AS3320 filter issues

[JK Sleepling]

Hi all
I' working on the same project as usual, building a midi-controlled effect mixer. Still considering myself a rookie, but came a loong way - thanks to you guys in here!

Todays issue is the final component in the audio chain. The lowpass filter.
This is build with the AS3320 VCF using this exact lowpass circuit from the datasheet 

The filter functions correctly regarding the controlled cutoff frequency, but it seems to somehow choke and distort the signal and also amplify digital noise. When bypassing the filter section, everything sounds clean and powerful.

Right before the filter I mix three signals together. And after the filter I invert/buffer the signal.

Here's some questions:
1) I use +-9V instead of +-15V as the datasheet example.
I should change the -9V resistors to something equally smaller like 130K, right?
Other things that I should change because of that?

2) Does the AS3320 invert the signal like the AS2164 VCA?

3) Does the AS3320 need post amplification?

4) Is there any other obvious problems in my design, that I've missed?

[ElectricDruid]

If you use the datasheet example, then changing to +/-9V is a bad idea - or at least, it might be the source of some of your problems. The components values (almost all of them) are calculated based on the +/-15V supply. That includes stuff around the stages (91K, 200K, etc) which are set to provide the correct quiescent current, not just the resistor on the negative supply. Changing the supply is going to push the filter stages further out of their happy zone. That may or may not be good, but it probably involves increased distortion.

Somewhere I've got a fuller copy of the datasheet that details the process of working out values, but I'll have to find it. That resistor on the negative supply in particular is something that I've never been able to get a clear answer about, or I've never understood the answers I've been given, perhaps. There's a zener regulator on the negative supply is what I understand, and the resistor sets the current drawn through the zener. Some people say the "safe side" is to make the resistor slightly too large. Other people say the zener won't regulate properly unless the current drawn is at least the minimum specified, so the safe side is the next value *down*, not up. So I haven't even got a clear answer about that.

1K5 is too big for -9V though, of that I'm certain. The formula in the CEM3320 datasheet gives a value of 787R for -9V. Whether you call that 750R or 820R is up to you. CEM call it a "current limiting resistor" which makes it sound like the larger value is safer. The AS3320 datasheet completely scrambles the formula and doesn't tell you anything useful. It looks like a typo: REE = (1.5 ÷ 2.2)KΩ ?!?

To answer some of the other stuff, no, I don't think the AS3320 inverts the signal, and the output can be as much as +/-12V on a +/-15V supply, so you might get +/-7V or so on +/-9V. So, no, I don't think you need post amplification.

Finally, have you spotted that the Cutoff CV works "upside down"? More CV equals lower cutoff? It's designed expecting an op-amp mixer on the cutoff CV input to mix the various sources of control voltage, so the fipped response makes life *easier* for that situation.

[JK Sleepling]

Okay, thanks again Tom!   

As I see it, I have two options: Change my bipolar supply to +/-15V or somehow find the values for using the AS3320 at +/-9V.

This is a guitar effect, so I'd really like the power supply to be 9V, then it making +/-15V may be a bit cumbersome I guess. I would need a charge pump and additional part I guess, as I still need 5V and -5V for my CV offsetter and digital board...
But using +/-15V may give me some more headroom through all VCAs and opamps. But I don know if the AS2164s takes 15V or then need other configurations as well? hmm..

My current power section look like this. Pretty simple.


I'm just thinking loudly here. But any suggestions are welcome

Alternatively, I have to find the correct resistors for the AS3320 +/-9V setup. I don't know which option is the easiet one. I have a PCB ready for the filter section seperately, so I could test it with other resister values at first. The problem is just calculating them then

[ElectricDruid]

Here's the details I've got. It turned out I'd actually filed it neatly! Fancy that!

The CEM datasheet is the full one for the 3320, which has some useful stuff. More useful still is the ICs for Electro-music article, which gives more details on the design procedure.
 

[JK Sleepling]

I also found this one:
https://www.bustedgear.com/images/datasheets/CEM3320.pdf

But my brain is still not capable of understanding what to change in case of other input voltages that +/-15V.

Anyways, I figured out my whole circuit design (incl. regulators, op amps and VCAs) should actually be able receive up to 15V DC input, and still do what it supposed to.

So I found a 15V out on my pedal power supply and fired up my thingy on that.

The filter section became a bit more quiet regarding digital noise. The sweep works fine as always sending -25mV to 166mv. But unfortunaltely it still distorts the signal badly.

I had a suspicion that the audio from previous parts of the circuit was too hot for the filter chip, but going from 9 to 15V didn't help. So it doesn't seem like a headroom problem. 

[ElectricDruid]

I had a suspicion that the audio from previous parts of the circuit was too hot for the filter chip, but going from 9 to 15V didn't help. So it doesn't seem like a headroom problem. 

It could still be. The headroom into the chip isn't that fantastic, and 9V to 15V isn't that big a jump, if, say, the signal was twice as big as it needed to be.

We need numbers. Sorry, but it's true.

[JK Sleepling]

When turning down the level going into the filter chip, the distortion remains the same. Even at levels below unity.

[JK Sleepling]

I use an electric guitar with lazy pickups for testing. They should maximum send 500mv (more likely between 2-300mV). This signal is buffered and send to a AS2164 which at maximum amplifies it with 20dB which is 10 x 500mv = 5V. I should have room for a +/-7V swing at +/-9V supply. 

[JK Sleepling]

I solved the problem! But I am unsure if it's an obvious solution that I just missed... 

After the filter I just added another buffer. That's it. And gone are the distortion.

The inverting buffer in the first schematic (of the initial post) was wrong, and I fixed it in a newer PCB version. This gave me the option of adding a (working) buffer or bypassing it.  With the inverting buffer on the sound was suddently clean and powerful. I guess I can do a non-inverting buffer here as well.

[ElectricDruid]

Ok, so it looks like the 3320 doesn't like that 10K load down to a virtual ground. There's an example in the datasheet using an inverting buffer, but it uses 100K input resistor. Possibly the 51K on the pin 8 Resonance input causes the problem, since it forms part of a divider with 3K6 to ground internally. It's too late on a friday night for me to think about that clearly right now.

Nice work for getting it sorted, any which way. If a non-inverting buffer is the way to go, that's all good.

[JK Sleepling]

Just a quick update from here: I've given up on the AS3320 filter. (Sigh!)
It sounds fantastic, but it's way too sensitive to noise on the power rails, and I can't seem to get my digital control quiet enough. By comparison, the AS2164 chips in the same circuit are dead silent.

So, I started looking for another way to create a voltage-controlled lowpass filter—and then I realized the answer was right in front of me. Thanks again, Tom (Electric Druid)!
https://electricdruid.net/multimode-filters-part-2-pole-mixing-filters/

The AS2164 would work well as a building block for a VCF. I'm already using two AS2164 (quad VCA) chips for volume controls but only need five of the VCAs, so I have three to spare. A 3-pole filter (without resonance control) could be built.

It works, and it's as quiet as I'd hoped.

[Mark Hammer]

The entire E&MM article for those who want it.

https://www.muzines.co.uk/articles/ics-for-electro-music/2596