Homemade Reverb

[puretube]

anyways: I recommend an E-H Holy Grail for instant satisfaxion.

[amz-fx]

I would love to see a layout for up to six delay chips and some modular filtering blocks.

I think the way to go would be individual pc boards that have 3 delay chips on them - so you could go with 3, 6, or even 9+ -  and a separate analog input/output board.

One sticking point is how to control all of the delay times...  how about a sextuple pot?  More practical is a PIC and some digital pots!  You could store a large number of presets with the PIC method.

regards, Jack

[scratch]

seeing as the PT2399 uses a resistor value to set the delay time, couldn't you set each individual chip to a different value using a fixed resistor and a series LDR modulated by single LFO, could even invert the LFO signal on a couple of the LDR's so the delay varies in the opposte direction of the other delay chips? I've been thinking of trying this myself, but I've got a couple of  projects that's been kicking my a** for a long time, I don't want to have too many things in progress at the same time.

[A.S.P.]

] information lost...

[Mark Hammer]

"Instant satisfaction" is one thing, and I suppose a stompswitch and a "more" control are all some folks want or need or can handle.

For the rest of us, reverb is not an effect, but a matter of creating virtual spaces by playing psychoacoustic tricks.  To do that equires an option to tweak more parameters than simply "more".

As an outgrowth of a much earlier thread (where the heterodyning issue first reared its ugly head in our discussions), I started planning/imagining what a simple 3-BB device could accomplish by creative use of feedforward and feed-back (regen) mixing.  The very same thing can be accomplished with a Princeton or other comparable digital chip.  Because it will get crazy quickly, we'll call the 3 delay devices Tom, Dick, and Harry (apologies to anyone with the same name).

Consider...

Tom is set to delay A, Dick to delay B, and Harry to delay C.  The three are in series, so that the first thing out the far end of the chain is the sum of delay-times A+B+C.  If the outputs of A and B are tapped and fed ahead to a final output mixer, then we get repeats at time intervals A, A+B, and A+B+C.

So far, so good.  Now feed the outputs of A and C back to the beginning (A), and feed B back to the begining as well.  What comes out at the end is now a first set of iterations/repeats/copies at: A, A+B, and A+B+C, plus a second set of copies at delays of A+A (recirculated A tapped from A), A+A+B+C (recirculated A tapped from the final delay stage), ([A+B]+A+B+C) (the C tap fed back to A but re-tapped from B), A+B+A+B (the B tap fed back to A), and...well you get the picture.

My guess is that the heterodyning possibilities are likely similarly complex and infinite, but the point is that one doesn't HAVE to think exclusively in terms of feedforward for taps, and global feed-back/regen from end point back to beginning.  If signals are permitted to travel part of the way and return (and we haven't even taked about feeding C back to B), via simple mixing stages at each junction. it doesn't take a whole lot of delay devices to produce a stunning range of virtual spaces.  It may well take a stunning array of POTS, but it needn't take a huge area of delay devices.

[Peter Snowberg]

he, who defeats* two clocks, doesn`t fear four more...

(*= sorry - don`t know the accurate english expression...)

Perfect word to use in English, Ton. (at least theat is the word I would use )

Jack, I think you're on to something there with ePots and a microcontroller. If you could get high enough resolution ePots to counter the "zipper" noise, you could use the microcontroller to do modulation too. Here's an alternate.... the time setting resistor is an element in a current path, so why not just make a simple programmable current source for each with a 12 bit D/A. TI has lots of multi-channel D/As in little packages for just such control applications.

Scratch, Nice idea modulating with LDRs in multiple directions!

Nice post Mark!

Modulation of the delay times is a little tricky, but setting mix levels with ePots is a piece of cake. Lets see.... 4 ePots for mixing the dry and each delay for output, and another 9 ePots for mixing the feedback/forwards.... oh yikes.... 

Time to buy stock in ePot manufacturers!

That is only 13 ePots and I'll bet there are some killer reverb tones in there!

I think LDRs for setting the PT2399s might be the easiest, but it might take trimpots to adjust the zero point and range. The CdS would probably do a nice job as a LPF for the control voltage.

A layout with three delays and this mixing could easily be done for using n-cards in parallel to make even thicker reverbs as Jack mentions. Reverb cards? There are lots of card connector systems out there now.

[gez]

Just taking things up a notch in sophistication, how about this?

http://www.elc.co.uk/toy-33128

I've actually messed about with one of these years ago when a mate brought one back with him from a binge weekend in Blackpool (like a UK version of Las Vegas but with fish & chips and done on the cheap).  They actually work really well.  Would be interesting to wire up some transducers to one...

[gez]

Whilst doing a UK google for the above (typed in 'Echo Toy' at first but got links to 'marital aids') I came across these things (scroll half way down):

http://www.kcb.co.uk/shop2/index.html?lang=en-uk&target=d49.html

I'd love to play harp through one! 

OK, dumb ideas...

[soggybag]

The delay time of the PT2399 is set by a pot to ground. Why not use an LDR for all delay chips and use an LED tied to a pot to set the delay times? No extra chips just a few discreet components.

Not sure if this was what Peter was talking about?

[SeanCostello]

In reply to Mark's post:

If you simply add the outputs of the 3 delay lines together, and send them back to the input, you will only get the mixing of echos for a very low feedback value. Once the feedback value starts rising, you will get oscillation at some frequency, long before you get nice reverb times.

The key for this type of reverb is to use a unitary matrix to mix the signals together. Read Miller Puckette's online book for details.

Sean Costello

[Mark Hammer]

What if simple highpass filtering (and I mean REALLY simple, like the value of a series cap before the mix-level pot) is used to take out some of the risk-for-resonance content in any recirculated signal?  Presumably what gets fed forward merits more bandwidth at the low end than what gets fed backward, if the goal is eliminating ugly resonances.

[scratch]

as long as the delays are not multiples of each other (a la MN3011), say 3,7,9 ms AND if the feedback signal is always smaller then the initial input, then the chances of getting resonances is pretty small? Guesstimating ...

[puretube]

keyword: "dispersion"

[RedHouse]

Can ya post a link to info on ePots?

[Eb7+9]

Jack, how about ...

- an optional gate on the input of circuit for "ping" removal ... like in the Rocktron verb
- send/return loop somewhere for sticking a delay in there for verb/delay mixes ...  ?!

---

I got one of those toy mics GEz ... oddly, they all have this little plate you can punch out !

~jc

[SeanCostello]

as long as the delays are not multiples of each other (a la MN3011), say 3,7,9 ms AND if the feedback signal is always smaller then the initial input, then the chances of getting resonances is pretty small? Guesstimating ...

As the system feedback gain gets closer to |1.0|, the chances of some resonance having a value greater than |1.0| (i.e. self oscillation) in a multiple feedback system as Mark described becomes 100%. Trust me, this stuff was figured out a few decades ago by people trying to develop good sounding reverbs. There are ways of combining multiple delay lines so that you can feed them back on each other, but you can't just sum them together and feed them back - you need to have separate feedback paths for each parallel channel.

Sean Costello

[SeanCostello]

keyword: "dispersion"

For spring emulation (and the physics of plate reverbs), this is true. Realistically emulating dispersion eats up a lot of cycles. The key word I would look for is diffusion.

Sean Costello