OT(?):Current Controlled Conveyors (CCCII) In Active Filter Design... Any Ideas?

[lovekraft0]

I've been seeing a bunch of new theoretical active filter architectures being implemented with CCCIIs in academic literature - anybody got any ideas on how to realize these (CCCIIS, that is) with available OTAS (13600/700 or similar), or even wih discretes? My math skills aren't exactly stellar, but these look like a great, if not exactly simple, way to make current-tunable universal filters. Probably more interesting to Synth builders, but there should be a few stompbox apps - any interest?

This paper gives one example, if interested.

[puretube]

fig.3 shows a discrete example...

[Sir H C]

You are not going to make figure 3 in the discrete world.  That is a pure matching circuit, and I would hate to be the one trying to match all those devices.  Even with a healthy amount of degen all around, it would be pretty hard to get successful.  I am not up on current controlled conveyers, know about the current conveyers where you have a voltage and a current in but not all nodes as currents.

[lovekraft0]

You are not going to make figure 3 in the discrete world.  That is a pure matching circuit, and I would hate to be the one trying to match all those devices. 

In practical terms, any idea how close is close enough? Is this going to be like matching JFETs for a phaser, or will it require a lot closer tolerances to get it working? Obviously, the idea is to implement it on a chip with matched devices, but I've got a couple of hundred 2N4401s and 2N4403s I could sort through, unless you think it's a futile effort.
It should also be obvious by now that I'm no EE, just another hack with a calculator and a soldering iron, so if this is totally impractical (short of having chips fabbed), save me some grief - I was just totally in love with the idea of a building block for easy electrically tuneable circuits (like gyrators, filters and such). I simply thought it might be slightly more interesting than what Stephen (Giles) refers to as another "...sodding fuzz box..." - apparently I was mistaken. 

[Sir H C]

First I would have to say that they are showing a sloppy "equivalent circuit" schematic.  No base current compensation and some other things (cascodes, degen resistors et al).  Looking at the circuit, matching is key to its operation, you will get dc offsets with things not matched.  Some of those multi-transistors in a DIP package might get you in the right direction (they are usually pretty well matched as are the devices in the inverters hence their use in phasers later on).

Also is missing a connection from the bases of Q3 and Q4 to the collector of Q11.  Otherwise the circuit will not work.

Other things, Y and X are relatively low impedance and have to be able to source/sink the whole current of Io.  I guess if those are just internal, that is not an issue.  Still looks like something cool to play with, might have to look at it later.

[lovekraft0]

Thanks - I had discovered the missing Q3/4 base connection by comparing this to a first generation (?) current conveyor. I put the (corrected) schematic together in my Circuitmaker simulator, configured as an amplifier (X grounded, Y  as input,  Z output, in parallel with a grounded resistor - Io current controls gain), and it works, although even with the (apparently) matched models for the 2n4401 and 2N4403, there's a fairly large DC offset at the output. If I get time, I may try to throw one together out of a random mix of the transistors I have on hand - let me know if you decide to fool with this, I'd be interested to hear your results.

[puretube]

of course, you could also play with discrete OTAs, for fun... 

[Sir H C]

Without cascodes and base current compensation, there are going to be big offsets.  If you are doing an filter where you don't care much about the DC component this is not a problem.