Broken Variac - Fixable??

[CarsonReidDavis]

I picked up this broken variac for the low price of free. It is a General Radio W2-S12-2. Any information on the specs would be welcome. There are many W2s on ebay, but this one seems to be a variant.

It has some broken wires. Any ideas on the best way to repair it?

Detailed pictures are in this album https://imgur.com/a/IF0ZLXV

[UPDATE] Variac size comparison in this album https://imgur.com/a/B4tsuq6

[Gregg]

You haven't even shown what abominations are hiding under the duct tape.  Maybe you are afraid to look, but to me it looks like somebody had a go at it and may have botched it so badly that repair is beyond consideration.
I'm sure it is more work than it is worth.  But there is a slight possibility that you could peel back the broken wires far enough to slide some small heat shrink over the longer end and solder a short length of new wire between the two ends.  Make absolutely sure that you don't cross any wires or short two turns together.
I would look into some UV hardening glue to keep the repaired section as close to original as possible using smallish dots of glue.  The open coils are part of the transformer cooling system.
Remember the Variac is a mains powered device and safety is tantamount.  I'm sure there will be lots of posts pertaining to safety.



 

[CarsonReidDavis]

Thanks for the reply.

I just removed the electrical tape and added a picture to the album. It appears you were correct with your guess about a previous repair job. The previous repair was using the tape as insulation. Doesn't seem like a terrible idea. Solder and then tape each layer apart as you go instead of struggling with heat shrink.

So do the wires need to lay down in their original configuration? How important is this?

There appear to be a total of 4 cut strands and 1 repaired strand.

[coromonadalix]

if you repair them carefully and properly isolate them  your variac may work again, at least i would try for the heck of it

Put an wire alongside the broken ones to sturdy them, shrink tube them and put an high temp varnish or sillicone to fix them properly ...

[Gregg]

Tape has two problems.  First it covers more surface area of the wires leading to potential hot spots; this is why I suggested spots of glue to keep the wires aligned and apart after repairing.  Second problem with tape is it can lose its adhesiveness and come off leading to shock and short circuit hazards.
The original wires were attached to the core with a thin layer of electrical grade varnish to keep them from moving against each other and shorting to each other or to the core.  Wires in transformers and other magnetic devices tend to move due to various forces.   
The wires don’t have to lie perfectly flat to the core, the further away the more loss of magnetic coupling and the more you should de-rate the amperage.  Try not to zig zag them across each other too much mostly because they may want to move against each other and wear away the insulation which would short adjacent windings and ultimately let smoke out. 
You could double up some kapton tape sticky sides together and put it next to the core in the groove to help align the wires to keep them in place as you solder. Then remove it
Variacs can be connected either direction, i.e. clockwise or counterclockwise to raise the voltage.  I would suggest making the lower voltage ranges the end away from your repair.
Put a suitable fuse in line before plugging it in and wear some safety glasses etc.   

[Gyro]

I would actually use the Kapton tape as the primary insulating medium for this sort of repair. It is thin (so reasonably thermally conductive), high temperature resistant, and doesn't plastic flow under pressure. You should be able to interleave or 'zig-jag' it between adjacent joins, that way it will be mechanically restrained. The voltage between adjacent turns is very low (in the order of 1-3V) but obviously you will have big problems if you allow any of them to short.

That repair under the tape was obviously done very badly. If you carefully strip and tin the wire ends, bridge and solder a length of same diameter wire across each break, then you ought to be able to make a very compact repair. Make sure you don't leave any sharp points.

Don't forget to fuse the output too.

[CM800]

I spot a whole load of missing enamal too:

I suggest, if you truly want to give it a go, give them space (make sure they arn't shorting) and cover them in epoxy.

else... rewind the whole damn thing.

[Ian.M]

Selective rewinding may be possible - the worst damage seems to be localised over about 25 turns, and doing a neat job of rewinding 25 turns is probably going to be easier than patching individual turns - *IF* that black compound isn't permanent glue between the toroidial winding assembly and the chassis.

Epoxy isn't ideal for this sort of work unless you get one that's suitable for high temperatures.   Many ordinary epoxies loose most of their strength well under 100°C, and some soften enough that they can simply be scraped off at that temperature.  Magnet wire enamel typically has a temperature rating between 105°C on the low end and 220°C on the high end.   If the manufacturer's data for the transformer includes a thermal derating factor, you can estimate the maximum possible surface temperature from it.

[Gyro]

Hard to know on this one...

Regarding the missing enamel, I agree CM800, as long as there is physical separation then adding setting insulation - I would go for Polyurethane lacquer, will be fine given the low voltage between turns.

As for partial rewinding, I suspect that the windings are securely fixed at the edges so that they don't get dragged around by the brush, I've seen Variacs where the space between the toroid and spindle bush was completely potted (maybe that's what is visible between the bottom edge and the chassis). There's also the problem of grinding the bare track section flat and level with the rest of the winding.

 

[Ian.M]

Yes - if its potted onto the spindle bush, rewinding isn't an option. 

I wouldn't be too scared of levelling the track - provided the new winding is well seated and glued down with an appropriate glue or lacquer, a fine file will do the worst of it, then protect the undamaged track either side with Kapton tape and stone it radially till its within a few RCH of dead flush then remove the tape and very carefully stone tangentially to smooth the transitions.  Finally clean up dust and swarf with a tack-rag and burnish the new section of track surface.

[james_s]

If you use epoxy, make sure you use something that doesn't soften when it gets hot, many epoxies do. Something like JB Weld may work to replace enamel, however do test it first to make sure it's not slightly conductive, I really don't know whether it contains metal particles.

[Ian.M]

Most JBweld Epoxies are loaded with steel particles.  I certainly wouldn't use them for this purpose as although they are generally insulating at low voltage, there's no way of telling if the path length through the epoxy between the steel reinforcement grains is long enough to withstand the maximum possible voltage between the winding and the core.

A high temperature clear or ceramic loaded epoxy could work, but I am unfamiliar with which products are likely to be available to you regionally in small quantities.

[james_s]

Maybe one of the ones meant for bonding concrete? Or if there's one that says it's suitable for high temperatures. Otherwise I'm not sure how to tell other than testing it. In my adventures of repairing and reverse engineering potted modules I've found that some epoxy gets very soft when heated, some other types stay rock hard.

[Grapsus]

What's the maximum output voltage ? Some variacs can go slightly over the input voltage. If it's the case, maybe it's possible to short the broken section and still have a usable range ?

[CarsonReidDavis]

So I have finally gotten around to fixing the Variac!

First I scraped off the insulation from each of the broken wire ends. Then I cut some tiny pieces of wire and soldered bridges between each of the 5 broken wires. I put heat shrink tubing on each repair to insulate it. I haven't done anything to repair the dented sections, nor have I applied any resin or glue.

I slowly fired up the variac using the output of a second variac just in case. It appears to work normally, outputting 122 volts on the main channel.

I have updated the previous album of pictures https://imgur.com/a/IF0ZLXV with a bunch of closeups from the repair process. Let me know if I've done anything so poorly that it needs to be redone.

Anyone know some good ways to test the viability of my repair, and possibly determine some reasonable max amp specs for the unit?

[Gyro]

Well it may not be pretty but it looks as if you've achieved a workable repair. 

If any of your solder joints fail, the heatshrink should keep the ends from flailing around, worst case a join failing open will arc but you don't normally run variacs unattended. Check that the any scratched enamel on adjacent turns isn't in danger of shorting. As long as the voltage adjusts smoothly and there aren't any nasty smells from shorted turns then you're probably good to go.

As far as current capacity, hard to tell. I see that the terminal panel says "Special Model, see instruction sheet" so no help there. Your best course is probably to try to match dimensions and weight to a known one.

Don't forget to fuse the input (conservatively) and the Output.


P.S. Make sure you haven't loosened any of the ends of the turns where they pass under the brush.

[james_s]

You could attach a thermocouple and gradually increase the current, keeping tabs on the thermal rise. Or calculate the total length of wire, and the resistive heating under a certain load.

[station240]

The max current will be whatever the diameter of the wire used in the transformer.
I have a book here with a current to wire gauge conversion table, intended for constructing transformers, but obviously would work in reverse.

At a guess this was a custom unit mounted inside a larger machine, would explain why it's custom.

[GeorgeOfTheJungle]

Anyone know some good ways to test the viability of my repair, and possibly determine some reasonable max amp specs for the unit?

Max current is given by the wire diameter: https://www.powerstream.com/Wire_Size.htm
Max power output is a function of output voltage, so at low output voltages less max power, at max voltage output more max power output: (max current times output voltage)

Looking at the pictures I would say between 10 and 16 amps (0.8..1 mm wire diameter).

[Ian.M]

Its nowhere near so simple.   There are just too many unknowns for any book of tables to help.

The enamel on the wire has a temperature rating for 10000 operating hours.  The I²R losses, that temperature rating , and the thermal resistance to ambient determine the maximum current the wire is rated for.
All the I²R (copper) losses in the wire and the iron losses in the core must be dissipated from the surface of the toroid.  This is complicated by the potted base, which is likely to be poorer at conducting heat away than convection from the winding surface in free air.   A further issue is the unknown wiper current rating.

The best way forward would be to determine the max temperature rating of similar Variacs of comparable dimensions from the same manufacturer.  Initially measure the cold condition DC resistance of the winding  and cautiously apply low voltage DC to the whole winding to heat it with a known power, while monitoring the surface temperature rise with an IR thermometer or a thermocouple or other sensor applied to the surface with a dab of thermal grease and shielded from the ambient with a small block of polystyrene foam.  Note the temperature when it reaches a steady state. 

Next power its winding normally from the line with no load on the wiper and either measure the input power if you have a sufficiently sensitive AC wattmeter or again determine the steady state temperature,   That power or temperature rise above ambient is the iron losses + that part of the copper losses due to the magnetising current.

Given the ambient temperature, the max winding temperature, the temperature rise per watt, and the unloaded losses (either as power or temperature rise), you can calculate the maximum copper losses it can tolerate.   That would give you a maximum current for the whole winding from I²R and the winding resistance at maximum temperature (calculated from the cold resistance using published resistance vs temperature tables or formula for copper) .   

Due to auto-transformer action, the current in the winding is locally at a maximum when the wiper is very near (but not at) either the N end or the tap where L is applied, and for a conservative rating you need to keep that current under the maximum current calculated for the whole winding.   The odds of the wiper rating being less than the winding rating are extremely small, and as long as the damage isn't near the ends of the winding or any tap, hot-spotting due to the nicks in the wire should be insignificant as the current through that section of wire will always be significantly less than the calculated limit.

If you've got access to a calibrated thermal camera, you may be able to push the ratings higher based on observing the wiper temperature rise and checking for hotspots with a known load and the wiper at half the input voltage, then re-checking at 90% of full voltage to get an estimate of how well local heating in the winding is dissipated over the whole surface by conduction through the core, then recaculating the maximum load current vs wiper position, so that no part of the winding or the wiper exceeds the temperature limit.

[GeorgeOfTheJungle]

Minus a -say- 10% to account for losses and a safety margin...? :-)

[Ian.M]

No. No need for fixed derating (10% or any other value) unless you are unsure of the maximum temperature limit. 

The above procedure determines the maximum power that can be dissipated safely in the whole winding,  subtracts the no-load losses, then applies that whole winding rating to any part of it.  As the load current of an autotransformer is the sum of the current in the two parts of the winding above and below the load tap (or wiper), which are in opposite directions, at intermediate wiper positions the copper losses will be much lower, falling to as low as 25% at half the input voltage.   When the wiper is really near the 0V end or the input tap, most of the winding will only be carrying a small fraction of the load current so the core will be much cooler and the few turns carrying most of the load current will be much better heatsinked so it already has a considerable and very conservative derating factor built-in.

However,  you still need to derate for ambient temperature.

[Harb]

Well it may not be pretty but it looks as if you've achieved a workable repair. 

If any of your solder joints fail, the heatshrink should keep the ends from flailing around, worst case a join failing open will arc but you don't normally run variacs unattended. Check that the any scratched enamel on adjacent turns isn't in danger of shorting. As long as the voltage adjusts smoothly and there aren't any nasty smells from shorted turns then you're probably good to go.

As far as current capacity, hard to tell. I see that the terminal panel says "Special Model, see instruction sheet" so no help there. Your best course is probably to try to match dimensions and weight to a known one.

Don't forget to fuse the input (conservatively) and the Output.


P.S. Make sure you haven't loosened any of the ends of the turns where they pass under the brush.

Load it up until you get smoke, take note , then continue with repairs

[CarsonReidDavis]

Thanks everyone for your input!

I have done no further testing on the small variac, however I have created an album with a size and model comparison between it and my other 3 variacs that all have known ratings. It seems to be smaller than my 0-135V 3A variac, so I assume it can handle less current.

I have a FLIR C2 camera which I might be able to use for future testing.

I have quite a diversity of variac sizes. Here is the album with the new pictures https://imgur.com/a/B4tsuq6

[Gyro]

It doesn't look too dissimilar to the 3A one when you take the different aspect ratio of the cores into account, but best to play it safe. The 3A one has much more attached metalwork than the GR one, I'm not sure how easy it would be to strip it down to a similar state in order to do a weight comparison. Another tell-tale might be a comparison of the area of the carbon brushes. The C2 would certainly be helpful, load them both up to, say, 1A at 50% and compare the winding (and brush) temperatures.

You might just want to case it up as a nice conveniently small 1-2A one though. You don't always want a great big chunk of iron clogging up your bench.

[Gregg]

 :-+Congratulations, it looks like a reasonable repair.  Just don't snag the loops of wire as you take it off or put it on your bench.  Some of the old school variacs had nice perforated metal surrounds to protect them from damage; maybe another project in the offing to test your skills.

[Ian.M]

That would be quite a bit of work.   Its a pure open frame panel mount one, so unless you have a suitable enclosure you can mount it shaft up in, at the very least its going to need a back plate connected to the front plate with four thick walled pillars and massive bolts or studding so it doesn't rest on the rotor or winding before you even consider how to wrap it in metal mesh.   Then you've got to make a scale for it and find or make a pointer knob for that shaft size, and the knob's got to be very well fitted to the shaft as the inertial and shock loads when the rotor hits its end stops can be quite high.   

If I had that set sitting on my bench I'd concentrate on getting the big one set up for general bench use. Its going to need a  terminal cover and preferably a voltmeter, ammeter, circuit breaker,  voltage selection switch (so you can use it as 115V in 0-270V or 0-135V), IEC C13/C14 output and input connectors and a load power switch.  It could also do with a pair of lifting handles and some rubber feet, as I bet its a heavy brute.

For a weight comparison with the 3A one, you don't need to fully strip it down though you should probably remove the knob.   Just measure the case metal thickness and measure dimensions to calculate the total area (allowing for flanges etc.) and check with a magnet if its steel or aluminium, then look up the density and calculate the mass of the case and subtract it from the total measured weight.  It will be near enough.