[sdiy] More OTA meas.

[ASSI]

On Mittwoch, 25. Mai 2005 00:42, Ian Fritz wrote:
> First I looked at another 13600 chip and verified the 9% nonlinearity
> for Iabc of .25 mA and .5 mA. 

When you measure the linearity, do you keep a constant Iabc and then 
vary the input or do you keep a constant input and vary the Iabc? If 
the Iabc is constant, then no error in the Iabc mirroring should show 
up in the result, so I guess you compared the output for constant input 
at the two different Iabc settings. What's the supply voltages, common 
mode voltage and differential input voltage for the inputs?

> Then I tried the idea of tying the 
> buffer input to the (-) rail to keep it from drawing bias current. 
> This had no effect whatsoever on the transconductance.

You can't cut off the suspected "current steal" that way (unless the 
circuit shown in the data sheet is completely phony). The Wilson 
mirror behaves correctly (i.e. all base currents cancel) if the Q2 
transistor is 3/2 the area of the other transistors (the mirror ratio 
into the Darlington is not exactly one). I would guess that the 
designers of the chips were able to determine that themselves. If a 
nonideal mirror ratio (i.e. base currents don't cancel or current 
densities are not equal) is responsible for the nonlinearity, things 
should get worse at elevated temperature.

> Finally I looked at the 3280 Iabc linearity.  Again, just considering
> .25 mA and .5 mA the nonlinearity measures 1.5%.

The allowable Iabc for the CA3280 is 5 times larger than the LM13600, 
but that wouldn't explain the CA3080 result unless it's data sheet 
underspecs the part in that respect. It might be worth a try to scale 
down the Iabc when using the LM13600.


Achim.
-- 
+<[Q+ Matrix-12 WAVE#46 Neuron microQkb Andromeda XTk sonic heaven]>+

Q series MIDI Implementation & additional documentation:
http://Stromeko.Synth.net/Downloads.html#WaldorfQDocs