Interesting Waveform Relationship

[MoltenVoltage]

That's great analysis, R.O.

Are you getting those values from some sort of circuit simulation program or are you actually building the circuit and testing it on a digital scope?

If it is a simulation program, what is it called?

[R O Tiree]

Thanks, MV.

The sim program is Circuit Wizard Pro, written by a company in Cambridge, UK and it can be bought from Maplins. It's really for schools/colleges, I guess, with a fairly limited set of components and no way to "roll-yer-own" as in LTSpice, for example, or any of hundreds of other SPICE analysis programs. Where it scores, though, is an intuitive GUI and the ability to turn pots, or alter any component's value on the fly while the simulation is still running. Want to find out about biasing trannies? Easy stuff, and you can see where things start to get ragged when you're trying to wring the last mV out of a gain stage and how badly it distorts when you try to do so. You don't have to stop and re-start and try to remember what you were seeing 10 minutes ago - it can be done in "real" time.

As to the curves it produced for this little essay well, having built several of these little monkeys (TS-808 tweaked), that's exactly what you see on the scope at the output of the gain stage. I'd often wondered why this sort of clipping was "soft" compared with clipping to ground and I was like a dog with 2 tails when I saw the simulated scope traces this afternoon. As I said, it was an epiphany, so I thought I'd share.

[Projectile]

Interesting. Thanks.

[MoltenVoltage]

Thanks for the info R.O.

I am looking for the simplest triangle wave generator circuit I can find, so I can start experimenting with the multiplier.  Here is one I have in a book:



Does anyone have something simpler?

[R O Tiree]

MV - first thing is that the log amp you posted is very cheap and cheerful. There are much better designs out there. If you can get access to "The Art of Electronics" (Horowitz & Hill) then there is a much better one at Chapter 4 on page 254 (diagram "O" in my 2nd Edition copy). If you want, I'll get it to you.

Secondly, although that oscillator circuit is "minimal", you still have the problem of how to output 2 triangle waves 90 degrees out of phase. A simple 1-transistor oscillator will do to get a square-wave - either that or a simple 1-opamp relaxation oscillator. If you run it at 4 times the frequency you want, then put in a little logic circuit to give you 2 square waves 90 degrees out of phase at 1/4 of the frequency, terminate each output with an integrator, then you're good to go.

Or you could program a PIC to do this trick...

[JKowalski]

There are simpler triangle wave oscillators out there, just google it and you'll find thousands.

How about a traingle wave -> comparator (1/2 wave trigger, trimpot fine tuning?) -> integrator? That would give you a decent 90 degree phase shift, from a triangle wave source, with just one dual op-amp. You'll need two more op amps for the triangle wave oscillator, so you can have the whole package to create a triangle wave and it's 90 degree cousin with one quad op-amp.

[MoltenVoltage]

Or you could program a PIC to do this trick...

That's like painting with Photoshop.  It's just not visceral enough for me, and you rarely have those "happy accidents" that inspire fresh ideas.

[JKowalski]

I have been experimenting with the simple op-amp/diode/resistor log amp in LTspice, and it seems pretty decent. I know what you are referring to, R.O., I remember seeing that one in TAOE, but for our purposes I don't believe we need much more then "cheap and cheerful".

However, there is the problem of designing a four quadrant multiplier circuit. A two quadrant won't cut it for this particular situation. And the only 4-quad analog multiplier chips (AD633 seems the most common) go for $5-7 bucks apiece.

A multiplier makes for a very nice VCA, by the way. Just multiply an LFO and your signal input, and you've got a way easy tremolo. Has to be four-quadrant again, however!

[Cliff Schecht]

The AD633 does a fantastic job of 4 quadrant multiplication, but IMO is prohibitively expensive for everyday use. The MC1496, on the other hand, is about $0.85 per single unit and is capable of 99% modulation with very low distortion. It's a somewhat tricky IC to design with because it's only got about 12 transistors inside with very little internal biasing, but this also makes it VERY versatile. I've used it at PAiA, as has John Simonton, with much success. I was actually going to use this IC for my perfect doubler (for the competition) but I couldn't find the time to finish that design. You should definitely check out the 1496 though, it's been around forever and is a well understood part.

The big trick with the IC is getting the input levels low enough to not distort everything. If you're looking for truly linear operation, the inputs have to be about 60 mV P-P max. There is a lot of gain available (this is set with a single resistor or pot usually) and if trimpots are used correctly on the inputs, the balanced modulation is near perfect.

[JKowalski]

Thanks alot cliff! I was unaware of the existence of that chip. That's gonna help so much. 

[gez]

I designed this a few years ago:



Relatively simple, doesn't require a scope to set it up and it outputs perfect triangles in quadrature if you use the recommended op-amp or a suitable replacement (output has to be symmetrical, preferably rail-to-rail).

Mouser sell the ICL7621. 

Although interesting work, it's going to be a lot of circuitry just to get a sine wave.