Possible power buffer for Victor's oscillator
For what is worth, I'd like to present measurement results of a headphone amplifier driving a load of 33R@3Vrms from Victor's oscillator.
The amplifier was developed by Shaq888, there are extensive measurements in his thread:
"The Crocodile"
Prompted by the excellent THD results the author presented, I've soldered one channel on the PCB Sergey shared, and made THD measurements with my DIY system. Apparently, I was able to come close to his results - the details are here:
"The Crocodile"
Re-reading the post, I now see that I made a mistake when calculating THD in the 5Vrms case - the value is rather about -140dBc.
Such an amplifier can easily drive 50 ohm attenuators. I tested it with a HP355c, but the THD jumped by more than 20dB, probably because of worn-out or oxidized switches.
Regards,
Braca
For what is worth, I'd like to present measurement results of a headphone amplifier driving a load of 33R@3Vrms from Victor's oscillator.
The amplifier was developed by Shaq888, there are extensive measurements in his thread:
"The Crocodile"
Prompted by the excellent THD results the author presented, I've soldered one channel on the PCB Sergey shared, and made THD measurements with my DIY system. Apparently, I was able to come close to his results - the details are here:
"The Crocodile"
Re-reading the post, I now see that I made a mistake when calculating THD in the 5Vrms case - the value is rather about -140dBc.
Such an amplifier can easily drive 50 ohm attenuators. I tested it with a HP355c, but the THD jumped by more than 20dB, probably because of worn-out or oxidized switches.
Regards,
Braca
A buffer is not a guarantee of extra cleanliness. Most of the time, the buffer itself is far dirtier than the controlling loop amplifier, and relies on the loop feedback of the controlling amplifier to get 'cleaned up'.
From what I understand, the OPA1656 has a very hefty output stage, so I would think that adding anything to it would degrade the performance with no benefit. These oscillators are designed to drive 1-10kΩ loads, so I'm not sure of the need for lots of power anyway.
I have used a THS3001 buffer within the feedback loop of op amps like the LME49710 and it does seem to help a little, but it's slightly finicky due to its CFB design and huge bandwidth. With a careful PCB layout it can work, but I don't have the ability with an APx 555 to say that it's truly better or not than no buffer.
From what I understand, the OPA1656 has a very hefty output stage, so I would think that adding anything to it would degrade the performance with no benefit. These oscillators are designed to drive 1-10kΩ loads, so I'm not sure of the need for lots of power anyway.
I have used a THS3001 buffer within the feedback loop of op amps like the LME49710 and it does seem to help a little, but it's slightly finicky due to its CFB design and huge bandwidth. With a careful PCB layout it can work, but I don't have the ability with an APx 555 to say that it's truly better or not than no buffer.
Last edited:
You probably mean Audio Precision, and even their top of the line $30k is not up to snuff to Victors, by a long shot.
When BC says it is hard, it is HARD ;-)
Jan
Right! I don't think you can compare this. AP can output all possible levels from 0..x volts and all possible frequencies from DC to 1MHz. And the THD and noise remains world class.
Can we update older versions by just changing the opamps ?
Or there is more to it ?
Thanks,
Patrick
Victor, a thought. Is the opamp the final limit for your oscillator? I have a couple of designs using the Groner/Polak dual opamp, which has been measured at better than -180dB @ 1kHz, and from calculations probably gets down to -200dB.
Would that benefit your oscillator?
Jan
No, not only the opamp limits the performance, especially at 1kHz. The capacitors also can make some effect (sometimes I need to remove "bad" caps from the boards), and the board mask quality, too. It is very hard to get the performance stable under -160dB. Not all the 1kHz boards can run that. I claim under -150db, but real limit is around -155dB. Now I am waiting for new PCBs, where some sensitive areas will be without the mask.
Awesome!
I'm currently building an active Twin-T to put your oscillators to the test. Won't be needing better than -150dB, but it can't hurt either
I find it interesting that the level control FET is not the limiting factor.
Have you tried other PCB material? Supposedly there is very low loss RF PCB material available.
Jan
Impressive performance. I suggest using broader FFT bins, though. 64-128 ks would allow you to capture the harmonics even if the oscillator frequency wanders a bit.
There's no doubt Victor's oscillator is good. Just make sure you don't let the FFT lie to you.
Tom
Rogers shouldn't be that expensive. It's a very common material. Whether it's lower loss than FR-4 is another matter.
Tom
There is the very small signal across the FET. Less than 0,5% of the full signal.
RF materials are very expensive, but the main problem was the quality of the solder mask , not the board material. So, soon we will see how the new boards run.
You’re right, I think the cheap Rogers is barely better than good FR-4 like Isola 370HR.
I agree with both of these observations.
There is much better PCB material than FR4 available from the likes of Rogers and Panasonic, but you are not likely to find it at the usual Chinese et al PCB suppliers.
Cheers,
Bob