Filter ICs

idlefaction · September 30, 2003, 10:17:23 PM

does anyone know of any current production filter-on-a-chip ICs? i'd like to find something that has VC for filter and resonance, and sounds good without sucking a ton of current, and can plausibly be run off a 9V without using a charge pump chip.

i'm thinking of getting hold of an SSM2044 or a CEM3328, but thought maybe there was something new i was missing. it seems weird nobody's making anything like this, that i can find anyway.

Linear and MSI both do something approximate, without resonance though. how likely *is* it that i can feed out back to in and get resonance that way? :/

Paul Perry (Frostwave) · October 01, 2003, 01:35:40 AM

The major deal about the SSM & CEM voltage controlled filter chips, was the octave per volt response.. that is harder to get, than the filter part! (if you wnat it to trac accurately).
I guess anything will resonate wiht judicious feedback, but the CEMs had trics to make it smooth, not just go suddenly rail to rail & blow your mind.

R.G. · October 01, 2003, 07:54:33 AM

Quotedoes anyone know of any current production filter-on-a-chip ICs? i'd like to find something that has VC for filter and resonance, and sounds good without sucking a ton of current, and can plausibly be run off a 9V without using a charge pump chip.

I personally would use the LM13700/NJM13700/CA3280/NE5517 dual OTA's in a state variable configuration.

The single chip does a current controlled band/hi/low pass state variable filter (like the Mutron). You make it voltage controlled by using a resistor in series with the Iabc. They're cheap, highly available, and work great on +9v. The app notes even show you how to do the state variable filter.

Mark Hammer · October 01, 2003, 08:17:21 AM

The other alternative are the various switched-filter chips, like the MF10 (http://www.national.com/pf/MF/MF10.html), though I can't speak to the quality of sound. Consider that as clock-driven filters, they are likely optimized for situations where critical bandwidth is far more important than audio quality within that bandwidth.

Perhaps Mike Irwin might like to chime in about switched-cap filter chips, although I know he has some family-related business to attend to this week. If you're out there Mike, and have a spare moment, what about those MF4 and MF6, et al?

moosapotamus · October 01, 2003, 02:24:59 PM

Excuse me for hijacking this thread, but...

Quote from: R.G.. . . You make it voltage controlled by using a resistor in series with the Iabc. . .

Would adding a resistor in series with the Iabc inputs of the phase stages in a Small Stone also allow it to be voltage controlled? In other words, could you then replace the stock LFO in a Small Stone with a more versatile LFO, like the pseudorandom or other multi-waveform LFO that has ~5V to 10V output, for example?

Thanks!
~ Charlie

Mark Hammer · October 01, 2003, 02:50:47 PM

Check yer mailbox Charlie.

puretube · October 01, 2003, 04:46:01 PM

Quote from: moosapotamusExcuse me for hijacking this thread, but...
Quote from: R.G.. . . You make it voltage controlled by using a resistor in series with the Iabc. . .
Would adding a resistor in series with the Iabc inputs of the phase stages in a Small Stone also allow it to be voltage controlled? In other words, could you then replace the stock LFO in a Small Stone with a more versatile LFO, like the pseudorandom or other multi-waveform LFO that has ~5V to 10V output, for example?

Thanks!
~ Charlie

I`d say: yes... but the Small Stone`s magic lies in the "hyper" waveform..

moosapotamus · October 01, 2003, 05:01:00 PM

Quote from: Mark HammerCheck yer mailbox Charlie.

Thanks, Mark!

Quote from: puretubeyes... but the Small Stone`s magic lies in the "hyper" waveform..

Yes. But...

I can't help myself. I was born to tinker. :mrgreen:

Thanks-a-bunch y'all!
~ Charlie

idlefaction · October 01, 2003, 09:54:25 PM

heh, thanks everyone

that's really helpful actually!

idlefaction · October 01, 2003, 10:15:35 PM

grief, it gets more complex.

i'm going to try and build a VCF using the MF-10 mark pointed me at - but now i need a clock as well :/

can anyone recommend a way of making a low-power-consumption variable frequency low-parts count square wave clock? i'm thinking CMOS, a 4047 prolly. heh. designing stuff is fun.

acromarty · October 02, 2003, 03:34:34 AM

I suggest you try a cmos 555 timer. They can be configured for a square wave oscillator with the frequency set by resistor and capacitor values, and the more recent cmos versions are very low power. They are in an 8 pin DIL package and run from a wide range of power supply voltages.

Mark Hammer · October 02, 2003, 09:46:27 AM

Actually, the Small Stone does NOT have a hypertriangular LFO waveform. It is the Ross phaser that does. The two phasers do share in common the use of OTAs for allpass stages, which may be easily converted from allpass to lowpass stages if the phase-filter mod interests you.

There is an older Ray Marston or Robert Penfold project posted around that uses an MF10 as a phaser I think. Try the Cloned Analog Gear site. They may have it.

Paul Perry (Frostwave) · October 02, 2003, 09:53:28 AM

Since the MF10 (and similar) have a clock rate 100 times the cutoff frequency, is there a problem (?aliasing) when the cutoff is low? Because then the clock is bang in the audio range. Not a problem for a lowpass maybe, but what otherwise? Anyone had experience?

puretube · October 02, 2003, 10:05:32 AM

hi Mark:
on the scope, both my old 70`s and the latest issue SS show traces
like the letter "U" repeated (like UUU , with the adjacent peaks of course touching each other, and a little less steep than printed here).
Isn`t that whats always called "hypertriangular"?
Or is it "hyperbolic"?

//www.puretube.com

Rob Strand · October 02, 2003, 10:48:46 AM

>here a problem (?aliasing) when the cutoff is low?

Yes, these are devices are analogue sampled devices, not unlike analogue BBD chips, and are subject to all the evils of sampled systems. You should have an anti-aliasing filter on the input and a reconstruction filter on the output - regardless of the filter frequency of the switched cap filter (SCF). It might seem odd that you need extra filters even though the chip itself is a filter but it's true. When you use the high cut-off to clock ratio settings this means the clock frequency is usually much higher than the frequencies of interest and you can often get away with simple first order filters on the input and output.

Nasse · October 02, 2003, 10:56:38 AM

My memory may be wrong, but some years back there was a Guitar tracking filter project in Electronics Today magazine (Great Britain) using MF10, filter clock was probaply based on 4046 phase locked loop. Some years... or was it ten or more years ago...

Mike I. · October 02, 2003, 11:59:20 AM

Rob is quite correct - it is often necessary to filter the input and output of the switched cap filter. Fortunately some of these (MF6CN for instance) have a couple of built-in uncommitted op amps that can be used to make fixed filters for the switched cap filter.

Clocking is another issue. For most of these the clock freq is 50x or 100x the cutoff freq. So, if you don't want the clock to go lower than 20 KHz (to keep it above audibility and simplify fixed filter requirements for the output) then the lowest cutoff freq will be 400 Hz for the MF6CN-50 and 200 Hz for the MF6CN-100. If you can tolerate a 10 KHz min clock freq then you get 200 Hz and 100 Hz for the respective fc's. Perhaps not low enough for an autowah application or for a general-purpose VCF. At the other extreme, most of these filters misbehave when clocked much above 500 KHz or 1 MHz (the max cutoff freq specified in the app notes is typically 10 KHz or 20 KHz). They do work quite nicely within these limits though - a cascade of two MF6CN yields a 72 dB/oct lowpass filter...
Regards, Mike

News:

Filter ICs

Mike I.