Jfet BOZ

Thanks Nelson... I am a real geek with ultra simplistic circuits

This morning before going to the office.. I switched the 2Sk170 for 2N3819s ..and measured the THD adn Voptimum.. I tried a few.. .. I did make a small change to the circuit so were njot able to make full comparisons..
 

Attachments

  • jfet boz 2n3819 variations.png
    jfet boz 2n3819 variations.png
    13.7 KB · Views: 2,707
It seems that the 2N3819 might be a cheap alternative to teh 2SK170....

Results for 4 samples of 2N3819 1 Vrms out at 1KHz:

sample 1]
0.18% THD, pure H3, Vopt = 21 Volt, gain=18dB

sample 2]
0.15% THD, pure H3, Vopt = 23.3Volt, gain = 18dB
at 300mVrms THD was 0.02%

sample 3]
0.127% THD, pure H3, Vopt = 26.6Volt, gain 18dB
at 300mVrms THD was 0.015%

sample 4]
0.120% THD , Vopt = 24.5Volt, gain was 19dB

Using sample 4 in the shunt feedback schematic I got these results:

gain = 3.1dB
Vopt = 24.1 Volt
THD at 1 Vrms out was 0.039%


chrzz,
Thijs
 
I haven't done any listening tests.. but it seems that the 2N3819 follows the same square-root transductance law as the 2Sk170..

Besides that lower gain (18dB v. 26dB), you can expect more noise.. but nothing to worry about... I would say the 2N3819 (selected for Idss => 8mA) would make an excelent low-cost alternative.

grt, T
 
Thanks, tschrama, for this ideas and tests, because this FET is easy available and cheap enough for a precise matching and selection.

Sorry to ask a stupid question, but did you select the 2N3819 for the right IDSS before this very interesting test? And how?

I posted before the relative similarity of the JFET BOZ and the TNTAudio Preamble. I also found several other audio circuits in old ELRAD mags using the 2N3819 (in 12V circuits, mostly with another transistor for low impedance output or CCS).
I wonder why this JFET - although widely used "elsewhere" - is so underrated here: Is it only because of noise? That should not be a big problem in a line level circuit. But I do not have an THD analyzer to optimize these circuits.
 
Hi lohk,

No, i didn't select the 2N3819 jfets at all.


After the tests I measured the Idss of the last two JFET (sample 3] and 4]. The Idss of two fo them were:7.39mA and 8.53mA, optimum Voltage supply was about 25 and 27 Volts.

Yesterday I measured them again in the exact same circuits as I used for the 2SK170 (the original Nelson Pass JFET BOZ), so with two 2K2 resistors:
They had the same THD as the 2SK170 with Idss = 8mA, but need a higher voltage supply (22.3Volts for the 2N3819, 16.3Volts for the 2Sk170).

I don't understand why the 2N3819 isn't regarded as a fine alternative for the 2Sk170 in line applications.

Although internal resistance is a bit higher, transductance is a bit lower, noise is a bit higher... the 2N3819 works absolutely fine in this application.


Goodluck,
Thijs
 
No listening tests yet... but I should defenitely do that ... the distortion character will change dramaticly when you adjust the power supply for low THD.. from allmost 100% second harmonic, to allmost 100% third harmonic....:att'n:

PS
I also wondered about a 2N3819 complementary... so a bought a few and just received 50 2n3820 JFETs in my mail box today... 0.10$/piece
 
So ... because this Optimum Voltage, 2nd harmonic canceling, thingy that Nelson suggests works so marvelously... I got thinking...:

I have just completed a Pacific RIAA pre (with the MC pre) amplifier last month or so, and judging from my measurements, the suggested supply voltages are WAY to high.. so I hooked it up to my HP 339A (my pride and joy).. here goes:

Power supply +24Volt, as the circuit suggests:
1mV input, 400mV output, THD = 0.3% , purely 2nd harmonic

now with optimum power Supply + 13.7Volt
1mV input, 340mV output, THD = 0.0485% , purely 3rd harmonic


Yep... :cool:
 
tschrama said:
So ... because this Optimum Voltage, 2nd harmonic canceling, thingy that Nelson suggests works so marvelously...


Can you draw me a picture? I'd like to see how you graph this out in order to cement, or jack-hammer, my ideas on how this is done.

I've been using the ID/VDS graph and adjusting my amp's values until the transistor's voltage/current falls on one of the graph's curves. Is this correct?

Many thanks,:)
 
Nelson Pass said:
Think about it as presented in ZV9. Working the load line is
something that the smarter tube guys have done for a long time.

:cool:

It's a great feeling when a newbie, like me, can learn something important.

Thanks Nelson,--this is so cool! When simming out a schematic that employs the LU1014D jfet, I've been staying on the lowest VGS curve and adjusting my voltage/amperage to match that curve.

That's what got me to wondering about the mosfets--could it be the same?

It's weird how much amperage some of the mosfets must have to stay on a VGS curve--I'm thinking IRFP044N as an example. When tracking the lowest (4.5V) curve, I have to have about 6 amps for anything over 10 volts. That's a lot of amps to add to the bias.

The big, bad situation I can't always resolve occurs when the bias wattage is around 100 watts per device! To keep the IRFP044N linear, I have to run high amperage, but low voltage... or switch to a mosfet that's more suitable.

Still, I have a bunch of these guys, so I'd better learn how to do this.:)