200W jfet mosfet amp, no gnfb but output error correction - from my archives

Status
Not open for further replies.
Jump to Latest
S

soekris

Member
Joined 2009
soekris.dk
Paid Member
#1
Found this nice 200W power mosfet amplifier in my archives. It's fully symmetric with Toshiba jfet input and current mirrors, Hitachi mosfet drivers and output devices, with bipolar cascode devices. The unique is that there is no global negative feedback and only about 20 db local feedback in the output stage, much like a Hawksford error correction circuit, just much simpler.... At that time I was also inspired by the Tandberg TPA3016/3026 which also was no nfb with hawksford error correction.

I designed this in 1991, tested the output feedback circuit on a breadboard, then designed and build a couple of boards. Then it stalled, don't really remember why exactly but probably something about drilling and taping holes in heatsinks by hand....

Now that Linear Integrated Systems are finally shipping p-ch jfets it could probably be built, maybe just a little updating needed, the Hitachi mosfet drivers are not available anymore.... Not really suited for series production with all the matching of parts required....
 

Attachments

#3
Deserves investigation..................does anyone have the Tandberg schematics for comparison? This might be the start of the best new thread in a while.:up::up:

Thanks Soren.
 
Last edited:
#8
RNMarsh said:
It looks interesting - esp the no global nfb and OPS correction. Did you ever model it and get some idea of the performance?

[I still have all those new mos/fets in a drawer]

THx-RNMarsh
Click to expand...

I don't really do sims, especially those fet models are never that good, prefer building up something on a breadboard....

But 4 sets of Hitachi power mosfet by themselve should have something like 0.2% THD at full power and 0.25 ohms output impedance. The 20 db feedback should then improve that to 0.02% and 0.025 ohm, which is perfectly fine in my opinion. THD can be improved with higher bias current, although cooling will then become an issue.... Rest of performance should be pretty good, with all jfets and mosfets.

I did some comparative testing of Toshiba/Hitachi/Bipolar output devices a while ago:

http://www.diyaudio.com/forums/soli...er-mosfets-2sk3497-2sj618-little-testing.html

I also found a set of Tandberg TPA3026A schematics, got them off the web some time ago, although I actually also have a paper service manual deep down in my files....
 

Attachments

  • Tandberg_tpa3026a_1.jpg
    Tandberg_tpa3026a_1.jpg
    134.1 KB · Views: 876
  • Tandberg_tpa3026a_2.jpg
    Tandberg_tpa3026a_2.jpg
    165.5 KB · Views: 852
#9
Thanks for the schematics Soren.

Maybe a good area to explore is the effect on frequency response with and without error correction an comparison to an amp with global feedback applied. I guess the comparison should be made with a resistive load as well as reactive one.

Jam
 
Last edited:
#11
UltimateX86 said:
HI soekris

Q25/Q26 are the output error correction ?

thanks
Click to expand...

Yes, Q25/Q26 is the error correction. The source resistors to be selected to get 2.5 mA Id at 0V Vgs. Jfets from GR group will probably work best, BL should also be fine but will result in less feedback due to larger source resistors. Not that much is needed, the idea is to get just a factor 10 improvement in Rout....
 
Status
Not open for further replies.
Top Bottom