Minimoog Filter Schematic Available (was Re: Design of VCFs)

[Magnus Danielson]

> Hi all,
> 
> I just posted the Minimoog filter to my web site, at
> http://www.seanet.com/~costello.  This is a 75K file, that should be saved
> and resampled to print out at about 9 1/2 inches by whatever, at 300 dpi.  I
> couldn't get it any smaller without losing all of the data.
> 
> This is the classic Moog ladder lowpass VCF.  It uses all discrete
> transistors, several pairs of which need to be matched for Vbe by hand
> (directions for this are also on my web page).  The advantage to this over
> designs that use pre-matched transistor arrays (Moog Prodigy, Source,
> Taurus, Rogue, MG-1 - hell, pretty much every Moog filter except the Modular
> 904A) is that all the parts for this filter can be found at Radio Shack, or
> whatever your local low quality electronics store happens to be.  All of the
> NPN transistor can be 2N3904, and the PNP can be replaced by the 2N3906. 

Don't forget that the two control transistors Q26 and Q28 should be matched to
be +/- 3 mV @ Ic = 20 mA (Not that it's a certain collector current). To find
a good match it's a good idea to by a smaller bunch of transistors.

Also, according to the schematics I have is it the control transistors Q26 and
Q28 that can be replaced.

Transistor	Should be	Can be
Q26		TIS93		2N3906
Q28		TIS92		2N3904

The ladder itself is built up of TIS97 transistors. There is no replacement
noted for those. Also, the lowest pair Q29 and Q30 should also a pair since
they form an discrete operational amplifier transistor pair.

The next four pairs (Q2 & Q3, Q10 & Q11, Q19 & Q20 and Q23 & Q24) are noted as
being TIS97S transistors (this is not found on the schematic but on the
replacement part list). There is no notification that these pairs should be
matched but I guess it woun't hurt.

Just so you know what should be there.

There is a pair of small schematics for transistor matching among the MiniMoog
schematics.

> P.S.  Does anyone out there have any opinions on the different resistor
> values used in the Moog filter ladder, and how they impact the sound?  I
> have a Realistic Moog MG-1, which uses 1K resistors for all of the resistors
> in the transistor ladder (instead of the 220/150/150/150/150/200 ohm values
> in the Minimoog).  The filter in this sounds GREAT - in my opinion, better
> than the filter in my Multimoog - but I can't figure out why.  Would the
> resistor values affect the cutoff range?  The amount of distortion?  Any
> ideas, anybody? 

Well, the resistor rope will create a set of voltages (around 8.0V, 6.5V, 5.0V,
3.5V and 2.0V) with quite low output impedance (compared to use of higher 
values). As the voltages inside the ladder change will the working point for
the transistor in the ladder change. Changing the values of these resistors
will change how the transistors react to these changes.

It's easy to fall into the beleif that the emitters of the transistor pairs
have fixed voltages since the resistors give fixed values and from those we
just remove those standard 0.75 V between base and emitter... but that is far
from how these transistors operate. If there where some fixed 0.75 V diffrence
there would the ladder not work at all. Recall that Ic depends on Vbe like 
this:

		     Vbe * q
	Ic = Is (exp(-------) - 1)
		      k * T

so this would make Vbe depends on Ic like this:

	      k * T    Ic
	Vbe = ----- ln(-- + 1)
	        q      Is

And when we sweeps the filter we change the standing current in the ladder.
This actually moves the peak value of Ic (the peak value of any Ic in the 
ladder
will be the Ic of the single transistor below the op-pair). Now when the 
capacitor sitting on the emitters of a pair decides for a voltage will the
transistors tune in for certain Vbe/Ic combination. We use this 
transconductance
to tune the filter. The resistors will set an offset resistance/
transconductance. Changing their values will change the ladder's cutoff
frequency for the same ladder current. However, it migth be of interest to
change the standing current as well in order to change the extreme upper 
operating point of the transistors. This can effect the signal amplitude range
of the ladder.

Please correct me if I am totallu of here, but I haven't finished figuring out
the ladder fully.

BTW. Have more people than me noted that the ladder filter is really a close
relative to a state-variable filter? If you hook in over another capacitor you
get diffrent filtering functions.

Cheers,
Magnus