Babelfish ᄅſ....or FW J2 on Steroids .... or Not your Father's J2!

[Zen Mod]

THD vs. Freq for all 3 iterations of interest

.....

 

[Itsmee]

Thank you, that's very interesting, especially when combined with your comments on presentation.

Looks like I'll have to build at least 7 variations - 3 with double output and one to satisfy curiosity - IXFN thingy up and 4 SS down

 

[Zen Mod]

naah

either two pucks up and 16 SS down ......... or nothing !

Bust or Glory!!!

 

[indra1]

... Looks like I'll have to build at least 7 variations - 3 with double output and one to satisfy curiosity - IXFN thingy up and 4 SS down

And then you'd be curious with RF exotics like the MRF151. 😀

 

[Dennis Hui]

naah

either two pucks up and 16 SS down ......... or nothing !

Bust or Glory!!!

Maybe a dozen input jfets? 😀😀😀

 

[Zen Mod]

no need , 2+2 are enough ....... or there will be even smaller THD ....... and we don't like that !



(besides , cascoded EF is there with a reason , so no need for oomphteent small critters ; catch is that each 2SK2145 is already double 🙂 )

 

[Dennis Hui]

Ah...good point. I was thinking more of single sissies like the 2sk170. 🙂

 

[Zen Mod]

even in that case , 2+2 is more than enough , considering that there is EF, actually driving output gate

of course , one can always piggy-back additional hundred or two

 

[generg]

Where is Generg with all his previous listening efforts.... it's all a re-hash 😀

I am here 🙂 and indeed inspired by the new discussion flared up, I looked after 7 years again more closely at the J2, with the help of Spice and real measurements.

Here my last guesses what the J2 and his variants do....

The OLG is fixed by two things.

1. the current of the CCS determines the load value for the J-Fets at the drains-

Lower current, more resistor load needed to get the bias voltage for the SS or IRFP. In this case the J-Fet stage makes more gain.

A simply way to calculate the gain of the J-Fet stage was shown by Nelson for the F5 some time ago.

Input Fet = .04 S, Degen = 10 ohm, Load = 500 ohms

Gain = 500 / (1/.04 + 10 ) = 500/35 = 14.3


We have no degeneration resistor here so we can possibly (?) do like that

I got for 15mA CCS current by the K170/F370 pair a value of 335R for 1.4A at the Semisouth.

So we have Gain J-Fets = 335/ (1/0.04 + 0) = 335/25 = 13,5x

For the F5 output stage Nelson stated

Output Fet = 3 S, Degen = 0.47 ohm, Load = 8 ohms

Gain = 8 / (1/3 + 0.47) = 8/.8 = 10

I supposed the gain is only done by the lower SS (may be wrong), I supposed 6S for R125, may be too high. Two 0.47 parallel as source resistor supposed.

so we have Gain SS = 8/(1/6 + 0.235) = 8/0.4 = 20

Total gain 13,5 x 20 =270x means OLG around 48dB.

I could confirm these values approximately with measurements.

So you see besides the transconductance, the drain resistor value for the J-Fets and the source resistor of the lower SS and its transconductance determine the OLG.

When we do the feedback network in a way to have 20dB CLG we get around 28dB feedback.

I wanted to have less feedback, maybe 20db.

So you must rise the source resistor of the SS and go even higher with the J-Fet CCS current. I think 20mA is the highest value for the four J74 one should use.

I took 1R as source resistor for the lower SS and stayed with the 15mA for the J-Fets.
So every of the four J74 gets around 4mA. A value where they seem to sound good.

In this way I got a gain of 13.5 x 6.85 = 92x means 40dB.

In real it worked! The OLgain of my "J2 try" is 40dB so feedback only 20dB.
A surprise for me was that the bias in the OS did not went down considerably by he 1R source resistor at the lower SS.

But the h2/h3 relation shifted. H2 was no longer my wanted 12-14dB higher than h3.

So I had to change at the SSs the two resistors a bit. The drain resistor of the lower SS and the source resistor of the upper SS.

You won't believe it I had to make the drain resistor 0R05 higher and the source resistor of the upper 0R05 lower to get the 14dB difference and keep the bias.

So you see how close Nelson works to get the best out of his "simple" circuits.

That makes a simple cloning nearly impossible.

It starts with the choice of the transconductance of the SS.
When Nelson really takes only 10mA for the CCS at the J-Fets, his SS seem to have a lower transconductance than mine

So as always, I am no expert and I learn here in the forum, with the help of ZM and some hints from Nelson. Nevertheless I might add the valuable hints and the breadcrumps to a false building...

🙂🙂

 

[generg]

Many thanks for the good wishes for my operation I have this month!

Brianco, thank you for your description of how it was with you, that gives hope!

😉

 

[Dennis Hui]

Wow. Thank you for writing this. Will have to read this carefully and try to absorb... 🙂

 

[EUVL]

6dB OLG missing due to the modulated current source.
Changing the source resistor will of course also change distortion spectrum.


Cheers,
Patrick

 

[generg]

Thank you EUVL.

 

[EUVL]

You can see Vgs and Yfs variations here :
https://www.diyaudio.com/forums/gro...tracer-matched-pairs-quads-4.html#post5892320
(post #188)


Patrick

 

[generg]

Thanks again!

🙂

 

[res07njc]

Thank you for sharing your notes and experience Generg, very generous .

(edit- and good health too, godspeed)

 

[Zen Mod]

Generg is always behaving that way

🙂

 

[generg]

Me and my assistant at kitchen table work.....

seems we are waiting for inspiration or dinner....

😀😀

 

[Zen Mod]

must be pic from yesterday afternoon , now it's little early for dinner , at least in our neck of wood 🙂

while you there , considering what you wrote in earlier posts , go look at graphs in #118

 

[indra1]

Your assistant looks c😎😎l. Glad to see you well. 🙂