Hi

I'm doing this excact lowpass filter (se datasheet example) with the AS3320 chip.
But I can't really figure out from the datasheet what the minimum and maximum control voltages are for the frequency control.
And further more - how the cut frequency corresponds to the given cv voltage.
AS3320 VCF Datasheet

Thanks

It's a "copy" of the CEM part. I remember when the CEM3320 was new, being annoyed at its datasheet.
https://electricdruid.net/wp-content/uploads/2017/06/CEM3320-VCF.pdf

There's a discussion here:
https://modwiggler.com/forum/viewtopic.php?t=258795

The figures you need are there, but *finding them* is perhaps a different story! I agree it's not obvious.

The basic Freq CV response is "Sensitivity of Pole Frequency Control Scale, Midrange", which is given as 60mV/decade. This corresponds to 18mV/oct. If we want to cover the full audio range and make sure the filter is fully open, we might want the cutoff to sweep from 10Hz to 30KHz. This is an 11.5 octave range, but upwards of 10 octaves is generally good enough.

Note that the datasheet gives the 60mV/decade figure within a range from a minimum of 57.5mV to a maximum of 62.5mV. This is worth considering. It means our 18mV/Oct figure could be as low as 17.3mV/Oct or as much as 18.8mV/Oct. This implies that we need between 199mV and 216mV to cover the 11.5 octave range. If you've got roughly 200mV variation at the pin, you're doing ok.

There's also a note at the bottom of the table suggesting that the best range for this control voltage is from "-20mV < VCFI < +160mV", which is to say: that's the most linear part of the range, so if you're aiming for 200mV, you should try and get it around that range - e.g. -25mV to 175mV or so. This isn't essential, it's just best practice for minimum error. You can quite well run the filter with CVs from 0mV to 200mV, but you'll get slightly worse V/Oct compliance. Unless you're trying to use the filter resonance for playing flute solos, you probably don't care too much.

I figure 13.5V for nominal range of frequency. We recognize this as the nominal max swing of an opamp eating 15V power (couple diode losses).

IIRC one of the control inputs is current, the external resistor sets the relationship between voltage and the control current.  The other control was defined as voltage.

I deciphered the chip behaviours in this thread and summarized the results,
https://www.diystompboxes.com/smfforum/index.php?topic=125885.msg1205318#msg1205318

Also, get the datasheet in the link I posted in that thread, it's much more comprehensive.

Quote from: PRR on May 10, 2024, 05:53:55 PMI figure 13.5V for nominal range of frequency.

*** WITH THE GIVEN VOLTAGE DIVIDER VALUES!!! ***

This is not the voltage at the pin. As Paul's notes say, the voltage at the pin only varies by 240mV because of the divider. Even this is probably an over-estimate. I suspect this ratio is set up for a 0-10V Frequency CV to give roughly 10 octaves of control, since that follows the 1V/Oct standard. As you go further to either extreme of this range, the1V/oct tuning will start to get a bit wayward.


Also Rob is right - the "Resonance CV" is really a current. The given voltage creates a current through the input resistor and the current is what matters. The datasheet suggests 0 to 100uA, so 0-10V into the 100K resistor shown in the datasheet. In practice, some synth designs use less current. The Sequential Pro-One uses a 0-15V CV from a pot through a 200K resistor, so that gives only 0-75uA, and I always found that I never used the control much above half way anyway, so even 50uA of current may be enough depedning on your tastes. You can use whatever CV range you like, but just scale the resistor proportionally.

Quote from: ElectricDruid on May 10, 2024, 06:56:22 PM*** WITH THE GIVEN VOLTAGE DIVIDER VALUES!!! ***

Quote from: JK Sleepling on May 10, 2024, 04:20:28 PMI'm doing this excact lowpass filter (se datasheet example) with the AS3320 chip.

(Underline added)

Cool! Thank you guys. That helped a lot!
I'm getting 0-5V from my DACS so I need to convert that to someting around -25mV to 175mV

Then i came up with this:


Sending 5V : 5V / (-130/5) = -0.192 V
New Offset: -5V / (-150/5) = 0.166 V

would give m a span from (-192 + 166 mV)  = -26mV
to a max of 166mV

or am i calculating this correctly?

Quote from: JK Sleepling on May 12, 2024, 08:29:06 AMam i calculating this correctly

Your calculations are correct but your -26mV output is a result of subtracting two larger numbers so it will be subject to the effect of tolerances.

I'd add a smaller -25mV to the zero input and have a 175mV for the 5V input.  I'm not sure how important the direction of the DAC output and the control voltage are to you.

Something else to consider is if the opamp output glitches at power up to 5V or -5V will that inject indesirable voltages or currents into the chip's control pin?

I was thinking of resistive mixer made from three resistors:
- resistor from vctl to ground
- resistor from vct to -5V to set the smaller -25mV offset.
- resistor from vctl to DAC in  ; maps 5V to +175mV
- add control voltage filter caps as desired.

This scheme removes the opamp and the need to consider the opamp offset voltages.

Alternatively, using a simple passive voltage divider *after* the op-amp mixer is a good idea for safety. E.g. you could output -0.25V to 1.75V after the op-amp, and then scale that down to -25mV to 175mV with a divider after the op-amp. This provides some protection for the chip. It also reduces the effect of any op-amp offset voltages.

Having the op-amp in there is not a bad idea since it allows you to mix other voltages in as well as your DAC control. It also flips the voltage so that the CV will work the way you expect. The "raw" CV input on the 3320 works upside-down (almost like they were *expecting* an op-amp mixer ahead of it... ) and more CV makes the frequency go down.

There are lots of ways that will work!

Right! I like the voltage flip. So here's a new version with 10x smaller input-resistors and a voltage divider

And another AS3320-thing - how should I handle the resonance CV-input if I just want resonance to stay neutral. Connect it to ground or let it float?

Quote from: JK Sleepling on May 13, 2024, 01:46:10 PMAnd another AS3320-thing - how should I handle the resonance CV-input if I just want resonance to stay neutral. Connect it to ground or let it float?

No, definitely connect it to ground. Floating is rarely good - the exception is stuff that has internal pull-ups, in my experience. Those look from the outside like they're floating, but they're not really. There's probably other situations too, but that's not the point here.

And note that that's "ground" in the chip's view, and it's expecting a bipolar supply. You haven't mentioned what your supply is, but if you're on a single-supply, it'd be the midpoint Vref, not the actual ground.

Quote from: ElectricDruid on May 13, 2024, 02:53:08 PMAnd note that that's "ground" in the chip's view, and it's expecting a bipolar supply. You haven't mentioned what your supply is, but if you're on a single-supply, it'd be the midpoint Vref, not the actual ground.

Thanks Tom! I have a +-9V bipolar supply so I'll just ground it to zero 

Is that 100K resistor still necessary then?

Quote from: JK Sleepling on May 13, 2024, 04:11:27 PMThanks Tom! I have a +-9V bipolar supply so I'll just ground it to zero 

Ok, great.

QuoteIs that 100K resistor still necessary then?

I wouldn't go with anything as large as 100K, but at the same time I wouldn't just short the pin to ground. Although as no current is flowing, it shouldn't really make any odds.
I can't justify this feeling technically though, so feel free to discard it. I'd probably use something between 1K and 10K, *as if* it was a pull-down or a pull-up, although it's not.
I suppose my gut feeling about this is based on the fact that if you had a pot providing a CV to the chip, even when the pot was turned down to Ground, there'd still be a resistor between the ground conneciton and the pin. So I'd probably want to keep that the case, even if it might not be entirely necessary. I suppose one potential justification would be negative offsets on the Res CV input - if that pin were to go a bit negative for whatever reason, then some current *could* flow, and if the resistance to ground were very small, that might produce enough current to create significant resonance. After all, we don't need much. But that's me trying to come up with after-the-fact justifications for something that I *feel* rather than *know*!!

Quote from: ElectricDruid on May 13, 2024, 05:07:53 PMI suppose my gut feeling about this is based on the fact that if you had a pot providing a CV to the chip, even when the pot was turned down to Ground, there'd still be a resistor between the ground conneciton and the pin. So I'd probably want to keep that the case, even if it might not be entirely necessary. I suppose one potential justification would be negative offsets on the Res CV input - if that pin were to go a bit negative for whatever reason, then some current *could* flow, and if the resistance to ground were very small, that might produce enough current to create significant resonance. After all, we don't need much. But that's me trying to come up with after-the-fact justifications for something that I *feel* rather than *know*!!

I'll go with your gut feeling anytime. Thanks man!