PT2399 filter

[bioroids]

Hi!

Did anyone ever used the PT2399 clock output to run a switched filter, like in the old MXR units?

I think is a good idea, gonna try it on the course of the week.

Luck!

Miguel

[Mark Hammer]

Can the clock be tapped in the PT2399?  I don't know.  But if so, bear in mind that you may have yourself some hellacious dividing down to do.  The clock signal for the BBD's in the MXR Analog Delay couldn't have been much over 100khz, even at the shortest delay, and that necessitated dividing down a couple of times to get something usable for the switched filter.  Given that the PT2399 likely conducts several times the number of operations per unit time, compared to a simple BBD (after all, it's not just moving things along the memory stream, but the A/D and D/A as well, and its delta modulation as well), my guess is the division circuitry will end up being more complex and physically larger than the delay itself.

From where I stand, far better, and simpler to just whip up a variable 2-pole lowpass filter that you can adjust independent of delay time.  That way you aren't forced to have bright-sounding short delays.  I suspect it will use less current as well.

The switched filter was a brilliant solution for its time, in that it did not force "compromise" filter settings as was typical of most other contemporary 3-knob units.  At the same time, recognize that it was a solution prompted by the assumption that if one was ONLY going to offer 3 knobs (delay time, repeat, wet/dry mix), one may as well have the optimum usable bandwidth attainable at any given delay.  Clever, elegant, etc., but it still hemmed you in creatively.

[Vsat]

Where deltamod  delays run into trouble is not with aliasing (the sample rate is inherently very fast, greater than 500 KHz for the Princeton and Mitsubishi chips), but with slew distortion. Deltamod only gives you wide bandwidth at small signal levels - the bandwidth decreases as the signal level increases.  Beyond a certain freq x amplitude limit, slew distortion will set in as the modulator will be unable to track changes in the input signal quickly enough. A steep lowpass filter on the input is not appropriate... rather some form of slew-limiting filter is needed.
Mike

[bioroids]

Well, there's a clock output (pin number 5).

I'm afraid the frequencies are a little high as Marks says (2 Mhz at 340ms of delay, 22 Mhz at 34ms ! according to the datasheet). It should be posible to make a filter at the right frequencies using this clocks... I think.

Mike, that's pretty good info. I been using the classic filters and even increasing the signal level going to the chip! for increasing signal to noise ratios. I'm gonna do some googling about slew limiters now.

Thanks!

Regards

Miguel

[Vsat]

Miguel,
If you want a tracking filter for the output you could use something like the MXR approach or a ready-made switched capacitor filter like the Texas Instruments TLC04 or MF10CN and similar offerings from Nat Semi or Linear Tech or Maxim. These switched cap chips usually operate so that the cutoff freq is either 1/50 or 1/100 of the clock freq. You will need a selectable or programmable divider if you want different clock:cutoff ratios.
Cheers, Mike

[bioroids]

That was my idea. I'm gonna check if I can get some of those chips around here, though I think I prefer the MXR way if it can be done.

Luck!

Miguel

[bioroids]

Where deltamod  delays run into trouble is not with aliasing (the sample rate is inherently very fast, greater than 500 KHz for the Princeton and Mitsubishi chips), but with slew distortion. Deltamod only gives you wide bandwidth at small signal levels - the bandwidth decreases as the signal level increases.  Beyond a certain freq x amplitude limit, slew distortion will set in as the modulator will be unable to track changes in the input signal quickly enough. A steep lowpass filter on the input is not appropriate... rather some form of slew-limiting filter is needed.
Mike

Hi!

I wonder if this make sense:
using some kind of inverse preemphasis: cutting highs before the delay and restoring them after. This is to avoid the slew distortion by reducing the amplitude of the higher frequencies. Then the highs are restored.

Obviuoslly this would accentuate the hiss, but I'm not sure if this kind of digitalization also has quantization noise (as in regular PCM). If not, then the hissing should'nt be a problem. In my experience with the PT2399, I never noticed hiss, but I did noticed distortion.

Luck!

Miguel