Hi,
I would like to seek your opinion on this crazy circuit concept I was thinking off.
I was wondering what it would be to use a Sziklai configuration for a single stage amplifier, somewhat similar to the use in an F2, F2J or L'amp.
Instead of using a bjt as the driver, we replace that with a Jfet. However, the source current of the JFET would not be enough to drive a power transistor due to insufficient hfe. One way to overcome that would be to parallel a whole bunch of small signal transistors to replace the one power transistor.
So the end circuit would be a 2sk170 driving 50 pcs of BC557C (hfe 400) with each transistor taking in about 20mA. Giving a total quiescent current of 1A.
Why would I want to do this. Essentially it will be a device with a transfer function similar to a 2SK170 and the current capability of a few amp.
I did think about IGBTs but their transfer characteristic is not interesting.
BTW, I have done other crazy concepts for these single ended circuits too. Such as putting 25 NPN darlingtons in parallel to share a load of 30mA each...
Your opinions are very much appreciated.
Oon
I would like to seek your opinion on this crazy circuit concept I was thinking off.
I was wondering what it would be to use a Sziklai configuration for a single stage amplifier, somewhat similar to the use in an F2, F2J or L'amp.
Instead of using a bjt as the driver, we replace that with a Jfet. However, the source current of the JFET would not be enough to drive a power transistor due to insufficient hfe. One way to overcome that would be to parallel a whole bunch of small signal transistors to replace the one power transistor.
So the end circuit would be a 2sk170 driving 50 pcs of BC557C (hfe 400) with each transistor taking in about 20mA. Giving a total quiescent current of 1A.
Why would I want to do this. Essentially it will be a device with a transfer function similar to a 2SK170 and the current capability of a few amp.
I did think about IGBTs but their transfer characteristic is not interesting.
BTW, I have done other crazy concepts for these single ended circuits too. Such as putting 25 NPN darlingtons in parallel to share a load of 30mA each...
Your opinions are very much appreciated.
Oon
Last edited:
yea, nice idea, (...but i never heard of "Sziklai"...)
scribble up a shematic and we can see what your thinking and
we can discuss if its feasable.
..BTW, Papa did similar thing,
...and you 50 pcs are really not enough...😀
http://www.firstwatt.com/pdf/art_beast.pdf
scribble up a shematic and we can see what your thinking and
we can discuss if its feasable.
..BTW, Papa did similar thing,
...and you 50 pcs are really not enough...😀
http://www.firstwatt.com/pdf/art_beast.pdf
Why not use a MOSFET instead of bipolars for the output device? MOSFETS don't need input current other than to charge/discharge capacitances. I did something like that (two stages instead of one) in my ZGNFB class A amp -- http://www.diyaudio.com/forums/solid-state/276375-single-ended-class-no-global-feedback-8w-mosfet-low-output-z.html
How it looks like. My proposal is to use jfet for the first transistor and 50 pnp bipolar in parallel for the second.... With a small resistor at the emitter of course to share the load evenly between all of them.
Oon
Oon
Sent from my D6653 using Tapatalk
2sk170 can only do about 5ma quiescent current due to their IV curve. Transistors can do a lot more, limited by their power dissipation...
Lsk170 is now $30 for 8 pcs... Not something you can do easily.
2sk170 can go from zero current to maximum current with a Vgs change of 0.4 V. That makes it quite different from many other FETs. It has really high gain.
Oon
Sent from my D6653 using Tapatalk
2SK170 and LSK170 are available right up to 20mA
Even bl/B grades start at 6mA
They are not limited to 5mA
I think I read that LSK170 is now available in D grade.
Even bl/B grades start at 6mA
They are not limited to 5mA
I think I read that LSK170 is now available in D grade.
For normal class A operations, you only run it on half Idss. Compared to bipolar with some small transistors, you can even achieve 1A maximum current. Only problem is maximum power dissipation is one watt.
In theory, if you decide to cascode a transistor it could be run at a few hundred mA.
Sent from my D6653 using Tapatalk
In theory, if you decide to cascode a transistor it could be run at a few hundred mA.
Sent from my D6653 using Tapatalk
Actually what you did was interesting. Now you have a transistor with the current capability of the Mosfet. But the transfer characteristics of a bipolar.
Oon
Yes, it is a neat trick (not my invention, I ran into the output topology in an old thread by AKSA here at diyaudio- thanks Hugh}. It has the transfer characteristics of a SMALL signal bipolar (high hfe and bandwidth), though actually much more linear due to the pairing. It also has an almost absurdly low output impedance when used as a follower.
And for paralleling output sets, bjt-powerMOS sziklais have the input characteristic matching of small signal bipolars (each acts almost like a clone of the others), so its easy to parallel many for more current or to spread heat out over a heatsink (very important for high power class A). Highly recommended.
And for paralleling output sets, bjt-powerMOS sziklais have the input characteristic matching of small signal bipolars (each acts almost like a clone of the others), so its easy to parallel many for more current or to spread heat out over a heatsink (very important for high power class A). Highly recommended.
Apropos of nothing....
Sziklai being a perfectly fine approach falls under the "2 stage" definition in
the topological dictionary.
Admittance to this category is fiercely guarded against marketing jargon, and
so far admits only cascodes and diff pairs, and maybe mu followers, as not
constituting a gain stage.
All arguable, of course.
😎
Sziklai being a perfectly fine approach falls under the "2 stage" definition in
the topological dictionary.
Admittance to this category is fiercely guarded against marketing jargon, and
so far admits only cascodes and diff pairs, and maybe mu followers, as not
constituting a gain stage.
All arguable, of course.
😎
Ah, but if used as a follower, the circuit output is right at the "first stage"'s emitter.... Making the big FET just a hanger-on, not really a stage on its own...
🙂
🙂
Hi Nelson,
You are absolutely correct that this is technically a two stage amplifier. I was revisiting the F2, F2J, Zen topology and wondering if something else could be done.
The SiC are no more, the curves of the MOSFET sucks. The Bipolar does not have enough gain to have a decent input impedance. THE IGBT sucks as much as the MOSFETs, might as well use the mosfets.
I was looking at alternatives. One way was to parallel a 50 small signal BJT darlingtons, using those with a hfe of 50,000. Which I had tried, and worked quite well. Another one whihc I am working on is the Sziklai. It is a topology that will give a Id, versus Vgs of a Jfet, just that the Id is now 500X that of the original. If you used 50 pieces of small signal bipolars in parallel. Unlike a typical 2 stage, the source of the input JFET is connected to the "source" of the hybrid device. Therefore if used in a degenarative resistor, it would be as if the the transistors weren't there.
I would like to add that I had built the F2Js and they are beautiful sounding amplifiers.
Oon
- Status
- Not open for further replies.