amp or pedal reverb, not spring tank

[jsantana]

Hi everybody. Do you know about, or have you even build, a reverb circuit not based in tank spring?. I have seen one in Mark Hammer's page, based in MN3007 and MN3101. Does anybody build it? (have you done it Mark?). How it sounds?. Is it really a reverb or delay effect?.
Thanks

[Arn C.]

http://hammer.ampage.org/files/reverb_sd.gif

This is the one I tried but it did not work for me.  Maybe it is okay and I did something wrong.  Maybe something is wrong with the schematic.  Maybe Mark Hammer can tell us if it works or someone else who built it.

Arn C.

[Mark Hammer]

It will not work as shown.  There are some errors in the drawing and enough missing information that if you aren't very familiar with these things you'll spend months trying to troubleshoot it because of what it *doesn't* show.  That's probably one of the reasons why Tom Gamble took it down.

Sorry.

If you do not already have the MN3007/MN3101 chip combination, I would recommend trying to build one of the PT2399-based delays that are posted at different sites.  The chip itself is inexpensive, and the design are tested and verified by many people here.  The sound quality will be as good as the all analog type.

Unfortunately, such delay-based devices will NOT sound much like a reverb tank.  Real reverb has many many different reflections, and a single fixed delay, whether produced by analog or digital means, cannot duplicate that.  You CAN duplicate it electronically but it will require considerably more complex delay circuitry than is probably worth it.  In that way, springs may *seem* big and bulky and complicated, but the duplicate the multiple different sound reflections quite well.  I am hoping to finally finish an inexpensive homemade mini-spring-reverb project that will fit in an 8" by 4" box and post it, but you probably shouldn't force yourself to wait that long.

The good news is that analog delays can be made to sound a bit more like reverb by a few simple mods.  

Normally, analog and digital delays are designed under the assumption that if you go for long delays you want the repeat to sound like the original, and if you go for multiple repeats you want them to sound like the original too.  So, there is usually little or no filtering in the feedback/recirculation loop, in order to maintain signal "quality".  Unfortunately, if you go for lots of recirculation and a short delay to mimic reverb, you get a very boxey resonant sound (with the resonant frequency corresponding to delay time), and the pseudo reflections are a little too "in your face" to feel like reverb.  More importantly, the repeats distract considerably from the main signal.

If you provide some simple lowpass filtering to the recirculated signal, though, each repeat gets progressively more filtered so that the sound gets more mute as it dies out, not entirely unlike what happens to high end in a reverberant space.  Doing this is as simple as finding the Repeat/Recirculate pot, and tacking on a medium value cap (e.g., .01uf to .47uf depending on the pot value and other circuit features) from either the input lug or wiper of that pot to ground.  This will provide VERY simple (6db/octave) high-cut of the recirculated signal.  It will have NO effect on the first repeat, but each additional repeat will be filtered yet once more, so a repeat of a repeat will be filtered twice as steeply.

You will find this makes attempts at reverb with a simple single fixed delay MUCH more livable.  If you want to get fancy, you can use a toggle to select a few different filter rolloffs for different sorts of reverberant "spaces".  You'll also find that you can crank up the recirculation more without it being a distraction, and that it will sound less box-ey.

Example.....

Consider Scott Swartz's AD-3208 (http://www.generalguitargadgets.com/v2/diagrams/ad3208_schematic.pdf).  If you rotate the schematic so that it reads from left to right, over on the far right you can see two 50k pots tapping off the output of the SA571.  One sets the overall delay level in the mix, and the other sets the amount of delay signal recirculated back to the compander (SA571) input.  There is a 1uf DC-blocking cap in that recirculation path to keep nasty DC bias errors from being magnified and screwing up the works, but that is the only filtering in the recirculation/feedback loop.  

Since sticking anything onto the input of that 50k feedback pot would also affect the delay level pot signal too, we don't want to stick the treble cut feature there.  A better place for it is after the pot.  You can see the signal goes back through a 100k fixed resistor to the SA571, and there is where we'll have fun.  That 100k fixed resistor can be replaced with a pair of 47k fixed resistors in series - not exactly 100k, but close enough.  If you run a .0033uf (3n3 or 3300pf) cap to ground from the junction of those two resistors, you will roll off frequencies from around 1khz and up at 6db/octave.  Switch it to .0039uf and that rolloff point changes to about 870hz (using Freq = 1/[2pi*R*C]).  A 4n7 cap gets you a rolloff at 720hz.  Somewhere in the 700-900hz zone is probably about right.

Although normally "better" design consists of more selective and steeper filters, in this case the shallowness of the filter works to our advantage because you want to still have something left to filter after the first 2 iterations.  As well, since the filter slope is pretty shallow, you still get at least *some* high end in the first repeat; not "crisp" but not unlike what you'd get in the real world from a nearby hard, flat reflecting surface.  Anything that comes reflected back to you from farther away will be considerably muted and in this case the 4th repeat will be much more mute than the 2nd, and the 6th repeat even moreso.  The result is that even with the feedback cranked, it still comes out as a kind of blurry ambient wash, which is what you want reverb to be sometimes.

Alternatively, you can tinker with the low end by playing with that value of the 1uf DC-blocking cap.  Drop it to, say, .1uf (100n), and you may start to hear successive repeats getting a bit "sharper" as it chops off more of the bottom in the same manner.  This is a very common effect in dub music.

With a 3-position toggle to select one of two lowcut settings (plus no lowcut) and one of two highcut settings (plus no highcut), you can get a remarkable range of ambiences out of a simple analog delay like the AD3208.  Again, not a Lexicon, an EMT plate, or a 6-spring unit, but one heck of an improvement over what an analog delay normally does when you try to fake reverb.

[jsantana]

Thanks Arn and Mark.
Marck, incredible and delightful theoric explanation. I have made a PT80 delay and I will try the mods that you said, what seems to be very interesting.
Thanks