Issue with my multimode VCF + Amplifier design

[CrackersX]

Hey everyone,

I'm back after designing my VCO, and now I want to complete my synthesizer with a voltage-controlled filter (VCF) that includes low-pass, high-pass, and all-pass modes, along with a non-voltage-controlled amplifier. This way, I can directly connect the output of my VCO to this new circuit.

I’ve put together a schematic based on the design mentioned in Electronic Druid (Craig Anderton’s Multiple Identity Filter: https://electricdruid.net/multimode-filters-part-1-reconfigurable-filters/), which uses an OTA. I think I’ve grasped the core concepts, so I started modifying and improving the circuit.

Here's what I changed:

I added a voltage-to-current converter at the input of the OTA’s control current, allowing me to control the cutoff frequency via an input voltage.
I replaced the op-amp with an inverting amplifier, where I can adjust the gain by changing a resistor.

Unfortunately, I haven’t been able to simulate the circuit properly because I'm running into issues with the OTA model. However, before troubleshooting the simulation, I’d like to confirm: does my design theoretically work?

If anyone has the time to help me refine and finalize this design (which might just be flawed), I’d really appreciate it!

Thanks in advance,
Crackers

[Andree Henkel]

OTA circuits usually are not easily be converted to OPV circuits.
Unfortunately the CA3080(A) OTA is long gone. Available NOS is largely overpriced.
The OPA860/OPA861 is not really a similar IC.

Good alternative is double OTA LM13700 - cheap and readily available. And improved compared to CA3080 by the buffer stages.
And with spec for filters in the audio range.

So I advice: try to convert your historic OTA circuit to using LM13700.

[Terry Bites]

OTAs have dreadful thermal drift. Not what you want in a VCO.

[CrackersX]

oh, I didn't knew that.
I also saw that concept about pole mixing https://electricdruid.net/multimode-filters-part-2-pole-mixing-filters/ but it seems hard for now.
Do you have any alternatives in mind?

[CrackersX]

I might stick to using an OTA since it simplifies the VCF a lot. I'll deal with the temperature later.
Is my design working, aside from the components, on the principle?
For example, with the amplifier over the buffer, won't it cause everything to saturate?

[Andree Henkel]

the Circuit around U9 doesn´t work as a voltage to current converter. It is just an inverting voltage amplifier. I assume you want gain of OTA AS13704 linearly controlled by your "modulation" input, but your circuit won´t work that way, current into the gain controlling "bias" input will be strongly nonlinear with applied voltage.

What you need is a controlled current source, with max current well below 2mA.

[CrackersX]

Ah,
I’m just going to change the design. I saw this one in the LM13700 datasheet.
I haven’t been able to get it to work in KiCad so far, but the design looks better.
It might be possible to achieve a high-pass filter with this, using well-placed switches and I will just put on inverting amplifier at the output for the amplifier part.

I don’t see the +15V and -15V inputs in the LM13700 model. Is that normal? the LM13700 is connected to those?
This might be why it’s not working.
Also, I’m not fully grasping the theory behind it. Are there any sites that explain it? If not, would you mind explaining it to me?

[Andree Henkel]

Yes you need the supply.

The OTA alias operational transconductance amplifier is simple to explain.

OTA circuits just work differently than OPA Circuits.
OTA circuits do not rely on feedback, that results in both inputs being at same voltage.
Instead you have a voltage difference at the inputs.

The output of the OTA is a current. That is is what transconductance means: voltage in, current out. Usually the transconductance is denoted as gm (m lowerscript) the fun thing with OTA´s is, that gm is not fix, but you can control it by changing the Ibias control current.

In case you connect a resistor between output and ground your OTA circuit gets a voltage amplifier. And in your case with Capacitor at output the circuit becomes an integrator. That easy way to build integrator stages, adjustable about typically 7 decades with Ibias is what makes OTA fun for oscillators, filters and the like. It is also very easy to do things like AGC (Automated Gain Control) Circuits.

OTA circuits are very common within analog integrated circuits, so there is a lot of more detail in special literature on analog IC Design. On Board level however they are less common. Except niche like audio effect circuits. Best you study the datasheets / application notes.

LM13700 datasheet is good start, contains many example circuits.
Main plot to understand is figure 8.

Example: you have 100mV at input, and Ibias is 10µA and 1k load resistor.
from figure 8 you get about 200µmho whith mho a funny way to denote 1/ohm also known as Siemens - unit of conductivity
so 100mV * 200µmho * 1kohm
the ohm and mho shorten, as well as m and k,
so 100V * 200µ = 20mV
not really impressive, but you could drive Ibias at 1000µA and would get 2V

Now change the resistor load to capacitor load and calculate
Uout = 1/C*Integral(I(t)), where I(t) is gm(Ibias(t))*Uin(t)

[Andree Henkel]

With your LM13700 symbol, it might just be an issue of how it is done in your kicad library, maybe there are several subparts, the two actual OTA, the supply subpart and two buffer subparts. or the Supply pins are hidden and automatically connected to some net name.

I don´t use Kicad, so I can´t tell what the issue is.

[Andree Henkel]

your circuit still has a problem:

the Ibias Input is refered to the negative supply (see figure 16 in LM13700 datasheet), while your control signal is refered to ground, so it still won´t cut it, this way you won´t get linearly control of gain with your control voltage.

[CrackersX]

Thank you very much! I understand how it works.
The LM13700 works like a resistor, I looked at the maths and it's so smart !!
I made a 3-pole VCF from the datasheet.

However, I don't understand four things:
-The purpose of R30 and R39 with the -15V supply.
-Why I get a bandpass output from the RC filter in the middle instead of a low-pass?
-Why R31 is 20k instead of 10k for the voltage divider, like R14?
-Why the feedback is taken after the darlington not before?
I added a sommator after to get a HP because Vhp=Vin-(Vbp+Vlp). Did I do it correctly?

Thanks for the answers! My VCF is starting to look great.