[sdiy] VCO compensation suggestion

[ASSI]

On Sunday 02 February 2003 23:07, Ian Fritz wrote:
> It consists of two cascaded OTAs.  It seems to me that if the first
> OTA is hooked up to include the linearizing diodes and the second is
> not, then the composite voltage gain is proportional to temperature. 
> If this is correct, then there is no need to build any additional
> circuitry such as the bandgap reference recently discussed.

That is correct, but there's no free lunch in it.

A non-linearized OTA already is almost a PTAT circuit, only that you 
look into the wrong end and get PTIAT (proportional to the inverse of 
absolute temperature). So if you servo the input of an OTA so that the 
output current is constant, the input voltage is PTAT. Which is what 
the circuit you propose does, provided you can supply a non-zero input 
to the first OTA that does not vary with temperature. The temperature 
dependence of the tail currents fortunately cancels (as long as both 
OTA are at the same temperature of course), so you can be a bit lazy 
about these as long as their ratio is constant. A more serious concern 
is that the output voltage should probably be kept below 30..100mV and 
that you cannot load the output, which requires a high-impedance input 
OPA with quite some gain (temperature independent and with no offset 
please) to actually get the signal you want out of the circuit.

The above is valid only if the current mirrors are perfect. In reality 
they aren't, but I'd rather not try to unravel these. I would either 
fall asleep before I'm done or make some bad mistakes (if the above 
does not not contain any already :-). I also haven't looked into 
frequency compensation, which is quite likely necessary for stability 
as we now have a feedback loop to worry about.


Achim.
-- +<[ Q+ & WAVE#46 & microQkb Omega sonic heaven ]>+ --

ForumTwo Q and microQ Soundset
http://homepages.compuserve.de/Stromeko/ForumTwo_Q.zip
http://homepages.compuserve.de/Stromeko/ForumTwo_uQ.zip