[sdiy] Difficult 2164 behaviour between different sections.

[David G Dixon]

Hey Tom,

I've used hundreds of 2164s and never noticed any interaction.  However, I
generally don't mix module functions on the same chip, so I couldn't have
seen the behaviour you've seen.

One thing: I make dual expo-linear VCAs from a single 2164 chip.  All my
layouts have a straight line of ICs down the middle with the power rails
running underneath.  When I first started making them, I would lay out the
VCAs in mirror-image on either side of the chip axis, so that the two VCAs
were sharing not only the 2164, but TL074s as well.  Then I noticed one day
that high-pitched signals were bleeding to the other channel.  So, I changed
the layout so that each VCA was on either end of the board, only sharing the
2164, but not sharing any other chips.  This completely eliminated bleed,
and I came to the conclusion that TL074 opamps bleed into one another on the
same chip, but 2164 VCAs don't.  (The TL07X datasheet notes 120dB of
crosstalk attenuation, but this is a big fat lie.)

However, this is not to imply that the controls for the 2164 don't interact.
So, I guess you've learned something fundamental about how to handle the
2164 -- don't mix different modular functions on the same chip when these
functions are particularly sensitive to voltage control.  For example, if
you've used two VCAs for the expo and tempco on a VCO, then you could still
use the other two as a linearized VCA to do linear FM on that VCO, since you
would then not notice the VC interactions (this is how most of the Dixies in
the world are organized).  But you shouldn't use the two spares for a
separate VCF.

Of course, the 2164 chip is only two or three bucks, so wasting one or two
VCAs on a chip really isn't the end of the world.  I know, I hate wasting
parts of chips too, but two bucks is not a big price to pay for peace of
mind and proper circuit functionality.



> -----Original Message-----
> From: Synth-diy [mailto:synth-diy-bounces at synth-diy.org] On 
> Behalf Of Tom Bugs
> Sent: Wednesday, May 30, 2018 2:16 AM
> To: synth-diy at synth-diy.org
> Subject: [sdiy] Difficult 2164 behaviour between different sections.
> 
> Hi,
> 
> I've been farting around with a big design for 18 months 
> already at least (!!) and turned up some behaviour in the 
> 2164s that I've not encountered before, so wondered if anyone 
> else on the list has such experiences or other comments - or 
> posting as a potential warning..
> 
> The design uses two 2164s, so 8 VCAs used as follows:
> 1 for Tri-Core VCO plus one for temperature compensation
> 2 for SV-Filter plus one for temp comp
> 1 for VCA plus one for linearization
> 1 for VC-decay
> Most of the circuit approaches are pretty standard and ones 
> which I have used many times before.
> 
> Having made what I thought was the final prototype and 
> overall enjoying the results, I then noticed that the 
> adjusting the VCA and VC-Decay settings were slightly 
> altering the VCO frequency by about 10cents each
> - argh!
> In that version the TriCore + Temp Comp were on the same 2164 
> as the VC-Decay and the VCA (not the linearization section of it).
> 
> I did another prototype sharing the 2164 sections differently 
> - so the VCO sections were on the chip with the SVF core (not 
> the temp comp for it).
> It solved the original problem - but then shifted it 
> elsewhere! Now adjusting the VCF cutoff would affect the VCO 
> freq (by 40cents)
> 
> Voltage refs are stable, power lines are solid and bypassed, 
> overall layout approaches are, I think, good - so I can only 
> now put it down to interaction between 2164 sections.
> I have identified that the output of the VCO 
> Temp-Compensation section is varying as VCF cutoff is 
> adjusted (again, without the voltage sources or the temp-comp 
> voltage varying).
> 
> One thing I have not massively investigated is the biasing of 
> the 2164 - there is the graph on the datasheet suggesting 
> that minimum CV bleedthrough should be achieved with 
> something like a 560k resistor. 
> Tried a bit, but without success.
> 
> I've already drawn out a new prototype with different 
> temp-compensation approaches, but I wondered whether anyone 
> has any smart ideas on something I may have missed.
> 
> Best! Tom
> 
> --
> BugBrand LTD
> UK company No. 07199808
> VAT No. GB 988 2629 57
> 1 Ninetree Hill
> BRISTOL BS1 3SB
> United Kingdom
> www.bugbrand.co.uk
> 
> 
> ---
> This email has been checked for viruses by AVG.
> https://www.avg.com
> 
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> http://synth-diy.org/mailman/listinfo/synth-diy