Some other Source Follower COnfigurations - Page 13 - diyAudio
Go Back   Home > Forums > Amplifiers > Pass Labs

Pass Labs This forum is dedicated to Pass Labs discussion.

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
Reply
 
Thread Tools Search this Thread
Old 4th August 2010, 06:24 PM   #121
EUVL is offline EUVL  Europe
diyAudio Member
 
Join Date: Oct 2003
I am afraid that you do not understand the "triode" mode in LU1014 fully and how the DAO circuit really functions.

Both the effective resistance of the cascode device (1/Yfs for FETs or Re for BJT) and the degeneration resistor contributes to a dynamic change in Vds that just counteracts the quadratic behaviour of the Id vs Vgs, resulting in the published linear behaviour around bias.

Some other Source Follower COnfigurations

The values in my published circuit were not there by chance. They were results of 20 hours of experimentation. It may look very simple in the end, but it is not.

Maybe you should try the original circuit one day.


Cheers,
Patrick



T

Last edited by EUVL; 4th August 2010 at 06:28 PM.
  Reply With Quote
Old 4th August 2010, 07:12 PM   #122
EUVL is offline EUVL  Europe
diyAudio Member
 
Join Date: Oct 2003
Let me add that the effective resistance of the cascoding MOSFET or BJT is not entirely linear. However, they are non-linear in a different manner; MOSFETs are close to low-order polynomials, BJTs to exponential functions. In general BJT has a lower Re at the same bias, so you can use a BJT at the cascode position and try to get back some of the dynamic Vds change by increasing the degeneration resistor. But then you are trying to replace a quadratic device by an exponetial device + a linear device. It will never be 1:1 replacement. And in doing so, you have increased the output impedance (by about 2 ohms).

Maybe it is better, maybe it is not. I have not tried, so I simply don't know.
And you would notice the difference more if you are using the follower to drive a low impedance (hence high current) phone, like Grado's or AKG's.

The jury is still out, as far as I am concerned.


Patrick

Last edited by EUVL; 4th August 2010 at 07:14 PM.
  Reply With Quote
Old 4th August 2010, 08:47 PM   #123
juma is offline juma  Germany
diyAudio Member
 
juma's Avatar
 
Join Date: Nov 2005
Location: Berlin
It seems to me that you missinterpreted the way your circuit works. There is no "dynamic Vds change". Vds stays constant regardless of signal, Yfs or Id change... There are microscopic changes due to inherent non-linearities of component itself but they are so tiny that they can be safely disregarded.

No need to spread mysticism about this circuit - it's a great headamp even without it
  Reply With Quote
Old 4th August 2010, 09:04 PM   #124
EUVL is offline EUVL  Europe
diyAudio Member
 
Join Date: Oct 2003
> Vds stays constant regardless of signal

You cannot be more wrong. I designed it; I measured it; I know. Vds only "stays constant" if there is no load (RL >>300R).
And there is no need to accuse me of spreading mysticism even if you could not get what I was trying to tell you.

If you would bother, go and read Nelson's article on ZenV9 again please, the part on cascode modulation (what I called dynamic Vds).
Compare his Fig.2 & Fig.4, and the schematics of the DAO, to be precise. And check the Yfs of the cascode MOSFET at 200mA.


Patrick

Last edited by EUVL; 4th August 2010 at 09:19 PM.
  Reply With Quote
Old 4th August 2010, 09:37 PM   #125
juma is offline juma  Germany
diyAudio Member
 
juma's Avatar
 
Join Date: Nov 2005
Location: Berlin
Temper, temper, kuhl bleiben bitte

ZV9 has nothing in common with DAO except the parts used - that cascode is completely different pair of trousers. In ZV9 elko at cascode MOSFET's gate is grounded - in DAO it enbles MOSFET's Source (JFET's Drain) to exactly follow the JFET's Source.

Anyway, there is no need for me to prolong this discussion, you can't make me believe in magic.

Again, lovely circuit

Last edited by juma; 4th August 2010 at 10:00 PM.
  Reply With Quote
Old 5th August 2010, 04:00 AM   #126
EUVL is offline EUVL  Europe
diyAudio Member
 
Join Date: Oct 2003
I hate to do this, but I am doing this for the other guys, not you.

As in ZV9, the DAO circuit I published has a cascode MOSFET, the gate of which is tied to a voltage reference (in case of the DAO the Zener diode and a cap in parallel). The other end of the Zener diode is tied not to the source of the LU1014, but to the lower end of the source resistor (ca. 3 ohm in the DAO). Thus,

Vds of LU1014 = V zener - Vgs(cascode) - I x Rsource ................(Ohm's Law)

Suppose you have a load of 20R (Grado), and you apply a signal of 1V pk-pk (sine). This means output power

p = V^2 / (2 * R phone) = 25mW so not excessive.

The peak output current i = 1V / 20 ohm = 50mA.

Thus the voltage drop at the source resistor (3R) will increase by 50mA * 3R = 0.15V.

The Vgs of the cascode MOSFET will also increase due to the increase in current, by an amount of

vgs cascode = 1/Yfs * 50mA = 80mV.

Thus, vds of LU1014 will decrease by 230mV. This vds drop is output current (hence signal) dependent.

It is precisely this decrease of vgs that causes the triode effect (the linear portion of the curve I published in post #26).



Of course this is not the whole story, as we have only been discussing quasi-static effects so far.

If you use the DAO with a 200R load (e.g. Sennheisser), you will see some overshoot when applying a 10kHz square wave. Where as if you reduce the load to 66R (e.g. AKG), the square wave is just about perfect. So it is not for no reason I keep saying that the published DAO is optimised for the AK701 / 702.

Which one sounds better ? That depends on the headphone (how much HF roll off it has). Can this dynamic behaviour be changed ? You bet.

A simple, naive looking circuit is much more complicated to design that it looks on the surface.
But no magic, just science intelligently applied.


Good day,
Patrick


(Magenta is input, red is output. Left is RL 200R, right is RL 66R.)
Attached Images
File Type: png DAO 200R.png (74.4 KB, 674 views)
File Type: png DAO 66R.png (76.2 KB, 653 views)

Last edited by EUVL; 5th August 2010 at 04:26 AM.
  Reply With Quote
Old 5th August 2010, 06:42 AM   #127
juma is offline juma  Germany
diyAudio Member
 
juma's Avatar
 
Join Date: Nov 2005
Location: Berlin
You got the cascode's Vgs part wrong and it's influence on JFET's Vds. You make a measuring mistake somewhere - check your setup.
Still, the mild triode character (output characteristic) of LU1014 is intrinsic and not dependant on cascoding device.
  Reply With Quote
Old 5th August 2010, 08:45 AM   #128
godfrey is offline godfrey  South Africa
diyAudio Member
 
Join Date: Nov 2009
Location: Cape Town
Hi Patrick

Thanks for the food for thought!

Just to make sure we're all on the same page, I've attached a circuit (copied from one of your earlier attachments).

As I understand it, the linearization mechanism you described above would apply even for a fairly simple follower e.g. just the Q1+Q11 part of the circuit, with a simple current source at the bottom.

Looking at the bigger picture, I see the same arguments applying to the linearity of the active source made of Q2+Q22.

One question, though - Doesn't Q3's non-linearity throw a spanner in the works, or is it relatively insignificant?

(Sorry if it's a dumb question, but I'm a total noob regarding fets)

Regards - Godfrey
Attached Images
File Type: gif dao.GIF (8.5 KB, 652 views)
  Reply With Quote
Old 5th August 2010, 04:39 PM   #129
EUVL is offline EUVL  Europe
diyAudio Member
 
Join Date: Oct 2003
> Just to make sure we're all on the same page, I've attached a circuit (copied from one of your earlier attachments).

You posted the one with the Taylor current source.
I was talking about the original circuit posted in Grey Rollins thread.

> As I understand it, the linearization mechanism you described above would apply even for a fairly simple follower e.g. just the Q1+Q11 part of the circuit, with a simple current source at the bottom.

Yes.

> Looking at the bigger picture, I see the same arguments applying to the linearity of the active source made of Q2+Q22.

In the TCS circuit, yes. Absolutely.

> One question, though - Doesn't Q3's non-linearity throw a spanner in the works, or is it relatively insignificant?

Very good question. It does.
So in theory it is better to use a high gain device there, such as 2SA970, as in the original circuit of the TCS in tubecad.com.
This would minimise the effect of Q3.

However, I had the 2SJ103 lying around, and I actually tried it. I ended up getting a good working point, where I traded a bit of 2nd harmonics for reduced high order (3rd, 4th, 5th, ...). So I did not bother to try PNP afterwards. And I would suspect that a PNP will get me more high order harmonics.

But feel free to experiment and report back.


Regards,
Patrick

Last edited by EUVL; 5th August 2010 at 04:42 PM.
  Reply With Quote
Old 5th August 2010, 06:24 PM   #130
godfrey is offline godfrey  South Africa
diyAudio Member
 
Join Date: Nov 2009
Location: Cape Town
Thanks

I'm not planning on building anything now, though - just learning.
  Reply With Quote

Reply


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
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Source follower and common source output stages slogan2112 Pass Labs 2 9th October 2008 08:19 PM
Bootstrapped FET source follower ? ash_dac Solid State 11 14th November 2006 08:19 PM
Aleph-type current source, but source follower this time tschrama Pass Labs 4 29th July 2005 12:55 PM
Source follower electret mike -> use with current source? capslock Solid State 1 28th February 2003 12:54 PM
Source Follower Enlightenment SteveS Pass Labs 3 20th June 2002 09:58 PM


New To Site? Need Help?

All times are GMT. The time now is 09:42 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2