Revisiting DIY Reverb Schematics

[SparkyMart]

After reading various threads for Box of Hall, BYOC, and Belton based circuits, I wanted to see if any more recent designs / schematics exist.

Things of note:

From pro sound article:  http://bobbyowsinski.blogspot.com/2010/04/secret-of-abbey-road-reverb.html#ixzz3hIzK6R3F
Regardless of whether the reverb was from one of Abbey Road's fine chambers or one of their plates, the secret is in the high and low-pass filters placed on the reverb send. These were set to roll-off everything below 600Hz and above 10kHz, since physical reverb tends to have trouble with low and very high end frequencies. A lot of low end on the verb tends to muddy everything up as well, with no audible benefits.

While there might have been a real necessity to use a bandpass filter set to those frequencies back in the day, rolling off everything below 600Hz and above 10kHz still works well today regardless of what kind of verb that you're using

I have seen only HPF and shelved HPF configurations used, some with the HPF freq corner set way too low (should be 600-700 Hz - the Belton circuit diagram serves a good guide re component values). 

No design that I know of configures the op amp ahead of the Belton brick as bandpass filter stage - not sure why since this only involves adding a cap across the feedback resistor.  Based on the Abbey Road Studios info, placing a 1 nF in parallel with the 22k commonly used would give a 7.2 kHz LPF corner freq.

Footswitch wiring may be straightforward for hard bypassing the reverb, but I wonder if there's an alternate way of doing this, since the dry version of the input is available.  Guess there likely will be a change of signal level either way, so perhaps simpler is the answer.

Kudos to whoever came up with the voltage follower stage for a buffered VREF since there is otherwise an unused op amp going to waste. 

If anyone can point me in the direction of up-to-date, debugged schematics that may address some of the above, please let me know.  Otherwise, may go at this as a 'clean up' proj of sorts.

[Mark Hammer]

Both of those upper and lower bandwidth parameters are pretty easy to address.

In the example below, IC1A serves as a buffer and IC1B "conditions" the signal for feeding the Belton unit.  As shown, there is no treble rolloff, but the bass rolloff, set by R2/C2, is 1064hz; a bit higher than your spec from the article.  If you wanted a 600hz rolloff, C2 would need to be .012uf.  A (roughly) 10khz rolloff would be provided by putting a 390pf cap in parallel with R3.

The R2/C2 pair only provides a single-pole rolloff (6db/oct).  A steeper filtering of bass could be provided by sticking a 6n8 cap between each of the Reverb-level pots and R5/R11.  This would repeat the 600hz rolloff.  Since the Belton is an electronic, rather than mechanical device, my sense is that the impact of pre-and-post filtering is different than it would be with a spring or plate; namely adding up to the same thing.

[R.G.]

There are multiple references to this in resources from the Paper Era. The one I'm most familiar with is the info on reverb in the National Semiconductor Audio Handbook from 1976? 1977? Which advocates both high and low frequency rolloffs, as well as some info on just how to drive the inductive coils of reverb tanks. Anderton's  Stage Center reverb has some of it built in.  I don't have a good handle on an add-on circuit to do the limiting. It's down in the class of circuits I'd call "glue". I remember some references about using high and low filtering as making reverb sound more like what you get out of a real reflective chamber.

There's an issue lurking in what you're thinking, I believe. A bandpass filter circuit isn't really suited to the kind of thing you're describing. Bandpasses all integrated as single multiple-order networks are really single frequency resonances with widths that are 1/Q. The half-power points fall out of the math, and aren't really chose-able separately.

You need a band limiting filter, which is really one high pass and one low pass where you can pick each separately. You're looking at filters much more like speaker crossover filters. They're "band-pass" filters, but different from the second- and fourth-order real "bandpass". For this use, you want separate high and low filters.

I am not sure I understood all you were saying. I'm just guessing at it.

[PRR]

Bass reverb IS important to fine concert-hall sound. Most of the great halls have rising reverb in bass. This makes music more warm and solid. Some of the disappointment in post-1930s concert halls is changes in interior finish leading to poor bass reverb.

All the fake reverbs are too small. Even the fine rooms under Capitol and EMI are not large enough for multiple bass waves. Acoustic lines trade-off size and loss, accepting crummy bass. The Hammond spring goes boingg. I have built PC-based reverb settings with useful bass, but the mini-chips won't come close.

And mostly we don't care. (Also with electric bass we have plenty of rumble in the room.) So we pick a bass-cut. This will depend a lot on the technology. The Capitol rooms can be run a little deeper than the smaller EMI rooms. Hammond used ~300Hz bass-cut; Fender used 500-600Hz for springs. Bi-Amp had a heavy-EQ spring which attempted to run to 100Hz; IMHO it was rarely useful that low (you had a 5-knob EQ so you could pick your mud-up point).

Most reverbs have treble problems. Sizzley, ringy. Rooms usually need a few hanging towels (or a high-cut). The Hammond goes into a series of resonances. The low-price digital delays sound awful without a hard high-cut.

So leave places for caps and tune by ear. It isn't any simple theory.

[Mark Hammer]

The problem with springs is that they are mechanical devices that resonate, and resonate differently depending on how hard they are pushed/jiggled.  That is a major source of their unique appeal.  But it is also a source of why they don't/can't sound like the concert halls Paul describes.  If anything, feeding reverb pans with too much bass ends up generating irritating midrange, since the spring is set in motion by it and cannot actually reproduce that bass accurately.

On related matters, I finally dug up the 1983 ETI reverb-project article I was looking for that incorporates a variable optical limiter in the driver stage.  It has a balanced input and output (depending on use of mono or stereo plug), NO means for mixing dry and wet (presumably this is a semi-pro outboard unit to use with a mixing board), and a "Decay" control that provides feedback from the post-pan output stage to the pre-pan driver stage.  What Fender has traditionally called a Dwell control simply adjusts how hard the springs are pushed, resulting in faster and longer decay. This unit doesn't rely solely on the initial jiggle to make the spring vibrate longer.  It actively prolongs the vibration.

Hopefully, I can scan this at work tomorrow, and post it.  It includes a PCB pattern/layout, which you'll need to paste into a graphics package and create a mirror-image, since it predates the toner-transfer era by a wide margin.

The author (Bill Markwick) writes: "How does it sound?  Well, it won't sound like an EMT plate or an AKG spring.  There are limits to the El Cheapo route, you know.  On the other hand, it sounds remarkably better than the average spring line that you find in small mixers and instrument amps."

[Mark Hammer]

Found and scanned.  I need to do a bit more digital retouching of the PCB mask and will post later this evening.

[samhay]

>If anyone can point me in the direction of up-to-date, debugged schematics that may address some of the above, please let me know.  Otherwise, may go at this as a 'clean up' proj of sorts.

The basic Belton/BTDR application note has been pretty thoroughly debugged in the sense that this is what most commercial and DIY BTDR designs are closely based on, and they both seem to work and be quite popular.

As far as playing with the pre- and post- filtering of DIGITAL reverb, I think some of the 60's folklore should be taken with a grain of salt - I think that is what others are saying.
A BTDR, breadboard, and a few spare hours aught to clear that up.

I have had a fairly thorough play with the BTDR-2 chips. In my experience, you can significantly lower the HP filtering into the brick without it getting muddy*. This is in the context of a guitar - which is naturally rather bandpass limited - and I don't know how well this will work for a bass or orchestra. LP filtering at the output of the brick seems to be the most useful - to me - tone control, and this is what most designs with a tone control have implemented.

*My SaT reverb has 1-pole ~30 Hz HP and 3-pole ~7 kHz LP filtering before the brick and sounds pretty neat: http://www.diystompboxes.com/smfforum/index.php?topic=107226.0

[Mark Hammer]

As promised, here is the reverb project from ETI.  Destined to be a favourite! 

I flipped the PCB patterns around so that they are now suitable to print off for toner transfer.  Note that a number of the traces were digitally altered to replace diagonals and curves, that came out jagged during the scanning, with straight lines and right angles.  I think all the parts placement legending is legible.  If not, pop me a note and I'll check the original.

Page 1  http://img.photobucket.com/albums/v474/mhammer/Reverb1_zpsqsjyqgdh.jpg
Page 2  http://img.photobucket.com/albums/v474/mhammer/reverb2_zpsthfuueft.jpg
Page 3  http://img.photobucket.com/albums/v474/mhammer/reverb3_zps44kot7mr.jpg
Page 4  http://img.photobucket.com/albums/v474/mhammer/reverb4_zpslbgwb1ee.jpg
Large-size PCB mask for main board  http://img.photobucket.com/albums/v474/mhammer/ReverbPCB_zps4gl58x0a.png

[PRR]

> the reverb project from ETI.

Thanks. Hate PB ad-storm, so pulled them out to a pair of PDFs on a calmer host. Sorry for the really grotesque folder URL:

https://drive.google.com/open?id=0B3Lt-dOXyP3-fnpPR21IbE1lekI3V09TalVVMDJ4a19JYi1tbTlCRzliLTlVbmVxZ1pKMFk

ETI-reverb.pdf ------------ 2.3MB PDF file, 4 pages
ETI-reverb-PCB.pdf -------- 600KB PDF file, PCB mask alone, large scale
ReverbPCB_zps4gl58x0a.png ---------- 400KB PNG file, PCB mask alone

The "Large-size PCB mask for main board" file seems to be scaled for yard-long paper. I tole the PDF generator to do it that way. Many folks will find the PNG file more handy?

[noisette]

Thanks, neat circuit, very interesting (differential?) driver circuit, but isn´t the return path inverting because of Q7??
What benefit to use transistors for a high gain recovery stage?

Questions, questions

I´m using spring reverb a lot and find a series hp/lp with individual cutoffs very handy for tailoring reverb to the dry signal (like in the drawmer noise gate, which is cool for reverb anyway!)

[Mark Hammer]

Thanks, Paul.  Appreciated.

I kept the resolution of the PCB layout higher to give folks a choice.  I figured the text and schems came out legible enough at lower res.

The article notes that a CLM6000 could work in place of the unit originally specified, which suggests a variety of suitable replacements, like the NSL-32, et all.

[PRR]

> isn´t the return path inverting because of Q7??

What is the phase of a signal delayed 38mS multiple times? IMHO, 'random' for all practical purpose.

Q7 *and* Q8 is a follower (gain of 30).

The differential drive connections are not marked. If you think it inverts, flip 'em.

> What benefit to use transistors for a high gain recovery stage?

Possibly lower hiss than '741-type opamps. (That recovery amp is still hard to beat for ~~~20K sources.)

[noisette]

What is the phase of a signal delayed 38mS multiple times? IMHO, 'random' for all practical purpose.

Q7 *and* Q8 is a follower (gain of 30).

Ah, I see... I am thinking about a spring reverb circuit for use with a synth.
And thanks for PDFing too!