Matching 3080s - and new project (VCO)

[Haible Juergen]

	>Maybe it is a good
	>idea to overdrive the differential input, this should be checked. 

	I'd say yes, because the gm of the differential pair can't have
	much tolerance. So overdriving the input would be ok.
	Suggestion: - input to GND, + input to resistor divider 15k / 1k
	from positive suppy. Fixed resistor to the output, and just
	measure dc output voltage. Do this for two different Iabc's
	at least.

	>If gm is known, it should be not neccessay to match, instead you
can
	>change to respective cap so that the product gm*C is constant.

	But what about the input impedance of several Iabc inputs ??
	At high Iabc, external resistors will equally distribute the
current.
	But IMO for low currents you either need multiple current sources,
	of match for impedance of the Iabc input, or live with a low
frequency
	error. (That last one is aparently what most circuits do (;->) )

	If you need only two matched otas, I highly suggest a 13700.
	Just tie the Iabc inputs together. Don't even need resistors.
	I'm currently building a VCO with Triangle core and a separate
	integrator for Saw that is synced to the first one. Easy with a
	13700.
	I plan to build a VCO with the following features (sorry for sudden
	change of topic (;->) ):

	Genuine Triangle and Saw waveforms.
	Triangle will be used to derive PWM and a VCS-3-style Sine with
	waveform control (No more glitch here !)
	Saw will be used to derive a VCS-3-style Ramp with waveform control
	(the most versatile waveform for LFOs, IMO)
	There will also be a built in Sample and Hold (triggered by the sync
	pulse that resets the saw integrator. Useful for LFO applications
	of course, but will also serve as a crude form of ring modulator
	at audio rate.
	Linear and exponential control, of course, with thru zero FM.

	I have a lot of different VCOs meanwhile which all have their
benefits,
	and my intention is to combine my favorite features in one compact
	new design. I'm sure you will like it.   8-)


	JH.