Anyone know of a design for a -simple-tunable- low distortion sine wave oscillator (audio frequency range)

The old Hewlett Packard audio oscillators use multi gang variable capacitors for tuning. They got a range of 10:1 for each switchable range from 20 cycles to 40 Kc.

I have a J. W. Miller Band Pass TRF AM band receiver that uses a four section variable capacitor for tuning.

I used it in making off-the-air FCC compliance frequency measurements.

And different capacitance per section. TRF used the same capacitance per section.

tions

y to

x 12

ugly

he

rces

LUT

ho

Thanks, I've only built 'crappy' Wein bridges stabilized with light bulbs. I could tweak them up such that the harmonics would almost disappear on the SRS 720(?). -90dB. but it wouldn't last.

George H.

adios

Some variable caps had slots cut in the outer plates so you could calibrate the top and bottom of the band, and then adjust the tracking by bending a plate section one way or the other.

-Bill

Hint: x=-d^2(x) or x=-x(double dot); from analog computer daze.

All of that could be in one micro..but i think plain analog would be better.

The outer plates were usually segmented, so you could bend them outwards a little to adjust the tracking. These tuning caps also used to have mica compression trimmers to allow two-point tracking adjustment.

You can do a one-off really fast when you need to. I built a really nice low noise LDO regulator this afternoon out of parts from my travelling junk box. Took about two hours start to finish. Dead bug rules!

Cheers

Phil Hobbs

Presumably you could predistort the digital data to null out harmonics, if you had some way to get the feedback info.

The cheap ones have 10 or 12 bit dacs. One of my guys figured out the optimal circuit to convert the differential DAC current outputs to a single-ended signal with minimal distortion. It's non-obvious. We're getting -60 dB to 32 MHz, but nowhere like 1 PPM.

John

The HP3312A has a 0.1 to 13 range on the dial, uses sawtooth generator; squares it for that O/P and a 6 diode-resistor array for each polarity for generating the sine wave; spec is

..and you will -love- the zipper effect...

Why filter a square wave, when one could filter a triangle wave which is half way there (like the HP generator i mentioned)? Better yet, use a generator that is mathematically sine/cosine?

The W.W. II vintage ARN-6 Direction Finding Receiver had a really nice variable capacitor in the front end. The front end was TRF and there were five or six identical sections.

Some? Only the really cheap ones didn't. Still, the differences between the RF and OSC sections tracked at the IF difference.

You might take a look at the Tektronix SG505 design. It used a dual 10k pot to go from 0.9 to 11 in four bands. Covered 10 to 110 kHz. The spec between 20 Hz to 50 kHz is 0.0003 % typical, 0.0005 speced.

If you want GPIB control, get the SG5010.

The 505 runs about $200 - 300 on ebay from time to time.

tm

They track well enough for a Wien bridge. I used to have a boxful of dual gang caps, and I built at least 3-4 working Wien bridge oscillators using them. It was pretty easy to get 10:1 frequency range on the cap, and switchable bands to cover from 1 Hz to 100 KHz using resistors was no problem. The tricky bit back then was to get the thermistor controlling the gain with a long enough timeconstant so it didnt add to the distortion at the very low frequencies.

I threw away my last good cap, a 20-1000 pf dual gang Philips unit, calibrated to 1% or better. Tracking was better than 0.1% over most of the range. Those were the days when electronics was made by precision mechanical engineering. Although I guess you can say an IC is precision engineered too, but on a different scale.

I can't. Which is the point for me. As I said I could live (happily) with .005%. Which I should say means .01% is liveable. The problem I have is tuning it without stuffing it all up. The earlier thread pointed to some linear design notes. One for bridge circuits that were right in the range I wanted and another one gave a design where the output distortion was in parts per billion - somewhat overkill for me.

zipper effect?

Don't you have a lot of trouble filtering out the residual clock frequency?

I was thinking analogue mainly for ease of construction and I'm happier in that world - but its not an absolute killer. If I found the right analogue solution it would be simulated by next week and a prototype within the month (then the hard part of making it work). Digital would take me longer

Found the pots. Thanks. I'm trying for specs.