Tube DIY Asylum

Do It Yourself (DIY) paradise for tube and SET project builders.

Return to Tube DIY Asylum

Has anyone had success with "Partial Feedback" recently

114.78.68.180
Posted on October 25, 2009 at 14:30:11
Ray Moth
Audiophile

Posts: 1962
Location: Jakarta, Indonesia
Joined: November 10, 2003
Contributor
  Since: May 20, 2009
 I know this has been discussed in the past, but I also know that inmates of this asylum are always experimenting.

I'm talking about the plate-plate NFB as applied by Gary Pimm in his 47-PP (at http://www.pimmlabs.com/) and analysed by John Broskie (at http://www.tubecad.com/march2001/ et seq.)

It seems this short-loop feedback arrangement has a lot to offer if you can get it to work, but operating points are tricky. I've tried simulating such an ampp in PP using 6AU6 and EL34 (pentode mode) without much success - I find that I run out of headroom on either the voltage or the plate current (or both) in the 6AU6 stage that feeds the OP stage as a V-I converter, with characteristic clipping.

So, I'm wondering if others have found this type of circuit useful?

RE: Has anyone had success with "Partial Feedback" recently, posted on November 18, 2009 at 15:52:53
ottoamps
Audiophile

Posts: 103
Location: Seattle
Joined: February 6, 2005
 call me larry late for lunch here,
its been a while since I built anything new using the partial feedback technique but I played around with it extensively (as well as doug's E-Linear) both in simulation and on the bench in SE circuits.
I recall the problem you're experiencing, and as I recall it was only a problem when using blocking caps. While direct coupled as Gary did it, the driver has plenty of headroom. Of course, direct coupling is not for the faint of heart.
of all the tubes I tried in this config the HY69 was my fav.... but darn things kept dieing. Not sure if it was quality control on those tubes or something to the circuit, as no other tubes exhibited similiar failures.
I also liked the 6V6, 2e22, 2e25, 307a and 47. Interestingly I didn't care for the 1624 that Gary went with for his Tabor.
In the garage I have some 4e27 that I meant to try. Been thinking about that again lately and you MAY have just inspired me to pick that one back up.
My 811a amp derailed much of my building. It was TOO good.

A side note.... IMHO partial feedback sucks with horns. Too low a zo. Much nicer on those cruddy, less "tube friendly" speaker types.

RE: Has anyone had success with "Partial Feedback" recently, posted on November 19, 2009 at 01:42:27
reVintage
Audiophile

Posts: 120
Location: Eskilstuna
Joined: June 19, 2008
 "I recall the problem you're experiencing, and as I recall it was only a problem when using blocking caps"


Hey O,
Could you be a little more specific about that?
Brgds
Lars
RE: Has anyone had success with "Partial Feedback" recently, posted on November 20, 2009 at 19:11:20
ottoamps
Audiophile

Posts: 103
Location: Seattle
Joined: February 6, 2005
 LOL
Sure I can try, but they've touched on it elsewhere in this thread. When the feedback resistor is also the EXCLUSIVE load for the driver (ie no load resistor or current source to B+) the driver must be able to swing enough current to "drive" the resistor fully positive or negative. In a DC coupled amp with say 400v on the output tubes plate, and 100v on the driver's plate that gives you 300v to span. A high value resistor of say 100k would give you 3ma idle current to drop that 300v. To swing a 200v change in plate voltage on the output, the driver would then have to swing 2ma into the resistor, leaving 1ma.
In the case of the same amp, but cap coupled you would only need to drop from 300v to 100v, so 200v, the same 100k resistor then only idles at 2ma, and to swing the same 200v peak the driver must completely shut-off.
This is why most cap coupled amps use the extra resistor or ccs, to up the idle current a bit, but this in turn messes with the feedback effect.
Hope that clarifies at least a little.

EDIT: After reviewing the thread, I see you're probably way beyond me, so that explanation was probably a little too simple. Oh well :-)

RE: Has anyone had success with "Partial Feedback" recently, posted on November 21, 2009 at 07:32:57
reVintage
Audiophile

Posts: 120
Location: Eskilstuna
Joined: June 19, 2008
 Hey,
Got it thanks;-). Aha, you where refering to the (OS)type, getting all its anode current through Rfb. Lets turn it around as that type has few chances to work properly as they instead will lack driving current. I say this even if Gary Pimm has done it, using the very unlinear 6AU6A as driver.

Zin of a typical Plate-Grid feedbacked output tube is frightfully low at 2-4kohm. This means we will need current to drive it, normally the suitable pentodes need 10-20mA Ia to work at their best with such a load. So adding current through a gyrator(CSS wonīt work) will be the solution without upsetting feedback. And it will of course work both DC or cap-coupled. Check Michael Kosters work both here and at diyaudio.

But if we do a partial-feedback(plate-grid Lite) where we add more feedback oops around the amp, OS will absolutely work as Zin will be a lot higher. But my way of seeing the Plate-Grid feedback, is that it should eliminate the need for global feedback.

Noticed you didnīt like 1624. I was actually just about to buy a batch as I wanted to go DHP. Maybe I should back out on that deal;-)?
Brgds
Lars
RE: Has anyone had success with "Partial Feedback" recently, posted on November 21, 2009 at 12:45:03
ottoamps
Audiophile

Posts: 103
Location: Seattle
Joined: February 6, 2005
 As to the 1624, only back out of the deal if its not a good one! Nothing wrong with the tube, but yes I would suggest trying a few tube types before you settle on one, as they do sound different (even though they shouldn't, I know its the old objective science vs. subjective taste). you may love its sound.
Also all my work was in directly coupled single ended circuits, pp and/or with gyrators and caps will likely alter the flavor considerably.
To be honest I don't believe the low input impedance of the "power handling" tube is a bad thing. Actually, the lower the better. It permits the driving pentode to do what it does best, swing current not voltage. Ideally the driver/output interface should show as little AC voltage as possible. The AC signal is created across the feedback resistor. (the top end at the plate of the output swings, not the input.
The overall effect is quite amazing. Sounds just like a triode... well sort of... a really clean triode, not much 2nd, so if you like a syrupy sweet 2A3 sound, raising the input impedance to the output would be a good thing. I'd say rather than any other feedback, using a resistor in place of the gyrator would be beneficial coupled with a fairly high value feedback resistor. This essentially hobbles the circuit allowing more "pentode as a voltage amplifier" character to shine through. Nobody will ever convince me that's a bad thing. Pentode distortion sure sounds great on my guitar amps. :-)
RE: Has anyone had success with "Partial Feedback" recently, posted on October 28, 2009 at 09:20:51
Michael Koster
Industry Professional

Posts: 585
Location: Bridgeville, CA
Joined: October 24, 2007



Hi Ray,

This is how I determine the OP for the driver with plate feedback.

Here is an example load line for the pentode loaded by the output plate feedback current. Taking the voltage between the output plate and the driver plate (output grid) across the feedback resistor gives you the plate current of the driver. Do this for the quiescent point and the 2 signal swing end points, and you can trace the driver load line.

Edit: With the shunt resistor in parallel with the driver anode, you increase the slope of the load line.

Cheers,

Michael

RE: Has anyone had success with "Partial Feedback" recently, posted on October 27, 2009 at 09:52:44
thermionic addictions
Audiophile

Posts: 434
Joined: June 16, 2006
 Pete Millett used this in his "Mighty Midget" amp. I just finished building this little amp and it sounds great.

Pete also has for download a paper writen by O.H. Schade that discusses
this method, a very interesting read.

Cal

RE: Has anyone had success with "Partial Feedback" recently, posted on October 28, 2009 at 14:32:12
Ray Moth
Audiophile

Posts: 1962
Location: Jakarta, Indonesia
Joined: November 10, 2003
Contributor
  Since: May 20, 2009
 Hi Cal,

The presence of R2 (61.9k) in the Mighty Midget takes away the problem, because it provides a handy partial load for the driver. My difficulty has been trying to do without R2, as in Gary Pimm's solution and as explained by John Broskie.

RE: Has anyone had success with "Partial Feedback" recently, posted on October 27, 2009 at 00:26:03
reVintage
Audiophile

Posts: 120
Location: Eskilstuna
Joined: June 19, 2008

This is how it could look! Just a raw model of what I mentioned in my previous post. Clipping will take place in the output stage at ca 40W, not in the driver. R8, R9 can be between 0 and ca 5k and changes overtone spectrum.
Brgds
Lars
RE: Has anyone had success with "Partial Feedback" recently, posted on October 27, 2009 at 05:21:34
Ray Moth
Audiophile

Posts: 1962
Location: Jakarta, Indonesia
Joined: November 10, 2003
Contributor
  Since: May 20, 2009
 Interesting, but don't you think there might be a problem having a pentode (6AU6) with CCS load? A sort of battle of wills between the two concerning what the "constant current" should be?

RE: Has anyone had success with "Partial Feedback" recently, posted on October 27, 2009 at 05:59:05
reVintage
Audiophile

Posts: 120
Location: Eskilstuna
Joined: June 19, 2008
 Hey Ray,
"problem having a pentode (6AU6) with CCS load"

Read the schematic again, probably you posted before you gave it a closer look ;-)!

In my first post I talked about gyrators. The sources above the 6AU6īs are high-Z, fairly constantvoltage, and could maybe be called AC-CCS. So there will be no battle. Another solution could be to use inductors.

The 6AU6īs are loaded by 2-5k from the "partial feedback" plus the optional R8/R9.


Brgds
Lars
Sure. With DHP 47 PP like Pimm's but different overall design., posted on October 26, 2009 at 15:39:10
kurt s
Audiophile

Posts: 78
Joined: October 12, 2009
 I applied NFB from plate to cathode and drained the grid DC current sent there into the output of a SE-PP interstage. Balanced current from each 47 into the secondary of the IT.

The primary circuit was a ECC99 parafed into the primary of the IT. It was a single stage single-ended tube that kept up some 2nd order harmonics.

Overall this easily is the best sounding PP amp I heard and my ex wife now owns. It beat out the 45 PP due to extra sweetness, not definition or drive. It most sounded like a SET amp for a PP. And it was class AB with it being very deep into A before switching over to B. Only limitation for bias was the power transformer was overheating for me on class A. Put out a sweet 5W/ch.

Another reason for fewer class A PP's is heat and lower power. Some is not what customers want here, and marketing isn't all that dumb about what people will buy. Unfortunately, good stuff is hardly marketable, except to esoteric cults with some inside feel for what they like. This group is a cult.

And yeah, everyone has to have a remote with that. Getting up is just not human anymore.

"Seeing is believing, but hearing requires a panel of judges and double blind testing."

-Kurt
RE: Sure. With DHP 47 PP like Pimm's but different overall design., posted on October 27, 2009 at 05:25:40
Ray Moth
Audiophile

Posts: 1962
Location: Jakarta, Indonesia
Joined: November 10, 2003
Contributor
  Since: May 20, 2009
 Hi Kurt,

Do you think the class of operation necessarily has to be Class A for this type of feedback to work? I've been simulating it in Class AB1 with fixed bias, but maybe I'm making a mistake doing that?

RE: Sure. With DHP 47 PP like Pimm's but different overall design., posted on October 29, 2009 at 12:21:19
kurt s
Audiophile

Posts: 78
Joined: October 12, 2009
 It was stable in all types of bias conditions because this NFB loop is so quick acting that the margin is huge. Have to have it wired correctly and engineered correctly, though.

-kurt

RE: Sure. With DHP 47 PP like Pimm's but different overall design., posted on October 27, 2009 at 13:23:23
PakProtector
Audiophile

Posts: 9953
Joined: May 14, 2002
 hey-Hey!!!,
I have a pair of Class A E-Linear amps and a pair of AB. The partial FB works in both Classes.
cheers,
Douglas

Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
RE: Sure. With DHP 47 PP like Pimm's but different overall design., posted on October 28, 2009 at 14:35:47
Ray Moth
Audiophile

Posts: 1962
Location: Jakarta, Indonesia
Joined: November 10, 2003
Contributor
  Since: May 20, 2009
 Thanks, that means I haven't been beating my head against a wall!

heh-heh-heh...:), posted on October 29, 2009 at 03:54:44
PakProtector
Audiophile

Posts: 9953
Joined: May 14, 2002
 -----Thanks, that means I haven't been beating my head against a wall!

good thing, eh?

But seriously, the amps still behave rather like a bridged pair of SE amps. The FB doesn't really care if one phase is cut off, though it looks better for me if it isn't happening( though for nothing to do with the FB circuit ).
cheers,
Douglas

Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
Westrex used it alot......, posted on October 26, 2009 at 08:08:52
grhughes
Audiophile

Posts: 1872
Location: Central US
Joined: February 17, 2004
 in their motion picture audio recording amplifier modules; RA1474 and others. Feedback around small local gain stages and little global feedback.Ray Hughes
"I think the journey should be just as enjoyable as the destination." GRH
RE: Has anyone had success with "Partial Feedback" recently, posted on October 26, 2009 at 07:18:37
reVintage
Audiophile

Posts: 120
Location: Eskilstuna
Joined: June 19, 2008
 To use the 6AU6 you must run it at 8-11mA/150V/150V and feed it by a gyrator as this way the tube will only be loaded(ca 2k) by the EL34. You should not go lower than Rfb=100k. With 6L6 you can go 68k as it has lower Gm.
Brgds
Lars
I first used this concept in a pcb I designed to replace the..., posted on October 26, 2009 at 04:47:44
Allen Wright
Manufacturer

Posts: 4737
Location: Schaffhausen
Joined: January 31, 2002
 ...highly UNRELIABLE output module of the tube output amps (AEC?) that looked like NAD rip-offs back in the early 90's, that soldered (!!) EL34s in the pcbs!

I thought it sound great but the project's backer said it sounded thin, and the project flopped.

I have tried many levels of feedback from a ratio of driver anode load R to f/b R of 1:100 right dowen to 1:1, and of course you have to adjustthe driver's current for each change.

My Oz partner (Joe Rasmussen) also worked with it and he got the "thin" reports as well. he solved it by adding a second f/b loop from the output anodes to the cathodes of the driver - as well as the anode to anode loop. But this required a delicate balanced act to get the two loops to harmonise, and when I then tried no NFB at all, I promptly forgot about any of this NFB malarky, and have never looked back.

Regards, Allen (Vacuum State)

RE: Has anyone had success with "Partial Feedback" recently, posted on October 25, 2009 at 16:38:31
PakProtector
Audiophile

Posts: 9953
Joined: May 14, 2002
 hey-Hey!!!,
I like this one with the E-Linear simplification. Now it does behave a bit like you describe if the tap location is not correct. That usually dictates tap locations less than 30%, but the amps I built around a Heathkit W6m iron set deliver 60W v. 70 as implemented with a grid-current capable driver. Can't say as I miss those last few Watts either as the amps are powering folded bass horns of very high efficiency.

The E-Linear implementation with 6AC7 pentodes is very simple. I've got some more tweaking to do with their g2 voltage and bypassing( and if I'll even bypass them). I suspect I'll get a bit of an improvement w/o the bypass as the summed screen current increases and the voltage drops a little bit( and the gm along with it). Not sure if the lowered gm of the phase with increasing plate voltage will be a big deal, but the phase with increasing current should be a good thing.

I've not found any special sensitivity in these amps; fixed bias finals and RC coupling to the input stage behaves very civilly.
cheers,
Douglas

Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
RE: Has anyone had success with "Partial Feedback" recently, posted on October 27, 2009 at 05:42:14
Ray Moth
Audiophile

Posts: 1962
Location: Jakarta, Indonesia
Joined: November 10, 2003
Contributor
  Since: May 20, 2009
 Douglas,

I confess I don't know what E-linear is. If it involves the use of UL taps on the OPT primary, I can't try it because my OPTs don't have such taps.

RE: Has anyone had success with "Partial Feedback" recently, posted on October 27, 2009 at 13:25:59
PakProtector
Audiophile

Posts: 9953
Joined: May 14, 2002
 Hey Ray,
Pete Millett has a nice article on his site about the circuit in SE implementation.
cheers,
Douglas

Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
RE: Has anyone had success with "Partial Feedback" recently, posted on October 28, 2009 at 00:56:19
reVintage
Audiophile

Posts: 120
Location: Eskilstuna
Joined: June 19, 2008
 Hey Douglas,
If you havenīt noticed, Peteīs new PP "Engineerīs Amplifier" is using the exactly the technique Ray is asking about. Albeit with TV-tubes.
Brgds
Lars
RE: Has anyone had success with "Partial Feedback" recently, posted on October 28, 2009 at 14:48:41
Ray Moth
Audiophile

Posts: 1962
Location: Jakarta, Indonesia
Joined: November 10, 2003
Contributor
  Since: May 20, 2009
 Alas, it's not exactly the same because it has those 30k resistors (R50 and R55) that help to feed the 6CB6 plates, i.e. the tubes are not reliant solely upon R29 and R47 for their operating point. I have the feeling that those parallel plate loads are diluting the partial FB - or am I just being silly?

I'm familiar with plate-plate NFB and I use it in cross-coupled NFB in my all-differential 6SL7 - 6SN7 - EL34 (triode-connected) amp. However, the 6L6 plates get miost of their current from conventional 120k plate load resistors. Only a fraction of their plate current comes from the cross-coupled FB resistors, which are connected to the OP tube plates.

RE: Has anyone had success with "Partial Feedback" recently, posted on October 28, 2009 at 23:18:14
reVintage
Audiophile

Posts: 120
Location: Eskilstuna
Joined: June 19, 2008
 "Alas, it's not exactly the same because it has those 30k resistors (R50 and R55) that help to feed the 6CB6 plates, i.e. the tubes are not reliant solely upon R29 and R47 for their operating point. I have the feeling that those parallel plate loads are diluting the partial FB - or am I just being silly?"

R50/R55(or driver anoderesistors, in my schematic they should be ca 28k) makes the load harder on the the wimpy 6AU6 and distortion will much higher against using the voltage-controlled gyrators. Also 6AU6 is a unlinear tube but this is compensated by the "inverted phase" unlinearity of the feedbacked EL34. This is what we call distortion-cancellation. Grid-plate feedback should be of the magnitude that gives you "triodestrapped" Zout.

Think of it like this: You want to use the Rfb to feed all the current to the drivers like in a Pimm amp. Wonīt work in this case. The EL34 needs in the ballpark of 100k Rfb to work like we want with triode characteristics and pentode effiency. The 6AU6 wants maybe 8mA to work properly and be capable to drive its 2-4k load(input of feedbacked EL34). But 800V(100k*8mA) isnīt realistic so we have to add current from somewhere else.

With the above mentioned "gyrator" we are not loading the driver, only supplying the currrent not fed to it by Rfb. It you get less current from the Rfb you get more from the "gyrator" and viceverca. MSo no problem if you want you want to try different Rfb values. Michael Koster also has done an extremely smart solution for driving A2 this way.


Brgds
Lars
RE: Has anyone had success with "Partial Feedback" recently, posted on October 28, 2009 at 03:11:12
PakProtector
Audiophile

Posts: 9953
Joined: May 14, 2002
 hey-Hey!!!,
I did take notice. I have been working on a TV tube amp for a while now, and it is nearly ready for assembly. Those horizontal outputs are very nice looking tubes, and remarkably inexpensive compared to the usual audio power finals. Looking forward to firing it up soon.
cheers,
Douglas

Friend, I would not hurt thee for the world...but thou art standing where I am about to shoot.
RE: Has anyone had success with "Partial Feedback" recently, posted on October 25, 2009 at 15:02:17
Gingertube
Audiophile

Posts: 409
Location: South Oz
Joined: October 8, 2004
 Ray,
I'm persevering with the Baby Huey's and there have been upward of 50 of these built by various folks around the world, a couple of guys are on their 3rd or 4th BH build (I'm on my 6th and 7th).
Also see Pete Millet's new amp here:
http://www.diyaudio.com/forums/tubes-valves/151206-posted-new-p-p-power-amp-design.html
Cheers,
Ian
RE: Has anyone had success with "Partial Feedback" recently, posted on October 28, 2009 at 15:04:23
Ray Moth
Audiophile

Posts: 1962
Location: Jakarta, Indonesia
Joined: November 10, 2003
Contributor
  Since: May 20, 2009
 Hi Ian,

Yes, hat's the sort of thing I was talking about. You've obviously made a success of it. I just couldn't seem get the hang of it with pentode drivers and fixed bias for the OP tubes, though.

Incidentally, in my model the 6AU6s are preceded by a Williamson-style 6SN7 voltage amp and concertina splitter. I'm getting close but so far no cigar.
"a Williamson-style 6SN7 voltage amp and concertina splitter", posted on October 29, 2009 at 16:35:14
Allen Wright
Manufacturer

Posts: 4737
Location: Schaffhausen
Joined: January 31, 2002
 Why? Have you never heard of KISS?

Quad made a very successful amp with a pair of pentodes and the output tubes - one of the most elegant and sophisticated designs ever. Less is normally far more!

Regards, Allen

RE: "a Williamson-style 6SN7 voltage amp and concertina splitter", posted on October 30, 2009 at 19:57:44
A9X
Audiophile

Posts: 32
Joined: September 4, 2006
 "Why? Have you never heard of KISS?"

Like the RTP line stage?

It sure doesn't look KISS, but is actually..., posted on October 31, 2009 at 03:43:00
Allen Wright
Manufacturer

Posts: 4737
Location: Schaffhausen
Joined: January 31, 2002
 ...as KISS as it can be, and still work at it's performance level.

The signal path is just the one CF tube, the rest are : 1/ a cascoded constant current source for the cathode load, to have the CF tube operate under constant current conditions, and

2/ a bootstrap tube for the anode to keep the CF's tube under constant voltage conditions with signal.

Doing both these things make a CF sound completely transparent yet maintain it's very low output impedance.

And of course all doubled up because it's balanced.

Einstein said: "Make things as simple as possible but no simplier"

A9X? Do you own one of those classic Oz monsters? Was just looking at some pix I took of Brocky winning at Bathurst in one.

Regards, Allen

RE: It sure doesn't look KISS, but is actually..., posted on October 31, 2009 at 17:27:08
A9X
Audiophile

Posts: 32
Joined: September 4, 2006
 I understand how it works Allen (got Valley/Wallman about the same time as RDH4), and have built two as well as some SVP and (my design) variations on the line stage side and even a cascode 6H30 SVP phono first stage for the Decca, which IIRC came from a suggestion from you. I still reckon the RTP linestage is over complicated, however we have different ideas on correct solution.

Mine are currently in parts in boxes in my workshop until I get the TT's back online in a few months.

Unfortunately I don't have an A9X. They are too damn expensive here and even getting a base model to turn into replica, or at least my interpretation of one, ie EFI and 6 speed manual, is getting hard and costly on top of the mortgage.

The moniker came as I was consolidating the various ones I had scattered on different forums - this was the week Brockie died. I picked A9X because of the love of the hatches from a boy and his win at Bathurst in 79 in one.

RE: "a Williamson-style 6SN7 voltage amp and concertina splitter", posted on October 30, 2009 at 00:43:45
Ray Moth
Audiophile

Posts: 1962
Location: Jakarta, Indonesia
Joined: November 10, 2003
Contributor
  Since: May 20, 2009
 Sure, Allen, I've heard of KISS. However, I wanted to keep the phase splitting and the partial feedback drivers separate, for now at least.

I've never heard a Quad amp but I've seen mixed opinions of them. In any case, Quad didn't use partial FB.

RE: "a Williamson-style 6SN7 voltage amp and concertina splitter", posted on October 30, 2009 at 01:49:40
reVintage
Audiophile

Posts: 120
Location: Eskilstuna
Joined: June 19, 2008
 Hey Ray,

At what current did you run your 6AU6 in your unsuccessful attempts? 6-10mA is a must. A V-I converter must be able to supply enough of that stuff.

If you want a PI before the drivers, why not use a transformer?

Actually I did a sim with transformer before the drivers and it worsened the results. Probably due to the distortion-cancellation wasnīt as effective as when going LTP. Maybe the same with a concertina......
Brgds
Lars
RE: "a Williamson-style 6SN7 voltage amp and concertina splitter", posted on October 30, 2009 at 17:25:37
Ray Moth
Audiophile

Posts: 1962
Location: Jakarta, Indonesia
Joined: November 10, 2003
Contributor
  Since: May 20, 2009
 6AU6s: 2.3mA each plate, 0.8mA each screen. Cathodes have separate resistors, i.e. not a differential driver stage. I finally managed to get it working, after a fashion, with -60v on the junction of the gridleaks and cathode resistors.

Incidentally, I also tried your approach of putting an FET in each plate but that made it oscillate.

RE: "a Williamson-style 6SN7 voltage amp and concertina splitter", posted on October 31, 2009 at 01:25:28
reVintage
Audiophile

Posts: 120
Location: Eskilstuna
Joined: June 19, 2008
 "2.3mA each plate, 0.8mA each screen. Cathodes have separate resistors, i.e. not a differential driver stage."

It wonīt work at that low current. Remember it is current input, not voltage! You must run them at 6-8mA/each, 150V/150V.

"Incidentally, I also tried your approach of putting an FET in each plate but that made it oscillate."

Donīt tell me you just followed my schematic ;-)? Off course you have to add gatestoppers, 220-1k.

If you donīt add the gyrators, you must add anoderesistors to get the necessary Ia. Ref. Milletts latest PP.
Brgds
Lars
RE: "a Williamson-style 6SN7 voltage amp and concertina splitter", posted on October 31, 2009 at 02:28:11
Ray Moth
Audiophile

Posts: 1962
Location: Jakarta, Indonesia
Joined: November 10, 2003
Contributor
  Since: May 20, 2009
 I did add gate-stoppers of 4.7k each. What puzzles me at the moment is how providing extra current from a gyrator (or other source) is supposed to help if the extra current is required to go through the NFB resistor (if I understand you correctly).

RE: "a Williamson-style 6SN7 voltage amp and concertina splitter", posted on November 1, 2009 at 02:12:42
reVintage
Audiophile

Posts: 120
Location: Eskilstuna
Joined: June 19, 2008
 Hey Ray,

You donīt need to take any DC-current at all through Rfb, you can add a cap in series with it if so. Then you take all current through the gyrator/anodechoke/anoderesistor. Whatever way you do it you still need a total Ia of 6-10mA for the driver.

Look at a gyrator as a Ra with an extremely high Z. In other words it doesnīt load the tube at the same time as it is supplying the necessary Ia and Ua. Rfb can be looked upon as part of the load as Rfbī= Rfb/(Gm*RL+1). Gm is for EL34 and RL is for half Raa of the PP OPT.


Brgds
Lars
RE: "a Williamson-style 6SN7 voltage amp and concertina splitter", posted on November 1, 2009 at 08:50:07
Michael Koster
Industry Professional

Posts: 585
Location: Bridgeville, CA
Joined: October 24, 2007



Hey Ray and Lars,

Now I think the load line diagram I posted up at the top of this thread might be helpful. I think you need to get the quiescent current of the driver up where it can accomodate the current swing in both directions. The driver needs to operate in class A in a more or less linear part of the load line.

Michael

PS Using Lars' formula will get you in the ballpark but note that Gm is not constant. That's why I use the graphical approach here, but spice should give the correct answer also.

The driver Gm decreases as the output Gm increases. There will be some optimum quiescent point where the driver and output distortion partially cancel.

RE: "a Williamson-style 6SN7 voltage amp and concertina splitter", posted on November 1, 2009 at 23:45:30
reVintage
Audiophile

Posts: 120
Location: Eskilstuna
Joined: June 19, 2008
 Hey Michael,

Your method would work if we assumed that lowest distortion at the driver(your point with 5k load for the D3a is extremely well choosen for that purpose) was the main goal. Unfortunately it isnīt always the case and we have to rely on distortion-cancellation.

In the case of the highly unlinear 6AU6 we first of all need to find a workingpoint that can deliver both current and voltage. So ballpark figures comes in handy ;-). But we anyway need to choose a point where the curves are fairly equidistant.

In my example above we can assume 25W at somewhere between 1-2% THD, when the 6AU6s probably are 5-10%.

If we ran D3a at 20mA+ in this circuit it would probably win hands down due to better drivecapability and that the 6AU6 is beyond "acceptable unlinearity" at larger swings. A sim using Steve Benchīs D3a model verifies this.
Brgds
Lars
RE: "a Williamson-style 6SN7 voltage amp and concertina splitter", posted on November 2, 2009 at 08:22:28
Michael Koster
Industry Professional

Posts: 585
Location: Bridgeville, CA
Joined: October 24, 2007



Hey Lars,

You can use the graphical method to dial in the nonlinearity also. If you want to predict the distortion cancellation accurately you would need to use spice, right?

I would go with g2=150V on the 6AU6 to enable a higher plate current regime. Anwhere from 6mA-10mA Iq with +/= 2.5 mA swing at the 6AU6 anode should work fine. In the attached graphical example, driving a 6L6 type in AB1 with about 25V pk on the grid and ~350V Pk anode swing would put Rfb in the ballpark of (25V + 350V)/2.5mA or about 150K ohms.

Then you could vary the driver Iq between 6mA and 10 mA or thereabouts to tune the distortion cancellation.

Input sensitivity would be 1.1V p-p for full output, and can be decreased by using an unbypassed Rk on the 6AU6. This also linearizes the I/V characteristic and changes the distortion cancellation.

If you need different drive voltage, either increase the swing of the driver or increase Rfb.

cheers,

Michael

RE: "a Williamson-style 6SN7 voltage amp and concertina splitter", posted on November 2, 2009 at 09:00:19
reVintage
Audiophile

Posts: 120
Location: Eskilstuna
Joined: June 19, 2008
 Hey Michael,

You are spot on the choosen working point from my example above ;-). Adding 68ohm unbypassed cathode-resistors above the CCS seems necessary to get an acceptable result (anyway in Spice environment). There are quite a few 6AU6 curves out there and they are not always alike. Which one is this?

Rfb of 150k for a 6L6/807 will probably not do at all. I get indications of 47-68k depending on OPTZ. My idea is to choose Rnfb to get in the ballpark of the same Zout as when triode-strapped. Also seems like a lighter load on the driver makes distortion-cancellation less effective.

I usually begin designing for SE as this shows more of what you do wrong ;-). Suspect you do the same, Michael?

I have two candidates on my bench, a ST70 and a pair of QuadII. 6F12P(pentode) or E280Fs, also in stock, will be used. Will keep the CFB on the QII with a little less plate-grid feedback around the KT66s.

Brgds
Lars

Page processed in 0.216 seconds.

[ Top Top of Page ] [ Contact Us ] [ Support/Wish List ] [ Copyright © 2009, Audio Asylum, all rights reserved ] [ Go Green ] [ Take5 ]
[ General ] [ Speakers ] [ Tubes ] [ Vinyl ] [ Digital ] [ Hi-Rez ] [ Video Asylum ] [ Cables ] [ Tweaks/DIY ] [ Music ] [ Films ]