Richard is extremely lucky in that Cool CA is pretty much an EMI low zone. Googlemap his address which is on the avatar and you'll see what I mean.
He can get away with a lot more than most of us and has a much better chance of getting a interference free measurement. Demian has to be much more resourceful, he lives in the Bay area, nuff said.
Cheers
Alan
who grew up about 50 miles from where Richard lives now.
He can get away with a lot more than most of us and has a much better chance of getting a interference free measurement. Demian has to be much more resourceful, he lives in the Bay area, nuff said.
Cheers
Alan
who grew up about 50 miles from where Richard lives now.
Richard thanks. It was due to your suggestions some years ago that i converted my test setup to using isolated and filtered AC - i found this to be essential.
The cables are I guess more a case of guilding the lilly.
Yesterday I redid the tests with the unused ends terminated with 100r termination. I found that the starquads made a very worthwhile difference (look at the 50hz component which is down 17db or so) and so I will be converting all my test cables to the as I go along. Of course ymmv
That looks very worthwhile. I will make up some star-quad test cables also.
For just incase. I can take some to Bangkok office/facttory also... big city environment.
THx-Richard
Last edited:
Low distortion audio oscillator
Now we have star quad sorted I would like to continue on the path to an oscillator.
I have a query about Mike Renardson's state variable circuit
The feedback around the FET looks "sub-optimal" to me, the 10k and 3.3k will divide by 4 but I expect 2, have I just missed some point?
Best wishes
David
Now we have star quad sorted I would like to continue on the path to an oscillator.
I have a query about Mike Renardson's state variable circuit
The feedback around the FET looks "sub-optimal" to me, the 10k and 3.3k will divide by 4 but I expect 2, have I just missed some point?
Best wishes
David
Now we have star quad sorted I would like to continue on the path to an oscillator.
I have a query about Mike Renardson's state variable circuit
The feedback around the FET looks "sub-optimal" to me, the 10k and 3.3k will divide by 4 but I expect 2, have I just missed some point?
Best wishes
David
There is no four quadrant multiplication in the multiplier. You might have trouble leveling this at frequencies above 10kHz. The oscillator is at risk of going to saturation.
It's not obvious but the opamp has a gain of 2.
I'm not sure about David's comment about the attenuation/gain range of the FET.
There is also the issue of common mode on that opamp which can contribute distortion.
As you noted the FET controls the level in opposition to a fixed amount of positive feedback.
Bob's leveler is not so different from Mike's but is more easily conceptualized as a four quadrant circuit.
I have slowly realized that a true four quadrant multiplier is an obvious solution, only about 9 years after Samuel published it
.Not cheap and noise is an issue, as Samuel noted, but tolerable for my application.
I have looked at this lately because my balanced dual FET idea turned out to be not so simple once I tried to add distortion cancellation.
I could use Scott's photovoltaic driver idea to float the distortion cancellation but then I need two dual FETs, which seems a bit clumsy.
I could use two NSL 32 opto-couplers in a balanced circuit but I am not sure about their continued availability.
Hmmmm.
Not that obvious to me, seems it depends on the feedback around the first leveler op-amp and the values aren't specified.
I will compare it more carefully to Viktors' circuit, which does divide by 2, and think a little more.
How Bob's circuit works out to produce the correct ratio is quite obscure to me.
I could try to simulate it but would like to be able to work it out, any simple explanation?
Best wishes
David
Last edited:
Assume for sake of analysis that the first leveler op amp output is at 0V. The circuit then resembles a gain-of-two non inverting amp.
The FET Vds is divided by 4 and applied to the non-inverting input of the FET-driver op amp. Its non-inverting voltage gain is 2. Av= 1+(4K7/4K7). The overall voltage gain for the distortion-reducing loop is 2/4 or 1/2. The inverting gain (from the integrator output) is -1.
Viktor's circuit divides Vds by 1/2 and applies it back to the gate. His FET-driver stage cleverly combines the integrator with the FET driver. The non-inverting gain is 1 so the net distortion-reducing feedback to the FET is also 1/2. There is an added gate resistor to limit current and possibly for capacitive balance.
You probably looked at that and an knew the gain is 2 with out calculation. Nothing wrong with the shortcut of 1+ inverting gain.
Still in production...got some recently:
https://www.digikey.com/product-detail/en/luna-optoelectronics/NSL-32/NSL-32-ND/5039800
Yes, I checked too.
But Cadmium is not RoHS compliant and the CdS photocell production has been passed from company to company, now Luna, previously Silonex, previously Vishay was it? then National I think?
So I suspect they could be orphaned, like many other non-mainstream parts.
In particular the famously now unobtainable R53 thermistor (at least this was no loss, I have one and it's horrible
).I have also read that the opto-couplers don't match that closely, so distortion cancellation may be more limited than one would like. Anyone have data on this?
But they do make it simpler.
At the moment the multiplier, FET and opto options don't have a clear winner.
Multiplier is more expensive and a bit noisy but simple.
Opto is a bit suspect on future availability and on distortion cancellation but fairly simple and fairly cheap.
FET is a bit more complicated if I want balanced distortion cancellation but fairly cheap and no RoHS cloud.
VCA also looks reasonable.
Eventually I expect Davada's MDAC will win as DAC/ADC takes over the world.
I just hoped for a cleverly simple solution to do it quickly but I haven't found one yet.
If anyone has a nice circuit I would love to see it.
Best wishes
David
Last edited:
I did actually, briefly, consider this because the low distortion has obvious appeal.
But the backlash and wear as the pot hunts over a small section of track makes it impractical, as I am sure you realize
Best wishes
David
Not necessarily. A servo pot can have essentially no backlash and 1,000,000 cycle life or more. http://www.jameco.com/Jameco/Products/ProdDS/2201519.pdf Normally they are pretty expensive but not always: 5127R1K-L.50: BI Technologies : 7/8'' Round Wirewound Potentiometer<br/> 1 kΩ 1 Watt, 5% Single-Turn Panel Mount : Passive Components $7.
A valid question is how much change is needed once the loop is stable.
A valid question is how much change is needed once the loop is stable.
But the backlash and wear as the pot hunts over a small section of track makes it impractical, as I am sure you realize
Best wishes
David
I agree this approach is doomed.