"new" oscilator core
Martin Czech
czech at Micronas.Com
Thu Oct 19 15:26:57 CEST 2000
Following the linear oscillator core discussion, I propose
this circuit (see attachment).
This is by no means a new circuit, perhaps this very structure may
be new.
The target is not cost reduction, but linear triangle performance
You can't take it literal, because e.g. the logic inverters will
have to match to their PMOST/NMOST operating voltage, all
these level problems, and, well ideal voltage sources do not
exist, and V1 and V2 have to be derived from a single voltage
in order to make sense:
Vb=Vref2-k*V
Va=-Vfre2+k*V
(Vref1 = - Vref2), k scaling constant, V ground referenced input
This means two additional opamps. Instead of a voltage input, a current from
expo source can be feed in also.
Possible problems:
-servo loops might get a large kick if the switches toggle
I'm not much afraid of that, if switching is much faster than
opamp, depends on what the opamp input stage will see
in the moment of transient
-switching may lead to strange "offset like" problems, or nonlinearity
the could be a serious problem and very difficult to fix,
ie. this would be the end of this idea then
-capacitive spikes coupling from switching signal
perhaps compensated due to opposite direction (inverters)
-the usuall offset voltage problems, though you can buy arround 100uV
today from the shelf. Typical max input voltage should be 1V,
so that's 100uV/1V ->0.01% error. Not that bad.
-resistor mismatch will give symmetry problems
easily reduced to 1% - 0.01% with from shelf metal resistors (belt)
and DMM
-reference voltage noise
-addtional computational errors introduced for deriving V1 and v2
from a single voltage referenced to ground
-cost (well, not really, 4 OP07 or 2 LF412)
I guess I will not try it, as I'm preparing to move house (appartment)
and then the usuall problem : no time.
Now I take my flameproof suit, sit back and wait...
m.c.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: osccore.jpg
Type: image/jpeg
Size: 25759 bytes
Desc: osccore.jpg
URL: <http://synth-diy.org/pipermail/synth-diy/attachments/20001019/472276c0/attachment.jpg>
More information about the Synth-diy
mailing list