Low-distortion Audio-range Oscillator

[davada]

Originally Posted by RNMarsh

Your on a roll.

I noticed there are electrolytic caps and trimmer controls that are not part of any feedback... for example the trim controls dropped thd by 6 db when replaced by better parts. Now, I see a pair of tantalum caps back to back as coupling caps. Well, that configuration reduces the cap distortion BUT we are after thd levels below what tant caps back to back provide.... I replaced the two tantalum back to backs (A3 C206,207) with a small 50v film cap and the third harmonic dropped another 6dB. No opamps have been changed, yet... waiting for adapters for the IC's to be mounted.

The THD (source plus analyzer) is now already 25 db below the -80db Full Scale, so we need a meter switch for X10 or -100dB F.S. range. When the opamps are changed, this will be more feasable (with the lower noise and distortion). Just need to trim the 2H down to much lower level. -Thx RNMarsh

I get the same on this 339a but only in certain settings. I wonder why. Try this in the overlapped next multiplier range. Like x100 at 10 on the left dial for 1 KHz and see if you.
get the same. Also try stepping the input range down 10dB and see if you get the same.

To have 1Vrms into the input of the notch filter you need to input -10dB of oscillator level.
Remember the input amp of the analyzer kicks everything up 10dB.

[RNMarsh]

Perhaps dc into the LDR will change with range switching caused by dc offset on the opamp outputs to vary to the LDR. Perhaps another cause of thd. Need to check. -RNM

[1audio]

The ST1700A does the same thing with internal distortion limits moving with input level and range. Not an issue when the DUT is .1% THD. Big issue when the DUT is .001% THD.

The QuantAsylum guys were quite responsive and helpful.

[richiem]

Rick -- note that the quad amp used by HP for the filters is a real stinker noise wise, and will need to be replaced with something much quieter if you're going to get lower distortion readings on the meter.

[RNMarsh]

Originally Posted by richiem

Rick -- note that the quad amp used by HP for the filters is a real stinker noise wise, and will need to be replaced with something much quieter if you're going to get lower distortion readings on the meter.

Yes, i believe this is so, also. We had noticed this early on. i have the replacement quad IC ready to go into it. in fact, I'll do it now. Thx-RNM

[RNMarsh]

Originally Posted by RNMarsh

Perhaps dc into the LDR will change with range switching caused by dc offset on the opamp outputs to vary to the LDR. Perhaps another cause of thd. Need to check. -RNM

Possibly requiring ac/cap coupling to the LDR (?) if it is dc sensitive (THD).

[scott wurcer]

Cleaned up my supply and added trims on the notch (>70dB). First the AD797 and second the AD844 just in case you thought a CFA would not work here. I suspect the thirds are down to the lightbulb's residual. This is a great find for anyone that simply want's to spot check a design, 30dB better than the 327 lamp oscillators that I built 20 yr. ago. There are two spurs left from my display drivers (easily ignored). It does take 10 or more seconds to settle, this goes hand in hand with the distortion. The correction is add 10dB to seconds and 5dB to thirds.

[davada]

Originally Posted by scott wurcer

Cleaned up my supply and added trims on the notch (>70dB). First the AD797 and second the AD844 just in case you thought a CFA would not work here. I suspect the thirds are down to the lightbulb's residual. This is a great find for anyone that simply want's to spot check a design, 30dB better than the 327 lamp oscillators that I built 20 yr. ago. There are two spurs left from my display drivers (easily ignored). It does take 10 or more seconds to settle, this goes hand in hand with the distortion. The correction is add 10dB to seconds and 5dB to thirds.

Does the color of the lamp matter?

Cheers,

[scott wurcer]

Originally Posted by davada

Does the color of the lamp matter?

Cheers,

I'd say probably not.

[davada]

Have a close look at where the 2nd H is now. This is at the expense of higher frequency performance. The distortion at it's best at 1KHz and 1Vrms into the notch filter.

I forgot to mention, in the list of what's been done, I put a 50k trim across A3R48 and trimmed the total R to 10k. This forces the LDR to operate at a lower resistance much the same as putting a trim in series with the LDR. It has the same effect.

New stuff.

I removed capacitor A1C46. Cut the trace at right at the pin5 of A1U1 and added a 33 ohm
cheapo $0.06 carbon composite resistor in series with pin 6 and the output trace.

Tweaked the Jfet distortion trim and this.

I'll see if I can fix the high frequency performance with out losing the 1KHz.

Cheers,