speaker jack shorting or not?

[jmaf]

The wire is a bitch to solder, enamel has to be burnt on a gas flame first and then the blackened remains (which still cling tenaciously to the copper) have to be scratched off with a sharp box cutter.

I use a Dremel with a fine stone grinder tip on it. A single slight pass by the enamel and the wire's shiny copper shows through. I have a butane minitorch that I used to use for that but I don't anymore.

[jmaf]

Has anyone tried 'resitive tethering' ie adding a 330K or so resistor from the power tube anodes to ground?
This is something recommended by Kevin O'Connor.
Supposedly it's better than the string of diodes often used.

Cheers
Shane

I've never tried it, but the resistors will surely have an impact on tone whereas the diodes only leak microamperes. Ken Fischer used the string of diodes and nobody's complaining about the tone of his amps

[markr14850]

FWIW, a bunch of people over on AmpGarage have reported some unpleasant distortions that seem to be caused by the diode strings, and go away when they are ripped out.

[R.G.]

FWIW, a bunch of people over on AmpGarage have reported some unpleasant distortions that seem to be caused by the diode strings, and go away when they are ripped out.

I'm guessing that the diodes that were causing unpleasant distortions were not correct for the application in some way. These would include:
- diode(s) not properly rated for the reverse voltage
- diode(s) damaged by earlier spike events
- diode mods improperly selected or installed

The thing about protection circuits and devices is that if they're properly designed and installed, they do *nothing* in normal operation. Protection events can sometimes damage protection devices - after all, the protection stuff should be simpler and cheaper to replace than what it protects.

The fact that there are distortions which go away when the diodes are removed is itself evidence that the diodes were not doing the intended thing. In this application, the diodes are always reverse biased, and do not break down or flash over, as people think. They are intended to be nonconducting unless their side of the primary goes below ground, at which point they go into forward conduction and clamp both sides of the primary to no more than B+ across it. The output tubes simply cannot make this happen in normal audio operation. Well, OK, except for that grid-conduction-bias-shift-class-C thing, but that's not normal operation either, and in itself causes audible distortion.

Humans being what they are, they think that if removing the diode "fixes" the amp, then the intent of the diodes was wrong. This is a simple thing to think - and wrong. For some ideas about how the human mind works like this, see http://failblog.cheezburger.com/thereifixedit

[jpfamps]

@ R.G. : the wire I use has such a tough high temperature Epoxy Enamel that starting with sandpaper is real slow.
Surface is so hard and smooth that sandpaper sort of "slides" over it.
Burning takes care of that.
And on thick wire (1 mm to 1.40 mm) a sharp cutter blade is faster and more precise than sandpaper.
Although on 0.65 mm and thinner yes, sandpaper is preferred, a blade can nick the wire.

"Self soldering" wire, by Law, *must* be coloured different.
Usually Pink or Green Fluo, the typical wire you see on Wallwarts, Sega power supplies and such, where they just insert the wire in a PCB hole and simply soldering (with an iron, don't think wave soldering lasts long enough) removes it, leaving behind perfect shiny copper.
Same wire in a "big" transformer may be a fire hazard and on a motor, which overheats when stalled or under load, plus vibrating and spinning at high RPMs under heavy load, is criminal.
But now wire manufacturers supply (on customer's responsability) self solderable with almost transparent colour, which can cheat you.
Look at these transformers which I usually rewind:
Left is 80's Twin Reverb Power Transformer.
Big, heavy, very high quality, wound with good classic wire, fully mounted (EI iron + covers) and then they filled it with Polyester Rosin throuh a cover hole.
And indestructible transformer (unless you short the load and use a nickel as a fuse, the only way to damage them) ; this one still works, I had to rewind it (actually made a new one) because it's 110V only, we have 220V and the amp owner was *sick* of carrying a 12lB autotransformer.
Right is a modern Marshall one (ok, 1992 , then's when they became cheap minded), you'll clearly see the typical bright Pink wire.
Center is a MB *Triple* Rectifier OT (6x6L6), I was very disappointed when I saw the carbonized bright pink and green wire inside .... although they used paper, not plastic bobbins and added a thin strip of wood to take up free space.
Very "ecologic", very "green" .... and very dead.
By the way, I also rewound the OT for the MB combo shown below.

Marshall transformer closeup:

MB transformer closeup
Unfortunately my camera chose to focus on the Tolex, instead of on the coil top.
You can see it has all the "Mojo" stuff: wound on brown paper and cardboard ... but with the "un-Mojo" wire


@ trem:
Yes, it's expensive, but remember one of them went through *3* original Marshall OTs, the customer was desperate; for others just one was enough.
As of the MB one, it had to be ordered from USA, its weight made shipping expensive, it had to clear Customs (paying Duty, of course and probably bribing somebody) and the delay would be at least 2 weeks, so my customers were *happy* to pay and have the amp working within the week.
By the way, I have these here because I preferred to build new ones
I'd say: study those you have.
If you have to work at home for any reason and have time available, it might become an interesting side line.
I'm sure some people must prefer their vintage 40's or 50's Fender/Gibson/Silvertone/Ampeg/Magnatone transformers rewound on the same cores, "the original way".

Juan, excellent post.

I'm always surprised how little good info there is on the net about transformer construction.

It seems that transformer construction took a nose dive in the 80s. I've seen many transformers with no insulation between the layers.

I've always assumed the reliability of older transformers was due to this additional insulation, which of course adds to the cost, not only in the construction of the winding, but presumably due to the fact you have potentially to use a larger stack to get the bobbin in the winding window.

What is the prevailing view of using paper vs polyester interwinding insulation?

A transformer winder I spoke to, expressed the view that paper reduced tracking as it had a rough surface compared with polyester.

Also, does the paper have to be thicker than polyester tape for the equivalent insulation, and what effects would there be on distributed capacitance? Presumably the thicker insulation will have lower capacitance assuming equal dielectric constant (I've not been able to find a dieletric constant for paper, but I would be surprised if it larger than polyester).

Thanks in advance for any answers.

[loudthud]

Look at post 27 in this thread: http://music-electronics-forum.com/t28096/

Under hard overdrive the coupling capacitors to the power tubes take on additional charge (in the form of voltage) from grid current. This shifts the bias into class C territory where there is a short time when neither tube is conducting. The plate current of one tube cuts off abruptly, the other tube is not conducting yet and the result is an inductive spike from the OT.

I neglected to say that the inductance of a speaker can also create a spike. I have done this with a solid state amp and used an MOV to clamp the voltage to a safe level. In fact it's unclear to me if the spike in a tube amp comes from the OT, from the speaker or both. The spikes across the speaker can be hundreds of volts. Do the math to see what that tranlates to on the tube side. Yikes!

[J M Fahey]

In fact it's unclear to me if the spike in a tube amp comes from the OT, from the speaker or both.

I think it's both, but anyway voltage is voltage, no matter where it comes from

[Steve Conner]

There are two possible kinds of spikes.

One kind is created by the OT magnetising inductance, or the speaker's voice coil inductance. These affect the two halves of the primary symmetrically. One end tries to spike below ground, the other tries to spike over twice B+. The "flyback diodes" catch these by conducting forwards and catching the negative-going end of the primary a few diode drops below ground. Through magnetic coupling, this clamps the other end at 2x B+. The flyback diodes are behaving exactly like the diodes used to protect BJT output devices in a solid-state amp.

The second kind of spike is created by a sudden cutoff of current through the OT's leakage inductance. The flyback diodes can't catch these spikes in the forward direction, because by definition, they come about through a lack of magnetic coupling between the offending half-primary and the rest of the transformer. So, the diodes probably suffer reverse breakdown and clamp the spike to 3000V or whatever the total PIV is. Modern avalanche rated diodes can withstand this to some extent, but too much will roast them.

There is no analog to this second kind of spike in a solid-state amp, because the two halves of the output stage are connected directly together with unity coupling.

[kleuck]

I neglected to say that the inductance of a speaker can also create a spike. I have done this with a solid state amp and used an MOV to clamp the voltage to a safe level. In fact it's unclear to me if the spike in a tube amp comes from the OT, from the speaker or both. The spikes across the speaker can be hundreds of volts. Do the math to see what that tranlates to on the tube side. Yikes!

That's why Zobels or Conjunctive filters are made for. And i my own limited experience, depends on both the OT and speaker.

[R.G.]

A transformer winder I spoke to, expressed the view that paper reduced tracking as it had a rough surface compared with polyester.

Polyester film is indeed "slicker" than paper, and has a lower resistance to bending at the ends of a layer. So it may present a skills problem to winders.

Also, does the paper have to be thicker than polyester tape for the equivalent insulation,

The voltage breakover for papers is highly variable, depending on the impregnation, calendering, and so on. One layer of paper cannot be relied on for insulation, as even the best papers have some distribution of voids and conductive contamination. This info comes from a book on the practices of winding paper-foil capacitors. The capacitor guys would never use less than two layers of paper, so the chances of two voids happening one on top of the other was miniscule. The voltage breakdown rating is therefore not all that comparable in the raw state. Several references said that paper by itself had little breakdown strength, but could be made much better by impregnating it with various stuff.
Some numbers I found: Paper - 40-60kV/mm, 1200V/mil; kapton tape, 7700V/mil; distilled water - 65-80kV/mil

Also see http://www.vias.org/eltransformers/l...ers_03_06.html. In fact, if you're interested in transformers, you may want to read that whole online book. Rueben Lee has been quoted many, many times.

I did find some data on wire insulation. The enamel used on wire is something like 6000 to 9000 V/mil. Magnet wire has three versions of film thickness: single, double, and triple thickness. The thickness is something like 0.5 to 1 mil, so each wire has 3000V to 27,000V of insulation on it, and between two wires, you get 6000V to 54000V of insulation *if the winding process or heating/vibration/yada/yada does not break the film. The wire insulation seems to be 3-10 times the insulation of a paper layer.

and what effects would there be on distributed capacitance? Presumably the thicker insulation will have lower capacitance assuming equal dielectric constant (I've not been able to find a dieletric constant for paper, but I would be surprised if it larger than polyester).

Paper is a variable. Google found me several quotes on it, with it being mostly 3-7. It depends a lot on finishing and impregnation, as well as unintentional contamination, which paper ...loves... to soak up. Paper is a felted mass of compressed fibers plus gook within the openings between the fibers. Polyester film has a DC of 2.45-4.3. Not a lot of difference. Most of the insulators are in the 2-5 range, with exceptions like alumina and some other metal oxides.

[J M Fahey]

Agree and add, about

What is the prevailing view of using paper vs polyester interwinding insulation?

Polyester is *much* better , of course.
Best of all, is that you can safely use a much thinner sheet, so it actually does "the paper thing" better than paper
The general consensus is that paper is not the insulator by itself, but simply a separator, guaranteeing a certain layer to layer distance, and the actual insulator is oil (or dreaded PCB).
Now that applies to *big* street power line transformers, and in our area, of small portable transformers, that job is taken by insulating varnish.
But the proper application is:
1) oven dry the transformer, at least a couple hours.
2) let it cool (another couple hours)
3) while still warm, impregnate. If possible, use vacuum, to ensure varnish reaches real deep.
Such a transformer will be very robust.

[jpfamps]

Agree and add, about
Polyester is *much* better , of course.
Best of all, is that you can safely use a much thinner sheet, so it actually does "the paper thing" better than paper
The general consensus is that paper is not the insulator by itself, but simply a separator, guaranteeing a certain layer to layer distance, and the actual insulator is oil (or dreaded PCB).
Now that applies to *big* street power line transformers, and in our area, of small portable transformers, that job is taken by insulating varnish.
But the proper application is:
1) oven dry the transformer, at least a couple hours.
2) let it cool (another couple hours)
3) while still warm, impregnate. If possible, use vacuum, to ensure varnish reaches real deep.
Such a transformer will be very robust.

Thanks for the info.

I was always under the impression that polyester would be the better insulator, and of course the heterogeneous nature of paper makes it difficult to be very specific about any of the parameters.

However, using paper implies greater distance between the layers, and hence you would imagine lower distributed capacitance, which I would again assume will affect the performance of the transformer.

[R.G.]

However, using paper implies greater distance between the layers, and hence you would imagine lower distributed capacitance, which I would again assume will affect the performance of the transformer.

]
Yes, it does. However, it's very much a secondary or tertiary effect. For audio transformers of common voltage/impedance matching ranges the Really Big Deals are the primary magnetizing inductance and the leakage inductances.

In designing high performance audio matching transformers, there is a hierarchy of design tasks. I'm doing this off the top of my head, but it's something like:
1. Get a big enough core area and window area lamination and stack to handle the power needed and low frequencies to be passed. This is a first guess, and will often have to be iterated upon if you take many of the extra steps below in search of higher performance.

2. Calculate the primary turns for the primary inductance and the resulting secondary turns for the necessary power and/or impedance transformations. Fit the wire to the window, taking into account margins for easy manufacture and interlayer insulation for ease of winding and primary/secondary isolation, if applicable.

3. Decide your interleaving scheme to meet the leakage inductance needs set by the frequency response requirements. This will mostly affect the stack of interlayer insulation and interconnection schemes before final lead attachment. The interleaving scheme will also effect balance of both leakage inductance and wire resistance in the sections. It is best to have balanced leakage inductances from/between all primary and secondary windings. However, perfect balance is not always possible. The interleaving scheme usually doesn't eat up a whole lot more of the winding window than not interleaving.

4. Recalculate wire and window fit and and margins, and if this doesn't go inside the estimate you picked in (1), go back and get more stack or a bigger lamination, then work back to this point interatively until you have a scheme you think works.

5. Calculate an estimate of the frequency response based on the primary and leakage inductances, then estimate the self-capacitance of the various windings and the winding-to-winding capacitances from the sectionalizing exercise. If needed, subdivide the windings into sections. The most-elaborately-sectionalized audio output transformer I've ever seen in the "Williamson" OT, which has two sections. Sectionalizing will affect leakage inductance and resistance, both in absolute amounts and in balancing. Sectionalizing will also make window fit worse, because now each section has to have end margins, and these all come out of the total window length, which makes for smaller wires or a bigger window or a bigger stack to get less turns, or all of the above. If you blow your core size or stack limits (best not to go above 2:1 stack to tongue length for most winding), then go back to (1) and work through it iteratively again.

The point of that long winded narrative was that winding self capacitance may be ignored til last, if even then.

It is *very* likely that no guitar amps have sectionalized balanced windings unless someone either copied the Williamson, or got a commercial sectionalized one by semi-accident. Non-interleaved designs had no thought to self capacitance at all; interleaved ones may or may not have, very unlikely for guitar. Maybe for tube hifi.

Just guessing, but I think it's not too far off accurate.

[Steve Conner]

For DIY transformer builds, I always recommend using a really big, high quality core. If you're only making one or two units, the cost of the materials isn't so important. You'll end up with plenty of magnetising inductance and a nice spacious winding window, so a better chance of nailing the design first time instead of having to scrap and rewind.

Of course that may not be so good if you're trying to make something that sounds like a cheap and nasty guitar amp OT.

In guitar amp work, the high frequency loss from distributed capacitance is far less important than the loss due to the speaker. Leakage inductance comes second, I think. In tube hi-fi it's a completely different story, you're looking for 2 octaves more high end and up to 10x more NFB, and the low-pass filter formed by distributed capacitance and leakage inductance is your worst enemy.