Discrete OTA?

[Kit]

What would you say is the major drawback when using OTA as compressors?

Some stompbox-comps uses the CA3080 for gaincontrol. Might not be good enough for "pro" audio. Dunno.

The topology is similiar to vari-mu, exept OTA´s feed of current instead of voltage. Vari-mu being tubes, has the advantage of headroom but other than that......

So, a BJT pair with a JFET in the tail. CV to gate...wadda ya think?

Something like this...

Note that this is not the full schematics. Sidechain and output stage are omitted for clarity.

 

[jdiamantis]

A simple OTA gain cell like the 3080 has an input-output transfer curve that crosses into distortion at about 50mvac input.

In order to linearize this transfer function, you can set up your circuit with 2 of these devices; one in the feedback loop of an opamp, the other feeding this circuit. Alternately, you can use a diode linearization network on the input, to "pre distort" the signal, resulting in a linear transfer curve.

Chips that used this type of circuit, were the CA3082A (no longer available, I think), used in Orban's early processesors, and the LM13600-LM13700 series (also, possibly no longer available).

The gain transfer curve of these devices is linear; a decade change in gain current provides a decade change in output.

As far as using them in pro equipement, considering the availability of these chips, you'd be better off using a modern VCA from That or Analog Devices. If you're looking for a different gain transfer curve, try FET's; but they're not without their problems either.

Re your circuit; it looks like it would operate as an FET voltage divider attenuator rather than an OTA. Which is not a bad thing.

Have fun.

 

[Kit]

Thanks, John.

As far as the premade chips are concerned, i agree that the classic "blackmer cell" is hard to beat for low price and low distortion.
But sometimes a little "crunch" can be a good thing....

A simple OTA gain cell like the 3080 has an input-output transfer curve that crosses into distortion at about 50mvac input.

How do you mean, while compressing or when its just passing AC?
And whats the cause of this?

Re your circuit; it looks like it would operate as an FET voltage divider attenuator rather than an OTA. Which is not a bad thing.

The basic idea i had was to use the FET as a variable current source beacuse of its high Z input, but i guess one could scale up the sidechain impedances and use a BJT.

 

[jdiamantis]

Here's a good ap note from Intersil that describes the phenomenon, and offers a circuit to linearize the input circuit; it's similar to the CA3280....

http://www.intersil.com/data/an/an6077.pdf

Here's the data sheet for the CA3280...

http://www.intersil.com/data/fn/fn1174.pdf

and here's the ap note for the CA3080...

http://www.intersil.com/data/an/an6668.pdf

Enjoy!

 

[Kit]

From the app. note regarding gain control:

Many manufacturers who have
nothing better to offer propose the use of a four quadrant
multiplier. This is analogous to using an elephant to carry a
twig. It may be elegant but it takes a lot to keep it going!

Funny guys....

Ok, lots of great info in the links. Thanks.
Time to study.

 

[peterc]

What about a discrete VCA, something like the Minimoog's?

http://hem.bredband.net/bersyn/minimoog%20VCA.htm

There is also the VCA in the Drawmer gates (& others) that is discrete.

Peter

 

[Kit]

Drawmer, eh.....

Is it this one?

http://img.photobucket.com/albums/v517/uncle_boko/DL231inputVCA.gif

 

[bcarso]

[quote author="jdiamantis"]... and the LM13600-LM13700 series (also, possibly no longer available).

[/quote]

The 13700 is still around and about, still done by National last time I looked and second-sourced as the NJM13700 by JRC; and the Philips after-Caen-fire-discontinued NE5517, which was essentially the same part, is made now by On Semi.

Some of the problems with these: old old low performance geometry PNPs, small NPNs with limited current capability and large rbb'. No ballasting R's in the emitters of the Wilson current mirrors.

All of these make for high noise. Nonetheless, I have used them and gotten about 90dB S/N with consumer-acceptable distortion---but it was hard work.

Why use them at all? Cheap cheap cheap.