
But what about flashdrive/jump drive...?  Would it be possible to use flash memory instead, like a USB stick, and use it for the memory? USB drives these days are quite cheap at that, but does anyone know where I could start looking for information to do a project like that?

[Processaurus]

With a 4 GB memory stick you could get delay times of up to 6 and 1/2 hours!

[R.G.]

Let's play math for a minute.

CD quality sound is sampled at 16 bits per sample, 44,000 times per second. That's two bytes per sample, so 88,000 bytes are needed to delay sound by one second. Converting that into bits, you get 704,000 bits you need to remember per second. Call it a megabit per second, or 131kB. Delays longer than, say, 5 seconds are not very useful as other than raw looping. I personally would find it hard to use delays longer than one second for other than looping, but then I'm not Robert Fripp.

1Mb memory chips are cheap. You have to buy static RAM because you can no longer get dynamic RAM in sizes that small. Alternately, the cheapest, smallest DRAM chip you can get in current production is far more than you need.

The crucial issue in making delays is not that you can't store the bits - it's how complicated it is to get the bits into and out of that memory. The issue is the A->D->A and how to cope with the data rate in and out of the memory. It is quite difficult to do data in/out more simply than static RAM.

[liddokun]

I see. So I guess it's completely out of my reach huh?

[Mark Hammer]

Okay then.  We have chips like the PT2395 which can make use of an external 256k DRAM to produce 800ms delay.  Is there any way to fool it into using higher capacity chips to provide more than 800msec delay, or is the 9-bit address bus the kiss of death?

[Seljer]

not sure how much it applies to audio, but one of my professors was working on a recording system for accelerometers to be used with athletes, recording the movement to compact flash cards and he had a bunch of issues with buffering since he couldn't write it fast enough

[liddokun]

The usb flash drives have 4 connections, 2 of which we really don't need to worry about. 1 is 5vdc, second is ground, and then there's data negative and data positive.  I guess I need to really dig through research and find a way to get data into and out of there, and at a speed that is usable.

[tommy.genes]

Okay then.  We have chips like the PT2395 which can make use of an external 256k DRAM to produce 800ms delay.  Is there any way to fool it into using higher capacity chips to provide more than 800msec delay, or is the 9-bit address bus the kiss of death?

You could theoretically use additional logic to add more address lines, but at that level of complexity, you might as well get into uP programming and do away with the PT239X chips altogether. The addressing would really set the practical upper limit on delay time for a Princeton-type chip if it wasn't for...

Piggybacking DRAM chips as discussed HERE and HERE

-- T. G. --