Gsoz-t

[flg]

I've been working to improve the SOZ and ZV7-T design. So, to that end, I've come up with the idea pictured below. Let me first protect myself somewhat... No, it's not something new, maybe just several old valuable ideas put into one. I like to call it Grand Son Of Zen T.
This is still a Work In Progress
But, I was wondering if some of you would be interested in helping along the design process. The values pictured are of coarse "ThumbNail" as N.P. would say. But, I beleive it to be a funtional idea. And, I hope a good sounding one.
Any thoughts or comments would be greatly appreciated
Thanks.

[Babowana]

Very interested in with the CCS, cascoding and coil loading.
BTW, what controls amp gain . . .

[flg]

Babowana, mostly the 1 ohm R between the sources. Like PaPa SOZ
I think the gain may be higher than I expect though. That being the case, a little larger cross coupling resister value might be appropriate???
I don't think my Pspice model is good enough for the JFET yet, but I will try a simulation soon. I've been playing around with a version that only has an IRFP244 as a gain device. It seems to be a small improvement over the ZV7-T. Also, I'm having trouble actually sim'ing a real xfmr. Inductors work O.K. but when I try to use a real Xfmr, I get the dreaded, Does Not Converge!
Hey, maybe you could help me with somethin... I would like to convert my OrCad Capture Pic to something the Forum accepts that is of better visual quality. I havent been to sucsessful at figuring that out They look O.K. on all my graphic apps

[Babowana]

Originally posted by flg


. . . the 1 ohm R between the sources.

Yeahhh . . . . . . and the two 0R6 . . .
So, the gain depends mainly on the speaker impedance . . . ?

[flg]

Well, I'm not sure mainly is the best description but yes. The .6 ohms have a big effect also but so does the crosscoupling ressitor. Like SOZ... ZV7-T also suffers this problem even with F.B.

[Magura]

You can get away with 2*80mH inductors.

Magura :-)

[Nelson Pass]

Very likely you can eliminate the 0.6 ohms and couple the
cascode reference to the other side of the 1 ohm, or even
make the 1 ohm into an optimal value based on distortion.

Also, note that while the cascode modulation straightens out
the 2nd harmonic, so does the differential, and you may find
that entirely different values for cascode modulation are called
for. This means a distortion analyzer.

[flg]

Thanks About the Inductors, I just figured getting this big should get the response down pretty low. At least a few octives above the 8 ohm point would be nice. I have a little help from the experts. They should get back to me in a few day's. I asked for matched windings of 270mH at less than 1 ohm DCR and >3A Isat. 2 samples is all they'll go for... But, that's all I need...
Nelson, thank you for the suggestions. But, do you mean a form of cross coupling the cascode refrence??? "to the other side of the 1 ohm"???
Yes, my MOSFET version seems to have mostly 3rd harmonic in the simulation. I have not been successful simulating with the cascoded JFET yet.
Well, I have access to an HP339 now but, Hmmm, I was hoping to get a new job soon and I don't know if I will have that benifit much longer...

[flg]

I finally got a model for the JFET working nicely in the ZV8 circuit So, I tried it in the GSOZ-T and all I get is some switch mode looking output (Oscillation City). So, I replaced the Inductors with R's. 35 ohm R's and 100V +V. I still could not get a decent output until I used a 220p from D-G on the cascode transistors. Lots of gain and low THD levels. But, that's not the answer. Especially with an inductor back in there. Caps cannot be used in this position with the Inductors, without creating an oscilator
I took Nelson's suggestion of removing the .6 ohm Source R's. It also did not seem to like the cascode refrence on the other side of the cross coupling 1 ohm???
I Also noticed a mistake in the way the I Sources are fed by the bootstrap. I split them into two seperate voltage sources

[flg]

I worked on the oscilation issue but in doing so I ended up with a simpler design. The simulated performance is better. I was able to use a real transformer model as the load, rather than just inductors. And, I have adapted 1 of Nelson's suggestions with good results also.
I'm not sure the Pspice jfet model will actually produce the "Triode Character". So, this may be the reason the cross couppling of the cascode refrence is still not showing a benifit
I guess I can improve the model or build this thing and play with the distortion analyzer and the 1.2 ohm resistor