Thanks.
Do you expect a group buy on some boards then? I'm just trying to get an idea on where this thread is leading. The goal, in other words.
Well I suppose if one of us makes a break through boards may become available.
You have to clear what you're looking for.
Are you looking for inexpensive medium performance full range or expensive high performance full range? What about a price target?
I like this one:
http://www.users.on.net/~glenk/thd/generator.pdf
which appears to be a variation of Bob Cordell's SVF oscillator.
but I'm willing to build an HP 239A clone (bridge T) also.
So, I guess the former comes under the expensive high performance full range and the latter is less expensive less performance, maybe? Why not both? Why not start with the former and make improvements to it (which I don't have any idea what those would look like)?
Okay then put an SVO on a separate board from the AGC and multiplier.
This way you have an option for future modifications.
Boonton 1121 AGC circuit
I have attached the Boonton 1121 AGC circuit. This circuit has fast settling and works without any time constant changes from 10 Hz to 140 KHz controlling a SVF oscillator.
Point c is the sine output, point b is the cosine output and point a is the dc voltage back to the AGC element (an analog multiplier in this case). The absolute flatness is in the .01 dB range across its full range. I don't know the settling time but its fast enough to use for swept measurements. An informal test from 500 Hz to 20 Hz showed no obvious change in level.
It may seem complex but all the other solutions are similarly complex and I would guess the cost of an extra wafer on a switch for range caps would not be much less expensive.
The HA2625 opamp is uncompensated and they used the compensation connection to limit the output swing to work with TTL. You would need a separate 5v supply for this circuit.
I have attached the Boonton 1121 AGC circuit. This circuit has fast settling and works without any time constant changes from 10 Hz to 140 KHz controlling a SVF oscillator.
Point c is the sine output, point b is the cosine output and point a is the dc voltage back to the AGC element (an analog multiplier in this case). The absolute flatness is in the .01 dB range across its full range. I don't know the settling time but its fast enough to use for swept measurements. An informal test from 500 Hz to 20 Hz showed no obvious change in level.
It may seem complex but all the other solutions are similarly complex and I would guess the cost of an extra wafer on a switch for range caps would not be much less expensive.
The HA2625 opamp is uncompensated and they used the compensation connection to limit the output swing to work with TTL. You would need a separate 5v supply for this circuit.
Attachments
Do you mean a good double balanced mixer, or a beat frequency oscillator? Beat Frequency Oscillator 700a Equipment General Radio
I ws thinking Gilbert Cell multipliers might be better suited to audio - but not certain of practical limits - obviously Spice perfectly matched Q models will give over optimistic results
and what are we practically aiming for -120, -140 dB for < 20? kHz harmonics + IMD, Spurs?? from 24/192 aduio DACs generating input sines in 50-100 kHz range
and what are we practically aiming for -120, -140 dB for < 20? kHz harmonics + IMD, Spurs?? from 24/192 aduio DACs generating input sines in 50-100 kHz range
Last edited:
Yes. . .
I think there are several goals here. Mine is characterizing ADC's (Victor's oscillators are probably enough for any available ADC). Others may be looking for a very good general purpose oscillator for less that the $1,000 or so of a KH4400 or equivalent. And its an interesting technical puzzle/challenge, like high end audio, which may have far exceeded any actual utility.
I think there are several goals here. Mine is characterizing ADC's (Victor's oscillators are probably enough for any available ADC). Others may be looking for a very good general purpose oscillator for less that the $1,000 or so of a KH4400 or equivalent. And its an interesting technical puzzle/challenge, like high end audio, which may have far exceeded any actual utility.
Hi Demian,
Thanks for this.
There isn't much on the schematic that isn't obsolyte. They don't make sample and hold amplifier IC's like they used to. I'm not sure one could find suitable replacement.
Nice circuit though.
Everything but the smp-11 is still current. AD has a number of possible replacements and the Intersil part may work in both slots. and its function is less demanding than the first SHA. The ssi is all still around in various form.
The point is that its quite possible to make a suitable agc circuit that doesn't need tuning.
I have a similar solution that is mostly discrete if you prefer. It would be cheaper as well.
The point is that its quite possible to make a suitable agc circuit that doesn't need tuning.
I have a similar solution that is mostly discrete if you prefer. It would be cheaper as well.
This leveling loop does have a proportional integrator time constant, or it would not be fast settling. It's again a (though rather basic) variant of the sample-and-hold detector, with gated integrator.
As in many threads on this board, there is a strong tendency to avoid any clear design goal. Necessarily we're just wobbling around, and its unlikely that anything useful will ever emerge from the mud...
There are tons of FET switches available, and caps and opamps too. That's all you need for a suitable sample-and-hold. The true problem is not about parts; it's about knowledge. I'm ready to give detailed support, but I'm not ready to just hand-in a working design.
Samuel
Last edited:
I'm not an expert, but I would say that if an oscillator was better than 1ppm THD with noise @ -130dB or so, over the range of 10Hz to 100kHz, then that would be a pretty good oscillator.
This is an excellent suggestion. In fact I went one step further, and separated AGC and multiplier too. This makes it very easy to improve either one, and try out new ideas without wasting too much money.
dirkwright, have you looked at the SG505 now? I think this would be a very easily clonable design, with very good performance.
Samuel
I'd be perfectly happy if you had boards and/or a schematic available for use. I suppose you make money off of this kind of work though so maybe that's why you're not willing to share something like this. I'm Ok with that.
I can make my own boards no problem. I just need to know what to make. Given the support on here for an HP239A or a Bob Cordell clone, that's all I really have to work with that isn't really complicated. I'd have less confidence of success with a complicated circuit vs. a more simple one on my own. I like the simplicity of the HP239A but most here feel that an SVO like Bob's is a better oscillator.
I meant for the Booton circuit. Of course there are plenty of parts available for a THSH and numerous ways of going about it. I just haven't found one that I happy with the performance of noise wise.
OK, so you do have your own oscillator up and running.
I downloaded the SG 505 file but have not studied it yet. Thanks for reminding me.