Advantages of tertiary feedback winding

Hi - could someone enlighten me a bit ...

What are the advantages of using a transformer tertiary winding for feedback, over a 16ohm tap in a typical mid power (30W) mullard type configuration ?

I used the Hammond 1645 with it's tertiary winding for feedback to the driver cathode and had good results lowering THD, output Z, noise and extending the flat freq response curve.

I've also used the 16ohm tap on the secondary for similar benefits.

Just wondering what if any extra advantage the tertiary winding method ..  confers

Thanks

Nearly decided on the Hammond 1645 with feedback winding (again)  for my next homebrew stereo monitor amp.

Probably with 6BQ5s parallel'd  in push-pull maxing out at around 20Wrms all up, keeping it moderate

So - to keep it mixed up and trying different things out, I want to

- all 'balanced' from the input transformer thru to the driver tubes, with an interstage traffo
  ie. just like a vari-gm limiter

- stereo pair bloc with a seperate, matching psu bloc

- screens switchable from tube regulated supply (to triode mode)
- dc heater for pre, ac for finals
- fixed bias + cathode bias both
- tertiary winding feedback, balanced like!

I may also try some other groovy things such as a pentode constant current source in the preamp.
Maybe even an SS ccs.

Just to try things out.
The toob heads are always banging on about ideal current sources in differential pairs

AND also, not to forget, negative HV rail stuff. Is it useful in a toob amp ?

-----

So - that's what I've got in the planning stage!

If I understand correctly what are you looking for, is the ~70% tap of the transformer to make "ultra linear" amps IIRC, there were also approaches using resistive divider or capacitive divider (with some resistors so a mixed one in fact) if you don't have the tap in the winding. In this way you get some NFB and you get better response, yes, using the NFB network from the secondary makes something similar but I think one of the points of doing with the tapped transformer leads to a better performance because it keeps the polarization of the second grid in an optimal point, making the tube behave better rather than only correcting it's bad behavior with feedback, (speaking bad and quick)

JS

Do you want your Load and NFB connected together?

When you run 1,000 foot speaker lines all over a school, you probably want them totally floating from the amplifier. So you use a heavy winding for the Load, and add a light winding to get NFB from.

Same thing for broadcast long-lines (remote pick-up, remote transmitter, network feeds).

When your lines are short, you may not mind Load and NFB connected together, you save 3 clams not winding another winding+leads. And arguably this is "better" because your NFB senses the EXACT signal on the line to the load, not a 99.9%-coupled replica. (Tertiary NFB will not sense drop from the copper resistance of the load winding.)

PRR said:

Do you want your Load and NFB connected together?

Click to expand...

That is certainly a key question.

I do believe the isolation of the feedback signal from the load is probably a good thing.
But only if it is in a cost effective package; and one is looking to explore different techniques.

Thanks PRR  for the example of the long lines - makes good sense.

joaquins said:

If I understand correctly what are you looking for, is the ~70% tap of the transformer to make "ultra linear" amps IIRC,
JS

Click to expand...

For  this build, for the primaries of the output transformer, I am will be using a switch to the finals' screens to allow an ultralinear connection (43% traffo I think)  or a regulated (lower than plate) screen supply.

I find the ultarlinear connections I've done to date are very much 'triode-y'.  (I need to do some more investigations here - hence the exploratory nature of this build).

But I'm more interested in precisely setting the screen voltage, with regulation, for this build.

The crux of my conundrum is regarding feedback around the transformer-to-driver \

- is it better to tap the secondary (at say the 16ohm connection) OR use the  provided tertiary winding.

Sounds like load isolation is the next-in-line benefit, but in my short-cabling situation, probably is small-ish effect.

For studio environment or a guitar cab I don't see any advantages on isolation. For the cases PRR states there are some pretty obvious reasons.

So, the thing here is to have three windings so you use one just for the feedback? I thought the need of feed backing a transformer was a thing of the past, maybe you can use (with a situ able cap in the middle) use the anode of one of your output tubes (the most convenient polarity one) as signal to feed back. You are relaying on your transformer being good, but you don't need an extra winding on the transformer and you are insulated. Also some of the transformer distortion will be sensed since you are measuring the voltage in it's primary.

For the ultra linear configuration and still get your desired regulated voltage at the grids you could AC couple the tap to the grids and a resistor from the grid to the regulated PS, accounting it as a RC filter you should get a good compromise. Maybe it has been done before, I don't remember any amp doing so.

JS

I may have missed it in my quick scan:  the tertiary winding isolates from noise induced on the output line.  May or may not be a problem, probably not in a power amp with a short run.  Could be a problem in a distributed system, easily a problem in a small signal amp (preamp/program). 

emrr said:

I may have missed it in my quick scan:  the tertiary winding isolates from noise induced on the output line. 

Click to expand...

No, it doesn't really. Well it isolates from CM noise but is it a problem? CM noise is generally not a problem on loudspeaker lines, because a loudspeaker is an excellent balanced load. And all the differential noise on the speaker line is transmitted almost entirely to the tertiary (as well as the primary). The only sensible reason is when the output must be galvanically isolated from the circuitry. I've used tertiary windings in distro amps a lot, just because one never knows what it'll end up be connected with.
Another example is McIntosh MC series, where the tertiaries are in the cathode circuit (in fact there are both a load AND part of the NFB.

Thanks for comments - it's all food for thought. Looks like some fun to be had, although by no means necessary, with the feedback winding.

- the McIntosh output transformer feedback/load windings are certainly a major discussion on it's own for sure

While on the subject of old-style feedback windings ... what about 'field coil' speakers.

I have a tube hifi el84 amp (sansui OEM into Australian market circa 1966-67 I think) that has input for a speaker system each L, R with a feedback coil return-to-amp. 

It operates like a servo control circuit, I think. On the front panel, there is a control for it ('M.F.B') which is essentially a 'bass drive' control. Something to do with adaptive speaker/amp damping factor.

Or something like that!

Makes me want to review my control theory from the early 80s.

Could also be useful in determining whether or not my homebrew 'fairchild 660 inspired limiter'  is massively overshooting with the blast of CV or not  ;D

Here's an example schemo Marantz 5 which does the tertiary winding for feedback.

Subsequent Marantz 8 uses even more comprehensive dual tertiary windings 

My previous studio monitor monblocs were based on this, sans tertiary and with some  Eico thrown in for good measure!

They used around 10dB of NFB using an R/C network fed from the 16ohm secondary winding tap.

abbey road d enfer said:

emrr said:

I may have missed it in my quick scan:  the tertiary winding isolates from noise induced on the output line. 

Click to expand...

No, it doesn't really. Well it isolates from CM noise but is it a problem?

Click to expand...

Like I said, doubtful a problem in this case, could be a real problem in a remote amp feeding a telephone line. 

> what about 'field coil' speakers.

That term usually means an electromagnet where we now use a permanent magnet.

Before the discovery of Alnico, ~~WWII, permanent magnets were lame. Electromagnets were well-developed. Also electromagnets could be used as part of the B+ filtering, though that didn't last long once Alnico got affordable.

Certainly not 1966.

What you are describing is a servo connection. Because a voice-coil's motion loses contact with its amplifier above ~~200Hz, they mostly work on the bass. Super-odd to find that on external speakers. Too much wiring. The amp and speaker must be made for each other.

> tertiary winding for feedback.

No. NFB is R27 R28 across 4 and Com. Com is nominally grounded. (There's a dotty zigzag which I am ignoring for today.)

The function of the Bk/Yl and Bk/Wt winding is not clear, but probably a super-sonic tweak. With a gain of over 1,000 in the front and closed-loop gain of 10, there is a LOT of bandwidth and phase to control. R25 C11 point to something above 200KC. C13 seems absurd, MC? A big lump of copper on iron can have all sorts of strange low-RF effects, and someone treated all of them.

re field coil - thanks for the explain.

A big 'oh' here. I never realised that electromagnets could be used that way especially with B+

That (field coil)  is course, very different from the speaker servo capability of my 'Encel M.F.B Stereo Phonic'. (the MFB stands for 'Motional Feed Back') - actually, the amplifier was manufactured in Japan, by  "Daiwa Corporation."

Apparently 'motional feedback' was done by Phillips as well in the early 70s.

here's a link to a SS version from early eighties  http://www.circuitstoday.com/motional-feed-back-amplifier

I don't have the matching speakers but even without them, the front panel alters the bass response (on a 7 pos switch) with regular speakers.

It's a well built hifi tube amp. I did minimal repairs on it and it's been my daily hifi for around 8yrs.
Sounds really very nice, with quite big output traffos for the 6BQ5 pp finals.

All 'miniwatt' tubes from the day.

http://www.radiomuseum.org/r/encel_mfb_stereo_phonic_csm_40.html

---

re the Marantz 5 feedback network : right, OK - a 'super tweak' 

I thought the M5 was a bit odd, seeming to  have both a tertiary winding connected to a fb network,  as well as a secondary winding tap connect.

Hmmmm, I recall playing with a power supply connected to a field coil speaker while repairing an old jukebox - changing the voltage changed the tone quite noticeably.  Wish I had payed more attention that day...

Here in Argentina a company makes a set of speakers with a small 'headish' thing which is the control, or they sell they amps with this built in. They control the sensibility of the speaker and replace the attenuator used in tube amps so you can crank the level up without loosing your head in the way. Tonekeeper they call it, here it's a link (in spanish of course) http://www.jakeamps.com.ar/tonekeeper.php

I don't know if this has been made, it's the first time I saw it, few years back, 2010 IIRC.

JS

PRR said:

> what about 'field coil' speakers.

That term usually means an electromagnet where we now use a permanent magnet.

Before the discovery of Alnico, ~~WWII, permanent magnets were lame. Electromagnets were well-developed. Also electromagnets could be used as part of the B+ filtering, though that didn't last long once Alnico got affordable.

Certainly not 1966.

Click to expand...

There is still a french company named Audio Sound Machines that manufactures excitation speakers: they were known as Gougeon for about 50 years. They cater to the fairground market. There is no website, they conduct only word of mouth business.

mjrippe said:

Hmmmm, I recall playing with a power supply connected to a field coil speaker while repairing an old jukebox - changing the voltage changed the tone quite noticeably.  Wish I had payed more attention that day...

Click to expand...

Changing the excitation current changes the electrical damping of the speaker (Qes) in an inverse manner.
Increasing the current gives a better control of cone motion, which generally results in perceived "less bass".

abbey road d enfer said:

mjrippe said:

Hmmmm, I recall playing with a power supply connected to a field coil speaker while repairing an old jukebox - changing the voltage changed the tone quite noticeably.  Wish I had payed more attention that day...

Click to expand...

Changing the excitation current changes the electrical damping of the speaker (Qes) in an inverse manner.
Increasing the current gives a better control of cone motion, which generally results in perceived "less bass".

Click to expand...

Thanks Abbey, makes perfect sense!

One of the advantages of a tertiary winding is that you can pass dc through it if the transformer is gapped. This eliminates a blocking capacitor in the NFB loop and hence removes a pole and a zero from the transfer function, thus aiding stability.

Cheers

Ian