[sdiy] Discrete OTA

[Olivier Gillet]

I've built VCFs with LM13700s and had to use DC-blocking cap on the
inputs and outputs. For example:
http://mutable-instruments.net/static/schematics/Ambika-SMR-v04.pdf

They were needed on the input because the input signal comes from a
DAC and is unipolar. I can imagine something similar might happen in
other synths in which the output of the waveform mixer might not be
DC-free - and thus more likely to make the OTA soft-clip in one
direction.

They were needed on the output because the output goes right into a
VCA ; and a DC-offset on a signal fed into a VCA will cause the VCA
control signal to bleed through.

So you can build an OTA filter without DC-blocking caps of course, if
you consider that it's the job of the waveform mixer upstream and the
VCA downstream to deal with DC-offset problems. As soon as you piece
together more blocks, the problem has to be addressed...

It really depends on the buffers you use, too. If you use a cap + an
op-amp follower (or, more conveniently, an op-amp in integrator
configuration), you will have less DC-offset problems than if you use
the darlington buffers built into the LM13700s - and you could
probably get rid of the AC-coupling cap in the VCA input. In an
earlier version of the Shruthi I had no blocking cap, and a trimmer to
trim out whatever DC offset was left after the filter - though most
people never bother adjusting it and couldn't hear the difference, or
misadjusted it on purpose to get more "punch".

Olivier


On Mon, Apr 7, 2014 at 4:24 AM, David G Dixon <dixon at mail.ubc.ca> wrote:
> I don't know, Andy... I just call 'em like I see 'em!  I've never put a DC
> blocking cap in a filter circuit in my entire life, nor have I ever seen the
> need for one.  I only built one filter with OTAs (the earlier version of the
> Dr. Octature cascaded four-pole) and it works very well without DC blocking
> caps.
>
>> Hi David,
>>
>> Can you please elaborate as to why the 33u is needed after
>> each sallen key? As you can see from the schematic:
>> http://machines.hyperreal.org/manufacturers/Korg/MS-synths/sch
>> ematics/KLM-307.GIF
>> , the design actually doesn't use dc blocking caps after each
>> stage of the sallen key, they only have them after the last
>> stage. I would have thought there was enough DC blocking from
>> the negative feedback to not need them at all, the final
>> 2SC945 x 2 transistors already distort a lot anyway, so a
>> tiny amount of DC input here isn't going to matter much, and
>> the amp adds dc anyway which needs blocking before the final output.
>>
>> So the only thing I can think of is they are protecting the
>> 4558 from DC, since there is a 10u / 33u DC blocker just
>> prior to each one of those. So is this a case of protecting
>> 4558s rather than a need for an OTA + dc adding buffer design
>> needing AC coupling? The IR3109 roland designs are basically
>> an OTA + dc adding buffer (in this case a p-mos
>> buffer) and if you check the schematic of the Juno-6 or
>> similar there is a single 10u DC blocking cap at the input,
>> and output, but nowhere else in the filter core or resonance
>> feedback path, and the filter will work just fine if they
>> were removed and operate on DC signals.
>>
>> --Andy
>
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at dropmix.xs4all.nl
> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy