Maker Pro
Maker Pro

JFET Power Amplifier With Output Transformer

rogerk8

Jul 28, 2011
179
Joined
Jul 28, 2011
Messages
179
You are an interesting man, I give you that.

But here is the fact, I have carfully tuned bias for the MOS to just about conducting (yet they do not conduct), I did this with a simulated "transformer" of 2X18 Ohms as "primary windings" (its the first oscilloscope graph above that shows obvious crossover distortion).

Then I changed to my ferrite OPT and got the above "perfect" results (using 12dB of feedback).

While I am very uncertain why there suddenly is not any crossover distorsion I have an idea why it is so, the use of 18 Ohms as load per MOS means that the loadline is much more flat than the effective MOS load by the OPT which is some 6 Ohms per MOS so if you imagine this loadline to be more "flat" at 18 Ohms than at 6 Ohms, MAYBE I am hitting a higher trace in the I/V diagram due to "Early Voltage" so that in the case of my ferrite OPT and its load the MOSs conducts a little more.

If this is true I do however think that the increse of bias current should almost be negligeable much because the difference in load actually is not that much while rp is quite high.

It should NOT give the remarkable improvement of the crossover distorsion as above.

At least this is what I think.

Best regards, Roger
PS
It suddenly strikes me that I must be wrong, if there is no signal the reflected impedance is infinity, the DC load (the OPT copper resistance) is however then the load so I am going from 18 Ohms to some 0,5 Ohm...
 

Audioguru

Sep 24, 2016
3,656
Joined
Sep 24, 2016
Messages
3,656
An output transformer that was used 60 years ago should have no DC current in it but have an AC load. Without an AC load then its reflected impedance is close to infinity then the crossover distortion will not show until it is properly loaded.
 

rogerk8

Jul 28, 2011
179
Joined
Jul 28, 2011
Messages
179
I have now checked the bias and to my great surprice there is no drain current!

I measure 0,5mV over each primary, the DC resistance of the primaries is kind of hard for me to measure because it is so low but we can do an estimation:

Now I really wish you had LaTex...

Anyway, here is the estimation:

Lp (poloidal circumference i.e the length of one turn)~4cm
Np (primary number of turns): 90
Lw (length of wire)~90*4cm=360cm=3,6m
rho (resistivity of copper)~1,67E-8
Dw (diameter of wire): 0,7mm
Aw (area of wire): pi/4(Dw)^2=0,38mm^2=0,38E-6m^2
Resistance=rho*Lw/Aw=0,16 Ohm

Drain current is thus around 0,5mV/0,16=3,16mA~3mA

To me, this is Class B.

Best regards, Roger
 

WHONOES

May 20, 2017
1,217
Joined
May 20, 2017
Messages
1,217
No. Class B has no current flowing in the output stage in a quiescent state. Even a small amount of quiescent current in the output stage makes it Class AB.
 

Audioguru

Sep 24, 2016
3,656
Joined
Sep 24, 2016
Messages
3,656
This amplifier uses obsolete Mosfets that have an extremely low gate-source "threshold" voltage. Normal modern Mosfets have a threshold voltage of 2V to 4V for a very low output current of 0.25mA but this 2SK175 Mosfet conducts 100mA for a gate-source voltage of only 0.65V.
 

rogerk8

Jul 28, 2011
179
Joined
Jul 28, 2011
Messages
179
No. Class B has no current flowing in the output stage in a quiescent state. Even a small amount of quiescent current in the output stage makes it Class AB.
I don't agree with you because everything is relative as someone legendary wise said.

So if maximum working current is in the order of 6 Amps a rediculously small amount of bleeding current of some 3mA is called Class AB?

I don't buy that.

I don't care about exact definitions, a Class B amplifier has very low or even no quiscent current, that is how I see it.

The point being made is that a Class B amplifier may be driven by batteries for quite a long time, a Class AB amplifier can not.

And this is the design goal I have, to make my batteries last as long as possible, thus Class B.

Best regards, Roger
 
Last edited:

WHONOES

May 20, 2017
1,217
Joined
May 20, 2017
Messages
1,217
If there is current flowing in the output transistors, they are biased on which is class AB regardless of what you think, The only time it could not be so is if the current is due to leakage which would be in the order of a few μA.
If you want to generate crossover distortion there should be no quiescent current flowing in the output stage.
 

rogerk8

Jul 28, 2011
179
Joined
Jul 28, 2011
Messages
179
I understand what you are saying but I do not understand the mechanism that is why is there no crossover distorsion when only a few milliamps is flowing?

What happens?

I think that the drain current is sharply cut-off and what's more is that the transconductance (i.e the vertical distance of the rp) is lower at cut off so you have the mechanism that as the level is reaching cut-off it does so in an unlinear manner because it can't smoothly cut-off due to difference in transconductance above and close to cut-off.

How far from the truth am I?

Best regards, Roger
PS
If you haven't said that part about leakage current I would have said "what about an insolator, there is none in the universe with infinite resistance".
 
Last edited:

WHONOES

May 20, 2017
1,217
Joined
May 20, 2017
Messages
1,217
To get crossover distortion you want no quiescent current flowing in the O/P transistors as this means they are biased on. This probably applies to the previous stage as well.
 

rogerk8

Jul 28, 2011
179
Joined
Jul 28, 2011
Messages
179
The previous stages is Class A and while current consumtion is less than some 3mA this does not matter.

What matters is the power stage and that it should not consume too much quiescent current, 3mA which I seem to have right now is however considered ok.

Best regards, Roger
 

WHONOES

May 20, 2017
1,217
Joined
May 20, 2017
Messages
1,217
You still won't produce crossover distortion like that because the output has been biased to class A and AB.
 

rogerk8

Jul 28, 2011
179
Joined
Jul 28, 2011
Messages
179
I can't say that I WISH to produce crossover distorsion, I just say that it is remarkable that such a low bias as 3mA eliminates it.

What I wish to do is to create an amp that can run decently (that is, not-so-much-crossover-distorsion) at so low a current as possible, that is my wish.

Today I did however come the conclusion that:
1) My ferrite transformer does not work (core saturation at some 600Hz at maximum output level)
2) My iron-powder transformer does not work (AL is too low, requires too many turnes that cannot fit)

Next step for me is some C-cores I bought from the world's leading transformer manufacturer here i Sweden, Lundahl Transformes, I will build FR2 bobins for these cores and handwind them for guitar amp purposes, not less than 150Hz that is.

Best regards, Roger
 

Audioguru

Sep 24, 2016
3,656
Joined
Sep 24, 2016
Messages
3,656
Your amplifier is very unusual with obsolete Mosfets that have an extremely low threshold voltage and an output transformer.
 

WHONOES

May 20, 2017
1,217
Joined
May 20, 2017
Messages
1,217
If you insist on using an output transformer, you should use the largest one you can get to avoid saturation. Laminated cores are best for lower frequency operation.
 

rogerk8

Jul 28, 2011
179
Joined
Jul 28, 2011
Messages
179
I am now on the verge to start to wind my transformer.

While I am designing a winding-machine which may be functional within a week I will wind this transformer manually.

I will wind it manually because there are really few turns needed (and the wire is rather thick).

I am aiming at some 2T@150Hz which gives me 2X60 turns, the bobin will be some 46 inner mm so with 0,7mm Cu this gives a maximum of 66 turns for one layer in each bobin/leg, inductance is then some 0,5H.

I have measured the C-core with regard to my_r and I am attaching the results as well as some pictures of the core.

Best regards, Roger
PS
I am seriously thinking of building the bobin using 2mm plywood, I have some FR2 patches but it is difficult to saw and the glue will have to be two-component epoxy which is rather messy, plywood can be glued using ordinary "white-glue" and the fit is rather strong, the only drawback for the plywood is heat BUT there really isn't so much hreat.

I love wood :)
DSCN4153.JPG
DSCN4154.JPG
DSCN4189.JPG
my_r.PNG
 

Attachments

  • DSCN4189.JPG
    DSCN4189.JPG
    127.8 KB · Views: 1
Last edited:
Top