Descret OTA was "that Japanese site"

[terry michaels]

Message text written by INTERNET:jhaible at metronet.de
>Very good points. I don't know for certain why he didn't use the opamp
integrator
approach. I got one single email from Dave Rossum when I asked about
the 2040, but not a second one when I asked for further details.
He mentioned that they experimented with zener diodes to increase
the level shift of the buffers, but dropped this idea. He also mentioned
they used these minimal transistor count OTAs because of a limited
number of transistors available on chip. Maybe he had the choice between
internal darlington buffer or external opamp for the same reason.
Who knows.
But as a fact both the old data sheet and this email from a few years ago
only mention the limited voltage swing as a restriction, not a feature.
And the Prophet 5 was built that way, too: avoiding higher levels at the
filter.

So I still tend to believe that this "feature" was unintentional. No one
mentioned it, the first applications didn't make use of it ... I am
curious which synth really was the first to make use of it ! (Any ideas ?)

Speaking of ARP and opamp integrators: Do you have a feeling (or numbers
even) how much using opamps as buffers actually slow down the circuit ?
I vaguely remember an EN article about that. Interesting enough, ARP 
split the capacitors in two parts in one of their filter modules, one
in the feedback loop of the opamp, and one from the -input to GND.
I guess that was to increase the noise gain of the opamp, so they
could use a fast opamp without getting in trouble with stability.

JH.
<


Hi Juergen:

One possible reason for the darlingtons in the SSM2040 is it might have
been made with a bipolar process that didn't include a way to make a JFET. 
The op-amp integrator approach really needs FET inputs to work properly. 
Or maybe they didn't have a good JFET op-amp macro to insert into their IC
design.  Or maybe they wanted to keep the die size down.  A darlington
buffer is much smaller that an op-amp.  We can only speculate.

I looked through my files and I couldn't find the ARP filter module you are
referring to.  Maybe you can e-mail me the module number.

Without running the numbers, I think modern op-amps (LF351, LF356 etc.) are
plenty fast enough.  Plus you have a capacitor across the op-amp. creating
an integrator.  The integrator slope will be far below that of the op-amp
phase compensation slope.

Terry Michaels