MOTM

It seems my comment and Paul's response about getting the FFB out has 
caused a flurry of FFB orders. [ Paul: Feel free to send me ten 
percent! :-) ]  Since I'm sure people will want to make informed 
decisions, I've taken the liberty of reproducing my review of Paul's 
FFB, which appeared (with the format all mixed up) in AH last year.  
Paul's answer is also reproduced here, so be sure to read that as 
well.

Mind you, I don't think FFB's make much sense nowadays. They were 
created in Moog's day because a parametric FFB would have been too 
costly and impractical.  But people still seem to want them, so 
hopefully this detailed review will help.

Johnm


MOTM-450 Review:


Yes, I know: the MOTM-450 Fixed Filter Bank from Synthesis Technology 
isn't shipping yet. But it will be shipping ere long, and what good 
is reading a review after you've bought the damn thing? Since I had 
the good fortune to test the prototype and am in a mood to dish, 
perhaps you will humor me and actually read this review. But before I 
grace the world with the opinion it so deserves, let all be reminded 
that my synth of choice is Synthesizers.com and my only MOTM module 
thus far is a sub-octave mux (love it). Fortunately, the two systems 
use identical power supplies and are pretty much compatible, so I was 
able to test the FFB with no problems.

I have heretofore reviewed the dotcom FFB, so we won't rehash that 
again. However, some comparisons will be made. How could I not, what 
with two Texas FFB's sitting side-by-side begging for a shootout? So, 
armed with my trusty Scopemeter and my trustier ears, I set about 
putting the 450 through its paces.

SOME BACKGROUND

The MOTM-450 follows the venerated Moog 907A filter bank in that it 
includes an LP filter, an HP filter and eight bandpass filters. But 
whereas the Moog FFB used BP filters based on an LRC topology 
(inductance, resistance, capacitance), the MOTM FFB uses FDNR filters 
(frequency dependent negative reactance). This is the way to go if 
you want to emulate an LRC filter without using inductors, which are 
expensive and bulky and a general bitch to deal with nowadays. 
How much the MOTM filter sounds like Moog, I don't know. I'm not even 
sure that's important. What is important is that the FFB does what it 
sets out to do and does it well--namely, providing a choice of fixed 
frequency bands for spectrum enhancement and formant synthesis.

Controls on the MOTM-450 include a Wet/Dry mix pot and a 
manual/remote bypass switch with LED.

THE GOOD STUFF

When all of the controls on an FFB are set to zero, you should hear 
nothing, ideally. I guess that makes the 450 ideal in that regard. 
With all controls set at minimum and the mix set to wet, there was no 
detectable output, even with my amplifier cranked as loud as I dare. 
Bugbears such as bleedthrough, hiss, and line noise must have been 
out having coffee or something because hey sure weren't at the 
output. The filter was also nicely resonant, lending an excellent 
instrumental quality to otherwise electronic sonorities. 

I didn't take a measurement, but my ear detected no attenuation of 
the filtered signal, other than what would be expected from the lack 
of overall spectra. What you dial is what you get, and at the output 
level you expect.

The mix control was nice, but I hardly used it, given that my 
interest was in purely filtered signals. The LED was also a nice 
touch, as I like knowing at a glance whether something is active or 
not.

The LP and HP sections performed as expected. The HP section gave 
particularly crisp results.

THE BAD STUFF (not all that bad, really)

The module was a real power hog. For some bizarre reason I neglected 
to measure the amperage, but suffice it to say that when I hooked the 
FFB up "hot" it created a noticeable spark at the connector. I 
actually had to disconnect six of the eight dotcom VCOs just to run 
the thing. 

The .com oscillators output a 10v p-p signal, which caused clipping 
in the MOTM FFB unless attenuated. The LP section was particularly 
prone to clipping and would not handle a signal greater than about 3V 
p-p without distortion. The max input for the remaining sections was 
about 3.5V p-p, beyond which was obvious clipping. This deficiency 
was not serious, however, because a 3V p-p signal still represents a 
lot of juice. If the filter were noisy, this would have been more of 
a problem because SNR would be lessened. But given that the noise is 
essentially zero, having to deal with lower than normal signal levels 
(by .com standards) really wasn't an impediment.

My ear told me that some of the MOTM FFB center frequencies were off 
the mark, and measurements with a frequency counter confirmed this. 
At a target center frequency of 350 Hz, for example, the dotcom FFB 
resonated at 353 Hz. The MOTM FFB weighed in at 422 Hz. At a target 
of 700 Hz, the dotcom was resonant at 701 Hz (!). The MOTM FFB was 
resonant at 802 Hz. Paul had some explanation as to why the MOTM FFB 
was so far off the mark, but the brain cell with that memory is now 
dead. He did say that the frequencies in the production version 
would probably be changed to reflect the center frequencies called 
out on the panel. I hope that change was made because not to do so 
would impact the usefulness of the FFB. To get the primary formant 
for an oboe or bassoon, for example, you need to be able to dial up 
500 Hz. Any other value won't quite cut it. 

One more thing: I think the switched bypass should be attenuated, so 
as not to be so jarring when switched from filtered to non-filtered 
output. But this is a minor nit that can be worked around easily 
enough.

SHOOTOUT TIME!

So how did the MOTM FFB stack up against the dotcom FFB?

The MOTM FFB offers bands that are more resonant than what the .com 
FFB offers, but the difference--while noticeable--was not dramatic. A 
sawtooth processed at 1 kHz with a touch of reverb sounded virtually 
identical on both systems (when the .com FFB Aid module was used).

The MOTM FFB offers 7 bands only, while the .com offers 12. From the 
standpoint of synthesis, this isn't a major advantage for the .com 
because the ear is less sensitive to the higher and lower frequencies 
offered. All of the "business" frequencies are covered on the MOTM 
FFB, and the HP and LP filters can be used to take care of the rest. 

The dotcom FFB costs less than the MOTM FFB and the center 
frequencies called out on the panel were far more accurate than on 
the MOTM FFB (see above).  However, my dotcom FFB has a serious 
bleedthrough problem that spoils the filtering effect. Roger Arrick 
claims to have a fix for this bug, but I haven't gotten around to 
taking him up on it. Since I don't know how quiet the dotcom FFB 
would be with the patch, I can't make a comparison on the 
bleedthrough. But I doubt that ANY design could match the low noise 
floor of the MOTM FFB.

Interestingly, the MOTM FFB reminded me a lot of the Doepfer FFB, 
which was also nicely resonant. But while the Doepfer FFB offers far 
more ranges and costs much less than the MOTM FFB, it is rather noisy-
making the MOTM FFB a better choice, even at the higher price.

CONCLUSION

The MOTM FFB is expensive and offers fewer bands than other designs, 
but what it does it does quite well. Fans of "CD quality" analog 
circuitry will be sent into an ecstasy at the astonishing quietude of 
the 450. This is an excellent filter bank that MOTM users with some 
cash to spend will certainly want. 

Konkuro Rating is K:4 (on a scale of 1 to 5)

johnm

*************************************************************

Just a few comments on John's through review:

a) the filter bank does use more power than any other analog MOTM 
module, around 180ma. The average MOTM module draws about 25ma. I 
could use lower power quad op amps, but they are 4X the cost of the 
ones in there now.

b) the prototype tested is actually 'missing' 1 frequency band at 
2100Hz. I ran out of pc board space :) But it will be included on the 
production module.

c) about the filter band labels: they are the same as the Moog 907, 
BUT....it turns out that Moog didn't use the *center frequency* for 
his filter labels, he use the *lower 3dB point* on the bandpass. I 
didn't realize this until too late. So, the production version will 
tweak the capacitor/resistor values to shift all the filters down 
slightly to match the center frequency with the label. This is how 
the .com and Doepfer did it, and this is what users expect. I will do 
likewise.

d) the prototype gain is indeed set wrong for full 10V pk-pk VCO 
driving signals. I was using my bench test signal generator and Audio 
Precision, both default set to 1V Pk-Pk. This is a 1 resistor tweak.

Paul S.