Bucket brigades for Paul.

[Kassen]

[deknow]

[Kassen]

[paul e.]

[paul e.]

kassen looks like you were already posting an explanation of bucket brigades as i was posting the above...reading

[Kassen]

You can look at them as a chain of s&h modules based on charges. Each one samples the last one by taking it´s charge each time the clock fires. The clock is just a analogue osc, running at the sample rates. Lower clocks give you more time but less samples per second. This obviously needs a anti-alsiassing filter.

The fun bit is that there are small differences between the s&h´s (which we call buckets because it´s like puring water from one into the next in a way) and of cource some energy is lost each time we we do this. This give them their grungy and "alive" character which is missing even from digital systems with variable clocks. You can experiment with the G2´s clocked registers and see how close you can get....

Since delays are usable as a building block for other stuff you can also have flangers and choruses based on these little chips, the DM-100 can switch to chorus mode, the CS-100 has a build in chorus with it´s own chip. Ibanez included the best one I ever heard in the UE-405.

[paul e.]

interesting...thanks

[Afro88]

Maybe I'm missing something, but if it's using s+h modules to delay the signal, you would need a helluva lot of s+h modules!!

So I guess this means that when you change the delay time, you change how many s+h modules are being used... interesting...

The variable clock bit sounds cool too. Is there any chance someone could post a quick sample of theirs? I'm very much into delay effects, especially dubby over-feedback analogue delay's

[mosc]

[paul e.]

[blue hell]

Posted a G2 example at http://electro-music.com/forum/post-43023.html#43023

Jan.

[Afro88]

Hahaha, that dreadful noise hey mosc?

Thanks for the explainations guys, and the patch - makes much more sense now. And Paul, I'd love to hear it when you get the chance

[Kassen]

[Afro88]

[Kassen]

[blue hell]

[Kassen]

[mosc]

Engineers use the term "discrete time" for such sampled analog systems. All of the Nyquist theory applies by the way. The only thing missing is quantization noise, but of course you get analog distortion and sampling errors in exchange.

Indeed, the nice thing about such systems is that the user has access to a variable clock, where in more common digital delay systems, the delay time is controlled by chaning the number of memory locations.

Good digital systems, Kyma, Eventide, and even the G2 allow one to use both methods.

---

This has been a good discussion. Maybe a Wiki would really be of value at electro-music.com. There a many topics like this that might be better in that format rather than as a forum conversation.

[Afro88]

Yep, I'll just stick to using my ears when buying a delay pedal

Nah, I get what you're saying - there's a fine line between sampling voltages at audio rate intervals and converting voltages into the digital realm where both time and amplitude are given discrete values on each sample.

I'd probably agree with you and say that the BB delay pedal is analogue, simply because the voltages are never completely converted into digital domain of 0's and 1's, they remain as voltages the whole way through. For some reason that really appeals to me - an analogue device imitating an aspect of a digital one, even if that wasn't initially the case when it was released.

[blue hell]

Interestingly the technology did not die. Another word for BBD is Charge Coupled Device (CCD), you still find these in image collectors - but there after analg shifting the data is digitized, voltage wise, of course.

It is a compact way still to deal with information on chips, just one charge cell and some coupling to the next one vs an 8, 10 or even 16 bit memory location and some special xfer glue (or a processor).

In recent neuro chip deveolpment there seems to be some role for charge transfers through/over a chip again, in this case to model the brain's chemical reactions.

Jan.

[ian-s]

A couple of points, Digital delays don’t change time by altering the number of samples in the buffer, except for maybe a global range or max delay time. They use a constant size circular buffer and adjust time by applying fractional indexing to the head and tail. Depending on the interpolation used, this will create cumulative errors resulting in distortion and noise. There should be no difference in behaviour to a variable clocked BBD, just less distortion, clock bleed and noise.
Because you only hear the value output from the last bucket, it should not be difficult to commute the effect of 511 (or whatever) previous buckets. Statistically, what is the difference between accumulating 512 random errors and just adding 1 random error with the same average value? The non linearity should not present too much of a problem.
If someone doesn’t already do a good digital BBD emulation, it is probably due to the market, not technical issues.

[elektro80]

I have heard there might be some BBD emulations coming for the UAD-1

I have owned several of the early BBD devices. The sound was warm, full and very useful. Main problem was the noise. The usefulness was limited due to this noise. Some of the later products were far better.

[mosc]

The filtering you experience in a BBD boxes isn't due to the characteristics of the devices themselves, but rather the filter that the designer puts into the feedback loop. Rob explained this at EM2005 quite well in his talk describing his emulation of a tape delay unit. His tape emulation is very effective, BTW.

All of these effects can be emulated in the digital domain. You can buy whatever box you like for the characteristic sound, but keep in mind the characteristic sound is because of choices made by a human designer, not because of something inherent in the technology of the memory elements.

Most digital memories in use today are dynamic RAM. I've yet to hear people say they like the sound of systems made with static RAM better than dynamic RAM, but I'm sure there are people who will make the case.

[ian-s]

[Kassen]