Hawksford style error correcting output - diyAudio
Go Back   Home > Forums > Amplifiers > Tubes / Valves

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
Reply
 
Thread Tools Search this Thread
Old 29th May 2004, 04:59 PM   #1
diyAudio Member
 
smoking-amp's Avatar
 
Join Date: Dec 2001
Location: Hickory, NC
Default Hawksford style error correcting output

I just posted this over on the solid state forum in a thread on low feedback amplifiers. Thought it might be of interest here too. This is a tube output stage idea based on M. Hawksfords error correcting output stage. (M. Hawksford JAES, Vol 29, Jan/Feb 1981 pp 27-31 on the web too at: http://wes.feec.vutbr.cz/~gratz/Download/Hawksford.ZIP )

I have fitted the Hawksford idea to a partial cathode feedback circlotron output stage ( or "Elliptron") but it can be easiliy adapted to any P-P (or even SE for that matter) output:
Attached Images
File Type: gif hawksfordpp.gif (8.4 KB, 916 views)
__________________
I want a Huvr-Board!!
  Reply With Quote
Old 29th May 2004, 05:47 PM   #2
diyAudio Member
 
jan.didden's Avatar
 
Join Date: May 2002
Location: Great City of Turnhout, Belgium
Blog Entries: 7
Hi Smoky,

I am very interested in the Hawksford stuff at the moment, I'm working on an output stage with that applied. As you surely know, the big difference between 'normal' feedback and the Hawksford error correction (which is of course also a form of feedback) is that with 'normal' feedback we try to maximize loop gain to have as much feedback gain as possible. With the Hawksford topology, the part of the output that is fed back is ideally exact 1. You take the difference between the stage's input and output, and that is exactly what you add to the input to make the stage gain perfectly 1. That is also why Hawksford says that basically it is only useful in stages that are already close to a gain of 1, like EF output stages in SS amps.

So, after this long (for me) ramble, how do you do this? Not being a tube guy, I'm not familiar with the gain levels etc in output stages. Do I get it correctly that you take the signal part from the 's' tap and then return it in the correct ratio via the R1 resistive divider?

Jan Didden
__________________
If you don't change your beliefs, your life will be like this forever. Is that good news? - W S Maugham
Check out Linear Audio!
  Reply With Quote
Old 29th May 2004, 06:03 PM   #3
diyAudio Member
 
smoking-amp's Avatar
 
Join Date: Dec 2001
Location: Hickory, NC
Default How it works

Basic operation is as follows. V1 and V2 are the normal output tubes in circlotron-like configuration. (The screen taps on the transformer are used for the cathode connections here to reduce the amount of grid drive compared to a full circlotron connection.)
In normal operation, the cathodes of V1 and V2 will behave as cathode followers with the plates driving the other section of the transformer. The cathodes will have signals slightly less than their respective grids due to the grid-cathode drive signal required to enable current flow. So, depending on V1,V2 transconductance and load impedance, the cathodes will have a slightly reduced, say 90%, replica of the grid drive signals.

Now we add V10,V11 for error correction. The resistors R1 and R2 monitor the input V1 grid drive signal and an inverted replica of the cathode signal. They are proprtioned in such a way so that if the cathode is properly following %90 of the V1 grid drive, then no signal appears at the V10 grid and no action is taken. If however the cathode is not correctly producing %90 percent of the V1 grid drive (load impedance variation, crossover dist. in class AB, etc...) , then a correcting signal is produced on the V10 plate. Resistors R3 and R4 are sized so as to just produce unity correction gain. Likewize, V11 corrects V2 operation in a symmetric manner.

When tuned up correctly for unity correction gain, variation of the output impedance has no effect on output signal. A simple way to align/test this is connecting another power amplifier output thru an 8 Ohm power resistor to this amplifiers output and putting a square wave signal into the other amplifier. R3, R4 (and the V11 resistors) are tweeked until no square wave signal appears on the V1,V2 cathodes or plates.

Don (of course, if you don't get it right, you may get lots of SMOKE! )

Hi Jan,
I just saw your post as I was about to post this. Yes, I just use the cathode follower portion of the signal for the error monitoring, so is close to unity gain.
__________________
I want a Huvr-Board!!
  Reply With Quote
Old 29th May 2004, 06:53 PM   #4
diyAudio Member
 
jan.didden's Avatar
 
Join Date: May 2002
Location: Great City of Turnhout, Belgium
Blog Entries: 7
Got it. I think. Wil print the schematic and study it.

Interestingly, my very first amp I built when working for Philips in the 60-ies (my second name is Methusalem). That was.. a circlotron with 2 times 807 tubes. Of course, working for Philips made it easy to get those power transformers with 3 separate hv supply windings.
This amp was OTL for the 800-ohms speakers Philips was pushing at that time for the purpose (the fabulous 9710-series). As soon as the first power transistors (OC29) came out we salivated over them and forget all about tubes. Sigh.

Jan Didden
__________________
If you don't change your beliefs, your life will be like this forever. Is that good news? - W S Maugham
Check out Linear Audio!
  Reply With Quote
Old 29th May 2004, 07:14 PM   #5
diyAudio Member
 
smoking-amp's Avatar
 
Join Date: Dec 2001
Location: Hickory, NC
Hi Jan,
I guess I neglected to mention on this thread that I haven't actually built this yet! So don't take it as gospel for a working design. Will try it out soon though.

I have a modified version/idea of the circlotron/elliptron which uses just a single power supply too, but have yet too try it out, the parts are sitting ready on the bench collecting dust so far:

Elliptron Schematic

Circlotron amps

Don
__________________
I want a Huvr-Board!!
  Reply With Quote
Old 29th May 2004, 08:38 PM   #6
diyAudio Member
 
smoking-amp's Avatar
 
Join Date: Dec 2001
Location: Hickory, NC
Default design limitations

The error monitoring points come from the opposite "S" ( ultra-linear Screen taps) tap on the Xfmr. to the applicable cathode, in order to get an inverted cathode signal, and so depends on the Xfmr. having good coupling between the "S" taps. (At least each output tube IS driving current through BOTH "S" to "CT" to "S" sections, so first order coupling should be good, we're not talking an isolated secondary here.) This limitation may require some bandwidth limiting in the error correction circuit to prevent phase shift or oscillation at high frequency. The present circuit does however use BOTH the direct cathode signal AND the opposite "S" tap signal simultaneously, due to its symmety, to form its output. ( due to the differencing action of the LTP) So any small phase difference (besides the expected 180 degrees) between the "S"s is diluted.

The Hammond 1650T Xfmr. that I am using for the circlotron/elliptron output has the "S" to "S" section wound on the core as a single interleave section so should be pretty good. But until I construct this, I won't know how well this works. If this proves to be a serious problem, then I may have to re-configure the V10/V11 diff. amp. connections to work with the direct, non-inverted, cathode signals only. Probably can not still be done for both output tubes using the same diff. (LTP) amp. though.

Don
__________________
I want a Huvr-Board!!
  Reply With Quote
Old 30th May 2004, 01:47 AM   #7
diyAudio Member
 
smoking-amp's Avatar
 
Join Date: Dec 2001
Location: Hickory, NC
Default further iterations

Just some notes and further thoughts.
The signal sources driving C1 and C1' inputs (in solid state designs anyway) are usually low impedance (emitter followers, cathode followers or low Rp triode here). But I think Jarek's design (in the SS thread on no feedback designs) has shown that this is not necessary, but does make calculating parts values for unity gain a bit more difficult. However, depending on the previous stages output impedance significantly for gain setting, instead of on R3 (and other factors), opens one to the problem of varying plate resistance of the previous driver tube with signal level (particularly for triode plate output drivers). So unity feedback gain may only exist at one signal level.
It may be helpful to put a resistor in series with C2 and C2' and /or use degeneration resistors in the LTP (V10 & V11) cathodes to adjust feedback gain. One would like to maximize the value of R4 (and R4') so as to minimize loading on the previous driver tube providing the signal to C1 ( &C1' ). Likewise, R2 and R1 (& R2' and R1' ) would like to be high resistance to minimize driver loading. Here, its just the ratio that matters fortunately as long as V10 (& V11) maximum grid leak impedance is not exceeded. May be some noise considerations too for very high resistance values. I'm sure actual construction of this circuit idea will find some more useful mods too.
By the way, the input driving signals to C1 (& C1' )must be from a class A stage (ie, 100% duty cycle) rather than from a class B driver. The V10, V11 LTP requires balanced inputs of 100% duty cycle to perform its function correctly. The output stage, V1 and V2 however can be Class AB or Class A (determined by grid bias level).

Don
__________________
I want a Huvr-Board!!
  Reply With Quote
Old 30th May 2004, 06:42 PM   #8
diyAudio Member
 
smoking-amp's Avatar
 
Join Date: Dec 2001
Location: Hickory, NC
Default Oops! a few clarifications

After getting a good nights sleep and reviewing what I posted yesterday, I see some things I need to correct.

The ratio of R1 to R2 sets the nominal gain expected at the cathode point (relative to the V1/V2 grid drive signals) as mentioned, typically something like 90% or 0.9 gain for a cathode follower. One wants to set this at the nominal gain of the output tube cathode into the reflected load impedance, so that the error correction circuitry is normally passive. The error correction circuitry should only be actively modifiying the drive level when deviation from the norm is occuring. (due to speaker impedance variation, crossover distortion, etc...)

The gain of V10 loaded by R3, R4 and the output Rp of the preceding driver tube (likewise for V11 ....) sets the maximum OPEN loop gain (for the error correction feedback loop). Not a unity gain as I had suggested before. Higher open loop gain enables the circuit to approach unity (well, er, 0.9 gain in this case) CLOSED loop gain more ACCURATELY.

Deviation of the Rp of the preceding driver tube could then influence the open loop gain if R3 does not dominate, but would only indirectly have a smaller influence on the accuracy of closed loop gain. Deviation of the open loop gain could however influence stablility of the error feedback loop if the phase margin of the loop is low. Possibly causing oscillation bursts at signal peaks.

For this reason, one wants to set the nominal error corr. loop gain at a safe level, and a resistor in series with C2 (and C2' ) or cathode degeneration resistors for V10/V11 may be useful if needed to lower gain.

Also, for the two amplifiers connected by an 8 Ohm power resistor test, one would be adjusting the R1/R2 (and R1'/R2' ) ratios, (not R3 and R4 as suggested earlier) to minimize feedthru of the square wave from the 2nd amplifier. Hope I got it right this time!

Don
__________________
I want a Huvr-Board!!
  Reply With Quote
Old 30th May 2004, 08:00 PM   #9
diyAudio Member
 
smoking-amp's Avatar
 
Join Date: Dec 2001
Location: Hickory, NC
Default Re: clarification

My previous post #8 timed out while I was editing it and it has several mistakes, so please disregard # 8. (the open loop/ closed loop talk in post #8 only applies if the boost signal is applied in a way that does not affect the reference input for grid drive level, just the actual applied grid drive, and is a different scheme from the Hawksford style design considered here) Here is the correctly edited version:

The ratio of R1 to R2 sets the nominal gain expected at the cathode point (relative to the V1/V2 grid drive signals) as mentioned, typically something like 90% or 0.9 gain for a cathode follower. One wants to set this at the nominal gain of the output tube cathode into the reflected load impedance, so that the error correction circuitry is normally passive. The error correction circuitry should only be actively modifiying the drive level when deviation from the norm is occuring. (due to speaker impedance variation, crossover distortion, etc...)

The gain of V10 loaded by R3, R4 and the output Rp of the preceding driver tube (likewise for V11 ....) sets the error loop gain (for the error correction feedback loop). This error correction gain needs to be UNITY so that the V1 ( and V2) drive input signals are only boosted enough to correct the deficiency and no more. The unity error loop gain and the forward cathode follower gain (0.9 here) are distinctly different entities.

Deviation of the Rp of the preceding driver tube could then influence the error loop gain if R3 does not dominate. Deviation of the error loop gain could also influence stablility of the error feedback loop if the phase margin of the loop is low. Possibly causing oscillation bursts at signal peaks. But the main effect of deviation from unity error loop gain would be to allow distortion thru (or even generate distortion if > 1).

For this reason, one wants to set the nominal error corr. loop gain accurately at the unity level (adjusting for minimum feedthru using the two amplifier test with connecting 8 Ohm resistor and square wave signal). A resistor in series with C2 (and C2' ) or cathode degeneration resistors for V10/V11 may be useful if needed to lower gain. Some frequency dependant rolloff of the error corr. gain may be necessary if phase margin (mainly of "S" to "S" in the Xfmr.) is a problem at high frequency.

Hope I got it right this time!
__________________
I want a Huvr-Board!!
  Reply With Quote
Old 30th May 2004, 09:14 PM   #10
diyAudio Member
 
smoking-amp's Avatar
 
Join Date: Dec 2001
Location: Hickory, NC
Default Not my day!

I just noticed that I had the plates of V10 and V11 connected up backwards in the 1st post. Was doubling the distortion instead of nulling it. Heres the corrected schematic:
Attached Images
File Type: gif hawksfordpp.gif (9.2 KB, 650 views)
__________________
I want a Huvr-Board!!
  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
Hawksford output error correction circuit? Jorge Solid State 96 12th December 2011 11:31 PM
Marantz CD63 KI SIG, Error 12, Sled Error problem sisik Digital Source 1 9th November 2008 03:12 PM
Output devices for class-d, error amp rtarbell Class D 0 16th December 2005 04:05 AM
Another error correction output stage. Mr Evil Solid State 63 14th February 2005 11:26 AM
Whats worse, 1db error in response curve or a phase error of 6? primalsea Multi-Way 5 13th May 2003 01:24 AM


New To Site? Need Help?

All times are GMT. The time now is 03:29 PM.


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