You can't do better than the amplifier/passives by themselves. I am using one op-amp and have reduced the impedances so that the load is now <1k so this is not the ultimate. I don't know if I have the energy to build the two op-amp version with composite op-amps, and I certainly don't have the interest in swaping passives. I am sure you could make a single frequency source as good as any.
Ed, cascading does not help the op-amp in the second stage will just add the same distortion. I'm pretty sure with the sound card all the harmonics of the source are gone.
In this case the filter modifies the phase of the sidebands so that is why I think the PM is converted to AM. It was rather dramatic 10's of mV of amplitude noise, far worse than the source but the spectrum only shows magnitude.
Last edited:
How is it that other desings have lower measured thd? Even ones with AD797 as osc. It must be that that some cancellation is going on, then. ??
BTW - Can I build your circuit? I would like to duplicate your design idea and measure. can I get your schematic pls? And, which sound card and software did you use?
Thx-RNM
Last edited:
Since I have not posted (or accurately measured) the THD I don't catch your drift. I was too tired last night when I finally fixed my ground loops. With Samuel's discrete composite amps maybe it could be better. As I said I am playing with the one op-amp Wein bridge circuit which has common mode signal voltage and the op-amp is driving <1k Ohms so this will be no better than the distortion of the op-amp under those conditions. I simply want to show that the op-amp and passives are the only source of distortion left.
The simplest possible solution to the one frequency test tone is all I'm after here. Schematic later.
OK. Got it.
BTW-- after looking at a lot of 'prior art' and measuring them and poking around/tweeking etc as shown in this forum....I came to a similar conclusion about the osc and passives. In fact, no other upgrade does more to lower distortion than change osc opamp and sometimes passives on older osc. designs.
Next would be the compound osc/amp circuit for best osc. topology (and with harmonic cancellation?).
Thx,
RNM
BTW-- after looking at a lot of 'prior art' and measuring them and poking around/tweeking etc as shown in this forum....I came to a similar conclusion about the osc and passives. In fact, no other upgrade does more to lower distortion than change osc opamp and sometimes passives on older osc. designs.
Next would be the compound osc/amp circuit for best osc. topology (and with harmonic cancellation?).
Thx,
RNM
Last edited:
I think a lot of us here are waiting for such a result and appreciating your effort and sharing it with us. Simple, yet very, very good is a blessing for many. Thank you Scott!
Scott,
You know perfectly well the series suggestion was with the high distortion source. Now since you are using a sound card, the next trick would be to use it in stereo with the same sine wave but 180 degrees out of phase going into a pair of low noise Wein filters that use parallel opamps to lower the noise and increase the output drive. When you use those as a balanced output you really are cooking. (Of course there are fewer uses around these threads for a balanced output.)
You know perfectly well the series suggestion was with the high distortion source. Now since you are using a sound card, the next trick would be to use it in stereo with the same sine wave but 180 degrees out of phase going into a pair of low noise Wein filters that use parallel opamps to lower the noise and increase the output drive. When you use those as a balanced output you really are cooking. (Of course there are fewer uses around these threads for a balanced output.)
You would be much better off with a divided down crystal. perhaps 5 or 10 MHz divided down and a 10 KHz crystal filter (if you can find one) for the output. The Q of the crystal will be very high but its mechanical sensitivities may undo all the good work. Crystals are by nature piezoelectric vibration sensors.
True the Q determines the attenuation of the sidebands but the phase flip is nearly fixed over a range of Q's. Examining the real and imaginary components of the FFT confirm the rotation of + or - 90 degrees that turns the FM into AM. Probably a curiosity.
Somewhere I have a 1 KHz GR tuning fork. I may try to find it this weekend. I'm not sure it can make a big difference but its worth looking into. Similar to this: RARE Tuning Fork Oscillator by General Radio Engineering Marvel | eBay
Probably best to stick with coils, BTW don't expect heroic THD. One of those very high Q ringing "Zen" rods or even a Tibetan singing bowl might work.
Thanks David, you've just reminded me of all those Bulova watch ads from 40 years back!