Bucket brigade analog delay

[loogie]

I'd like to try to build an analog delay based on the bucket brigade chip.  It doesn't appear these are in production anymore although Maxon claims the use of their "proprietary" technology so someone must be making them.  Does anyone know of an equivalent device currently in production?

Also, I was wondering why fewer capacitors couldn't be used in some sort of a circular queue?  One clock would recycle the queue and another would periodically empty it.

I know that NOS BBDs are available.

[PRR]

Bucket brigades are an abomination to audio. They existed only because we had nothing better--- in 1972. Looking back... they never did decent "delay", the "reverb" was passible only if too subtle to make much difference, and you always knew a BBD was in the line because of the background swish.

Darth Vader, that's what they do best.

Early ones were 256 bit, not because that was a lot of bits, but because a 256-stage amplifier has horrible distortion. Even though each stage is a "cathode follower", stack enough pretty-clean stages in a row and it gets awful-dirty. The 1024 and 4096-stage devices came only after distortion was reduced somewhat.

Aside from the chip, the in/out/clock support systems are pretty hairy.

You can feed the output back to the input; nearly all complete BBD systems had a knob. It multiplies the awful sound. It does not give longer delay, it gives multiple repeats (and re-distortions) of the short delay.

ADC-RAM-DAC is a better scheme. Holtek HT8970 HT8972 -- Voice Echo are simple implementations.

[loogie]

The idea springs from the mind of friend who claims there is a resurgance of interest in analog devices of all types including BBD derived delays.  He finds the noise and general sonic degeneration appealing.  He wants to get away from the clean predictability of digital effects. 

I need to do a lot more reading and I'm pretty naive (sometimes a plus), but do you really need 256 or 512 or 1024 capacitors?  In my simple mind I envision (conceptually) one timing loop for your standard off the shelf delay of the 80's:  In one end of the chain, out the other and then back in until the required delay is achieved.  What if there was an inner loop as well?  So 100 trips down a 10 capacitor chain would be the inner loop -- approximately equivalent (conceputally) to one trip down the 1024 cap chain. This would be repeated by the outer loop until the desired delay reached.

Is there a minimum number of buckets required to properly (in so far as it goes) sample the signal?

[PRR]

The signal is sampled on the first cap.

The clock rate must be twice your highest audio frequency, and preferably far beyond hearing.

The delay is the number of caps divided by the sample rate.

Pretend you sample CD quality, 44 samples/second, and had 1000 caps.

The maximum single-pass delay, in fact the maximum "storage", is 1000/44000 or 1/44 of a second or about 22 milliseconds or about 24 feet of air-path.

Haas-effect tells us delays <35mS behind the main signal are almost un-heard.

You can recycle. But a single 22mS loop is musically pretty useless. You often want a continuous process, or at least longer than 22mS. What do you do with the next 22mS of time? Mix it with the first 22mS? Then after 1 second you have 44 different time-slices going round the loop. This is a somewhat effective many-echo "reverb" effect. Or would be if not for the garbage of 44 passes through 1000 buffer amplifiers.

BBD boxes used to sell for $49. Even Radio Shed had one. eBay Item number: 160303838 may be one, cheap enuff to play with.

[loogie]

I can see the flaws in my logic -- compounded  by the flaws in what follows.  At a tempo of 120 and 4/4 an eight note played for its full duration would need about 10 - 1024 BBDs at 44khz in order to sample it fully. Then it would need more BBDs to travel through before you could actually delay it.

So a sampling rate of say 4khz which would fit in most of a guitar signal might be possible.  One BBD could handle it and several more would supply the delay.  So really long delays would mean you'd have to slow down the clock even more or add BBDs.  My idea would be hopeless because the buffer would be the sum of the samples.  Conceptually.

Or something like that.

I have two of those old Radio Shed devices.  Probably a slow clock and not many buckets.  They seemed like miraculous devices to us at the time.  Sound awful now.

[PRR]

> 10 - 1024 BBDs at 44khz

About a quarter second.

> Then it would need more BBDs to travel through before you could actually delay it.

It would already be delayed the quarter second.

You can, if you are quick, play one-quarter-note and loop it "forever".

The distortion of ten BBD is large. The note won't sound good for long.

Recall how a BBD works. Recall a "bucket brigade" to pass water to a fire. Instead of a fire, pretend the first bucket-filler puts in water proportional to some interesting quantity, such as audio sound pressure. If the bucket were simply passed down the line, that quantity would be preserved. But it is electrically simpler to "copy" from one bucket to the next. Imagine the water were poured from one bucket to the next to the next to the next... It has the same problem as the "telephone" game, where a secret is whispered ear-to-ear and comes out totally different.

> sampling rate of say 4khz which would fit in most of a guitar signal

Over 2 seconds if clocked at 4KHz. The audio bandwidth is then 2KHz (or less), which is pretty dull.

I think BBDs are a pimple on the ass of audio. I was there when the first samples came out, and I have used some of the last and "best" BBDs. It was once marginally useful to have "reverb" in a 400 Watt 12 channel powered mixer you could carry with one hand. I don't think I used the "reverb" twice.

Don't let my bad attitude stop you. But go in with your eyes open. (And IMHO, with your ears shut... but that defeats the point, don't it?)

BBDs can make tolerable "flange" and "chorus" effects. And very effective Darth Vader / "vocoder" distortions. Because they are "useful" in those roles, vintage BBD chips are hard to score today, people are hoarding them.

I hate tape delay too, but tape is MUCH more flexible and less nasty sounding than BBDs.

[loogie]

Word has it that someone in China is making BBDs for Maxon.  We're investigating. 

I use a Lexicon MPX100 which is all digital of course and I love it.  Small money well spent.  But I also have a couple of oil can delays and I find the sounds they make mesmerizing.  Noisy, dull and beautiful for some reason.  Each one is driven by a 7199 and the circuits are similar, still each has its own distinct sound.  This variability also pleases some guitarists. 

The BBD project will be a great learning experience at the very least.  Maybe I should try a modern version of the oil can?  I have a gallon of the stuff...

[loogie]

I wonder if it would be possible to use a CCD or CMOS sensor?

[jerrydyer]

the new mxr is that  Carbon Copy .. per Bob Cedro

[jasperok]

http://www.generalguitargadgets.com/index.php?option=com_content&task=view&id=48&Itemid=26

in the middle of this project at the moment

may be of interest?