Running a 2A3 at 5K

Andrew · Post by **Andrew** » Wed Sep 14, 2011 8:12 am

I seem to recall that higher value I quoted was a pair of much regarded AE silver amorphous, tho' I can believe the real value could be lower than that since our method was not as water tight as the one you suggest.

You'd be quite surprised at how we found out how much capacitance was involved, we found out but listening to amp without the lower leg of the secondary connected to ground and it still worked - took a bit to figure out what was going on.

It will be interesting to hear from Nick, but I'm guessing that 150nF was Nick also measuring the same AE's; since they were his interstages brought along for an experiment.

Andrew

Post by **Nick** » Wed Sep 14, 2011 8:32 am

Yep, same pair of transformers. Inductance 50H, LF response just fine. The coupling between primary and secondary seems to only be covered in some models, but its there all the same, and I think bifilar wire may be the exception to some of the models rules.

http://d4magnetics.com/page6/page6.html

No cleverness involved on our side, we just found a effect in practice we didnt understand, so went looking for the theory to explain it.

Paul Barker · Post by **Paul Barker** » Wed Sep 14, 2011 8:37 am

OK so what now doesn't make sense is that your lf response is fine. This alrms me. You are saying that the capacitive coupling effect is there, but if you run the numbers on that the lf response should be very poor. So when you listened to it as a capacitor coupled stange was the amplitude the same and what was the lf response? Nick your link is nothing special, I could link you pages and pages of formulae but I haven't the time to waste. Everyone knows what interwinding capacitance is. What is important is to evaluate it's effect on transformer design.

Paul Barker · Post by **Paul Barker** » Wed Sep 14, 2011 9:09 am

Andrew wrote: much regarded AE silver amorphous,

Can't you see the flaw if they are highly regarded and yet when connected as a transformer as designed (not incorrectly as you say you have found your theory proven) they would in your theory have two response curves at LF, the transformer design theory response curve and then your capacitive coupling to grid choke response curve. The latter (if it was of significant amplitude relative to the transformer coupling to consider) would degrade the LF performance of the former.

And yet people like Langford Smith Terman and Flanagan (who is arguably the most comprehensive source on transformer design today) do not factor capacitance in their LF models.. Yet as I keep saying all show it plays a significant role in HF behaviour.

All that said, you have put me off bifilar winding as a credible technique by your figures.

ps if you unplug your loudspeakers you will hear your music through your output transformers. But it isn't hifi and it isn't relatively significant when modeling your transformer.

Andrew · Post by **Andrew** » Wed Sep 14, 2011 9:13 am

Spice provides us with a possible model for what's going on, but not how closely that matches reality.

Red line cap, green line inductor, as Nick said we didn't do anything clever, in fact, just the opposite then we went looking for a way to explain what we observed.

Andrew

Paul Barker · Post by **Paul Barker** » Wed Sep 14, 2011 9:23 am

I can't read the legends is the red line the capacitor and the other line the transformer?

What therefore is the relative amplitude of the two (and therefore can the capacitor contribute when the transformer is coupled as a transformer)?

i.e at 20hz what db down is the transformer curve and what db down is the capacitor curve and what db down is the combined effect.

You haven't yet proved to me that you have an LC coupling of significance when the transformer is modelled to be used as intended.

Now if you put your leakage inductance in the sim (to measure that short the secondary and measure the primary inductance) and take it higher to the point where the real issue of the Cww relative to LL at HF interaction takes place I might find your graph useful.

Why not add another curve which shows the magnetostriction output I referred to above? becaue it is not relatively significant. Is your effect?

But sadly I can't read the scaling.

BUT if what you are saying is that the db's are relatively close, you are telling me that you are increasing the problem in the behaviour of your transformer at LF beyond that normally encountered by a transformer where Cww is better designed out by careful manipulation of the two parasitics, which I am tired of re-iterating.

If I wound a transformer with greater Cww than 25nf I would cut the winding wire off it and try a different configuration. What you might have discovered is that your transformer capacitively couples with such a significant degree as to effect the performance inside the audio spectrum. ps I wouldn't use it in a feedback amp, and with that much capacitance I wouldn't couple it to the output of my DAC.

When you two have finished playing with Spice send your results to Flanagan h'ell need to bring out a new edition. I just feel sorry for poor old Terman and Fritz never having Spice.

Post by **Nick** » Wed Sep 14, 2011 10:03 am

Whatever Paul. No point in saying anything more, it would just be waisting your time further.

Paul Barker · Post by **Paul Barker** » Wed Sep 14, 2011 10:23 am

Nick wrote:Whatever Paul. No point in saying anything more, it would just be waisting your time further.

That's ok to say but if anyone is genuinly interested in transformer design and the practical implication of thes findings of yours, the questions I have asked need a reply.

"What therefore is the relative amplitude of the two (and therefore can the capacitor contribute when the transformer is coupled as a transformer)? "

"put your leakage inductance in the sim "

to test the effect at LF to midband in real terms I would be quite happy to wind a low capacitance lf transformer of the same primary L as yours, when you tell me it, (dc resistance we can equalise using a series resistor to balance the one with lower resistance) and put it at LF head to head with your bifilar transformer in a real situation, then compare the LF frequency response of the two. then if your theory is significant to transformer design it will be demonstrated. But if it is found, then what it will inform me is that it alters LF performance for the worse.

Post by **Nick** » Wed Sep 14, 2011 10:29 am

Nick your link is nothing special, I could link you pages and pages of formulae but I haven't the time to waste.

which I am tired of re-iterating.

When you two have finished playing with Spice send your results to Flanagan h'ell need to bring out a new edition. I just feel sorry for poor old Terman and Fritz never having Spice.

Paul Barker · Post by **Paul Barker** » Wed Sep 14, 2011 10:33 am

nick what I mean by your link it is a stating the obvious nothing link anyone could find any number of showing exactly the same on the web.

Find me some real in depth information on transformer behaviour without googling cuting and pasting a quick link which contains no useful explanatory text or formulae derivation.

steve s · Post by **steve s** » Wed Sep 14, 2011 10:35 am

Paul Barker wrote: i.e at 20hz what db down is the transformer curve and what db down is the capacitor curve and what db down is the combined effect.

this is getting very intersting pauL... i think you have discribed (to me) the well know effect( to me at least) of many output transformers ability to thicken the bass up ( due two response curves at low frequency ?)
i knew about phase shifts but never fully understood the reasons
phase shifts on there own would not thicken the bass up, (they just change the phase) but additional outputs related to inductance/ capacitance would ?
rather like a ported speakers output.
if have i got the basics of what you are saying?

i have quite a stock of output transformers, funnily enough the more thickened bass types all suffer loss of detail higher up...

if i've got the right end of the stick though...
but.. what your discribing, if my understanding is correct, was not apparent to me in the AE transformers i have heard, i'm talking in the main about output transformers though,

Paul Barker · Post by **Paul Barker** » Wed Sep 14, 2011 10:42 am

ah but Steve you are talking about a different beast, they have found this effect in a bifilar wound transformer with an inordinate amount of capacitance. My argument is that the effect is insignificant. If they can prove it is significant then they have proved their transformer is not something I would design for.

What it becomes is a tone control.

your experiences are with transformers which probably suffer Hf performance due to the greatest enemy of HF coupling Leakage inductance. Probably due to kindergarten winding arrangements of primary 1, entire secondary then primary 2. pr at most 3 primaries and two secondaries. In this type of transformer you can easily make plenty of bass by winding enough primary inductance. The clever bit is getting so many turns not to leak. the more you aim to prevent leakage the greater your capacitance becomes.

this is before I start on the LF performance of core material. we'll not go down that road. Except to say that in single ended transformer terms the LF performance of the traditional materials used by Leak etc is best.

Post by **Nick** » Wed Sep 14, 2011 11:35 am

Paul Barker wrote:nick what I mean by your link it is a stating the obvious nothing link anyone could find any number of showing exactly the same on the web.

Find me some real in depth information on transformer behaviour without googling cuting and pasting a quick link which contains no useful explanatory text or formulae derivation.

Here is one,

http://lib.tkk.fi/Diss/2003/isbn9512265 ... 265877.pdf

Andrew L will have access to more papers but I don't have the IEEE account to read the stacks.

For example

http://ieeexplore.ieee.org/Xplore/login ... ision=-203

Paul Barker · Post by **Paul Barker** » Wed Sep 14, 2011 11:40 am

Can you summarise them and show your conclusions and how they prove your case please?

Please answer the above questions?

Cressy Snr · Post by **Cressy Snr** » Wed Sep 14, 2011 1:01 pm

Blimey!

audio-talk

Running a 2A3 at 5K

#31

#32

#33

#34

#35

#36

#37

#38

#39

#40

#41

#42

#43

#44

#45