Is there a less noisy OTA than the CA3080?

[lowvolt]

I have a old analog synth.  A few of the stages use CA3080 OTAs in them, and they're kinda noisy.

Is there another chip that can be used in place of these old relics?  Something less hissy?  

Everything is socket mounted so changing ICs is a breeze.

Any help?  Thanks ...

[R.G.]

Yes. But not pin-compatible.

You might try just newer parts, as modern semiconductor processing is better than it used to be. New LM3080s might help, as might replacing/cleaning the socket contacts.

There are quieter parts, but they're generally dual OTAs with buffers, like the LM13700, NE5517 (?), and CA3280 in 16 pin packages. There are some other random ones like the Rohm BA-something-or-other that are hard to find and only may work.

[lowvolt]

Shoot, I should have mentioned that I've nearly totally refurbed this synth.  The OTAs in it now are ~new~ Harris CA3080s, and the sockets have been cleaned as best I could using a bit of DeOxit on old fried chips and inserting/removing them to clean the pin sockets.  It wasn't perfect but it sho' nuff was a damned sight better than what I started out with.

What do you make of the ones Small Bear is pedalling now, they're said to be made from the Intersil masks.  I wonder how accurate that statement is?  And what of all these "Harris" CA3080s on ebay?

Isn't one of the big name comps using CA3080s these days?  MXR I think?  I wonder what's loaded in those things ... or do users of such comps just tolerate the noise?

[Mark Hammer]

The general problem with CA3080s is that they distort too easily.  This requires serious attenuation of the input signal the chip sees, and then compensatory boosting of the chip's output.  The chip itself is not horrifically noisy, but using it in the most inhospitable circumstances and least effective way results in the worst possible S/N ratio.

[lowvolt]

Well I'll give ya that.  Many of these older synths have internal AC audio voltages between 5 and 10 volts.  The VCA's aren't actually ~amplifiers~ as the acronym suggests (Voltage Controlled Amplifier) .... they're actually just attenuators, usually at the ass end of the signal chain.  Some of the filters use these same ICs, as well as some panning/fading circuit blocks.  All of which are dealing with these fairly high audio voltages.  The VCAs and the panners/faders are pretty noisy, so are the filters but I sorta give them a mulligan simply because resonant filters are a bit hissy by nature.

If I route the signal path straight from the oscillators' outs (skipping the filter and the VCA) the synth is pretty quiet (as are most other analog synths I've used) but as soon as the VCA is involved there's hisssssssssssssssssssssssssss

[Mustachio]

I can't weigh in on the technical aspects like RG and Mark.

But I can offer a lil cheap advice on what I did recently building a mutron micro V. It originally used a ca3080 I replaced with an LM13700 and was trying to get both sides of the OTA working together. Since I socketed the IC I was able to use jumper wires to a bread board to test pins and different IC's and bridging both halfs.

You could do the same here since  you have it on a socket. Its pretty easy to look at the pin outs of both chips and connect the corresponding pins. This way you can try different OTA's and posibly running both sides of a dual OTA. Could easily make a small baby board to host the LM13700 or any larger package OTA for it. And A/B test to see what sounds best to your ear.

From what I remember these OTA's expect a very low impedance ( I think Ronan this to me).

Good luck getting it quieter.

[lowvolt]

Thanks .. I think I'll need all the good luck I can muster to get these CA3080 type circuits to be less noisy.  I'm considering actully replaciing the circuits involved with other circuits that use entirely different schemes rather than attempting to polish a turd.  A shiney turd is, well, still a turd.

[PRR]

> Many of these older synths ...  The VCAs aren't actually ~amplifiers~ ... actually just attenuators, usually at the ass end of the signal chain.

There is a different topology which works better when you have large near-constant signals and want to cut them down a variable amount.

In fact it used to be common in TV audio (the broadcast audio is very steady at the detector, you want to fade it down a large degree for late-night viewing).

It's late, my Google search isn't finding it. I'll try to find examples.

BTW: if you break-down far enough to use digital chips, a ratiometric DAC and a PIC to convert CV to bits might be a great "voltage control" audio attenuator.

[PRR]

Here is a very simple (and flawed) high-performance voltage varable attenuator:



CV=+5V gain=1 (0dB)
CV=0V gain=0.5 (-6dB)
CV=-5V gain=0 (typ -50dB)

Distortion in all cases is dang near zero (well under 1%) and can be improved by wrapping an opamp around Q3. Matching is not critical. Hiss is reduced along with signal. Ideally it goes to infinite attenuation, though infinity never happens because everything leaks.

Gain Law is *linear* not exp/log as sometimes used for synths. The usual converters apply.

Main flaw as shown: it thumps on rapid CV change. Q2 collector swings from +5V to +15V +10V to +15V. One major use is TV volume controls, which do not have to be fast, can be slugged to 0.1 second slew. Also a large gain-change on TV is usually rare and dramatic, so VCA artifacts may be overlooked. In synth work you may want faster changes all the time.

The fix for this flaw is to duplicate it, but feed Q3b a "no audio" bias, connect CV bases together, and connect the collectors so that current in R2 does not change with CV. In IC technology the extra parts are trivial.

There used to be RF mixer chips with very similar circuit. I do not recall part numbers, or which ones had the required pinouts.


Further Reading: Malcolm Hawksford has delved deeply into this topology.
http://www.essex.ac.uk/csee/research/audio_lab/malcolms_publications.html
Low Distortion Programmable Gain Cell Using Current-Steering Cascode Topology

[lowvolt]

Sweet!  It looks like it may be adaptable.  It would have to be able to deal with square wave (or pulse wave) modulation for "choppy" tremolo type CV.  Sometimes the modulation frequency can get into the low-audible range depending on what type of patch I'm setting up.  And I suppose I might be able to do without the EXP/LIN switch, especially if the trade-off is a lot less noise.

Here's one of my synths.  I built the cab by adapting two cabs into a single unit, the aluminum center strip is to be used for multiples jacks as part of a normalizing project.  Five VCOs that double as LFOs (two of which have fully mixable waveforms to create just about any analog waveshape imaginable, very handy and fun when used as an LFO), three VCfs (two of which are state variable that provide LPF, HPF, BPF, and NOTCH all simultaneously, one of those two is outfitted with a filter mixer that mixes all four states into a unique filter, the thrid filter is a Moog type "transistor ladder" filter), three 4-stage Envelope Gens, various VCAs and panners and routers, a ring modulator, and a friggin partrige in a pear tree.  It all runs on +15vdc and -15vdc, as well a +5vdc and -5vdc.  Control voltages run from 5v up to 10v.  If I run the oscillators straight out of the 4ch mixer this thing is stone silent.  The two State Variable filters aren't TOO noisy, but that Ladder filter is noisy however since it's also a resonant 6db/12db/18db/24db filter fully capable of self oscillation it's ~forgiven~ .. it's just so fargin FAT sounding!  But the VCA and Pan/Fade modules could stand to be less noisy.  The VCA is obviously quiet until it is opened up either manually or when sent a gate from one of the EGs.

This is easily the best sounding and most capable analog synth I've ever owned .. been playing synths since 1977.



Thanks for the help ... I'll keep researching things!