FET-hex explendit amplifier

In fact, John Linsley Hood proved these vertical mosfets by Toshiba to be the most linear devices, more linear than BJT and LatMosfet, when used in circuits (not per se) in his famous experiment with classic Lin topology.

I would not contradict JLH, (which would be silly), but I think the rest of us often do not interpret correctly what people like JLH or Nelson Pass say and write. I very much doubt that when he said "in circuits", he meant in every kind of circuit imaginable.

Having said that, I agree that this would be worth exploring, and in fact I bought 4 each from NicMac with something like that in mind. Except for the cost of shipping them from Italy to the US, the price seems reasonable for what you get, in terms of tested and matched parts, and he ships VERY promptly, and well packaged.

The main issue I see is controlling a layout and the rest of the parts well enough for a valid experiment which is repeatable by other people, and consistent in comparison testing...

Although my experience with audio is limited, I have been through this in other (instrumentation-type) applications, and getting consistent, and repeatable results that depend on the linearity of power semiconductors, and especially complementary pairs of power transistors, is a big job. By that I mean an order of magnitude bigger that the effort I have put into the through-hole version of VSSA/PeeCeeBee.

Not trying to be discouraging, b/c I would like to do it myself, :D, but just realistic...
 
I would not contradict JLH, (which would be silly), but I think the rest of us often do not interpret correctly what people like JLH or Nelson Pass say and write. I very much doubt that when he said "in circuits", he meant in every kind of circuit imaginable.

Having said that, I agree that this would be worth exploring, and in fact I bought 4 each from NicMac with something like that in mind. Except for the cost of shipping them from Italy to the US, the price seems reasonable for what you get, in terms of tested and matched parts, and he ships VERY promptly, and well packaged.

The main issue I see is controlling a layout and the rest of the parts well enough for a valid experiment which is repeatable by other people, and consistent in comparison testing...

Although my experience with audio is limited, I have been through this in other (instrumentation-type) applications, and getting consistent, and repeatable results that depend on the linearity of power semiconductors, and especially complementary pairs of power transistors, is a big job. By that I mean an order of magnitude bigger that the effort I have put into the through-hole version of VSSA/PeeCeeBee.

Not trying to be discouraging, b/c I would like to do it myself, :D, but just realistic...
Pete,

That's good (critical) thinking. I am just retelling some old stories. JLH did that experiment because people like Doug Self claimed that vertical mosfets (and mosfets in general) are unsuitable for linear applications like audio power amps because of low basic linearity. Doug used some simulation programs and JLH wanted to see for himself so he built the same circuit with different output tranies, measured and showed that Toshiba's vertical mosfets are more linear in that particular circuit than both excellent BJTs and IGBTs by the very same Toshiba. The whole thing started because Toshiba wanted to prove that (than) new IGBTs are the best.
 
Today I have soldered vas - 2sc4793/2sa1837 with 22pF compensation cap, results bellow.
 

Attachments

  • vas 2sc 2sa 22pf.png
    vas 2sc 2sa 22pf.png
    126.3 KB · Views: 1,395
  • vas 2sc 2sa step 20kHz 22pf.png
    vas 2sc 2sa step 20kHz 22pf.png
    112.6 KB · Views: 1,361
  • vas 2sc 2sa.png
    vas 2sc 2sa.png
    132.2 KB · Views: 1,338
  • bjt vs fet.png
    bjt vs fet.png
    388.3 KB · Views: 1,301
Last edited:
Yes instead off bd139/140...
These come in different speeds (see suffix), so it would be helpful if you posted which ones you are using. So, BD139/14016STU are fairly fast, and available from Fairchild. Without the -16 suffix, perhaps not so fast, but they still seem to give a good result in your test. KSA1220/KSC2690 is also available, and proposed in the PeeCeeBee thread as alternative to KSA1381/KSC3503.

I built my version of PeeCeeBee with unmatched KSA1381/KSC3503, with good results so far (although I would prefer to use a true complementary pair in the future).

Here is a link to some test results, and below my post there are more scope pics from another through-hole VSSA version by Nikosokey.

(My tests were done with 47uF compensation cap)

Square Wave Test
 
These come in different speeds (see suffix), so it would be helpful if you posted which ones you are using. So, BD139/14016STU are fairly fast, and available from Fairchild. Without the -16 suffix, perhaps not so fast, but they still seem to give a good result in your test. KSA1220/KSC2690 is also available, and proposed in the PeeCeeBee thread as alternative to KSA1381/KSC3503.

I built my version of PeeCeeBee with unmatched KSA1381/KSC3503, with good results so far (although I would prefer to use a true complementary pair in the future).

Here is a link to some test results, and below my post there are more scope pics from another through-hole VSSA version by Nikosokey.

(My tests were done with 47uF compensation cap)

Square Wave Test

Yes bd were -16 the high gain ones, they are doing good job but only up to +/-40V
 
Hi borys,

Want what be the suitested combination for +/-35V rail :

- all 2sc4793/2sa1837
or
- ksc3503/ksa1381 and 2sc5171/2sa1930

Marc

For +/-35V rail bd139-16/bd140-16 (they have hfe around 200) for vas stage and driver 2sc4793/2sa1837.
I have made some tests with above transistors and they perform very well.
I am checking other cofigs cause there will be 4 output pairs version.
The all 2sc4793/2sa1837 version is performing very well too.
 
borys
can you let us know what is the total bias per channel with the BJT version.
What I would like to know if you have any bias drifting under the warm up period or when you push the amplifier hard.
My version uses darlington power transistors, at low bias under 50mA the drift acceptable but when we bias the amplifier around 150mA the bias drifting play bigger role.
I do love the sound of the amplifier a lot, but these bias movement make the amp a bit unstable.
Same bias movement I got at the M. Hiraga to.
Some people wrote with BJT output these topology has these issue.
I would like to know what is your experience
These issue we have only with BJT, with LatFet rock solid..
Thank you
Greetings Gabor
 
borys
can you let us know what is the total bias per channel with the BJT version.
What I would like to know if you have any bias drifting under the warm up period or when you push the amplifier hard.
My version uses darlington power transistors, at low bias under 50mA the drift acceptable but when we bias the amplifier around 150mA the bias drifting play bigger role.
I do love the sound of the amplifier a lot, but these bias movement make the amp a bit unstable.
Same bias movement I got at the M. Hiraga to.
Some people wrote with BJT output these topology has these issue.
I would like to know what is your experience
These issue we have only with BJT, with LatFet rock solid..
Thank you
Greetings Gabor

Cold amp I have 68mA, very warm amp I have 90mA - but at the moment I have very bad contact with radiator (the mikka pads under o/p tranies are dry without silicon paste) - cause I am removing the pcb for tests many many times.
Bias stability - after I finish the project 100% this will need only one maybe 2 days to get sorted.