Cloning the ca3080

You'll have terrible thermal compensation ( = drifty offset and gain, plus thermal time constants) without monolithic construction. The representative circuit seems to be current mirrors and a diff amp, which will work fine. Doesn't seem to be anything remarkable about it, no bandgaps or Widlar-esque craziness.

Tim

We could make a copy ;-) ...Jim Thompson

Probably fakes, but maybe...they're good fakes? :)

schreef in bericht news: snipped-for-privacy@z11g2000yqz.googlegroups.com...

Digi-Key lists the CA3080E from Intersil though they seem to have no stock.

Though the design seems pretty simple, building a thing like that in discretes may become a real nightmare.

If you really need thousends, ask Jim. He seems to have some experience with that kind of designs :)

petrus bitbyter

Wow, that was shortly after the era of nuvistors :-)

You might want to call these guys and ask how many they are holding:

And these guys:

Haven't dealt with either company lately and I don't know what the difference between the A, M, non-dash or other versions might be. Why does it absolutely have to be the CA3080?

I'm curious about the last point as well - as far as I know if one doesn't use the linearizing diodes and output buffer for all intents and purposes one half of an LM13700 is equivalent to a CA3080. According to:

It makes nice voltage controlled filters

Ok, it's a nifty chip, has bias control:

But can't one do most things as well with a Gilbert multiplier or a gain-steerable video amp? If Steve wants to verbatim build another round of a yesteryear design then I could see the need. But to obtain thousands, that'll be slim pickings.

Because "Its Yesterday Once More" . I'm going to need thousands of analog, sine wave VCOs, and 8038, 2206, and max38 are gone. Cost is a factor and microprocessors are unwelcome for this application.

So I looked at the analog synth stuff , and 3080 has bandwidth to spare...

Since this product may have to run for a decade, roll your own starts to look good... But at estimated 5K a year its hard to justify a foundry run.

Steve

Don't know what the pricing is, but these guys still make LM3080 and CA3080s:

I used 3080s in a tunable formant filter for analog speech synthesis about 30 years ago. It worked fine, but watch out for amplitude scaling. Without the linearization diodes (LM13700) the 3080 is truly a small-signal device, you'll start to pick up distortion with input differentials less than kT/q.

We did try to make an "improved" version using the linearized parts, but it sounded different and people didn't like it as much.

steveg

The CA3080 had the problem that the transconductance varies from unit to unit too much.

You can make about as good of a variable gain by using a JFET to control the gain of an amplifier. If you get one designed for voltage controlled resistor service, you may be better off:

If you use a dual you can use a little feedback circuit on one to make the right gate voltage for the other and get the resistance control to be within about 20%.

You're going to have a helluva time getting the device matching/tracking necessary.

What's the end use? Offline if you like. ...Jim Thompson

Depends on how urgently it's needed in the marketplace and thus how much customers are willing to plunk down. At 5k/year a custom chip would cost serious two-digit, considering the amortization of IC design and masks over 4-5 years.

I don't know why uC are ruled out because that would be the way to go these days. Another option is to run an oscillator at a few hundred kHz or a MHz where you can use varicaps, and then mix that down to audio.

I just did a Google search and Google suggested "CA3080 replacement".

It would seem others have a need to replace this part as well.

With over 10,000 hits, there may be close some replacement out there.

hamilton

That's SO extortionate! The CA3080 has five transistor pairs, no MOS or multilayer metallization, no resistors, only handles a milliamp of current... What lame designer or design center would have to 'amortize the IC design' over a period of years to handle this little item?

Aren't there semicustom routes, too?

After prototyping, I replaced the linear design with a Atmel and told my boss just to buy a 10 year supply. Goal was no RF noise for a instrument app. I liked the linear, and the idea was to have to do as little redesign as possible during production.

Steve

How about using an LT1228 to do the same sort of function?