sig gen

[Andrew]

Hmmm, I suppose if the diode is conducting on on only half the cycle it might explain what you are seeing.

Andrew

[Nick]

I would have assumed that diode would conduct when the input goes -ve, so you have a 10k load on the +ve half and a 5k load on the -ve ones.

[ed]

From my limited knowledge of electronics am I right in thinking that the diode shown on the input network makes the circuit present a different input impedance to the positive side of the wave from that of the negative side. If so, wouldn't that introduce harmonic distortion in itself?

I would have assumed that diode would conduct when the input goes -ve, so you have a 10k load on the +ve half and a 5k load on the -ve ones.

isn't that what I said...

so, would it not introduce some harmonic distortion, or distortion, some of which would be harmonic by luck?

Any idea what function the diode serves. My simple understanding is that I am looking for a zero crossing detect..so I compare the input sig with 0v and when it detects the op amp saturates and immediately outputs the Vcc value, or thereabouts.

[ed]

there is now more.....I'm replying to my own posts now, maybe somebody can call for the chaps in white coats......

ok, I scratch my head for the reason for the diode so I measured all dc voltages with and without signal....and decided to remove it.....

the signal and the measurements stayed rock solid on the scope.
the thd changed to this picture shown below:
once the diode was removed switching the LM339 circuit in and out of the output made no difference to the signal whatsoever.

result!...don't care what it was for it's not going in the final version..

[ed]

I am deducing that all the harmonic distortion shown in the previous pic is from the wien bridge alone.......

here is the signal from the oscillator and the comparitor

[Andrew]

The diode will only conduct if you're over a certain voltage, right?

So as long as you are under that voltage nothing will happen regardless of whether the diode is in circuit or out, won't it?

The question is what will happen if the input is over the voltage and how will that impact the comparator?

Andrew

[Nick]

From my limited knowledge of electronics am I right in thinking that the diode shown on the input network makes the circuit present a different input impedance to the positive side of the wave from that of the negative side. If so, wouldn't that introduce harmonic distortion in itself?

I would have assumed that diode would conduct when the input goes -ve, so you have a 10k load on the +ve half and a 5k load on the -ve ones.

isn't that what I said...

Yes, but as you prefixed it with "am I right in thinking" I answered confirming what you had said.

[ed]

The diode will only conduct if you're over a certain voltage, right?

So as long as you are under that voltage nothing will happen regardless of whether the diode is in circuit or out, won't it?

The question is what will happen if the input is over the voltage and how will that impact the comparator?

Andrew

well spotted that man....curses!
at 941 mv RMS the comparitor produces another pulse which lasts right up to max output voltage....this obviously doubles the frequency...

back to the drawing board

Yes, but as you prefixed it with "am I right in thinking" I answered confirming what you had said.

apologies I missed that in my haste.....got to slow down a bit...it's time to start reading and find out how the rest of the world do this thing instead of wildly guessing.

[Andrew]

Why not try two diodes, one to 0v and one to V+? They should conduct on both halves of the cycle then.... or add a potential divider to allow wider range of inputs?

Andrew

[jack]

If you are using an audio oscillator to check for distortion in an amp, the oscillator has to be as pure as is possible...

Rod Elliot has a pretty good write up of the pros & cons of various audio oscillator designs at http://sound.westhost.com/articles/sinewave.htm

My favorite has always been Jim Williams' ultra-low-nose design (0.0003% THD, i.e. 3ppm!) found in the Linear Tech app notes (AN43, page 29, "Wien Bridge-Based Oscillators", figure 47 - I never got this to be very stable), or the oscillator section of Bob Cordell's DIY THD meter - http://www.cordellaudio.com/papers/buil ... yzer.shtml - nods here to the HP339A and the HP200A (up to the HP200CD) (based on Bill Hewlett's MSc thesis).

An updated derivation of these has been written up at http://www.janascard.cz/PDF/An%20ultra% ... 0%20dB.pdf

As an aside, whilst I was having problems with loop stability of the Jim Williams design, I emailed the great man, more in hope than expectation. He replied, in great detail, within a day. What a gent.

EDIT: Photo of my testbench in Sept. 2003 - the small board being intimidated is the Jim Williams design from AN43.

[Andrew]

Well, there is this historic doc

http://www.hp.com/hpinfo/abouthp/histnf ... patent.pdf

My favorite 200A story is that several were sold to Disney for ~$77 so that the cinema's planning to show Fantasia could have their sound systems "tested and approved" by Disney's engineers.

The other story is when the first ones were built in the garage the two baked the paint onto the chassis using Lucille Packard's oven, she subsequently is on record saying the Sunday roast never quite tasted the same again.

Andrew

[jack]

The other story is when the first ones were built in the garage the two baked the paint onto the chassis using Lucille Packard's oven, she subsequently is on record saying the Sunday roast never quite tasted the same again.

That's part of HP's official history - http://www.hp.com/hpinfo/abouthp/histnf ... ents/0002/

The early analogue pioneers were genuine characters - Bob Pease, Bob Widler (absolute crazy dangerous guy), Jim Williams etc. Pease & Williams wrote excellently too...

[ed]

back to basics...this is much simpler than the example on the datasheet and has no diodes. I've tested it top to bottom 10mv-1.7vRMS and 15hz-21khz. It adds no measurable distortion and shows no 'features'.

fingers crossed.....now to try xr2206 and see if it's any cleaner than the wien bridge. Just to make life complicated I have a vacuum thermistor, which Rod Elliot raves about, so if the xr2206 doesn't show promise I can always build another wien bridge with it .... which Rod says has least distortion.....

[jack]

... xr2206 and see if it's any cleaner than the wien bridge.

The Exar 2206 is a synthesized sine wave and is quoted at 0.5% THD - is that really good enough for this application? https://www.sparkfun.com/datasheets/Kit ... 020808.pdf

[Mike H]

IIRC the resistors and caps values for the bridge must be perfectly matched, and track together, for the lowest distortion. ANY mismatch adds distortion. I think for this reason my own Wien null sig-gen uses a dual-gang tuning cap instead of dual-gang pot, only way they could get the distortion reasonable over the adjustable portion of each range.

If using a dual gang pot you will probably have to add a small value trimmer pot to tune for lowest distortion. I once made a Wireless World mag THD meter that used a Wien bridge notch filter to take out the fundamental, it had to have this trimmer pot to tune it out as much as possible.

Also IIRC there was a Heathkit sig-gen that DID NOT use a Wien bridge BECAUSE the distortion could not be made low enough that way, instead, some other network was used, can't remember what it is called - crossed T filter I think. Summat like that.

HTH (?)