What's inside an analog 4-quadrant multiplier?

Get it back, or submit a bill. :>)

Ahem, a few pages may not do it. Barrie wrote 81 pages in Chapter 2, a translinear tutorial, and 58 pages in Chapter 6, a current-mirror masterpiece. That's 139 world-class pages in this 640-page book, and arguable the best pages. They are among his best writings, SFAICT.

I was going to check the Tomazou book, but I loaned it to a colleague who saw fit to read it at home :(

I did find a few pages that are OK for public consumption (i.e. not proprietary) and have contacted Tim by email about faxing them. If anyone else is interested please post here and I'll contact you. I don't have a scanner so fax or mail are the only ways.

Steve

Nobody else has stepped up to bat, so I'm going to give my answer: logical ANDing of PWM streams?

(I believe this is described in the MIT Radiation Lab series on waveform operations.)

Tim.

That either has to be disqualified as not really analog or we have to increase the number of methods.

You can multiply with an MG set. On a generator RPM * FieldCurrent * K gives the output voltage.

We may have to increase it anyway because you can do it electro-thermally.

It should be possible to multiply using a couple of tungsten filaments as the nonlinear elements.

Certainly CdSe photocells. Or mosfets in their saturation region.

Somebody (Beckman maybe) used to sell a polychrystaline MOV-sort of resistor that had an exponential i/v curve, sold specifically for analog computation.

John

curves]

Those are all interesting implementations, but they aren't a new mathematical identity like Max was asking for. They're just varying implementations of the exponent/logarithmic circuit.

One interesting thing about the CdSe photocells is that they're very linear in electrical response while being nonlinear in optical response. That has very real advantages in many applications.

Tim.

CdSe cells are similar to CdS but have faster response. They were once popular as choppers, among other things. HP made a cool DC voltmeter that went down to 1 mV full-scale (or 1 uV? I forget) that used them to chop and demodulate, all clocked by a synchronous motor and a spinning slotted disc in front of a lightbulb.

When *I* was a kid, I generally used photomultiplier tubes, or germanium transistors with the tops of the cans cut off.

I made an IR detector once that used a flashlight reflector and a black-painted germanium transistor, measuring just the Ge leakage current. Now that I think about it, it probably accidentally used thermal runaway as a positive-feedback gain enhancement mechanism. It could easily detect the heat of your hand from six feet away.

John

CdSe? I've heard of CdS, although I can recall, as a kid, dismantling Selenium rectifiers and using a single plate as a photocell.

...Jim Thompson

Cadmium selenide photocells are very similar to cadmium sulfide ones, except that they have a larger infrared response. I can't remember if they have a higher or lower frequency response.

This isn't a great reference, but it shows the spectral responses.

That is one of the many electro-thermal methods. You can also use the fact that the power in a transistor is the product of voltage and current. There are many thermistor methods too.

I think that we've got the number of methods up to something like 6 so far.

Circuits involving SQUIDs produce outputs that are A*sin(B). If B is small, this becomes A*B.

For small changes, the photcell signal in an atomic clock is the product of the frequency offset and the lamp brightness.