Passive tri to sine converter

RLC lowpass, or tuned LC bandpass, even better.

Unless you consider diodes to be "passive."

John

--- For a single frequency around 100Hz, try this:

IN>------+ | [13µF] | [100µH] | +----->OUT | [10R] | GND>-----+----->GND

Here's the circuit file for Linear's SwitcherCAD:

Version 4 SHEET 1 880 680 WIRE -80 256 -80 112 WIRE -80 432 -80 336 WIRE 112 112 -80 112 WIRE 112 224 112 192 WIRE 112 320 112 288 WIRE 112 432 112 400 FLAG -80 432 0 FLAG 112 432 0 SYMBOL ind 96 96 R0 SYMATTR InstName L1 SYMATTR Value 100e-3 SYMBOL cap 96 224 R0 SYMATTR InstName C1 SYMATTR Value 12e-6 SYMBOL voltage -80 240 R0 WINDOW 3 24 104 Invisible 0 WINDOW 39 0 0 Left 0 SYMATTR Value PULSE(2 7 0 0.005 0.005 0 .01 100) SYMATTR InstName V1 SYMATTR Value2 AC 1 SYMBOL res 96 304 R0 SYMATTR InstName R1 SYMATTR Value 10 TEXT -360 360 Left 0 !;ac oct 256 10 1000000 TEXT -362 506 Left 0 !.tran 1

-- John Fields Professional Circuit Designer

I recall that Intersil ? made a function generator chip that used a 'diode ladder' arrangement to bend that triangle into a sine.

Graham

In terms of a working circuit, what is my best bet for simulating a sinewave from a triangle input using only passive components?

IOW no transistors or op amps.

Supply is 9V single rail. Voltage in is 5V. Frequency is lower audio range.

Thanks very much for any suggestions.

Jim Thomas

LC is surely the way to go if its fixed frequency. One RC stage can not give you sin, no matter what the R and C values.

NT

For the "lower audio" range, LCs are not highly practical. What's so horrible about active devices?

The simple way to do this is to build a sine wave oscillator and lock it to the tri wave, or (failing this) to build a highish-Q active filter. If it absolutely has to have no battery, run a voltage doubler off the tri wave and power the circuit that way.

Cheers,

Phil Hobbs

Damn straight. worked pretty well, THD < 0.5%. for a single cap was tunable over 2 decades in frequency. was second sourced in its day too, about 30 years ago. 8038? 8308? 8388? damn, it was 30+ years ago and my memory not is perfect.

John, shoudn't that be 200mH? Dick

If diodes are allowed, set up a few diodes along a voltage divider to progressively compress the triangle wave. HP did this with considerable sucess in their 3300A function generator.

Exar made a sine-wave generator, the XR2206, that did this too. We used to use them in modems.

Exar used the exponential I-V characteristic of a couple of diodes to round ('clip') the 'points' off a triangle wave. For lowest distortion, the input triangle amplitude must be carefully adjusted/controlled. Exar's XR2206 datasheet related their technique in splendid detail. Alas, I don't seem to have my copy (which is rare--I don't usually lose things).

I've used LEDs as the diodes to get a higher voltage output.

Cheers, James Arthur

The Exar part might be an 'improved' version They did that quite often.

It's still there.....

Graham

Cheers! Rich

Hey, thanks! I wasn't sure they still existed.

I don't have my *original* XR2206 datasheet, but I did find an old app note. Their method was not exactly as I described -- I'd mixed it up with another, related circuit.

Here's Exar's technique for converting triangle waves to sine waves:

. . Vcc >--+-------------+ . | | . | .-. . | | | . | | | R.load . | '-' . | | . | o--------------> Sine wave output . | | . |/ \\| . /\\ / >-------| Q1 Q2 |-----+ . / \\ / |>. .---+ o--o-\\/\\/\\----o | . | | Re | | . | | | | . Triangle wave | i1 | i1 | | . input | | | | | | . | | v | v | . | | | | . | === === | . | GND GND | . | | . | | . +---------------------------+

KEYWORDS: triangle to sinewave converter waveform waveshaping logarithmic

Re is adjusted to bring Q1/Q2 close to cutoff at the input's peaks, causing the transfer characteristics to become logarithmic there rather than linear. Consequently, the triangle's peaks are rounded off, and a low-distortion approximation of a sinewave at the output results. The accompanying info indicates THD of 0.5% can be achieved with a single adjustment.

Jim Williams reports a similar circuit he calls a 'logarithmic shaper'. His version has two tweaks, an input-level adjustment he calls 'wave-shape trim', and an input offset adjustment he calls 'symmetry adjust'. JW reports 0.35% THD.

Best Regards, James Arthur

Excellent, and there's the sine conveter in all its glory.

Graham