Insulation stress in tube output transformers

Something recently caused me to think back about my early days in electronics and belatedly wonder about a certain point regarding the construction of audio output transformers. I didn't actually wind any push-pull output transformers but I did practise designing them. I recall that bifiliar winding was recommended to get a tight coupling between the two halves of the primary. This means that, with tube amps, the two wires would be at a potential difference of several hundred volts at each peak of the audio cycle and yet be separated by only two thin layers of enamel - something like a mil each, maybe less, for wire sizes used in output transformer primaries.

I can't recall off the top of my head the dielectric strength of wire enamel. If my estimate of a few hundred V/mil is in the ballpark, it doesn't leave much in the way of a safety margin in a bifiliar-wound output transformer.. Or am I totally missing something here?

Thereupon you have stumbled upon what the engineers at transformer companie s do. Insulation thicker, less efficiency. Why don't you guys stop using al l those high voltages and making their lives smiserable ? LOL

I thought of something similar recently when I ran across an amp with Crown 's pateneted "Grounded bridge" configuration. If not familiar, look it up. Pretty cool. Another dmanear unservicable but cool thing Crown did. Single ended supply, full DC push pull output. Part of me says it isn't worth the effort, and then well, I think they just like trying new things, and get of f on actually selling them.

Of course voltage is alot higher with tubes. Holy shit, don't get Crown pla ying with tubes, it would be a nightmare. They are bad enough with transist ors. With tubes they would probably be modulating the filament voltage for some reason only they and someone on LSD could fathom.

taken apart a few coils/transformers. A couple of coils just for the magnet wire. (planning on a switching PS soon) and then a small SMPS transformer. Thart was alot newer and it was encased damnear in glue or whatever the st uff is. Seems like the got it hot enough the viscosity was low enough to ac tually get in between the turns in the windings.

My buddy was trying to unwind the windings to get the copper. There was mor e glue than copper actually.

But really people don't realize, when you got a plate voltage of 400 volts, when you crank up the audio it is going to twice that.

As far as the winding goes, it's like a reverse Occam's razor. Usually the hardest way is the best.

The heavier grades of "enamel" insulation are better than you might think. Essex Formvar "heavy build" insulation on 18-gauge wire has a "required performance, room temperature" of 5700 volts, and a "typical" of 11,300 volts. At rated temperature these drop to 4275 and 9200 volts, respectively.

The thinner wire gauges might have thinner insulation overall... but it seems to be possible to get "single", "heavy" (double), "triple", or even "quad" insulation on such magnet wire. Sometimes the layers are of different insulating materials, to give a particular overall performance characteristic.

So, I don't think that a bifilar wind with a few hundred volts between the pairs would be any serious challenge, as long as the maker pays attention to which variety of "magnet wire" is used.

Also noteworthy that few actually undertook such construction. McIntosh's split-winding output comes to mind. With no cost to spare, they went for it; such was also a necessity to avoid oscillation, since unfavorable resonant modes can act as a cathode input, plate output oscillator.

It's not generally desirable to construct a transformer with such high impedances this way, anyway, because the excessive capacitance results in a hugely reduced winding impedance at higher frequencies. In other words, it has too much parasitic capacitance. But more to the point, it looks like a parallel transmission line of perhaps 50 ohms impedance, so that for frequencies on par with the electrical length (there might be 1000 feet of primary in the thing??), the impedance bounces up and down around that center value. And for frequencies less than 1/4 wave (~200kHz?), the impedance rises at 20dB/dec, more or less acting as a capacitor. So that by the time you get to 10kohms, you're 200 times lower in frequency (~2kHz?). Give or take other humps and valleys due to the winding construction (whatever the inter-layer capacitance does, etc.).

The optimal construction is interleaved such that the winding self capacitance resonates with the leakage inductance at the desired system impedance, and at as high a frequency as possible. This gives critical damping.

If it can be constructed of transmission lines comparable to the system impedance, there's no capacitive cutoff (because it's matched), and indeed, if made of transmission lines of equal length, connected in parallel at one end and in series at the other, it will work at frequencies well above 1/4 wavelength. Not that this really matters for tubes, of course (have you shopped for 10kohm coax lately? :) ).

Tim

--
Seven Transistor Labs 
Electrical Engineering Consultation 
Website: http://seventransistorlabs.com 

"Pimpom"  wrote in message  
news:mc82fp$8ve$1@news.albasani.net... 
> Something recently caused me to think back about my early days in  
> electronics and belatedly wonder about a certain point regarding the  
> construction of audio output transformers. I didn't actually wind any  
> push-pull output transformers but I did practise designing them. I recall  
> that bifiliar winding was recommended to get a tight coupling between the  
> two halves of the primary. This means that, with tube amps, the two wires  
> would be at a potential difference of several hundred volts at each peak  
> of the audio cycle and yet be separated by only two thin layers of  
> enamel - something like a mil each, maybe less, for wire sizes used in  
> output transformer primaries. 
> 
> I can't recall off the top of my head the dielectric strength of wire  
> enamel. If my estimate of a few hundred V/mil is in the ballpark, it  
> doesn't leave much in the way of a safety margin in a bifiliar-wound  
> output transformer.. Or am I totally missing something here? 
>

In a past life I played a lot with things like Xenon flash trigger transformers, and wound many myself using 'double' insulated enamel wire. A hundred turns or so on a P18/11 potcore former was plenty to give 4-8 Kv output. Never had any problem with flashover within the winding, only to stray fingers on the way to the tube.

--
Regards, 

Adrian Jansen           adrianjansen at internode dot on dot net 
Note reply address is invalid, convert address above to machine form.

** Not a practical or necessary method, coupling from primary to secondary is more important. ** Yep - and that would soon prove fatal to the insulation. Voltage, time and corona discharge across tine air gaps are the killers.

The best transformer are vacuum impregnated after manufacture, to remove all air from between windings. FYI: I have measured up to 5kV peak on the plates of audio output tubes under overdrive conditions - this often proves fatal to even well designed ones.

** Yep they simply are not bi-filar wound.

Commonly, the primary is split into two or four windings and interleaved with secondary windings.

... Phil

On a sunny day (Sat, 21 Feb 2015 11:49:03 +1000) it happened Adrian Jansen wrote in :

Indeed, I wound a similar coil on a potcore for HV supply in a scope with a cheap East German CRT in the seventies. Also about 4 kV. I did isolate some layers with sellotape, and make sure the lead from the bottom layer passing through does not touch the top layer, add some plastic tubing. And make top layer HV side, no flash-over to core through the plastic bobbin.

So if you do that right you only get one layer voltage difference, some hundreds of volts.

Putting the 'ends' with 4 kV against each other is not such a good idea, wire insulation is easily damaged. A am sure it would eventually burn through. For even higher voltage maybe wind in sections next to each other. Look at the construction of real HV coils for how it is done.

I remember having to replace output tube sockets because they arced over when some nitwit overdrivit without a load connected.

On a sunny day (Sat, 21 Feb 2015 02:36:49 -0800 (PST)) it happened snipped-for-privacy@gmail.com wrote in :

You are lucky, the spark gaps formed by the tube sockets protected the transformer.

In the vertical output of TVs where you also get a big HV pulse during vertical flyback, transformers were used to drive the deflection coils. Those were protected by a simple VDR (Voltage Dependent Resistor). IIRC I did protect some audio output transformer with a VDR ?? People would disconnect the speaker and then overdrive it, probably because they heard nothing,.

So a couple of VDRs saves money.

** The simplest error will do that to a guitar amp -

just plug the speaker lead back to the input and advance the gain control.

Whether it takes out the OT, tubes or sockets is purest luck.

.... Phil

The builder of the audio amplifier was not up to his task. A Class B tube audio amplifier needs a spark gap across the output transformer primary. Every AM ham transmitter builder (in 1960's) knew it.

--

-TV

** A much better solution is to put high voltage diodes or diode strings from plates to ground.

Output tube plates should not go negative, fitting diodes prevents this and so prevents large plate to plate AC voltages in operation.

Some makers routinely fit 5kV rated diodes but I found you cannot retrospectively fit them to any amp you like.

... Phil

Please think twice: This is in the tube era. There were no high-voltage semiconductor diodes, and you should connect the diode cathodes and heaters to the tube plates, to swing along with the high voltage we're trying to protect, makes it challenging to handle the heaters.

--

-TV

On a sunny day (Sun, 22 Feb 2015 11:41:10 +0200) it happened Tauno Voipio wrote in :

That is not current, HV rectifier sticks were already common in TV, for example for focus (4kV) in early color sets, those consited of a ceramic tube with many litte disks in series.

And them diodes do not protect against _positive_ spikes. assymetrical signals that overdrive can cause those too.

Use VDRs.

And VDRs were around in the tube days, every Philips TV set with a tube vertical deflection output stage had those,.

Spark gaps should work too as Phil pointed out.

There are some other tricks too, but I am not sure I am, still today, allowee to disclose those, even if I came up with it.

Tauno Voipio wrote: >

** We are still in the "tube era". ** Reads like gobbledegook.

.... Phil

** Makes no senes and is quite wrong.

** No - I damn well did not.

Using spark gaps in a tube amp is bonkers.

.... Phil

On a sunny day (Sun, 22 Feb 2015 02:32:50 -0800 (PST)) it happened Phil Allison wrote in :

Beautiful senes?

Anyways, consider this. I think you proposed this:

• | primary | |---------------- anode |

-grid --- cathode / \ | --- /// | ///

It is clear that a sudden switdh off from Ia will cause a positive spike that is now not limited.

I prospose this, cheaper too:

------ + | | VDR primary | | ---| anode -grid cathode | ///

Sorry, that was Tauno Voipio who said that (just looked it up in google).

Oh I dunno, anything that limits the over-voltage should help. the cracking noise and ozone smell may help as a warning to the user to reduce volume, and connect some speakers.. :-)

** Not interested.

The *CONTEXT* of this thread is a push pull or class B tube stage

READ the FIRST post !!!!!!!!!!!!!!!

With a diode from each end of the transformer to ground, no spikes are possible.

** Only the first three words above are true.

Go away.

... Phil

Phil Allison wrote