Linear current source

Magnus Danielson e93_mda at snorke.it.kth.se
Thu Oct 24 21:35:00 CEST 1996


>   Fellow Scholars-
> 
>   (It's nice to see all this new traffic on the list - it's been so quiet
>   for so long)
> 
>   I have need (for several applications) to linearly control an OTA. Now,
>   the easiest way to control it and get a very large range of sweep seems
>   to be exponential - the PNP pair.
> 
>   The A of E has a "Good Circuit Idea" that is called "Howland Current
>   Source for driving current controled voltage-current devices(or something
>   like that)" - that's OTAs - right? So I built this, seeing a linear
>   relationship bewteen Vin and control current - but it didn't seem to do
>   anything when I hooked it up to an OTA. Do I need a resistor bewteen the
>   PNP and the OTA pin (I thought this might be the problem this morning -
>   so I didn't have time to try it)? Will this work for 3080/13600 type
>   OTA's where the control pin is at -V+.6?

Now, the Howland current source thing is certainly a way to accomplish the
current to the OTA, but probably would some kind of modification be necessary.

I suspect that the Howland current source you mentions have a drawback... it
will have an limited voltage range being allowed at it's output. This becomes
a problem since the current input pin of a typhical OTA (the 3080/3094 for
instance) will be to the base input of a NPN transistor with it's emitter
connected to the negative power rail. This means that the control input is
around 0.65V higher than the negative rail.

One possible way of solveing this problem is to use a current mirror with PNPs
up at the ground rail. Such a current mirror hooked to a to either a Howland
current source (inverted version of the one you have...) could be a solution
to the problem. Creating a current mirror with the resistor inside the OTA
I would say is not recommended, since it's hard to match the external resistor
with the internal one...

I don't like the idea of haveing an resistor doing the voltage jump since you
will jump so much in the current througth it (possibly several decades).

>   They say the "Howland current source" is quite imperfect in the body of
>   the chapter - but I guess it's OK to use for this application. The rest
>   of the current sources they talk about are ground referenced - so I guess
>   they wouldn't work for our usual OTAs.
> 
>   Does anybody have any good linear voltage->current source for OTA
>   circuits that can drive the OTA over it's full range?

Driveing OTAs has become a recent little problem of mine as well...

In the ASM-1 there is a op-amp and two diodes that drives an CA3080 with linear
control. I haven't put that schematic on the web.... yet!

Magnus





More information about the Synth-diy mailing list