[sdiy] op amp based filter

Sat Jul 24 18:08:21 CEST 2010

Hi Jürgen,

thanks for your reply!

On Sat, Jul 24, 2010 at 07:39, JH. <jhaible at debitel.net> wrote:
> This will work indeed.
> Run the opamp with rather high gain (set with feedback resistor divider) to
> bring the bandwidth down, and divide-down the input signal with a similar
> resistor divider to get unity gain overall.
>
> No idea why you would want to do this with a 741,

I have just went to it since it has 'low everything', but the gbw
isn't low enough here! It fails at being crappy enough ;-)

> or any other opamp of
> fixed bandwidth - it only makes a fixed filter at the price of high noise.

Do you think that using op amps with very low GBW will make it (much)
less noisy?

Notice there are op amps with a GBW as low as 2.7 kHz as listed by
Digikey, so they might not end up having any noise problems at all.

On the other hand, maybe an op amp with an external capacitor could
work? I understand LM101 is one such design but I was unable to find a
place to buy them *cheap*. National has them at around $4, that
doesn't look encouraging, but maybe someone owns a huge heap of NOS.
As a bonus you can have them in metal can format. Maybe there are some
other op amps with external capacitors - I wasn't able to find one
though (because what do you search google for? "op amp with external
compensation capacitor" finds nothing useful).

The LM101A datasheet mentions different compensation types - single
pole and two-pole compensation - on page 7. I assume this relates to
"how many poles the filter has" (for lack of better wording)

Also, "a little noise" might work in this case. Note that we're only
getting that noise at the input of the filter. and also that we only
get it at very low cutoff. Both of which really change the way this
noise affects the filter noise - instead of being wide-band noise, it
just adds "more bass", I guess.

> But use an quad opamp with programmable bias current for its input stage
> (like the now obsolete LM346), and you can build an OTA-based LPF without
> using OTAs. The opamp's first two stages act as OTA, the internal
> comensation cap acts as integration cap, and the opamp's output stage acst
> like the unity gain buffer in the classic OTA filter.
>
> This is still noisy, tends to have a limited lower cutoff range because of
> the tiny, tiny internal capacitors, but has been used in some Inatlian and
> Russian synthesizer circuits AFAIK.
>
> Basically, I know of 3 variants to get to the same basic OTA-filter
> structure:
> 1) Using an OTA and an extra buffer (the usual way)
> 2) Using a Norton amplifier (LM3900) and an extra input stage for variable
> gm (ARP did this)
> 2) Using an opamp that already has variable gm for its input stage (LM346)

So are you saying that every op amp is actually an OTA based LPF? (I
know little about op amps)

David,

On Sat, Jul 24, 2010 at 06:29, David G. Dixon <dixon at interchange.ubc.ca> wrote:
> One question: Why?

Simple: because you can! I don't see this sort of design anywhere
(although admittedly after Jürgen's post it doesn't feel as esoteric
anymore) and got curious why not if it seems it should work.

As an afterthought: you get a thousand types of op amps. Each one of
them is a fairly complicated electronics device (from "somewhat
complicated for its size" to "so complicated we'd never make one on
PCB") and has its own analog quirks and sound. The basic topology
around the op amp could probably stay the same and you could just
switch them in and check out what sounds good. This would mean you
suddenly have a big treasure trove of different sounds and can get
exploring; I wager a bet that because of the different possible
topologies it's a much bigger difference in sound than using e.g.
differently made transistors in a ladder filter.

Dave and Tim,
On Sat, Jul 24, 2010 at 08:18, Dave Manley <dlmanley at sonic.net> wrote:
> ^very low^programmable => Polivoks VCF

Thanks, that was a very useful tip. Looking at the schematic I wonder
if the second "programmable" op amp is being heavily over-driven
(going by the recordings of the filter I'd heard it does sound like
it). I guess instead of a programmable op-amp you could use a normal
op amp with a JFET in the negative feedback which would make it a VCA.
This would also let us use op amps specially picked for a very low
GBW. There are lots of op amps with a low GBW available in the
outlets, I guess their main uses are high-precision data logging
applications.

Tim, you got me really curious with your mention of a textbook application.

Cheers,
D.

> ----- Original Message ----- From: "cheater cheater" <cheater00 at gmail.com>
> To: "synth-diy" <synth-diy at dropmix.xs4all.nl>
> Sent: Saturday, July 24, 2010 6:07 AM
> Subject: [sdiy] op amp based filter
>
>
>> Hi guys,
>> if I took an op amp with a very low GBW, and set the gain so that the
>> bandwidth becomes the cutoff frequency I need, could I make a low pass
>> filter out of that? I would need to make up for the gain before and
>> after that op amp (with op amps too, but ones that are much higher
>> bandwidth of course), and I would probably have to chain several such
>> modules (say 4-8) one after another to get a strong effect, but is the
>> basic idea correct?
>>
>> The only 741 datasheets I could find mentioned a GBW of 4MHz, so
>> that's a bit too much - to have a cutoff of 4 Hz you'd need to have
>> gain of 92 dB, which could be a bit too much to ask of the system
>> without drowning in noise. I've noticed Digikey has a lot of op amps
>> with a very low GBW, even down to 2.7 kHz.
>>
>> Any ideas are welcome!
>>
>> Cheers,
>> D.
>> _______________________________________________
>> Synth-diy mailing list
>> Synth-diy at dropmix.xs4all.nl
>> http://dropmix.xs4all.nl/mailman/listinfo/synth-diy
>>
>>
>
>