opinions on HIFLI - Diode based phasing ?

[Haible Juergen]

Hi!

Have you ever seen a phaser that used ordinary diodes as voltage 
controlled resistors?
My first thought would be. "cheap design, bad SNR", but maybe it's not 
as simple as
that.
The famous EMS HIFLI (for which people pay insane prices at the 
moment) obviously
uses dides as voltage controlled resistors, but not only one or two 
per stage. There are
20 (twenty !) diodes connected in series for each stage, making a 
total of 240 diodes
in the 12-stage phaser secton. Insane ?
There must have been some reasons to do it that way. The LFO and ramp 
generators
in the HIFLI are full of CA3080's, so it would have been likely to use 
them for the
all pass stages as well. Can it be that the diode version is superior 
to an OTA design ??
What I can *imagine* (but far from being sure) is that 20 diodes in 
series lift the
maximum signal level (for tolerable amount of distortion) out of the 
opamp's
noise floor somehow.
And maybe using a large amount of diodes gets rid of doing selections 
on the diodes,
as there would be some averaging of tolerances across 20 diodes in 
series.
(The phaser is organized in 6 stages x2. The diode strings of 6 stages 
are directly
connected in parallel, each battery driven by an opamp buffered CV at 
the top end,
and by an opamp inverter at he low end.)

I am looking for opinions from two different sides:
(1) What do you electronic experts (Don, Joachim, Gene, synth-DIY 
members, ...) think of this
circuit concept? Would there really be some advantages in SNR, 
compared with
a OTA design ? I'd like to hear some opinions before I start soldering 
a whole
drawer of 1n4148's onto a large veroboard, for some reason  (;->)

(2) 	Are there any *users* of the HIFLI out there ? I know it's ultra 
rare, but as
I haven't heard one myself, other than on records, I'd like to know 
how it sounds.
Is the phasing *clean* (in therms of Noise and distortion), or is it 
just an interesting
"dirty" distorted phasing sound ??
I'd also like to know about the tracking of the suboctave generator. 
Does it
work as intended, or are there tracking problems?

Any input -to these questions, or general comments about the HIFLI, 
is
highly welcome!

JH.