[sdiy] reverb mad from multiple delays

Thu Oct 22 21:45:47 CEST 2009

 	<75A9EDDA-EB91-43AB-9F6B-B5DB325D862F at earthlink.net>

 	<908a4c120910211726v1026cc7bm92ce5635b5708374 at mail.gmail.com>

 <25ACA776-7DF7-468A-9363-6ABD7EAFA7AB at earthlink.net>
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0

(crossing my fingers and hoping my formatting doesn't come out super weird.=
..)

Most of the makers of early digital reverbs were notoriously secretive abou=
t their designs=2C for obvious reasons.=A0 I believe EMT even sanded off ch=
ip numbers.=A0 From what I've been able to gather=2C both classic and moder=
n digital reverbs are mostly based on some variation of the Schroeder rever=
berator: basically some network of allpass filters (not as in analog phase =
shifter=2C but more or less a feedback delay with flat frequency response) =
for "diffusion" or "density"=2C and comb filters (with adjustable lowpass f=
ilters in the feedback path) to form the body of the reverb.=A0 There's not=
 really any reason to stick to Schroeder's exact topology=2C but that's the=
 basic idea.=A0 More advanced designs might also include an IIR filter for =
"room type" or "early reflections" or something like that=2C and of course =
some equalizer/tone control stuff.=A0 The allpass filters increase the numb=
er of reflections and the comb filters determine the character of the rever=
b=2C whether it's metallic or fluttery or smooth or whatever.=A0 They're ca=
refully tuned to different lengths so that the overall network doesn't reso=
nate heavily at particular frequencies=2C but it's more of an art than a sc=
ience (though I guess something like MATLAB would be very helpful these day=
s).=A0 The outputs of the comb filters are panned right and left for stereo=
 reverb.

Specific implementations vary.=A0 Some designs are completely digital=2C us=
ing a dedicated multipler IC (I think the Quantec QRS does this) or a TTL-b=
ased coprocessor (as has been mentioned with the Lexicon 224).=A0 Others (l=
ike the ones by Ursa Major) used multiplexed DACs and ADCs with VCAs and fi=
lters to do the mixing and attenuation.=A0 In any case=2C the "algorithm" i=
s still mostly defined in software.

I used to wonder what the secret was that enabled the EMT 250 to sound bett=
er than a Quadraverb=2C or whatever.=A0 I guess it turns out that aside fro=
m the obvious (quality DACs) it's mostly a matter of careful fine tuning by=
 ear.=A0 The algorithms haven't really changed=2C and early reverbs didn't =
use any more allpass or comb filters than a basic modern DSP is capable of.=
=A0 Analog attenuation might be part of the special sound=2C but many fully=
 digital designs are just as highly regarded.=A0 Old designs also used low =
sample rates=2C with a bandwidth of maybe around 6 to 15 kHz=2C and primiti=
ve DACs=2C but I think this is more of a guiding principle than something t=
hat should actually be implemented: just filter out all of the treble.

More eccentric reverb designs include the Dynacord VRS23=2C which uses mult=
iple BBD chips to make an analog pseudo-reverb=2C and the Ursa Major SST-28=
2=2C which you can read about here: http://web.archive.org/web/200708090333=
10/http://www.theworld.com/~cmoore/SST282_History.htm

In terms of what you've got=2C the delays really need to be independently t=
unable to get a decent reverb.=A0 You could=2C however=2C use the 8 taps to=
 do something interesting.=A0 The Binson Echorec (old analog delay) uses 4 =
equally spaced taps and is capable of some effects that still sound pretty =
exotic.=A0 http://www.youtube.com/watch?v=3DApEdnkiEGOY

> From: mverbos at earthlink.net
> To: synth-diy at dropmix.xs4all.nl
> Subject: Re: [sdiy] reverb mad from multiple delays
> Date: Wed=2C 21 Oct 2009 20:47:00 -0400
>=20
> Thanks=2C I have that one.
>=20
> I'm looking for reverbs. The delay portion is something I have already =20
> gotten really deep into. It's the mixing=2C routing and delay times that =
=20
> I'm looking for.
>=20
> Mark
>=20
>=20
>=20
> On Oct 21=2C 2009=2C at 8:26 PM=2C Amos wrote:
>=20
>> I'll see if I can dig you up the schematic for the peavey 1310 -- it
>> uses a legion of DIP memory ICs and reads them out clocked by an
>> analog VCO with LFO modulation.  Purely a delay/flange in its original
>> configuration=2C but several shorter parallel chains like this with
>> independent clock oscillators would probably sound very lush.
>>
>> On 10/21/09=2C mark verbos <mverbos at earthlink.net> wrote:
>>> Hi there=2C
>>>
>>>
>>> I recently built a digital delay based on the Buchla 288. It has 8
>>> equally spaced delays=2C with individual outputs and sliders on the
>>> front panel. Some of the sounds I can get from this got me thinking
>>> about hardware reverb. I know that there was the 3011 BBD that was
>>> used to make reverb=2C but I think it could be done much better with
>>> digital techniques (of course I mean hardware digital=2C not the
>>> sensible DSP/micro solution =3B). There was a guy from Chile who did
>>> something like this with 6 PT2399 chips and it sounded pretty good.
>>> I'd like to do something like that=2C but with control over everything.
>>>
>>> Does anyone have experience doing this and/or have documents from the
>>> old studio units that did it? I'm thinking it would help immensely to
>>> see schematics for the EMT 250=2C Eventide SP2016=2C AMS RMX and whatev=
er
>>> other reverb units were being made before everything went software...
>>>
>>> Mark
>>> _______________________________________________
>>> Synth-diy mailing list
>>> Synth-diy at dropmix.xs4all.nl
>>> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
>>>
>>
>> --=20
>> Sent from my mobile device
>=20
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
 		 	   		 =20
_________________________________________________________________
Windows 7: It works the way you want. Learn more.
http://www.microsoft.com/Windows/windows-7/default.aspx?ocid=3DPID24727::T:=
WLMTAGL:ON:WL:en-US:WWL_WIN_evergreen2:102009=