Info requested on EMS Phase Frequency Shifter

[Dan Slater]

Sean Costello wrote:
> 
> Hi everybody:
> 
> Does anyone out there have any details whatsoever on the EMS Phase Frequency
> Shifter?  I know that it was also included in the Vocoder 5000.  I have only
> heard a little bit of tantalizing info - that the same circuitry is used
> both for phasing and for frequency shifting.  Anyone ever use one?  Better
> yet, anyone have schematics?  It would be cool to see if the circuitry is as
> weird as the other EMS gear I have seen or have heard about.
> 

	I have used the EMS phase / frequency shifter but unfortunately, don't
have any schematics. As I recall, the basic design is a conventional
Dome filter driving a standard dual analog multiplier / summer setup.
The interesting part of the design is the quadrature oscillator. It is a
discrete chip digital oscillator that uses weighted summations of Walsh
functions to produce the sin and cosine signals. This voltage controller
oscillator can go continuously from 1000 Hz down to about 1 cycle per 20
seconds or so. At the low speeds, the unit sonically acts more like a
phaser than frequency shifter but the configuration remains constant as
that of a frequency shifter. The phasing effect is due to the low
frequency shift value. If one were building this type of a frequency
shifter today, the oscillator would be probably much better implemented
with a numerically controlled oscillator chip such as those from
Qualcomm or other vendors.

	The design is quite clever and the unit sounds ok, but I personally
like other frequency shifters such as the Bode better. The EMS will not
do a through zero shift and the maximum shift is limited to 1 KHz. Also
at slow oscillator speeds, you can hear low pass filtered phase steps.
Internally, the unit consists of 2 PC boards, one is the quadrature
oscillator, the other is everything else.

> Thanks,
> 
> Sean Costello
> 
> P.S.  Anyone have Synthi Hi-Fli schematics out there?  I'd like to see how
> diodes are used for voltage-controlled phasing.
> 
> P.P.S.  Anyone out there know of any good digital frequency shifter
> techniques? I'd like to work on making Csound do this, but apparently the
> Hibert transform is difficult to implement in a successful manner digitally
> for this effect.

	Hilbert transforms can be implemented fairly straight forwardly in
digital computers but can be computationally expensive. Two common
methods are through a convolution or by performing a FFT, zeroing all
negative frequencies and then an IFFT. I have used the latter technique
for several aerospace microwave applications and at least there, the FFT
/ IFFT approach worked extremely well. 

Dan Slater (dslater at ix.netcom.com)