F5X -- the EUVL Approach

Jump to Latest
#864
I have PMs asking me about the 5.1R degeneration and matching between the K170 & J74 Idss.
For those who have Q sets from us, this is a non-issue, as we did all the work for you.

For those who want to build on their own, or copy this for single ended F5, suitably paired K170/J74s might be difficult to find.
In that case, you may wish to do one of the followings :

1) for F5X, where the X is floating, or the H configuration (as published in Nelson's F5T article), you may still use the 5R1 as is, as long as the 2x K170 has matched Idss, and also J74s; i.e. NN PP match. The junction at the sources of the complementary JFETs will not be at 0V, but have a slight offset. This does not affect performance much in the F5X.

2) for F5 single ended, if Idss difference deviates from the nominal 0.8mA, you will need to change the 5R1 resistor to a value which gives you a resulting Id of the J74 which is close to the Idss of the K170. That way, and with well matched MOSFETs, you will still get good (but less perfect) even harmonic cancellation and close to zero DC offset.

3) for F5 single ended with unmatched MOSFETs, just build and trim 2nd harmonics with the poti trick as published by Nelson. Or even better, ask Nelson for his expert advice in the F5 thread. 🙂


Patrick
 
Last edited:
#866
In theory you can. Especially if you are prepared to burn current and use low impedance resistors.
It only works properly when your input signals are perfectly balanced.

I have not try if that was your question.
If I have to do that I would at least use an active Ground by using something like a BUF634 with two resistors.


Patrick
 
#867
In any case class A will burn current anyway... but yes active virtual ground is better.

If everything is well balanced, there shouldn't be any current flowing in the ground?

Only source current of the fets will go to the ground?



If the input is not 100% well balanced is-it a problem?

I plan using 2 TDA1543 dac one for + and the other for -, there can be a mismatch in symmetry....
 
#870
I've been trying to find an example of why this unstable equilibrium does not work in practice.
I think I have found one.

If you study my CEN IV circuit,
http://linearaudionet.solide-ict.nl/sites/linearaudio.net/files/didden-la -vol2euvl.pdf
http://www.diyaudio.com/forums/digi...-sen-evolution-minimalistic-iv-converter.html

you will notice that I included two resistors R1, R2 which have no real function in the circuit.
And in theory the circuit will centre itself so that the top half will see half of V1 as the bottom half, by symmetry.
In practice however, the circuit will tilt to one side so that e.g. the top half gets all of V1.
With R1 & R2 the balance is restored and stable, even though their allow very little current through.

You can apply the same thinking to the F5X.
In theory it will balance itself without any Ground reference. In practice it doesn't.


Patrick
 
#871
can I use passive virtual ground? what would be the max value of the 2 resistors?

What feedback configuration will be the most stable and have the less THD?

H connected to ground

Pass H lifted from the ground

X lifted from ground (in this case only the gate resistor of the input will go to the virtual ground ?)


with P3 on each side to try null out 3rd harmonic?
 
#872
H & X lifted are the same, and marginally better with distortion.
Grounded X is most stable. But it is then essentially 2x Singled Ended F5.

P3 will NOT null 3rd Harmonics. Nothing will.

I already told you how I would use a single 40V supply.
I have not changed my mind, but you are welcome to experiment anyway you like.
And always welcome to come back and tell us your results.


Patrick
 
#874
P3 has no use in F5X if you match your NN & PP devices properly.
The intrinsic symmetry of the F5X will cancel even harmonics.

The 220R at the output provide a Gnd reference for the output.
You may also argue that it is unnecessary theoretically.

Try to understand the circuit, and perhaps do some simulation first.
Getting simple answers from me does not help you to understand how it works.


Patrick
 
#875
thanks a lot for help, then I will still try to find a centre tapped transformer.


Another question,
Can I use 2 J74gr in parallel with equal Idss of the K170 ?

Maybe the gm of it will be higher than the K170, then I need to follow your degenerating trick.

But in this case how to calculate the value of the extra source resistor? and what should be the Idss of the 2 combined J74gr? still 0.8mA more then the K170? or even more?

This is difficult for me to calculate... it is over my skills.

this info could help others to build F5 with j47gr still available on the market
 
#879
Over the years we have been asked many times for F5X PCB sets and matched devices.
We have declined every time because a set of boards without the case is no use.

One of the senior (in terms of experience) members here expressed a willingness to supply the case for a limited batch to DIYers.
So we wish to know if there is still any interest before we make the necessary preparations.

It will be a last-time GB limited to 10 sets.
They will include matched FETs incl. regulator FETs, PCB set and fully machined case.
You should be looking at a total building cost (fully finished) for a stereo amp of around 3000€.

I've opened an interest gauging thread at the GB forum.
Please indicate any interest there.
https://www.diyaudio.com/forums/group-buys/330034-final-gb-f5x-power-amp.html#post5602963

There will be no guarantees if we go ahead, but we shall try our best to make it happen, some time in 2019.

You can see some more nice pictures here :
Balanced F5X Power Amplifier withdrawn from sale - Classifieds - Audio - StereoNET

It is very likely that we shall also complete the Beta test of the F5X Preamp in 2019.
And then we shall offer that also some time in late 2019.
Everything is working, but a lot needs to be done before it can become a robust public project.


Patrick
 
You must log in or register to reply here.
Top Bottom