Re: Hammond registrations as DX7 patches

["Bruce A. Wahler" <bwahler@tiac.net>]

Jersey Shore Baby wrote:
> 
> Hi all,
> 
> A while back a fellow musician (and Asbury Park, NJ policeman) told me
> that he had found a way to convert the different registrations to 
> patches on a DX7 by somehow manipulating the DX7's sine waves or some 
> such thing.  Unfortunately, because he was on duty at the time, he 
> couldn't give me a detailed explanation.
> 
> Has anybody heard of this, and if so, can you tell me how it's done?

I've done this on a number of synths, including the Yamaha FM-series.  
essentially, you use the fact that Hammond Tibia voices are very close 
to sine waves, and substitute sine waves -- in this case, Yamaha FM 
sines.  In most cases, the best FM algorithm to use is the one which has 
all six waves in parallel, which will allow combinations such as 
888800088 and 848808008, but not 848848448 or 888888888.  your results 
will depend on your access to a real Hammond (or a good clone like the 
XB-2 or V3), and how picky you are about the little details.

The hardest part of the mess is finding an accurate calibration of the 
levels.  The upper drawbars -- and TG's themselves -- roll off a little, 
so creating 888000008 by setting all four sines to "99" will not really 
give you the same sound as on a Hammond.  Also drawbars provide a smooth 
AURAL transition from off to full-on, but the DX-7 and bretheren provide 
a smooth VOLTAGE transistion for 0 to 99.   While "99" may correspond to 
a drawbar setting of 8, "75" may not be a perfect setting of 6, etc.

In addition, while drawbar dropout (like a spinet has) can be simulated 
easily, true foldback (like a console Hammond) is difficult to mimic:  
it requires using fewer "virtual drawbars" and using multiple copies of 
the folded-back voices, set octaves apart.  While some of the newer 
synths overcome this limitation by using "active voice allocation" 
(waveform playback elements are kept in a pool and doled out as needed), 
the DX-7 is strictly 6 sines/key x 16 simultaneous keys.

Key click can be simulated by a sharp attack on each sine, followed by a 
quick decay to a somewhat lower level. (R1=99 L1=99, R2=99 L2=85, R3=0 
L3=85, etc.)  Percussion requires setting the proper R2 (about 1 sec./4 
sec. for fast/slow) and L2, L3, etc. to 0.  Vibrato, distortion, etc. 
are a bit harder to realistically simulate; I'd recommend using a good 
EFX unit and a tube preamp, respectively.  Leslie?  Forget trying to 
simulate it on the DX-7; use the real thing, or at least a good 
electronic simulator of EFX unit.

[To avert the flames:

This method WILL NOT provide a sound which will pass a blind listening 
test with a real Hammond.

It is, however, better than no Hammond sound at all, and a damn sight 
better than the sampled organ sounds available in most modern synths, or 
a filtered-square-wave approximation on an analog synth.]

I have begun a section on my web page to discuss this method in greater 
detail at:

        http://www.tiac.net/users/bwahler/organ/alt_hamm.html

I haven't gotten too far with it; the urgent need disappeared after I 
acquired my V3 and Leslie 251.  :-)  If there seems to be a bit of 
interest in the topic, maybe I can hurry it along a bit.

-BW