v2164 linear VCA problem

From circuitbending to homebrew stompboxes & synths, keep the DIY spirit alive!

Moderators: Kent, luketeaford, Joe.

Post Reply
User avatar
J3RK
Super Deluxe Wiggler
Posts: 6733
Joined: Tue Dec 22, 2009 12:46 pm
Location: Seattle

Post by J3RK » Thu Dec 28, 2017 3:56 pm

KSS wrote:
guest wrote:im glad youre verifying this in spice, it helps reassure me that its not just some artifacts in my setup.
This is pure gold. I'm sure Bob Pease and Jim williams just rolled in their graves.
Laughing. :lol:
Perhaps they're laughing. Perhaps not. I'm of the opinion that any tool that exists that helps me do what I want/need to do faster, easier, and with less revision work is a good thing. When I think of something I want to build, I lay it out, simulate it, and once that's looking close to the behavior I want, I'll lay out a test board. After building it, I see how it performs in the real world, tweak this or that, then go back and revise my design files, lay out an improved board, and that's that. Sure occasionally I have to add an iteration to that process, but still, as someone who absolutely hates bread-boards, I find this process fun and easy, and generally something that gets the results I'm looking for. I envy those that happily bread-board actually. In the end though, my first pass test PCB will perform nearly the same function, and is neater and tidier in my brain. I find that I make more mistakes on the breadboard, than in a PCB layout with attached netlist.

User avatar
Dr. Sketch-n-Etch
Super Deluxe Wiggler
Posts: 8114
Joined: Thu Jan 21, 2010 1:46 pm
Location: Vancouver, BC, Canada

Post by Dr. Sketch-n-Etch » Thu Dec 28, 2017 5:41 pm

J3RK wrote:
KSS wrote:
guest wrote:im glad youre verifying this in spice, it helps reassure me that its not just some artifacts in my setup.
This is pure gold. I'm sure Bob Pease and Jim williams just rolled in their graves.
Laughing. :lol:
Perhaps they're laughing. Perhaps not. I'm of the opinion that any tool that exists that helps me do what I want/need to do faster, easier, and with less revision work is a good thing. When I think of something I want to build, I lay it out, simulate it, and once that's looking close to the behavior I want, I'll lay out a test board. After building it, I see how it performs in the real world, tweak this or that, then go back and revise my design files, lay out an improved board, and that's that. Sure occasionally I have to add an iteration to that process, but still, as someone who absolutely hates bread-boards, I find this process fun and easy, and generally something that gets the results I'm looking for. I envy those that happily bread-board actually. In the end though, my first pass test PCB will perform nearly the same function, and is neater and tidier in my brain. I find that I make more mistakes on the breadboard, than in a PCB layout with attached netlist.
I think we were separated at birth. You have exactly described my own process, and motivations. For me, laying out a Doc Sketchy board is no harder than laying out a proper breadboard (the actual original reason why I came up with my Excel layout process), so why not just eliminate the middleman? Actually making a board takes me half an hour, and stuffing and soldering it is way easier than trying to poke those damned flimsy resistor leads into a breadboard. Plus, 90% of the time, my first board is perfect and goes straight into the module.
Power spent passion bespoils our soul receiver
Surely we know.

User avatar
J3RK
Super Deluxe Wiggler
Posts: 6733
Joined: Tue Dec 22, 2009 12:46 pm
Location: Seattle

Post by J3RK » Thu Dec 28, 2017 6:03 pm

Dr. Sketch-n-Etch wrote:
J3RK wrote:
KSS wrote:
guest wrote:im glad youre verifying this in spice, it helps reassure me that its not just some artifacts in my setup.
This is pure gold. I'm sure Bob Pease and Jim williams just rolled in their graves.
Laughing. :lol:
Perhaps they're laughing. Perhaps not. I'm of the opinion that any tool that exists that helps me do what I want/need to do faster, easier, and with less revision work is a good thing. When I think of something I want to build, I lay it out, simulate it, and once that's looking close to the behavior I want, I'll lay out a test board. After building it, I see how it performs in the real world, tweak this or that, then go back and revise my design files, lay out an improved board, and that's that. Sure occasionally I have to add an iteration to that process, but still, as someone who absolutely hates bread-boards, I find this process fun and easy, and generally something that gets the results I'm looking for. I envy those that happily bread-board actually. In the end though, my first pass test PCB will perform nearly the same function, and is neater and tidier in my brain. I find that I make more mistakes on the breadboard, than in a PCB layout with attached netlist.
I think we were separated at birth. You have exactly described my own process, and motivations. For me, laying out a Doc Sketchy board is no harder than laying out a proper breadboard (the actual original reason why I came up with my Excel layout process), so why not just eliminate the middleman? Actually making a board takes me half an hour, and stuffing and soldering it is way easier than trying to poke those damned flimsy resistor leads into a breadboard. Plus, 90% of the time, my first board is perfect and goes straight into the module.
That said, I've seen a couple of your breadboards, and they're quite beautiful. If yours are NASA grade breadboards, mine... Radio Shack dumpster. hehehe I always start off with proper lengths, clean bends, and then somewhere along the way everything turns into electronic component Twister. :deadbanana:

User avatar
neil.johnson
Super Deluxe Wiggler
Posts: 1692
Joined: Thu Apr 08, 2010 3:51 pm
Location: Cambridge, UK
Contact:

Post by neil.johnson » Thu Dec 28, 2017 8:01 pm

neil.johnson wrote:For those interested in the history of the Frey cell here is the original patent:
http://www.google.com/patents/US4471320

The second patent - describing a monolithic embodiment - is this one:
http://www.google.com/patents/US4560947
And here is a third Frey patent, this time assigned to Analog Devices, and on a first glance looks a bit like the 2164:
https://www.google.co.uk/patents/US5528197

Neil
Random ramblings :: http://www.njohnson.co.uk
SSI parts and more :: https://www.cesyg.co.uk

User avatar
neil.johnson
Super Deluxe Wiggler
Posts: 1692
Joined: Thu Apr 08, 2010 3:51 pm
Location: Cambridge, UK
Contact:

Post by neil.johnson » Thu Dec 28, 2017 8:22 pm

guest wrote:thanks for posting that. i was just fixing a bunch of serge modules from 1979 this past weekend, and they were chocked full of figure 4 and 5 from his patent. i was wondering about the history of the gain cell.
Oh there's quite some history indeed.
A good read is Baskin and Rubens's paper "Techniques for the Realization and Application of Voltage Controlled Amplifiers and Attenuators" from 1978:
http://www.aes.org/e-lib/browse.cfm?elib=2976
covers pretty much all the bases, one of which seems to come from the LM394 datasheet according to the references.

Neil
Random ramblings :: http://www.njohnson.co.uk
SSI parts and more :: https://www.cesyg.co.uk

v8pete
Common Wiggler
Posts: 90
Joined: Thu Jul 18, 2013 5:32 pm

Post by v8pete » Fri Dec 29, 2017 4:24 am

Nice info Neil, thanks for the links.

User avatar
neil.johnson
Super Deluxe Wiggler
Posts: 1692
Joined: Thu Apr 08, 2010 3:51 pm
Location: Cambridge, UK
Contact:

Post by neil.johnson » Tue Jan 02, 2018 9:07 am

I see "guest" has started a new thread here:

https://www.muffwiggler.com/forum/viewt ... highlight=

Neil
Random ramblings :: http://www.njohnson.co.uk
SSI parts and more :: https://www.cesyg.co.uk

User avatar
clorax hurd
Notorious Frog Licker
Posts: 234
Joined: Wed Mar 09, 2011 8:34 am
Location: Novy Svet

Post by clorax hurd » Mon Jul 30, 2018 8:39 pm

guest wrote:also, fun fact, the noise on a 2164 is lower for higher input impedances. i just ran a series of tests to confirm this. the noise gets really bad, to the point of being unstable below 10k, so i wouldnt ever go lower than 30k.
Sorry for excavation of this thread, but did you mean the resistor going to audio input pin? If yes, then shouldn't you also adjust the "compensation" filter values before measuring? Didn't you just measure moving the cutoff of the 560p+500R filter being lower, naturally also filtering more noise?

I'm also curious - How much current is going into 2164's audio input pin? In my simulations, that 560pF/500R filter does cut horribly into audio range, unless the pin would be sucking hundreds of uA. Is that correct? What does the input bias current (in nA range) in 2164 datasheet mean then?

User avatar
guest
Super Deluxe Wiggler
Posts: 5852
Joined: Mon Aug 19, 2013 11:49 am

Post by guest » Tue Jul 31, 2018 8:28 am

i adjusted the compensation filter, and it didnt help at all, regardless of what values i used. the issue was further down the line, if i recall, and there was no way to externally compensate.

[EDIT] now i remember, the noise isnt a stability issue, its an increase in the "noise gain" in the amplifier stage due to the lower input impedance. dropping the compensation resistor has the same effect, and makes the noise worse while not affecting stability in any way.

the clip point is 750uA (at least from a quick scan of this thread, might have been revised later on). 560pF and 500R gives a cutoff of 600kHz. the input bias current is the error current drawn by the internal difference amp, which gets subtracted from the signal current that goes in, making a small error. this is a pretty linear error though, so it doesnt show up as distortion. its only an issue for DC circuits.
openmusiclabs.com

User avatar
neil.johnson
Super Deluxe Wiggler
Posts: 1692
Joined: Thu Apr 08, 2010 3:51 pm
Location: Cambridge, UK
Contact:

Post by neil.johnson » Tue Jul 31, 2018 1:07 pm

The small resistor does the compensation, the capacitor merely shifts it out of the way of the band of interest (way above audio).

The input bias current is a DC thing, so shouldn't affect audio unless it is appreciably modulated by signal currents.

I have an SSM2164 circuit on the bench (first prototype PCB) that is giving me around 0.007% THD+N at 1kHz at 0dBu input. So you may want to check your circuit is behaving correctly.

I am currently playing with the much-improved SSI2164 from Sound Semiconductor :love:

Neil

[EDIT: corrected THD+N figure, as quoting from poor memory is often a bad thing! Corrected figure taken from notes and experimentally confirmed just now]
Last edited by neil.johnson on Wed Aug 01, 2018 5:18 pm, edited 1 time in total.
Random ramblings :: http://www.njohnson.co.uk
SSI parts and more :: https://www.cesyg.co.uk

User avatar
clorax hurd
Notorious Frog Licker
Posts: 234
Joined: Wed Mar 09, 2011 8:34 am
Location: Novy Svet

Post by clorax hurd » Tue Jul 31, 2018 1:24 pm

neil.johnson wrote:The small resistor does the compensation, the capacitor merely shifts it out of the way of the band of interest (way above audio).
You are right. I wasn't testing the circuit. Was just thinking about it as of a simple passive filter
signal -> 30K -> test point -> 500R+560pf -> ground,
That way it kills audio range, but I didn't realise, that the input impedance of audio input pin is very low. (probably can't be too high, if it's current in :doh: ). That load shifts the cutoff of the filter quite radically so now it all makes sense. Thanks for all the info & please, would you mind sharing some wisdom regarding my question below.?
guest wrote:
Cool, thanks a lot for explaining. You seem to undestand pretty well how the VCA chip works inside, could you please clarify one more thing to me please?

If we want to go from the same (buffered) source into multiple audio inputs on the 2164 chip. (like when building VC matrix mixer or router. same audio signal but different CV signals on each channel).

Then do we need to go to each pin separately through the resistor + adding the filter to each input?
Or can we instead go through just one resistor into all 4 pins if we make the resistor 4 times smaller value, so we get enough current for each pin.
And also we would use just ONE filter made of 4x bigger capacitor and 4x smaller resistor. Would that work?

One possible issue I can imagine is, that if impedances of audio input pins are influenced by CV, then we wouldn't have the 4x bigger current divided equally into the four pins.
But if the input impedance on those pins is (kind of) static, then it could probably work. (Or not?)

loki
Common Wiggler
Posts: 116
Joined: Fri Apr 25, 2014 4:20 pm

VCA History

Post by loki » Tue Jul 31, 2018 3:44 pm

A patent by Doug Frey you have missed: 4,823,093. He also has an interesting RMS circuit: 5,585,757

The source of the log-antilog 4 transistor cell is David Blackmer's patent 3,714,462. I was first shown this by some guys who had worked on automated mixing consoles at Quad 8.

The precursor to that, using diodes instead of transistors, is 3,518,578 from Alan Oppenheim and Tom Stockham.

When I was a student at the University of Utah in the mid-seventies, Stockham was in charge of the speech synthesis research they were doing for ARPA.

User avatar
guest
Super Deluxe Wiggler
Posts: 5852
Joined: Mon Aug 19, 2013 11:49 am

Post by guest » Tue Jul 31, 2018 4:47 pm

clorax hurd wrote:Then do we need to go to each pin separately through the resistor + adding the filter to each input?
Or can we instead go through just one resistor into all 4 pins if we make the resistor 4 times smaller value, so we get enough current for each pin.
And also we would use just ONE filter made of 4x bigger capacitor and 4x smaller resistor. Would that work?
you need to use a seperate resistor and compensation network for each one. each amplifier has its own internal feedback loop that it is trying to hold steady. if you tie them all together, they will just fight with one another, and wont be able to stabilize. they need to be isolated via those input resistors.
openmusiclabs.com

User avatar
Dr. Sketch-n-Etch
Super Deluxe Wiggler
Posts: 8114
Joined: Thu Jan 21, 2010 1:46 pm
Location: Vancouver, BC, Canada

Post by Dr. Sketch-n-Etch » Wed Aug 01, 2018 3:31 pm

guest wrote:you need to use a seperate resistor and compensation network for each one. each amplifier has its own internal feedback loop that it is trying to hold steady. if you tie them all together, they will just fight with one another, and wont be able to stabilize. they need to be isolated via those input resistors.
I just bought 1000 499-ohm resistors and a similarly large pile of 560pF caps just for this purpose. I also bought 1000 30.1k 1% resistors, all because of my fixation on the 2164. I'll never run out now.
Power spent passion bespoils our soul receiver
Surely we know.

User avatar
neil.johnson
Super Deluxe Wiggler
Posts: 1692
Joined: Thu Apr 08, 2010 3:51 pm
Location: Cambridge, UK
Contact:

Post by neil.johnson » Wed Aug 01, 2018 5:50 pm

guest wrote:you need to use a seperate resistor and compensation network for each one.
Well.... umm, no you don't always. I've just tried this on a circuit on the bench, and wired up a couple of gain cells in parallel with common CV. Stable as before, no oscillation at up to +20dBu input. And there's no difference in THD+N (if anything a very slight improvement due to the reduced thermal noise thanks to the lower input resistance).
guest wrote:each amplifier has its own internal feedback loop that it is trying to hold steady. if you tie them all together, they will just fight with one another, and wont be able to stabilize. they need to be isolated via those input resistors.
I'll need to check this with someone, but I believe that what's going on is that because the gain cell inputs share the input current, where the internal feedback paths sum their currents in to the same summing node, and the system remains stable. Considering the input components, the input resistor and compensator RC are scaled according to how many inputs you parallel up (two inputs: halve the resistors and double the capacitor, etc). The outputs are easy to parallel up.

Not sure how well it would work when the gain cells are controlled by different CVs though (the matrix mixer example given earlier), although I suspect they will behave fine. Best way to find out is to build it and see what it looks like on the scope (it'll oscillate at around 600kHz or so, or overshoot on square wave inputs).

Neil
Random ramblings :: http://www.njohnson.co.uk
SSI parts and more :: https://www.cesyg.co.uk

User avatar
guest
Super Deluxe Wiggler
Posts: 5852
Joined: Mon Aug 19, 2013 11:49 am

Post by guest » Wed Aug 01, 2018 8:28 pm

well, thats interesting. i built one up as well, and it doesnt oscillate. there is some odd oscillation and noise, but thats exhibiting with a single cell as well, so im guessing there is something else screwy with my setup. im seeing slightly better distortion performance with a single cell than a dual. and my quick and dirty setup has too much bleedthrough to really measure the full "off" of a single versus dual.

the output impedance of the input OTA blocks must be high enough, that the feedback currents can equally mix (or at least as equal as the impedances are matched). this makes the output determinate, and stable. if they didnt have any impedance, it would be unclear how much current should come from input A versus input B, in order to sum to the source current being fed in.
openmusiclabs.com

User avatar
guest
Super Deluxe Wiggler
Posts: 5852
Joined: Mon Aug 19, 2013 11:49 am

Post by guest » Thu Aug 02, 2018 11:39 am

just did a quick check with seperate CVs, and the CVs interact as a result of the inputs being tied together. this sort of makes sense, as the inputs are held at the same voltage, so the internal compensation amps will be driving the same compensation currents. so as one pinches off the VCA output current, all that current gets shifted to the other VCA. basically, the amplitudes doubles/halves on one VCA as the other closes/opens.
openmusiclabs.com

User avatar
Graham Hinton
Super Deluxe Wiggler
Posts: 3318
Joined: Fri Jul 02, 2010 6:28 pm
Location: England
Contact:

Post by Graham Hinton » Thu Aug 02, 2018 12:42 pm

neil.johnson wrote:I've just tried this on a circuit on the bench, and wired up a couple of gain cells in parallel with common CV. Stable as before, no oscillation at up to +20dBu input. And there's no difference in THD+N (if anything a very slight improvement due to the reduced thermal noise thanks to the lower input resistance).
When dbx could not produce enough dbx202C discrete hand matched VCAs in the '80's they made a dbx202x module which comprised eight dbx2150s in parallel. The object was to get the distortion of the monolithic part down to the level of the discrete one. Research we did later showed that there were diminishing returns after two in parallel.

Image

You can see a 9-pin SIP resistor pack for inputs, then the other pins (except pin 5) are in parallel with a 5534 driving the CV inputs.
(The dbx semiconductor division morphed into THAT Corp and the 2150 into the 2180.)

User avatar
neil.johnson
Super Deluxe Wiggler
Posts: 1692
Joined: Thu Apr 08, 2010 3:51 pm
Location: Cambridge, UK
Contact:

Post by neil.johnson » Thu Aug 02, 2018 5:18 pm

Graham Hinton wrote:When dbx could not produce enough dbx202C discrete hand matched VCAs in the '80's they made a dbx202x module which comprised eight dbx2150s in parallel. The object was to get the distortion of the monolithic part down to the level of the discrete one.
Ah yes, I've seen that discussed on audio websites.
Research we did later showed that there were diminishing returns after two in parallel.
In theory the SNR goes up by 3dB for every doubling of the number of VCAs. So the diminishing effect is very evident, and you soon run out of justification. Eight VCAs will improve the SNR by 9dB. But that is a theoretical number, and reality says things don't always go as well you hope.

Neil
Random ramblings :: http://www.njohnson.co.uk
SSI parts and more :: https://www.cesyg.co.uk

User avatar
J3RK
Super Deluxe Wiggler
Posts: 6733
Joined: Tue Dec 22, 2009 12:46 pm
Location: Seattle

Post by J3RK » Thu Oct 17, 2019 2:23 pm

Bouncing this back to the top due to other threads that it may be helpful for, as well as me wanting to read back through it again in detail.

:party:

Also, I never caught Graham's last post about the parallel 2150s. Interesting!
Are you lemon? Does your head come to a nub?

jorg
Ultra Wiggler
Posts: 898
Joined: Fri Apr 03, 2015 9:38 am
Location: East Coast USA

Post by jorg » Thu Oct 17, 2019 3:55 pm

punch in log(0) on your calculator; this will give an indication of why you can't use zero CV into the log converter.

User avatar
hox3d
Common Wiggler
Posts: 175
Joined: Fri Feb 02, 2018 6:30 am
Location: Grenoble, Alps

Re:

Post by hox3d » Sun Feb 28, 2021 5:36 am

I've been reading this thread for 2 days, great stuff here!

So, I'm trying to understand how to get a 4Q multiplier without using the (notoriously expensive) AD633. Maybe I'm just complicating things because it is really a stupid circuit, but I'd need 2 of them, so that's steep. I guess this is the occasion to use 2164's instead and experiment with the chip.
Dr. Sketch-n-Etch wrote:
Tue Nov 28, 2017 5:45 pm
Since my last post, I actually have completely redesigned my 4-quadrant multiplier design. It now only uses one linearized 2164, and it seems to work perfectly. Simulations suggest that, with 0.1% 30k resistors, it should give 65dB rejection of the carrier. It uses exactly half the circuitry of the original, so the same circuitry of the original makes a dual unit of the new version, which is huge because there are so many places where a dual multiplier is required.

Here's my layout of the dual unit I built yesterday; each side of the board is a separate 4Q multiplier.

Image
Hey Doc, I'd be curious to understand how you did this.
The previous iteration of the circuit (correct me if I'm wrong) did use two linearized VCAs, where you separate inverted and "in-phase" signal, and what you do is selectively mix either one or another in the output, with complementary CVs.

Now, if using only one VCA you need to have the "mixing" of the in-phase/inverted before the VCA, right? That's actually what Mutable Instruments does with Blinds, where the inversion is done before so you only need a 2Q multiplication.
Do you use a similar topology?

User avatar
neil.johnson
Super Deluxe Wiggler
Posts: 1692
Joined: Thu Apr 08, 2010 3:51 pm
Location: Cambridge, UK
Contact:

Re: Re:

Post by neil.johnson » Sun Feb 28, 2021 6:27 pm

hox3d wrote:
Sun Feb 28, 2021 5:36 am
Now, if using only one VCA you need to have the "mixing" of the in-phase/inverted before the VCA, right? That's actually what Mutable Instruments does with Blinds, where the inversion is done before so you only need a 2Q multiplication.
Do you use a similar topology?
The technique of building a 4Q multiplier out of a 2Q multiplier is pretty well established. The LM13700 datasheet provides a simple example:
https://www.ti.com/lit/gpn/lm13700
Figure 24.

With the linearised 2164 you simply sum in half of the inverted input signal into the output transimpedance amplifier. So it's the Irwin linearised 2164 plus an extra opamp. Total parts count is 1/2 of a 2164 plus three opamps.
Blinds does it in a similar way just with a few more opamps.

Neil
Random ramblings :: http://www.njohnson.co.uk
SSI parts and more :: https://www.cesyg.co.uk

User avatar
Dr. Sketch-n-Etch
Super Deluxe Wiggler
Posts: 8114
Joined: Thu Jan 21, 2010 1:46 pm
Location: Vancouver, BC, Canada

Re: v2164 linear VCA problem

Post by Dr. Sketch-n-Etch » Mon Mar 01, 2021 4:19 am

Yes, the input signal goes around the linearized VCA and straight into the output inverter, and an inverse of the input signal goes through the VCA and then into the output inverter. If the VCA is biased to unity gain for a 0V control signal, then you have a four-quadrant multiplier. I have arranged mine so that a control signal from -5V to +5V generates an equivalent control signal at the VC pin of the 2164 of 2.5V to 7.5V. This way, the VC pin is always positive, so there are no dead spots (if the VC of a linearized VCA goes below zero, the VCA shuts off -- we don't want that to happen in a multiplier).

Note that the control signal (Y) comes in through 200k, but the reference voltage (-5V) is put in through 100k. Also, the control signal is biased to +5V through 100k. This will pin the VCA to unity gain if there is no input signal. However, a 10Vpp control signal will only adjust the gain from 0.50 (at -5V) to 1.50 (at +5V). In order for the VCA to hit a dead spot, the input signal would have to be 20Vpp, and I don't generate any signals that big in my modular.

At unity gain, the inverted X signal cancels out with the non-inverted X signal going around the VCA. When the Y signal is -5V, the gain of the VCA is 0.5, and the resulting output is +0.5 of the positive X signal. When the Y signal is +5V, the gain of the VCA is 1.5, and the resulting output is -0.5 of the X signal. The 200k feedback resistor on the final opamp doubles the signal, restoring it to +1.0 and -1.0 of X.

The trimmer at the X inverter is adjusted to eliminate the output signal when no Y signal is applied. This is ostensibly to make the inverted X signal exactly -1.0 of the non-inverted X signal, but in fact it also compensates for several other small sources of error in the circuit. I use this circuit in my frequency shifter, which is why I have the AC coupling caps in there. I also hand-match all of the 200k resistors to within 0.1%, and also the 100k resistors around the VCA (at I_In, V_Ref, VC_Lin, and the bypassing one from the X input) are matched to be exactly 1/2 of the value of the 200k resistors. (The two 100k resistors on the X inverter need not be hand-matched, since any error in these is corrected by the trimmer.) With this extra level of care, I get fairly high-performing multipliers. In the frequency shifter, I can adjust the two trimmers to virtually eliminate carrier bleed. Also, I can't stress enough that the AC coupling caps are absolutely critical for error-free single-sideband generation in the frequency shifter -- this was a major discovery.

Here is the schematic:
Dual Multiplier Schematic.png
and here is the Linearized 2164 VCA schematic, for clarity:
Linearized 2164 VCA Schematic.png
Linearized 2164 VCA Schematic.png (6.41 KiB) Viewed 116 times
Note that the 4.7k resistor is there to prevent the VC pin currents from going too high if the VCAs go off (when the linearizing opamp goes to 0V, the opamp pegs at the rail, and thus sends quite a bit of current into the 2164, causing excessive power dissipation across the 5k VC pin impedance, and thereby causing the 2164 chip to get quite hot -- of course, this is not an issue in this circuit because the VCA is biased to unity gain and so the linearizing opamp never actually goes to the rail -- I just always put these 4.7k current-limiting resistors in all my linearized 2164 circuits anyway -- they do no harm.)
Power spent passion bespoils our soul receiver
Surely we know.

User avatar
Dr. Sketch-n-Etch
Super Deluxe Wiggler
Posts: 8114
Joined: Thu Jan 21, 2010 1:46 pm
Location: Vancouver, BC, Canada

Re: v2164 linear VCA problem

Post by Dr. Sketch-n-Etch » Mon Mar 01, 2021 4:31 am

The layout of the multiplier and summing amplifier board of my Freak Shift is shown below. This board contains all of the circuitry except the two Dome filters. The dual multiplier circuit is shaded in pink (the left half of the left-most TL074 is actually just a +/-5V reference generator). As you can see, this is a very compact circuit. I'm using back-to-back 47uF electrolytic caps to obtain the bipolar 22uF caps called for in the schematic.
FSLite Main Board.png
Here's a photograph of the Freak Shift, with that board on top of the stack, and the two Dome filter boards underneath (this is a Dotcom version, so the power pins are different from the graphic):
FreakShift.jpg
I'm super happy with how this module turned out.
Power spent passion bespoils our soul receiver
Surely we know.

Post Reply

Return to “Music Tech DIY”