pt-80 delay question

[csmatt45]

Listened to the audio samples on GGG. on the last one Scott shows you that the higher notes distort a bit on the repeats.

Just curious, is this something that also happens in a 100% analog delay, or is this a "flaw" in the pt-80?

read about a million posts regarding the pt-80, and I do know that the degredation and loss of highs on the repeats are what contributes to the "warm" sound of the analog delays. Just didn't see anything addressing this exact issue with the pt-80.

thanks,

Matt

[geertjacobs]

I think it's due to the nature of a cheap digital delay IC like the PT2399.
Check this post on analog vs digital delay.
You can call it a "flaw", but the price/performance ratio of the PT-80 is too good to see it as a flaw in my opinion...

[Mark Hammer]

Listened to the audio samples on GGG. on the last one Scott shows you that the higher notes distort a bit on the repeats.

Just curious, is this something that also happens in a 100% analog delay, or is this a "flaw" in the pt-80?

read about a million posts regarding the pt-80, and I do know that the degredation and loss of highs on the repeats are what contributes to the "warm" sound of the analog delays. Just didn't see anything addressing this exact issue with the pt-80.

I can assure you that analog distorts on repeats too. 

The bucket brigade principle involves an analog sample of the instantaneous voltage available at the input being passed along, through FET gates, from one teeny tiny cap to another.  Those caps are far from perfect, and the more stages the sample has to be passed through, the more drift there is from the original sampled voltage.  This is particularly true when the sample has to be held onto for longer periods of time, which will naturally be the case for analog delay units that are always clocked slowly.  Keep in mind that there may only be one BBD chip in the unit, but when your input signal has come out the other end and then been repeated 3 times, it has passed through 4 x 4096 = 16384 "handoffs" or "bucket passes", each one of them a little leaky and more likely to result in sampling error the longer the sample has to sit in the cap before being passed along.  There may well be some error introduced when the delay time is set short (e.g., 100msec) but since those repeats are essentially laid over each other, the "audio clutter" tends to mask the resampling-based distortion.  In other words, not only is it a little worse when the delay time is longer and the caps are stretched to their limit, but you are also in a better position to be able to hear that distortion more easily if there are audible gaps between repeats.

That is one of the reasons why I personally favour inserting some extra lowpass filtering in the repeat/regen loop, because not only does it make the repeats sound more natural (reverberant spaces in nature lose bandwidth with reflection), but it also reduces the content where such repeat-based error is most likely to occur: in the high end where amplitude is much lower and any leakage-related errors are likely to be more erroneous and more audible.

My sense is that, whether analog or digital, audible sampling-type error is most likely to be found in the upper reaches of the spectrum.  The long and the short of that is that if you don't have higher quality digital or analog sampling (and I don't see us as ever likely to have a new ultra-high fidelity analog delay chip on the market, regardless of how much the price/feature ratio keeps improving for digital stuff), your next best strategy is to aim for better lowpass filtering of both the principle delay signal AND the recirculated signal.

[csmatt45]

Thanks Mark,

I appreciate the post. I'm still left with the question of A vs. B. Meaning, hypothetically, if we were to set audio quality aside, and were purely trying to duplicate the sound of, say the AD-80, but were doing it digitally, would it be the same?

even simpler, If I or you had acsess to a vintage ad-80 and had a pt-80 next to it, hit that same high note as the sample on GGG, do you think the ad-80 would have that same distortion on the repeats? Or a very similar dist. on repeats?

just curious......

Matt

[RaceDriver205]

Mark, do you know as to whether its possible to remove the 'analog-effect' i.e. the treble-cutting on repeats. Specifically the PT80. Its useless when your aiming for 'steve-vai-stlye' delay applied to distortion.
Cheers RD205