Repairing Burned-up Switched Mode Transformer

[rossmoffett]

I'm in the process of repairing a controller for an exotic piece of lab equipment (turbomolecular pump) and I've narrowed down the problem to a 12VDC SMPS transformer.  After all of the silicon and capacitors tested good, I cranked up the supply slowly using a 100W rheostat.  That transformer which had a fan mounted conspicuously above it started letting out its magic smoke.

Is it worth trying to re-wind it?  Replacing the power supply board isn't really an option.  I really doubt the transformer is available (and if it were, wouldn't know how to spec it out).  This is a very limited edition run of Japanese manufacture, circa 1991. 

If I can't re-wind it, the next best thing from my perspective is:

• Reverse Engineer the power circuits
• Remove the busted transformer
• Use external 12VDC / 5VDC sources as needed to replace the burned-out section

Any comments on re-winding a small burned up SMPS transformer?  Thoughts on the above?

[Harvs]

Assuming it's an off-line transformer there's a few things you need to know in order to get the isolation correct.

Chances are the easiest (and probably not very expensive) option is to look up a transformer winding company in your local area and get a quote on re-winding it for you.

[BiOzZ]

a 12v and 5v switching power supply is not remotely rare and you can get them cheap off ebay (like this 4A one http://www.ebay.com/itm/Double-Output-DC12V-1A-and-DC5V-4A-Universal-Regulated-Switching-Power-Supply-/111153167634?pt=LH_DefaultDomain_0&hash=item19e13ebd12 )

but i have rewound SMPS transformers before ... you simply need to desolder the leads, place some wire on the top and bottom part of the Ecore and heat it up slowly in an oven and pull it apart to break the epoxy and rewind it with proper tape and proper mag wire than glue it back together

many problems can come from this like having it supply too much or too little voltage!

[BiOzZ]

also chances are who ever made the company did not make the power supply so if you can find any marks on the supply PCB (manufacturer, numbers etc)  you might find out that that was a common SMPS for the time and get a replacement! especially if its lab gear i dont think anyone would take the time to make a separate power supply, run it, test it exc when they can just buy a common one on the cheap and if there only making like 100 of them the cost of making one alone would be too much (i inturn for a place where we make SMPSs for other companies and do our own testing and getting it too pass FCC and UL standards)

you can also find an equivalent transformer probably somewhere on the internet unless they wound it them selves with some strange specs and id say thats probably not the case!

[flolic]

Transformer is shorted anyway. Nothing cost you to try to rewind it.
That should not be too big problem, if you can separate ferrite core...

[rossmoffett]

There is numbering on it.  On the top and side, the rather cryptic "0892" and "162."  That was unsatisfying, so I hot-air gunned it until I could get it off and I see:

TOKIN
OZ
25B

Most likely some Japanese company in business in the early 90's.  I can't find anything - but then, I've never bothered with SMPS before.  Always the easy-to-understand heavy iron.  Can anyone help me out?

[BiOzZ]

Yeah TOKIN is a company in japan there still arround http://www.nec-tokin.com/english/product/product_list.html we have done work for them before ... but can we get high quality top and bottom views of the power supply? i can either point out the make of that, find you a suitable replacement or a drop in replacement for the transformer if i can

[rossmoffett]

Here you go, if you're curious, but the power supply is definitely custom.  It's an Osaka Vacuum Turbo-molecular pump controller.  There are four boards, one for power distribution/fusing, one for power supply and three-phase AC motor drive, one for digital controls monitoring temp switches, run-time and etc and another for the display and pushbutton controls.

I traced pieces of it.  The power supply board you see pictured has 14 silicon devices.  Two are half-bridge diodes, collectively the full bridge rectifier, near the blue electrolytics.  That's the extent of the high voltage supply, this is directly distributed to the 12 H-bridge transistors.  When the power supply is at full power, there is a +200VDC supply and a -200VDC supply.  H-bridge switching doubles that to 800 VDC total to control the 40kRPM+ three-phase motor in the turbomolecular pump.

On the power distribution board, the hot leg of the 120VAC power input goes through a 33 ohm 5 watt power resistor out to the rectifier circuit.  A 12VDC relay bypasses the 33 ohm resistor so that it's not in circuit anymore - confusing at first, but probably designed as a safety so that if the 12VDC circuit doesn't ramp up AC power gets cut off when that 33 ohm resistor burns out (not designed for full power, drops 110 VAC).  That would make sure that the other 12VDC safety relays are operable all the time.  That resistor was burned up - my first clue.  I replaced it and it burned up again.  I removed it again and replaced with a 100W rheostat and slowly ramped up voltage to around 20VAC until I noticed the trafo smoking. 

I found that website in the meantime, that they merged with NEC.  Do you have any ideas about what product category this may be in?

[rossmoffett]

If you re-read my post you may come to understand that I placed the rheostat on the 120VAC supply to replace a power resistor that was failing, I didn't tap into the circuit with 120VAC willy-nilly.  I'm aware of why it burned up - I was trying to identify the point of failure because all of the silicon, resistors and capacitors checked out in the areas where I could see that there had been a heat problem.  Most likely a failure of the fan or plain hot operating environment caused the transformer to short some windings.  The turbopump it was mated to has a blown bearing as well, usually caused from heat stress.

Based on the NEC-TOKIN website, I think these trafos are custom wound based on a selection of iron and bobbin requests.  This is looking more and more like it is a good idea to just rip out the affected power supply components and use a third-party supply to compensate.

[BiOzZ]

yeah i cant find anything directly on the PCB that would show me a manufacture company they might have been asked to make it without the companys logo

your best bet here is to make your own power supply by looking at the outputs and buying equivalents (like that one i showed you earlier) and fitting it in maybe with some LC filters to make sure its nice and smooth but a power supply replacement should not reflect the performance at all as long as you make them the same voltage and enough current (or more current)

do you by any chance have a schematic of this supply?

[rossmoffett]

I'm going to look a little harder for a schematic of the supply.  Osaka has not been forthcoming so far.  If I can't get it, it's something I can trace out.  It's just a double sided PCB using 80's components, nothing too hard to figure out.