JFET voltage amplifier

[Banfi T.]

Hi there,

I have been working on my amplifier at the weekend and I ran into some
confusion according the voltage gain and biasing stage.

One might ask the question why I spend my time to figure out such things
(voltage gain stages), lot of such topology is already available on the
Internet...Well, the truth is, that my mind was concerning too much about
topologies just to build someone's idea. If it is possible I want to build
an amplifier which at least in part a bit mine...And of course I have my
reasons why I tried to make things like in the topology below.

- I do not like symmetrical designs, because 1) "n" and "p" channel devices
are never completely complementary to each other, we all know this, 2) I
would like to use JFETs beacuse of their tube like caracteristic and finding
complementary JFETs are even more trouble, at least for me. So assymmetrical
design is one point.

- I tried to avoid differential amplifier at the input, beacuse the concjept
of it is based on the similarity of the involved parts. Some special BJTs
(MAT02 e.g) can provide a good result, but again I can not mach JFETs so
easily. (Of course there is something like the MAT02 with JFETs, but it is
very expensive. Check it at Borbely's site...). So, avoiding differential
amplifier was the second criteria.

- I like to use grounded base topology beacuse the gate or base capacitance is
no more a limiting factor when driving the stage. It means fast operation
with wide frequency range providing enough margin to the negative feedback
and higher slew rate can be achieved in this way with the same amount of
driving current which equals to lower dynamic intermodulation torsition.
(Source: Mark Alexander - A current Feedback Power Amplifier.) So grounded
emitter topology is not welcommed.

- DC connection at the input and at the output as well. This may not needs
further explanations why.

- Next criteria is to use as less stage in the voltage amplifying process as
possible. In this case it means two. It is beacuse I decided to use the CEF
as the output stage - I explain it later why. I could not figure out any one
stage topology which provides at least 15 gain having its input at zero volt
and its output near to 1 Volts. (Of course the output of a bridged amplifier
can be anything providing that both output has the same amount of DC. But
again, building completly similar amplifiers is a question of part matching
which I want to avoid. Even with X-feeback it is recommended to select the
MOSFETs as far as I know...) So, a to stage topology is likely what we can
have.

And why I used CEF, which uses BJTs insted of MOSFETs (of course the second
BJT in the CEF can be a MOSFET, but do I have to pair them?), it is even
symmetrical, and the second BJT works as a grounded emitter? The short
answer is beacuse it can be operated in free "run", which means that I do
not have to inculde it in the global feedback. It provides low output
impedance and fairlry low distorsion. It is something I like and want to
use. Besides, with a multimeter I can easily pair BJTs according to their
hfe and as far as I know BJTs are much more complemetary to each other as
the MOSFETs. And one more point, which is maybe a bit controversial...I
think the CEF is the only type of output stage which can benefit from the
theory of the split feedback. I do not describe here what "split feedback"
means, anyone who is unfamiliar with it can easily find out what it means
with the help of the search button above...The topology below does not
include it, but it is easy to imagine how it fits there. The CEF stage
already there moves into the loop and an another one gets in paralell
staying out of the feedback. The last reason to use a symmetrical output is
thus in this way the output current is not limited in a way as in the SE
outputs.

(Warning for newbies! (including me) The information above is not qualified
to be truth, before beliving what I have just typed wait for the reactions.
I may wrong in some or every statment I made here.)

So, that is the background of this topology. Sorry for being long, but I wanted to avoid questions about why I tried to do it in this way.

And now heading for my problem: Is this working??? I simulated it with
Electronics Workbench and sometimes it manages to calculate everything, (and
then it seems to be a HiFi power amplifer) sometimes the transient behavior
and the Fourie is an impossible task for it. Strange. Even more strange that
the THD does not changes at the output wheter I inculde the CEF in the
feedback or not, wheter I connect a 4 Ohms load or not. First I thought,
that the topology is may be instable, but the fact that sometimes it can be
simulated is against that. I tried to simulate a Borbely like symmetrical
cascoded two stage differential amplifier based (you know...) voltage gain
and the program was unable to calculate it. (I have backengineered the
values of some topologies available on Borbely's site and I've even modified
one which was staded as a working concept regarding Mr. Borbely (We had a
mail change about it some years ago) and using my schools simulator it was
stable. Here, it does not worked at all.)

So before buying all the parts and designing a PCB for it I call all the
experts to help me about getting sure about this topology...(And hey, I
could simulate it as many times as I coudn't, so hope is dying last...)

Oh, I almost forgot. The cascoding is used beacuse the feedback signal
drives the JFET as a gronded emitter and I wanted to speed up things.
Besides, the simulator does not showed any interested in it. Nothing has
changed with cascoding that stage....

And at least, this is just a first draw, for example the DC offset is to solve here. (Maybe with a comparator/integrator conncected to the second current source)

Anyway, thanks reading this...

bt

[tschrama]

What did you say... do I see that correctly?

a complex emmitor follower

[halojoy]

Thanks Banfi T. for sharing your idea.
I have taken a copy into my documents.
I call it: banfi_amp_01.jpg - as this is version 1
Will study it.

I am sure you will get suggestions
on parts and eventual changes.

/halojoy - likes new ideas

[Banfi T.]

Sorry for spelling mistakes I made...

Between I decided that I try to get somehow an another simulator because I got errors simulating a simple differential amplifier (no current source, some emitter and collector resistance, the simpliest you could imagine...). Nonsense. To my greatest regret I can not progress with the design without a working simulator. Maybe the DC reference problem...

I am reporting back when I have something new to say, anyway, I am still waiting for comments...

Halojoy...tell me that you named it after me because you haven't seen this shematic yet!



(Which is of course not a great deal if it does not work...)

Ups! I haven't noticed yet that the value of the supply voltage is missing...It is +/- 20 Volts regulated, but of course it can be greater or less as well....

[halojoy]

That is a most forgotten thing
in amplifier schematics -
At what voltage, is the amplifier going to work.

I always use a specific Volt Value,
when I try to design an amplifier.

Sorry to say, on the web, That important Data is missing.

Even if an amplifier can work well over a wide area of voltages,
there is one Recommended, and Limits (highest, lowest)
This is given in the datasheets of Power Amp Chips.

But not by those so kinda "amp experts"
on their websites.
At least it is missing too many timnes!

halojoy knows - he has been around the www for awhile

[halojoy]

Banfi T

Have you made any changes, or is this still
your working version?

/halo

[Banfi T.]

Thanks for the question halojoy...

To be honest - and to my greatest regret - this project is sleeping now. The simulator I have is nothing but a big sh*t, I know this for now...Meantime I tried to get some JFET here locally and you wouldn't believe, but only one type of JFET is available here! No p-channel device!!

I wanted to build just the voltage gain stage in the lab of my formerly school to see how it performs, but without the parts it is very hard indeed...

However I still think, that it would work and have a very unique "voice". Besides I wonder about that no one had any comment about it. I know that a lot of smart and experinced people are around here every day...(including the One and Only)

But I'm not giving up my hobby of course. I have ordered today the parts of my next power amplifier. It is fully BJT based and voltage gain stages are cascoded. At the input I am using a MAT04 quad BJT. It will have an exotic look because cascode stage biasing is done via blue LEDs. (They have about 3.7 voltage drop...)

I am working on the chassis in this weekend and I am expecting to finish it within a month or so...I will post the topology and pictures then to a new topic.

cheers:
bt

[PCP]

Don't you think that a non-simmetrical amp is a special case of a simmetrical amp where P-side does not work at all? (just in DC)

I think any simmetry will help.

[halojoy]

Originally posted by Banfi T.
Thanks for the question halojoy...

The simulator I have is nothing but a big sh*t, I know this for now...Meantime I tried to get some JFET here locally and you wouldn't believe, but only one type of JFET is available here!
No p-channel device!!

I have ordered today the parts of my next power amplifier. It is fully BJT based and voltage gain stages are cascoded. At the input I am using a MAT04 quad BJT. It will have an exotic look because cascode stage biasing is done via blue LEDs. (They have about 3.7 voltage drop...)

cheers:
bt

I do not SIMULATE - reasons are many
but in short It is not reality - when it comes to parts

It is hard enough to find true Complementart Bipolars
JFETs - never!

If my Wth was higher (high Wth* have N.Pass & P.Daniel and some few others at www.diyaudio.com )

I would buy some MAT-0X devices.
It is the best input stage you can select - IN WHOLE WORLD!!

Consider 2 Red LEDS in series instead.
My research, as well as other on forum,
indicates that this gives best, most constant
output from Currents Source. When used with bipolar.
It should be in close cotact. Have almost same temp koefficient.
Green and other diodes 1N4148 did not work as well.
I wouldn't imagine BLUE either.
Use RED LEDs.

Only alternative would be same bipolar BC550C
used as an amplified diode.
You can select any voltage, with help of variable resistor,
and temp coefficient would be same.

JFETS are good as current source
for the LED (the voltage ref) in the bipolar Current Source.
That is MY BEST use of JFETs
Each thing in its best place!

/halo - gives you a Tangband Pic - just as a bonus!

------------------------------------------------------
** Wth
a concept invented by halojoy, where he say you can judge
the designers Wth, by examine theor constructions.
Wth is a limiting factor in designs, for most of peoples on this earth
It even makes different dinner tables look slightly
different, all around this world; Bangladesh or Guatemala ...
-----------
Wth stands for Wallet-THickness
and was first mentioned in "Heat Issues" thread in PASS forum.
Noone dared to post an answer - especially not Nelson, the Forum owner
Well, some poor people lack courage

[Werner]

Banfi,

look here for the demo version of a decent simulator that should handle your circuit well.