Transformer rewinding

[Ted]

If you have the time, you can sometimes match modern metric bobbins to older inch-sized lams. Buying a box of 50 metric bobbins cost me about the same as having a custom bobbin made for a B15 mains transformer. However I spent about a day looking through manufacturers catalogues to find some that fitted.

You can specify the temperature rating of the wire used in transformers.

Transformer Temperature & Insulation

Guess what? The higher temperature wire is more expensive. I have stuff made to Class H (180C) in the day job, and run them at 120C. They are sufficiently reliable that we don't need to collect reliabilty data for them.

As for older amps not eating transformers - old Fenders maybe. But AC30s and Marshall JMPs regularly do.

[R.G.]

This discussion is ringing some bells with me because I'm in the middle of the wind-a-new-driver-trannie project. I have a number of comments in no particular order.

If I absolutely have to get a transformer rewound I try to meet my transformer winder half way - stripping the transformer and separating and cleaning the laminations. There's not much point doing this if you can't get off-the shelf bobbins for your rewind. A lot of older lams and their associated bobbins are not available anymore.

Saving the laminations and bobbins is a reasonable objective if the end result is an exact replica. However, it's worth remembering that the older the transformer, the less difficult it is to make new parts, laminations excluded. A new paper bobbin requires some time in the garage carving a wooden plug form to wind on, covering it with waxed paper for a release layer, and then laminating kraft paper and high-temp epoxy or phenolic glues to make up a sufficient thickness of cardboard, and then sawing off the right length. You can even use glass cloth for some off the layers and have a REALLY durable replacement for the original. Rocket surgery wasn't invented until the 1980s.

You even get a chance to try it on your lams and find out that you need to make one more that works after your first practice run.

Transformer manufacturers are usually capable of designing you a transformer to spec. This is often the cheapest and most painless route. If you have a customer who sincerely believes that not using original 1950s laminations will completely alter the tone of his/her amp this may not be possible. I do notice that transformers designed this way are often significantly smaller than the originals. I'm happy to believe that this is due to improvements in insulation materials, varnish and laminations in the last 50 years. I've never had a new design transformer fail, but given the small sample size that proves almost nothing

You hit a really important point - small sample size, and the shadowy thing hiding behind it: survivors' bias. The storied reliability of the old amps with waxed-paper insulation and floor-sweepings impregnation come partly from using the surviving amps gently, and partly from not knowing what percentage of the old amps died along the way, as well as not knowing what they died from. Surviving 50-year-old amps may be the 2% of the original population that was lucky, or far out on the reliability distribution. Or only 2% may have died. Can't tell without knowing the complete history of the manufacturing runs, and that's lost to us.

I'm getting a clearer picture of why a rewind is so expensive, but the idea of a 12lb transformer (and all the others) uselessly sitting there with such a 'minor' fault as an internal short on one winding is frustrating.

That's because you're confusing the bulk materials with the embodied labor. When that chunk of iron and copper with a little bit of insulation seasoning failed, the insulation-seasoning failed. The iron and copper are cheap compared to the labor that was used to distribute that little bit of insulation at just the right places inside the iron and copper bits to make it all work. What's been lost is the trivial cost of the insulation, but all of the labor needed to put the insulation in the right places again. And now more than any time in the past, labor is expensive with respect to commodities. You're paying for new labor to sift a little insulation back into the right places.

It's tempting at first glance to take a simple approach, and say that the new MM transformers just aren't all that good. Maybe they are, maybe they aren't. I've had my share of QC control problems with MM products. They did a great fill for me on my first order of 24+ units to capture my business, but my second order of a similar quantity looked like it got filled with all of the rejects/seconds that they had lying around the shop. The result has been that I haven't gone near that supplier for well over a decade. Their transition to a boutique marketing scheme that was accompanied by insane price hikes has made it easy for me to continue to avoid them.

The failure rate for manually made single units is nearly always higher than for parts from a well-run manufacturing line. The manual approach will and must have a much higher variation of every variable than the manufacturing line. (I can already hear some of the rebuttals about this statement, with images of luthiers with small-lensed wire-rimmed glasses and white beards lovingly scraping the bouts of guitars in progress... )
That being the case, it's not surprising that the failure rate is higher for low-quantity runs. MM **had** to up their prices, because in a low-volume business, economies of scale go away. Not that it makes it any more palatable, but the marginal costs of low-volume business are higher, and the niche advantages of ancestor worship for guitars and amps allow it to exist.

but to be fair, it's probably not fair to blame MM when new production transformers fail when they're being abused. There may be a selection bias in how these OTs are being treated by end users. Since the advent of the attenuator, end users are a lot harder on amps today than they were decades ago. If someone has an original 1950s Tweed Deluxe, they're likely to be easy on it because of it's collectible value. OTOH if someone buys a new production MM transformer, it's viewed as a disposable item that can easily be replaced, so there is little impediment for the user to flog a new production OT to the point of failure. So maybe it's not a problem with inferior materials or construction. Maybe it's just that people are intentionally abusing the new iron because they're not worried about not being able to replace it.

Excellent point. The amps are used differently.

I do know the history of that particular amp and it's been cosseted from new due to the price in the UK and only used for a limited periods. It still has it's tags, original cardboard box and invoice. The owner is over 60 and it never gets turned above 3. Maybe it's just an isolated case - there aren't many reissue Tweed Deluxes around over here so I don't have any other comparisions.

That's what I alluded to in the survivors' bias thing. We know only a small bit of the complete picture, so blind-men-and-elephant is hard to avoid.

In the UK a MM transformer is a high-ticket item (price but not necessarily quality). 5E3 transformers appear to attract a premium and I wish UK manufacturers would build dimensionally-accurate transformer sets that appear to be commonplace in the US.

Some bright boy is going to figure out a way to get those moved from the US to the UK in a reasonable way and be able to take a part of the difference in price for his reward, I suspect.

I'm puzzled why so many OPTs would fail with shorted turns;

All transformer failures that don't involve internal thermal fuses begin as shorted turns, however they end up. In general, it takes a shorted turn to generate the heat needed to further damage the thing. A shorted turn is an insulation failure. If the insulation doesn't fail, the transformer keeps working. Iron and copper will work at temperatures usable for cooking many foods. It's always the insulation that fails, and from there the temps caused by the primary power lines being able to focus on a shorted turn lead to secondary insulation failure (and fires!), solder melting, and wires melted, then burned open locally by arcs.

The failure cascade always starts with either a local mechanical or voltage failure of insulation, or from thermal failure of insulation in a hot spot. From there, things can go south several ways.

to my understanding the voltage relationship between adjacent windings or layers is derrived from the total voltage across the winding divided by the number of turns. Now, even accepting the flyback effect of a PP amp effectively doubling the B+, this still makes the volts/turn fairly low on an OPT primary - certainly within the limits imposed by the enamel insulation and any paper insulation between layers. I'm making an assumption here that the transformer was designed with an effective safety margin and that it can operate continually at its rated output. Perhaps manufacturers are squeezing designs by reducing wire gauge and core sizes, or the wire is getting stretched due to high-speed winding techniques and this is introducing cracks in the insulation.

High voltage arcing that punctures insulation is quite rare, for the reasons you mention. If there is a weak spot in the insulation, it takes the confluence of that weak spot and high voltage in the same place for it to arc. Once it arcs, the arc temperatures burn the insulation around the failed spot and the heating continues after the arc is over.

Modern insulation films are in fact better in mechanical, chemical, and voltage resistance and more consistent than the films of the 50s and 60s. It may well be that the winding machines have become too simplified. Or, it may be survivors' bias. Hard to tell. Not using layer insulation has a part in it for all random wound bobbin trannies, though.

As Steve said - you want consistency? Use layer insulation. I'm specifying how do layer insulation inside a walled bobbin for the driver trannie. I may have mentioned this before in the never-ending argument about paper and plastic bobbins... ... but it's not the bobbin or layer materials that matters as long as they're non-magnetic and non-conductive; what matters is the position of every wire turn with respect to every other wire turn. Put the wires where they should go, and you can predict the result well. Scatter wind them in tubs on the bobbin and you get - well, scatter. This is one reason modern film insulation needs to be better. Scatter winding puts more mechanical and voltage strain on the wire insulation.

I did specify an OT for the Workhorse amps, and specified how the winding was to be done. The trannie manufacturer didn't like it because it took his most skilled operators a long time to make them. But as far as I know we've had zero (that's 0.00000) output trannie failures from the runs.

... which means not much because we didn't build them in hundreds of thousands, of course. But there were enough to see the margins of the distribution come up if we had been close to the reliability edges.

[bob p]

"I'm getting a clearer picture of why a rewind is so expensive, but the idea of a 12lb transformer (and all the others) uselessly sitting there with such a 'minor' fault as an internal short on one winding is frustrating."

Hey! I'm not the guy who said that!

[R.G.]

Ooops... sorry. I have often regretted that my typing is so bad that it ruins some of the thought and attention processes that have to run to tell my fingers what to type!

I'll go fix that.


Later:
mutter, moan, whine mutter...
Looks like the edit time has expired on my post. It was Mick, not bob.

[Mick Bailey]

Some interesting points being raised here and all helpful. Something I'll add is the failure will sometimes be at the junction of the winding with the connection lead - burnt through the winding immediately adjacent to the solder joint. This just means cutting through the outer insulation, re-soldering and then carefully re-wrapping and sealing with varnish. Not too many like this, but it's good when one comes along.

Then it's a case of trying to identify what caused the O/C in the first place, but in many cases (particularly with OPTs) there's no other fault.

[Ted]

Awful lot of actual engineering talk here. Can't we just go back to discussing the benefits of gold-plated mains plugs?

[Steve Conner]

Bah, gold plating is OK if you like that warm, golden, euphonic sound. I prefer a carbon nanotube coating for those inky black silences between notes. It's far more expensive than gold, so it must be better

[Mick Bailey]

Bah, gold plating is OK if you like that warm, golden, euphonic sound. I prefer a carbon nanotube coating for those inky black silences between notes. It's far more expensive than gold, so it must be better

Better still if you use a green marker on the end of each individual nanotube. Jimi's silence between notes on the mono recording of 'Axis: Bold As Love' never sounded so dark. The absence of tone is remarkable - haunting, in fact, like the sob of an orphaned child in the corner of a darkened room. Gold simply can't achieve this.

[Steve Conner]

Ahh, the heft and meat of mono... Congratulations Mick, you win a guest column in Stereophile magazine I'm off to scour the charity shops for 78s to put in my new record cleaning machine.

[R.G.]

Some interesting points being raised here and all helpful. Something I'll add is the failure will sometimes be at the junction of the winding with the connection lead - burnt through the winding immediately adjacent to the solder joint. This just means cutting through the outer insulation, re-soldering and then carefully re-wrapping and sealing with varnish. Not too many like this, but it's good when one comes along.

That probably started out as an assembly flaw - too-vigorous scraping of the enamel on the wire, or nicked wire on the lead-out. Possibly also a handling issue - carrying the tranny around by the wire leads. From there, heating cycles fatigue the metal at the stress concentration, reducing copper area, increasing local heating, increasing thermal stress...

Then it's a case of trying to identify what caused the O/C in the first place, but in many cases (particularly with OPTs) there's no other fault.

It is entirely possible that the problem is not inside the OPT at all, at least not all of it. Heat is a quantity that must flow out of the trannie, like water in a tank. The difference is that heat will increase the local temperature until the heat flow out equals the heat being generated. There is *no limit* to the temperature that can happen to balance the heat flowing out with the heat being generated. A grain of wheat lamp uses a trivial amount of power, but the filament has a hard time conducting or convecting the heat away. The only way the heat flows can balance is for the temperature to rise to white-hot so it can radiate away as photons.

If the inside of an amp is not well ventilated, the temp inside of the trannie rises even higher. Poor ventilation, even a coating of sticky, fluffy bar funk will cut down on heat out of the trannie and make the inside hotter. If it has a weak spot, one day - well, you know.

Hard to find those on the work bench.

[BLS Transformers]

Someone asked what eli-meli is. It's an electrical grade paper with melinex (mylar) polyester film bonded to one side. I use it for all my interwinding insulation (unless the customer wants something "vintage looking" then it's out with the old brown presspahn and SRBP bobbins).

[Mick Bailey]

I've now pretty much abandoned the idea of rewinds. After a fair bit of experiment, the conclusion I've drawn is that there's too much time involved as well as investment in materials to make it feasible, compared to what I can pull in off general repairs. No problem with the process, and the results have been 100% successful. But I'm not in the business of writing off my own time - it isn't a hobby.

Actually, the whole experience gives me a better appreciation of the work and what's involved. So that's two things off my list - speaker recones and transformer rewinds.