This is more along the lines of what i was looking for.... lets see if it can be applied here. yes, you did mention Amber doing so. How do they do it in schematic form? Thx-RNMarsh
Last edited:
As we are measuring outputs that can drive <600-2K Ohms or at least 10K, we dont need to use 100K (noisy) Z and other R's can be scaled down as well with the new opamps in place.
The dual-gate mosfet is for protection (CMode?). Shouldnt do much unless overloaded.
Protection of what? It's coupled to the center of the bias string clamp through 390pF caps.
There is a lot of 2nd H on TP101.
Cheers,
Attachments
I did take it out. It didn't do much so I put it back in. This is the first answer I've got for asking this question.
Thanks,
No, I think a better statement is that the FET wants to stay in the triode region well away from Vp. What you want is very high Vp which happens in general to go with long channels. The down side is that high Vp low Idss FET's are higher resistance and it's harder to make them be just a small portion of the feedback mechanism. The linearization tries to keep the FET looking like a resistor centered on the origin rather than off to one side, hence the trim tweaks the seconds. This trick is very old BTW well known when I started in 1973.
Thank you for the good timely info. yes, very old --- so old, i thought maybe something new has come up since way back when. Or at least some new insights garnered over the decades. I suspect that we can make amplifying ciruits better than we can measure them... or soon will be doing so If we cant over come these old issues with new methods/thinking. We'll look into alternative long channel fet parts that may be available now that werent back in the 339A development days. Thx moocho-RNmarsh
Last edited:
There is a sweet spot for local degenerative feedback where all of the second can be canceled.
I believe it is equal to the internal source resistance.
Most people use a pot to trim the value while watching distortion on a meter.
AN105 by Siliconix is a basic primer on the subject.
Myhrrhleine -- thanks... good idea -forgot about that one. Approx 1/transconduct. is a good place to start. Or a 50-100 ohm multiturn trimmer might do it. Thats easy to try and adjust other values if need afterwards for the oscillator. (just as SW did in his SW-OPA circuit for the input jFEts) Thx-RNmarsh
Myhrrhleine -- thanks... good idea -forgot about that one. Approx 1/transconduct. is a good place to start. Or a 50-100 ohm multiturn trimmer might do it. Thats easy to try and adjust other values if need afterwards for the oscillator. (just as SW did in his SW-OPA circuit for the input jFEts) Thx-RNmarsh
Last edited:
without output opa changed yet and no external notch filter (all internal gains are as stock) -- here is what i get from the THD guru known as the 725; at 1V rms and 1KHz --->
H2 = .00008%
H3 = .00008%
H4 = .00005%
H5 = .00003%
THD+Noise = .00012-15-ish% or approx -117 dB
-RNMarsh
Last edited:
And this is straight off the oscillator I presume?
That's really not bad. See if LME49710 helps with that.
It's about the same is what I get. So what's up with Victor's oscillator?
Maybe it didn't like the power supply. Tis possible. I need a new one.
I found that a big C for oscillator buffer amp feedback cap reduced disto by another 2-3dB.
At the moment 0.1uF. It's big and a smaller one would probably due but it was the only poly
I had at the time.
So let's work on the fet thing.
Cheers,
Last edited:
Its from the output binding posts on the front panel of the 339A !
yes, lets focus on that jfet.
Thx - RNMarsh
