[sdiy] Modern codecs in variable-rate digital delay?

[Richie Burnett]

Variable-rate operation of the CODEC should work fine Tom, provided you can generate the necessary phase-locked MCLK and SCLK clocks at 128x and 64x Fs or whatever the CODEC wants to see. Provided all of the clocks scale correctly with the sample rate the CODEC should stay happy, and its anti-aliasing and anti-imaging reconstruction filters should track the varying sample rate.

However, you might want to consider another possible option...

Keep the input and output sample rates fixed at something standard like 48kHz and just resample the data internally to whatever rate you want for your processing inside the DSP.

-Richie,



Sent from my Xperia SP on O2

---- Tom Wiltshire wrote ----

>Hi All,
>
>After my recent efforts building a fixed-sample-rate digital delay with only a dsPIC and SRAMs, I'm looking again at the problem.
>
>One of my original goals was to produce a functional clone of something like the Ibanez DM2000, which is a late 1980's 12-bit rack delay with companding. The significant thing about it is that it is a variable-rate sampling system. It has delay time modulation which is able to shift the sampling rate from a centre point of 64KHz up or down an octave (32-128KHz).
>I didn't manage anything like this, as in the end I used a fixed rate. However, I'm reconsidering.
>
>Would it be possible to use a modern delta-sigma-type codec with a variable rate like this? For example:
>
>	http://www.cirrus.com/products/cs4270/
>
>Many of these chips have an upper limit of 192KHz or so, so the rates are no problem. It would be possible to feed them data from a dsPIC also on a variable clock - this would give the same effect as the original system, where modulating the master clock causes *everything* to change rate.
>
>Is there any reason why a modern codec would not be able to cope with this? Would there be upper or lower limits on modulation speed? The dsPIC can use an internal PLL to boost its clock frequency, and that PLL would have an effect on the modulation that was possible. Do codecs commonly do anything similar?
>Another potential problem would be building a VC oscillator/clock that could reach the required speeds, since many chips seem to need master clocks of Fs x64 or more. That's getting up into the ten+ MHz range.
>
>It seems crazy that in this day and age it would be difficult to recreate technology from 25-30 years ago, but pretty much all modern systems seem to be aimed at a fixed rate, and variable rates are implemented by doing sample-rate interpolation rather than genuinely altering the rate.
>
>Thanks,
>Tom
>
>
>_______________________________________________
>Synth-diy mailing list
>Synth-diy at synth-diy.org
>http://synth-diy.org/mailman/listinfo/synth-diy