BBD to make a analog reverb?

[Wavelength]

Gang,

I was designing an elaborate tank reverb the other day when Dan from TPS asked me some questions about BBD's. It got me thinking that maybe using a couple of different BBD's and some summation that something could be made pretty well.

Has anyone considered this? I do digital stuff at work and while it sounds good I would rather not do that here.

Thanks,
Gordon

[Fancy Lime]

I have considered it, done a good deal of research into it, and then quickly abandoned the idea. Getting a decent reverb out of BBDs requires multiple multi-tap BBDs or even more single tap ones plus the accompanying clock circuitry. I'm sure this can be fun but making it sound anywhere near as good as a simple digital reverb or real spring reverb will be challenging. Nothing wrong with a good challenge, though. If you really really want to do this, I would suggest experimenting with a PT2399 based reverb first and then apply the lessons learned to a BBD design. With the PTs, you will have a lot less of the troublesome circuitry and it will sound fairly similarly to a BBD design.

HTH,
Andy

[ElectricDruid]

The MN3011 was the chip designed with this in mind - six separate taps at non-multiple intervals. That's something you'd find very difficult to do with even a selection of different BBDs.

Whether it's any good or not...I'm not sure. The Rockman Reverb was done like that, so maybe find a few Rockman videos on youtube and have a listen? It's certainly not the ony way, even if if you wanted that particular sound (it would be easy to match the same tap positions and mix in an FV-1, for example).

While it's not impossible to make an analog BBD reverb (we could even mimic the Belton bricks with BBDs if we thought that was a good idea) it's not simple and might not be a particularly fantastic sound, so you need to know what you're aiming for and what's most important. Great sound? Simplicity? Analog signal path? Those might be mutually exclusive!

HTH

[Mark Hammer]

Panasonic produced two multi-tap BBDs, precisely for this purpose: the MN3011 and the MN3214.  The 3011 had 6 harmonically unrelated taps along 3328 stages, and the 3214 had 5 taps along 1024 stages.  I've never seen a 3214 for sale anywhere, although I believe Yamaha made a pedal using them.

The difficulty with using BBDs to produce reverb is that, theoretically, you'd need to clock them all at different frequencies in order to mimic different reflection times.  "But but, but...", you say, "Couldn't I use a 1024 stage, 2048 stage, and 4096 stage chip to do that?".  Sure, but they are all multiples of each other so avoiding the fantastically realistic replication of standing waves, they would have to be clocked at different frequencies.  And once you have several high-frequency clocks on the same circuit board, you entertain the risk of heterodyning and all that annoying whining.

This is what made the 3011 such a fabulous chip.  One clock frequency to drive the whole thing and multiple taps.  If a person wanted to thicken things up a bit, you could not only use those taps as is, but feed them to, say, a couple of subsidiary BBDs.  The 3011 has taps at 396, 662, 1194, 1726, 2790, and 3328 stages.   Feed the 396-stage and 1726-stage taps to their own 2048-stage BBD and you now also have delays at 2444 and 3774 stages in addition.   Feed their combined signal back to the input and you'll have a pretty decent reverb, assuming their relative amplitudes are mixed in the right combination and with relevant filtering (early reflections are always brighter and louder than later ones).  The 3 BBDs could all be run at the same clock frequency.

Another strategy is to use several cascaded PT2399s, each set to a different delay.  There is also a risk of heterodyning by doing this, but their HF clocks are generally well above human hearing range so it may be less of an issue.  Clever use of splitter/mixer stages between the chips, feeding the output of chip A forward, and the outputs of chips B and C back, could yield highly complex reflections.

A much simpler strategy is something I like to do with all analog and PT2399-based delays, and that is add in some additional shallow lowpass filtering in the feedback loop.  The steep lowpass filtering on the wet path provides a final "master bandwidth".  I like to stick a single pole of RC lowpass in the feedback loop itself, such that bandwidth is progressively reduced below the master bandwidth with each repeat.  The trick is to pick a rolloff below the corner frequency of the master filtering.  So if the unit is spec'd for, say, 2khz delay bandwidth, you aim for 700-800hz or so.  Because a single pole of lowpass filtering doesn't remove very much, you'll get a barely audible reduction in treble on the 2nd repeat, a little more on the 3rd, 4th, etc.  It is NOT reverb, but as delays go, it starts to sound a little more like natural reverb because of the way in which later repeats are progressively darker.

How to do this?  There will usually be a mixing resistor after the Repeats/Feedback pot.  If that mixing resistor was, say, 12k, you could split it into a 6k8 and 4k7 in series, and run a cap from their junction to ground.  If the "first" resistor was 6k8, running a 33nf cap to ground would provide one pole of lowpass filtering beginning around 710hz.  If the resistor order was 4k7/6k8, then you'd run a 47nf cap to ground from the junction, yielding a rolloff beginning around 720hz.

Again, NOT reverb, but at shorter delays the repeats really start to smush together in a way that begins to sound "reverbish".

[Wavelength]

Mark,

Thanks! I was actually thinking of using a PIC controllers with multiple clock outputs into dual XOR gates to create differential clocks to the BBD.

I think I will instead take my tube (6418) clean boost and use some of that with one of those small tanks and create something that way. That seems like a better idea and more in tune with what I am doing now.

Thanks,
Gordon

[Ripthorn]

Mark,

Thanks! I was actually thinking of using a PIC controllers with multiple clock outputs into dual XOR gates to create differential clocks to the BBD.

I think I will instead take my tube (6418) clean boost and use some of that with one of those small tanks and create something that way. That seems like a better idea and more in tune with what I am doing now.

Thanks,
Gordon

Do you have a schematic for your 6418 boost? I've got a bunch of them kicking around waiting for a project.

I did up a quad PT2399 reverb last year that turned out quite well, if I say so myself. Doing it all with BBD sounds expensive, which is the whole reason I did 4 PT2399's: whole build (including hardware) was cheaper than a Belton brick or an FV-1 on its own.