# square to sine

Discussion in 'Electronic Basics' started by Sandeep, Nov 4, 2005.

1. ### John FieldsGuest

---
You're joking, right?

If you start screwing around with the time constant all you'll wind
up doing is changing the output amplitude and waveshape, which will
_never_ resemble a sinewave, no matter what you do.
---

2. ### John FieldsGuest

---
The OP's query asked for a circuit with a stable sine wave output
with a frequency equal to that of a squarewave input signal. Since
he specified a "square wave", I think we can safely assume a 50%
duty cycle and, therefore, disregard the even harmonics.

3. ### Bob MastaGuest

If you can get a good triangle, you can get a very good
approximation... easily as good as a standard "function
generator", which are typically 1% THD. Although the
8038 uses diode breakpoints for shaping, this is not
a very good method for individual use... too many adjustments,
even assuming you can figure out how to proceed with the

My method of choice is an overdriven differential pair.
You will end up with 2 adjustments: The amount of
overdrive (input level), and the amount of feedback.
If you feed the output to an amp and headphones,
you can adjust this by ear. (Though a scope helps
to get you in the ballpark.) You will be balancing
between 2nd and 3rd harmonic levels. 1% THD is
easy, and I have gotten as low as 0.25% on occasion,
though I don't know how stable that would be.
The by-ear method is better than a scope alone for
distortion adjustment. An FFT spectrum analyzer
is better still.

I have seen a paper where the authors used a matched
pair in a temperature-controlled oven and got
something like 0.001%. However, if you are shooting
for these sorts of levels, then the input triangle itself becomes
more of an issue. It's pretty hard not to have some sort of
glitch at the peaks.

Best regards,

Bob Masta