Go Back   Home > Forums > >
Home Forums Rules Articles diyAudio Store Blogs Gallery Wiki Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

Solid State Talk all about solid state amplification.

JFET input, MOSFET VAS, LATERAL output = Perfect!!
JFET input, MOSFET VAS, LATERAL output = Perfect!!
Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Thread Tools Search this Thread
Old 12th June 2011, 06:52 PM   #1011
GregH2 is offline GregH2  Australia
diyAudio Member
GregH2's Avatar
Join Date: Oct 2009
Originally Posted by mikelm View Post
I think it's best to choose the CLG you need and engineer the OLG to give the amount of FB such that compensation is hardly needed.

Yes, I think this is a good idea and it's how I've gone about my last few iterations. It can be hard though! Today I tried 200r of vas degeneration with the zvp to no avail! Crazy huh?

It's taken me a while to get the workflow clear in my head, but now I have it. The last circuit I posted embodies everything I have learnt, please note that I have dropped the jfet drain resistor to 200r to u the current to 5ma.

I think it will be a winner. Stable with no compensation, but only just. Needs a snubber. Right on the edge. A gain of 16. Perfect for me. Peferct square wave with zero ringing. 100khz of easy bandwidth. Overall I'm pretty happy. Can't wait to hear it.
  Reply With Quote
Old 12th June 2011, 06:58 PM   #1012
KSTR is offline KSTR  Germany
diyAudio Member
KSTR's Avatar
Join Date: Jul 2007
Location: Central Berlin, Germany

Does anybody notice that OLG in SWF's amp seems to be heavily depending on output current, as the bootstrap still loads down the "current portion" of the VAS gate swing with 1k5? So at low (or even most of the audio) frequencies effectively the dominant current in the JFET is a scaled-down copy of output current, not output voltage. That is an interesting property of bootstrapping a VAS load resistor from the output voltage, especially with a loudspeaker load. EDIT : Note effect of snubber on it.

With a CCS'd VAS this "current portion" would see the same light load than the pure voltage swing and output current wouldn't be reflected back much into the JFET.

I tend to like current-dominated control loops better...

And yes, ...really the most exiting thread here in a while...

- Klaus
  Reply With Quote
Old 12th June 2011, 07:49 PM   #1013
lineup is offline lineup  Sweden
diyAudio Member
lineup's Avatar
Join Date: Dec 2005
Location: the north
Originally Posted by mikelm View Post
I think it's best to choose the CLG you need
and engineer the OLG to give the amount of FB such that compensation is hardly needed.
That is the way, mikelm
To introduce a compensation cap effects not only stability.
We wish not have any comp cap.
Hiraga, JLH & Nelson Pass, all very little users of cap.
Classical amplifiers WITHOUT comp caps.

Schematic below is such an amplifier.
The feedback resistors are 3300/220/220uF.
I use no resistors for the IRF9610. No Gate, No source.
AC Analysis shows it should be stable without cap.
(2200/150 needs a small miller cap.)

I use IRF610 as Current source for the IRF9610 VAS.

This amplifier should have a good performance
and still probably a good sound.
Attached Images
File Type: png s_fetzilla_2c.png (8.8 KB, 556 views)
  Reply With Quote
Old 12th June 2011, 10:09 PM   #1014
Sheldon is offline Sheldon  United States
diyAudio Member
Join Date: Dec 2003
Location: San Diego
Originally Posted by swordfishy View Post
Yes, you are right, we are going around in circles a bit and this is mostly due to the fact that I am currently the only active builder (as far as I'm aware). Unfortunately I am perhaps not the best person to take things forward in a logical and systematic fashion. I have only been into this hobby for a short time and am still full of excess enthusiasm.
No apologies, it's an approach that has some advantages. Not very efficient if you know exactly what you are after. But I'm a scientist too and work this way often, particularly when something is new to me. I have a hard time sticking to a linear path. It's like diving into a pool and cruising around until I know the shape of things. When I feel have a rough grasp on the territory, and only then, do I buckle down and head for the goal line. If you get on the straight line path too early, you might miss a lot. Even if you don't understand at first what you are seeing, understanding can pop up when you don't expect it.

If you are in that half awake state early in the morning, and you are dreaming about this stuff, then you are doing it right.

I should add, that if you take wide ranging exploratory path, you should be careful to record as much detail as you can, and keep it organized. In the interest of full disclosure, this is not my strongest skill.


Last edited by Sheldon; 12th June 2011 at 10:17 PM.
  Reply With Quote
Old 12th June 2011, 11:38 PM   #1015
sonnya is offline sonnya  Denmark
diyAudio Member
sonnya's Avatar
Join Date: Oct 2001
Location: Denmark
Default Moment of truth!

Originally Posted by swordfishy View Post

I have just made yet another incarnation of this amplifier that I think looks very promising. You may recall that I wasn't impressed with the JFET front end and BJT VAS. Well that was when I thought I needed a lot more drive for the fets so I was using the BD140. This one uses the BC560, which I think the JFET will have much less trouble driving.

Note that it needs no compensation, just a very small output snubber to keep it clean with no load attached.

Check out that 20kHz square wave into a 8R load at nearly full power...perfect!

I need to get some more current through the jfet, but my current dual led bias arangement won't let me as I can't turn on the JFET hard enough to get enough current through a smaller drain resistor. Will add another LED tomorrow so I can get enough voltage drop across a smaller resistor (like 150R or so).

Anyway, will post listening impressions tomorrow. I think it could be quite good!
First of all, i am not trying to send the circuit to graveyard.

I have just put the circuit into the simulator. Or should we basicly.

I know that i have added input lowpass filter, DC voltage adjust, C5 is only 10uF and the Zobelnetwork is disabled (4.7Ohm + 1fF cap)

R13 is modified to 600 + 150R (RCOMP).

I have added a CCOMP of 1fF -> not active

I wanted to see how the circuit behave.
On the first picture the circuit oscillate and the load is 1MOhm pure.

On the second picture the oscillation is removed by adding CCOMP of 100pF.

The slewrate is assymetric in both sims and just little smaller with CCOMP added.

On the third picture the load of 10nF||8R is added + CCOMP of 100pF + Zobel is 4.7R+22nF.

The Zobel is needed to add stability into capacitive loading as SWF,

On the fourth picture the load of only 10nF is added + CCOMP of 100pF + Zobel is 4.7R+22nF.
  Reply With Quote
Old 13th June 2011, 01:13 AM   #1016
GregH2 is offline GregH2  Australia
diyAudio Member
GregH2's Avatar
Join Date: Oct 2009

Interesting simulations, this is more or less exactly how the oscillation presents itself. However, the stability is much better than your simulation - 22nf doesn't faze it.

Also, won't all amplifiers with a single ended VAS have an asymmetric slew rate?

Going to go test the stability in more depth.
  Reply With Quote
Old 13th June 2011, 01:34 AM   #1017
AKSA is offline AKSA  Australia
diyAudio Member
Join Date: Sep 2001
Location: Melbourne, Australia

That 100KHz square wave indicates that driving the laterals ain't such a big problem. With a few pairs, and/or hexfets, yes, driving with an EF (like Workhorse, and in one of my amps, the NAKSA) is a good idea. In this case, horses for courses....


Great post. Good to see engineers getting along with golden eared chaps.


All SE amps regardless will have asymmetric slew rate. As long as the lesser value is well over about 10V/uS this is no big deal. It does indicate that much of the distortion will be even order, but since this is (mostly) musical this is no bad thing.


Maybe run the jfet at 2/3 of its Idss? The transfer curve is extremely abrupt at low currents, which would create lashings of H2/H3, so 5mA seems about reasonable?

In your attempt to build a VAS without compensation (indicating lots of degen and a very low transconductance front end and maybe a bootstrap chucked in to help with top end rolloff) you might be going a bit light on lag compensation.

In a bjt in particular, the quality of the parasitics in the device, which largely center around the depletion layer, are appalling. It's a leaky cap, essentially, with lousy Q. Yet this capacitance helps with the miller capacitance and is augmented by the real miller cap outside the device.

Some lag compensation is no bad thing, particularly if it is high quality, and I recommend silver mica. By using a LC cap, you tend to play down the poor quality of the internal parasitic cap. The next issue of course is to choose a good transistor; I have some KSA1142 Fairchilds I can send you. Email me your address and they are yours.

Great transistor, BTW.


If the amp tests stable with 100nF//8R, then it's very, very stable, and will drive a 4" rusty nail. It's just a test.... Oh, and caps, particularly good ones, can be part of Sheldon's 'seasoning', it ain't all bad y'know!


Aspen Amplifiers P/L (Australia)

Last edited by AKSA; 13th June 2011 at 01:40 AM.
  Reply With Quote
Old 13th June 2011, 01:41 AM   #1018
keantoken is online now keantoken  United States
diyAudio Member
keantoken's Avatar
Join Date: Aug 2006
Location: Texas
Sonnya, I think LTSpice tends to be very pessimistic regarding oscillation. Obviously the circuit doesn't oscillate this way in real life. Also you left out the gate stopper on the Jfet. SWF may have taken his tests with the input filter, I don't know if he removed it.

Also, add some .5R resistance to your supplies and some filtering just to make it more realistic. Add appropriate series inductances and resistances to all caps. Then the simulation may be more indicative of real life.

KSTR, you've noticed it about "current-dominated control loops" too..? I have found indications in the simulator this is how it should be. As it turns out, this is characteristic of amplifiers where OLG is flat throughout the audio spectrum, that is, dominated by resistances instead of reactances "less OLG, less compensation". I think such amplifiers also tend to be more stable.

Swordfishy, to decrease OLG you should probably try degenerating the Jfet. Degenerating the VAS simply makes the Jfet work harder, that decreases OLG but also increases distortion. Maybe it's time you start playing with the gate stoppers?

Juma, a few mV can make a big difference for a BJT, especially considering that these mV will contain the distortion of the output stage. Of course it may not matter (depending on whether this distortion is small compared to other distortions) always but when in doubt I tend to go for the lower number.

SWF, look at your square shots closer. I see some ringing on the lower corner. This may be unavoidable though.

- keantoken
The Kmultiplier rail filter! -=- The Simple Kuartlotron Superbuffer!
Need something built, repaired or modded? PM me and ask what I can do!
  Reply With Quote
Old 13th June 2011, 02:04 AM   #1019
keantoken is online now keantoken  United States
diyAudio Member
keantoken's Avatar
Join Date: Aug 2006
Location: Texas
I don't see stereo as an illusion, but as an attempted hologram. After all, it's not an illusion if you recognize it's not real. The mind after all is the final sensory organ, which can extrapolate the reality even if the senses themselves are fooled momentarily. The technology obviously is not complete or practical in terms of successful results, but still there are ways to improve it. The blacklevel can be raised or lowered, sometimes reds will bleed, bright spots will encroach on subtler adjacent hues, and so on. It's never perfect but it can still be better or worse.

At first we tend to think that good amplifiers help to create and illusion of security and a relaxed atmosphere, maybe inject some endorphins to get you to loosen up... This may be seen as be a way of escaping reality. But what if it's an unnatural environment that continually turns on our stress response, and the music is seen as a way to bring stress levels back down to reasonable levels? Stress kills.

To me, I like an amp that retains the edge, except the edge is not edgy. I find that listening to such an amp relaxes me, but doesn't shy away from realistic presentation. It helps me move into a state of relaxed vigilance, where I can think more calmly and objectively.

This is the effect I'm getting from a class A Pass-like thing I have running right now. Just two IRF250s (2nF!!!) biased about 800mA with 10R power resistors. They came with 1K gate stoppers, I lowered them to 100R, and that made a big difference. I am considering raising them a bit.

Am I close?

- keantoken
The Kmultiplier rail filter! -=- The Simple Kuartlotron Superbuffer!
Need something built, repaired or modded? PM me and ask what I can do!
  Reply With Quote
Old 13th June 2011, 04:02 AM   #1020
WuYit is offline WuYit  Sweden
Join Date: Oct 2010
here is what I had in mind, a J-FET transconductance input stage followed by a common base transimpedance stage, essentially having the same properties as the conventional cascode, but the latter offers additional advantages.
In the Hawksford cascode the voltage amplification is executed by a common base amplifier all right, but the intermediate grounded collector and grounded emitter stages only cause distortion.

The main properties of the common base amplifier:

low input impedance
high noise immunity
high isolation between input and output due to
low feedback factor
high stability and wide bandwidth due to the absence of Miller effect
wide dynamic range
high input overload capability
precisely controlled voltage gain
current gain just below one (low power gain gives low distortion)
the same amount of current flows through the input and output and in some configurations through the signal source (giving low distortion)
no phase reversal between input and output
Attached Images
File Type: jpg 41.JPG (20.8 KB, 263 views)
  Reply With Quote


JFET input, MOSFET VAS, LATERAL output = Perfect!!Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
10 Watt Single End, JFET input - Lateral N-MOSFET output lineup Solid State 86 4th April 2015 04:54 PM
Lateral MOSFET, How do I calculate power output synonymous Solid State 4 18th December 2007 07:54 AM
Calculating VAS current for mosfet output bogdan_borko Solid State 21 11th April 2006 04:57 AM

New To Site? Need Help?

All times are GMT. The time now is 01:29 PM.

Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 15.00%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2018 DragonByte Technologies Ltd.
Copyright ©1999-2018 diyAudio