6L6GC AB2 Amp - Page 63 - diyAudio
Go Back   Home > Forums > Amplifiers > Tubes / Valves
Home Forums Rules Articles diyAudio Store Gallery Wiki Blogs Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

Tubes / Valves All about our sweet vacuum tubes :) Threads about Musical Instrument Amps of all kinds should be in the Instruments & Amps forum

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 1st February 2012, 10:00 PM   #621
SY is offline SY  United States
On Hiatus
SY's Avatar
Join Date: Oct 2002
Location: Chicagoland
Blog Entries: 2
Originally Posted by Johan Potgieter View Post

I am simply saying that what difference there is for me, is in the fact that parallel (Schade) feedback still leaves one with an intrinsic pentode, whereas UL leaves one with mainly triode characteristics - through whatever kind of feedback influence.
Is that true? Certainly the source impedance presented to the primary is reduced by the feedback factor. With a low effective plate resistance, I would think that the curves would be more triode-like, and maybe even resemble UL...?
"You tell me whar a man gits his corn pone, en I'll tell you what his 'pinions is."
  Reply With Quote
Old 1st February 2012, 10:22 PM   #622
diyAudio Member
Join Date: Oct 2004
Location: Adelaide South Oz
I think I may have posted these links on a previous discussion.

From Fritz Langford-Smith on Ultralinear.

Parts 1, 2 and 3 all in one file.

A better scan of some of this can be found here:

In these articles he addresses the 43% tapping suits all tubes myth.

He also states
"These points all indicate that UL Operation is distinctly different from either triode or pentode mode operation and should be regarded as a separate phenomenon"

I haven't re-read these articles recently but I seem to recall that he stated that UL Operation "observed" performance could not be adequately explained by feedback theory alone.

Ians own observations:
The one thing that people tend to ignore (maybe with good reason but maybe not) is that with UL the screen currents will contribute to the output.

One more "shotgun blast at the barn door" - UL feedback to the screens will NOT be as time accurate as a Schade Feedback connection due to the reactive nature of the output transformer.


Last edited by gingertube; 1st February 2012 at 10:36 PM.
  Reply With Quote
Old 2nd February 2012, 02:27 AM   #623
diyAudio Member
smoking-amp's Avatar
Join Date: Dec 2001
Location: Hickory, NC
"I am simply saying that what difference there is for me, is in the fact that parallel (Schade) feedback still leaves one with an intrinsic pentode, whereas UL leaves one with mainly triode characteristics - through whatever kind of feedback influence."

"Is that true?"

I can see where there would be some difference. In the pentode the forward gm is approx. 3/2 power (really more like 2 power for power tubes), while the neg. Schade feedback is linear. So the inherent forward distortion will take large amounts of Schade feedback to correct down to neglible levels.

In the triode on the other hand, the internal neg. fdbk is the inverse of the 3/2 power, and so in theory could correct the forward error with finite internal neg. feedback (just sets the Mu and Rp). In real tubes though, the g1 is more like 2nd power due to "grid proximity island effect", while the g2 is closer to 3/2 power (until g2 current becomes an issue anyway). So real triodes still distort some.

Possibly one could correct the Schade feedback to near triode mode by putting a thermionic diode from the pentode plate through a resistor to B+, and taking the Schade feedback from the top (plate) of the diode. The current in the diode would track the pentode if the load Z is constant. And the voltage variation on the diode would convert the feedback to the 3/2 power. (ie, when the pentode increases it's gm at increased current, the diode increases the Schade feedback in proportion) Interesting experiment anyway. Probably requires some specific diode conductance to track the pentode and generate enough correction. (ie, both pentode and diode need to be operating over the same exact portion of the 3/2 power curve) Likely have to use some flimsy HV dropping diode like a 2AV2 or something similar.

Or alternately, maybe put some resistance in series with the g2 and tap off the Schade feedback from the g2. (since the g2 intercepts approximately a constant fracton of plate current until plate V gets below the g2 V. This method depends on the load R being constant, to get g2 V to track plate V over the linear range) Might be a good way to prevent the tube from drawing too much g2 current too. Of course this would greatly restrict power output.

Last edited by smoking-amp; 2nd February 2012 at 02:55 AM.
  Reply With Quote
Old 2nd February 2012, 03:10 AM   #624
diyAudio Member
smoking-amp's Avatar
Join Date: Dec 2001
Location: Hickory, NC
One more version of Schade corrector:

Put a high current damper diode in series with the pentode's plate, and take the Schade feedback from the damper's plate. So it's self tracking (diode versus pentode) on the 3/2 power curves this way, but less efficient power wise.

A little inconvenient to float the damper diode that way. So I'm thinking maybe put the damper diode in the pentode's cathode circuit instead. That will lower the pentode's gm, but helps enforce 3/2 power law at g1. The pentodes plate still has the diode correction built in since the output OT is referenced to ground. Would have to simulate this to see how well it works I think. Ehhh... Probably can't get enough diode voltage variation these ways without killing the output power. So likely better to stick with the original diode scheme, merely appended to the plate (an R from diode plate to B+), approach.

I think I see the way now. A small thermionic diode from the pentode's grid to AC ground through a resistor. So: Diode's plate to the pentode grid, diode cathode to resistor, then that to - bias V. Then the Schade feedback R goes from pentode plate to the cathode of the diode, probably with a series cap to preserve the bias V. Now as the pentode turns on, the Schade feedback makes the diode conduct more (through the previous driver stage plate R) giving tracking 3/2 power feedback. Easy to configure, and efficient. Have to try this out. Will need something like a HV 2AV2 diode to get enough voltage swing correction on the pentode grid I think.

Last edited by smoking-amp; 2nd February 2012 at 03:40 AM.
  Reply With Quote
Old 2nd February 2012, 09:59 PM   #625
diyAudio Member
Join Date: May 2005
Location: Pretoria, South Africa
..... by this time it would have been far more convenient to just use a screen tap on the OPT


I am carefully considering. Look at it the other way round: Take an OPT where the pentode source resistance rp (without feedback) is low compared to the load (an almighty big OPT with a high load, suitable impedance ratio, or simply a constant current load). In a manner-of-speaking, the pentode will still present a set of pentode plate characteristics, not similar to that of a triode. Or if I use a very low load resistor for a triode (say 1 K for a 12AX7) - does the characteristics begin to look like a pentode's - I mean distortion-wise?

(And to those wondering: How does this matter if the driving impedance is lowered? I am thinking of the intrinsic pentode pattern of distortion, with more high-order products than those of a triode. Sure, one is diminishing those by a feedback factor - but so can one with a UL topology. Does that change their ratio to something less prone to listener fatigue? But as said I have to go think ... I have never looked at a constant current loaded pentode with a spectrum analyser. Maybe sit down with Spice before I go on ....)
  Reply With Quote
Old 2nd February 2012, 10:27 PM   #626
diyAudio Member
Join Date: May 2005
Location: Pretoria, South Africa
Originally Posted by gingertube View Post
One more "shotgun blast at the barn door" - UL feedback to the screens will NOT be as time accurate as a Schade Feedback connection due to the reactive nature of the output transformer.
... and I need to be careful that I am not standing at the other side of that barn door!

But, inasmuch as an output transformer is reactive (or better, its reactiveness plays a role in the audio pass-band), the plate will then also 'suffer' from the resulting phase angle, thus making the Schade feedback also prone to said reactiveness? (This simply means that the OPT design should be such that its reactive characteristics are negligible or at least small within the audio pass-band. This will naturally have to include leakage reactance regarding the screen taps - which is hopefully achieved in the better types of OPT. Aware of this I have looked at phase angles of screen taps relative to anodes, and have mercifully not found anything of sufficient magnitude to worry me.)

It gets worse when using the cathode feedback type of UL, as e.g. in the classic Quad II. Here the 'common' part of the OPT 'sees' a cathode follower action, i.e. working into a low load. Now one does have to worry about phase performance; this OPT design is a little more difficult - but I digress.
  Reply With Quote
Old 28th December 2015, 06:32 AM   #627
chrish is offline chrish  Australia
diyAudio Member
chrish's Avatar
Join Date: Oct 2003
Location: Sydney
Hello Everyone,

I am resurrecting this thread from the past. After building these amps, they have been performing with no problems for the last 5 years. I am about to start another audio project - some active speakers. To make the equalisation of the speakers easier, it will help if the amps I am using have similar gain. This 6L6 amp with the two stage diff amp at the front has quite a bit of gain. Looking through my posts, it was 12 volts output for 204 mV input, a gain of about 59. I think gain for most commercial amps is about 20. Regardless, I can measure and work out exactly the gain on the two amps.

My question, to get the gain in check, I can either put a voltage divider at the signal input of the amp, or I could try to implement some global negative feedback. If I were to try feedback, where would I inject the signal? I am a little confused with this as the DC offsetg circuit at the grid of the second triode in the first stage adds to the complexity.

Here is the final amp schematic:

Click the image to open in full size.
Attached Files
File Type: pdf 6L6 6SN7 MOSFET amp.pdf (70.3 KB, 62 views)
  Reply With Quote
Old 28th December 2015, 09:04 AM   #628
diyAudio Moderator
TonyTecson's Avatar
Join Date: May 2003
Location: Palatiw, Pasig City
you may also want to consider output plate to grid feedback....
Plate-Grid feedback: How to get better than DHT performance from cheaper tubes Schade Feedback
planet10 needs your help: Let's help Ruth and Dave...http://www.diyaudio.com/forums/plane...ml#post5010547[B
  Reply With Quote
Old 28th December 2015, 12:10 PM   #629
bigwill is online now bigwill  United Kingdom
diyAudio Member
Join Date: Dec 2004
Location: UK
In order to avoid upsetting the DC offset control, if you were willing to add an opamp input buffer you could bring feedback to the input grid via a 20K resistor (very likely along with a phase compensation network). The opamp would simply be a buffer to drive the resulting 1K input impedance, which a humble NE5532 will likely do just fine with very low distortion. Obviously not the purist approach but is probably the easiest way without upsetting your DC control which I imagine is quite helpful as the stages are DC coupled.

Somewhat perversely you could probably, with care and appropriate shielding, implement the whole thing external to the amp in a small box and avoid touching the amp altogether (apart from making sure the amp is inverting, though with opamp trickery you could even avoid that if you wanted), but you'd need a scope to debug it and tweak the phase shift network.

Last edited by bigwill; 28th December 2015 at 12:14 PM.
  Reply With Quote
Old 28th December 2015, 03:04 PM   #630
diyAudio Member
Tubelab_com's Avatar
Join Date: Jul 2005
Location: West Virginia panhandle
There are two simple ways to implement feedback and reduce the gain on this design.

GNFB from the OPT secondary can be applied to the grid of the second triode in the first stage.

Lift the grounded end of C14 and insert a small resistor ( 100 to 500 ohms) in series. One end of the resistor to C14, and the other end to ground.

Connect the GNFB resistor from the hot end of the OPT secondary to the junction of C14 and the new resistor. Start with a high value resistor (5K or so) and verify that it actually reduces the gain. You may have to swap the primary OR secondary leads on the OPT to get the phasing right.

Once you are sure that the feedback is indeed negative (reduces the gain) you can reduce the resistor value until the desired gain is reached. Most amps implement a compensation capacitor across the GNFB resistor. This compensates for the peaking in the OPT caused by leakage inductance. It is usually a in the 100 to 1000 pF range and the value is determined by observing the square wave response on a scope. The optimum value is OPT dependent, and your good OPT's shouldn't need a large cap. If you don't have a scope, try several caps in the 100 to 470 pF range and tune for smoothest treble.

As AJT pointed out, "Schade" feedback can be used. It was actually "invented" by an RCA engineer named Schade specifically for making the 6L6 sound more like a triode. It is supposed to only be used with pentode drivers, but small amounts work well with triodes in LTP because of the higher plate resistance in LTP configuration. In this design you would use a resistor from the plate of each output tube to the plate of the corresponding driver. Plate of U3 to plate of U5 and plate of U4 to plate of U6. The resistors should be of equal value, somewhere from 470K to 2 MEG or more, and must be rated for the entire plate voltage swing of both tubes. In this design that would be around 1KV. Resistors with this voltage rating are hard to find, so I use 2 or 3 equal valued 1 watt resistors in series. Say, three 330K resistors in series for 990K. This will increase the plate current through the driver tube by around 1 mA per side, causing the plate voltage to rise a bit. You can increase the value of the plate load resistors to compensate (R11 and R13) or simply bump up the current in the CCS on the cathode to bring the plate voltage back to 295V.

I'm not sure that enough Schade feedback can be applied to achieve the needed gain reduction (about 7 db) in this design, without compromising stability, so some combination, or GNFB only may be the best choice. That depends on your personal preferences, your musical tastes, and how much you want to "mess with success."

A lot has happened since this thread was started. My engineering career of 41 years came to an end, the facility has been sold and is shutting down. I had to move all of my stuff, plus Tubelab twice in the past two years. The first move was on rather short notice, and covered 1200 miles, so a lot of stuff was sold, given away, or trashed. I have recently moved into the house where I plan to be for the next 20 years, but it will be a while before a new lab is finished. I started with an empty 2000 square foot basement, so I have a long way to go.

While I was in the temporary housing, I laid out a PC board for the driver section of this amp. I also rebuilt the old breadboard I used for designing this amp.

The 5 year old design is solid, and the only changes were to replace the LED's in the offset adjustment circuit with TI voltage reference IC's, and I changed to 9 pin tubes (6CG7's) to make the board smaller. I got a batch of boards made and they should be available from the Tubelab web site sometime next year. I plan to build a few amp designs with these boards and document everything first, but I need a workbench for all that.

Here is a picture of the board, and my new breadboard. One of the output boards is missing because I blew it up trying to squeeze too many watts through too small a tube! It was a TV sweep tube experiment.
Attached Images
File Type: jpg Universal Driver Board_x.jpg (580.3 KB, 308 views)
File Type: jpg Universal Breadboard_x.jpg (612.0 KB, 263 views)
Tubelab, it's 5 year mission. To explore strange new tubes, to seek out new circuits and topologies, to boldly go where no tube has gone before......
  Reply With Quote


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
Red Hot 6L6GC oldmanStrat Tubes / Valves 26 29th May 2009 12:53 AM
Need help with 6l6GC choices BlackUnikorn Tubes / Valves 19 5th August 2008 01:39 PM
6l6gc SET, my first DIY alexmoose Tubes / Valves 3 27th April 2006 03:00 AM

New To Site? Need Help?

All times are GMT. The time now is 12:56 PM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright 1999-2017 diyAudio

Content Relevant URLs by vBSEO 3.3.2