E-core shorted by mounting screws watch how you mount them

[oz2cpu]

here are a few pictures from the same transformator,
the powersupply i play with is 30 years old,
the E-core is a 500VA size, all original, the top parts with original varnish,
when i removed one of the screws, the short got away, idle power dropped, and transformator ran icecold
the pictures are taken right after power on, just to give a good contrast to cold transformer,
idle power drop was 10W

the powersupply is : Elektro Automatik EA-PS 7016-20A LCD
if you own one of them, you better check your idle power, and check the transformator with a thermal cam.

[Benta]

Nice photos, but I've no idea what you are talking about.

[moffy]

That aluminium bar holding the terminals looks like a great place to induce eddy current losses sitting just above the windings.

[Ian.M]

The mounting holes are *supposed* to be in low flux areas of the laminations, but when you cheap out on the core area, bad things happen unless you insulate the bolts, e.g. sleeve with heatshrink and use fibre washers.

[oz2cpu]

The explain : IF all field only travel inside the iron, and down both sides exactly 50% to each side,
you will NOT see any voltage inducted in the screw area,
it is called Stray Flux, it is the flux that continue up and not perfectly out the sides of en E core,
see this picture of a perfect ideal E core, in this case you could put a wire thru the two top holes
and get zero voltage out..
But that is not how it works in reality.
and this is why all transformator manufactures who know what they are doing, put some type of isolation under the screws.

[oz2cpu]

measurement setup

[oz2cpu]

the yellow, wire, 1 normal turn = 938 mV
wow that is quite hi flux for a mains treansformator

[oz2cpu]

the blue wire is placed where the screws will perform an unwanted short
if not isolated, the voltage here explains the Stray Field
345mV for this "winding"

[moffy]

From your measurements on the blue wire it looks like the aluminium bar connecting the screws would be carrying the majority of the current.

[oz2cpu]

the current need to go all the way round to compleete the circuit
on the far side the circuit is only one or more the the transformator metal sheets,
on the front side it is the aluminium bar or maybe even also one or more of the sheets in parallel,
it was a very interesting day, I had a lot of fun tracking this down, and solving it.

it is still a bit wierd to me a non modified transformator can have this fault.

if any other got a simiilar powersupply ? it could be fun to know if they all got this bug ..
Elektro Automatik EA-PS 7016-20A
most likely all 70xx-xx models could potentially have this problem ?
also please check if you hear loud hum ?
I also have a cure for that problem

[Vovk_Z]

Yes, those tranformer mounting bolts can increase losses (together with aluminium mounting in your case), but you can't just get rid of it. You still have something to hold it together otherwise a tranformer will take itself apart.
Sometimes I replace steel bolts with stainless steel bolts as they are less ferromagnetic. And use some isolation like plastic washers etc.

[oz2cpu]

i did put back both bolts, just added isolation washers and heatshrink on the bolt

[Wolfram]

Interesting discovery! I've seen this issue with toroids, if they are mounted with a bolt passing through the top and bottom of a conductive cabinet, but this one is a bit more subtle. It seems like the shorted turn is intercepting the fraction of the center leg flux that flows outside of the screw holes, giving a similar fraction of the volts per turn.

Yes, those tranformer mounting bolts can increase losses (together with aluminium mounting in your case), but you can't just get rid of it. You still have something to hold it together otherwise a tranformer will take itself apart.
Sometimes I replace steel bolts with stainless steel bolts as they are less ferromagnetic. And use some isolation like plastic washers etc.

In this case it's not about them having losses from being ferromagnetic, but from them being conductive. The fact that iron is ferromagnetic is actually a benefit here, as it increases the effective resistance on account of causing the current penetration depth to be shallower, but that's not likely a major effect in this case. Mitigating it can be as simple as installing insulating washers or other hardware to break the loop, or using a non-conductive bracket on one of the sides.

[Ian.M]

A nasty trap for the unwary is that the transformer manufacture may have relied on the 'blueing'  of the laminations or the thin varnish on them to provide enough resistance to limit eddy currents.  Disturb the bolt and re-tighten it *without* adding insulation and you may break through the treated lamination surface triggering this problem.

[CosteC]

I suppose this manufacturer prioritised grounding of core over efficiency
It still would be good to connect core to PE. I am pretty sure one screw can or even should have good electrical contact to core, while three other should not. Quire easy to make it incorrect in manufacturing, aluminium bar was also very risky, all depending on lacquer integrity.

[oz2cpu]

Good point CosteC, I did ofcourse add PE to core again, when i was done will all my experiments,
the PE solder eye was not added on this picture attached
it show how I mounted the aluminium holder for the terminal block

[CosteC]

I actually need to take a look how it is done in transformers I got. Those are rare today
Nice job fixing this power supply.

[oz2cpu]

thanks a lot, I had a lot of fun and saw something new and unexpected

I made a little video about the whole powersupply
and also all parts of my transformer finds and solutions are explained here

https://youtu.be/YG6AYyzVNcY

the fun thing is i also found a few other wierd things in the power supply
like thermal isolation, where people who are normally smart will try to make good thermal connection :-)

see for your self and have a laugh

[CosteC]

Bit of comments regarding your video:
Insulating heatsinks from chasis - maybe chasis would exceed acceptable temperatures? Nobody likes chasis which burns your fingers (literally). There is legal limit for enclosure temperature for CE mark equipment too. Thermal design of this PSU is however not great...

Why transistors are close together instead of spread as wide as possible?
I cannot be obviously sure, but I suspect they put them together to keep them at similar temperatures so current sharing will be better. It is helping with thermal runaway.

Very nice video by the way. Impressive insight into construction and repair process.

[oz2cpu]

thanks a lot CosteC for your comment
Yes i am aware of the legal limit outer case temperature for CE approval,
I just feel if they lovered the series regulator voltage 10V to 5V (200W to 100W loast at 20A)
it is realistic to be under 50C on that large outer alu frame, the thermal connection could have been at least made to transfer halve of it that way,
the Fets use a very high value source resistor, compared what we are used to from NPN solutions,
the reason is fets need more feed back voltage change, in my unit all resistors looked like same temperature,
I think it was designed for worst case spread of VGS, this means you can replace any of the fets, and it will still share correctly,
this also means we could have improved the thermal by spreading the fets,

the funny think about the transformer : one of my former co workers, contacted me, i know they got at least two similar units in their factory,
i think he will now go check if they also got the same idle power loss, they also got thermal cam to check this fast and easy.

[CosteC]

I think most of your observations are true and there can be done a lot to improve this PSU.
Oryginal manufacturer either could not do it, or designed with really large safety margins.

High voltage drop may be for example dictated by need to survive momentary voltage drop on mains cycle without disturbing output. This is far easier done with SMPS than with linear regulators.
Or mabye just to compensate for AC line variation (207-253 V for 230 VAC nominal) - or maybe pre-regulator can take care of it? Not sure. Anyway high current linear regulator is technical challenge.

[oz2cpu]

the preregulator is a cycle pr cycle regulator and it deliver a quite nice stable voltage with low ripple,
how ever still too much ripple to use directly, so that is why the linear end stage regulator is needed,
your point about mains drop out, must be the correct reason why this high voltage is wanted in combination will that large capacitor bank,
it is the only way to solve drop out times, and good drop out performance was clearly a parameter we purchased our products for.

line variations is normally a slow thing, and will be solved by the preregulator
I did not store or save all the curves, but i remember it took a few cycles to handle its regulation at high load,
i think that is a wanted side effect, this way it is auto protected for inrush,
it is a little bit more complex than the eye saw at first glance :-)

[madires]

That's why you should get screw insulators. Small ones for mounting power transistors on heatsinks or larger ones for transformers.

[oz2cpu]

follow up-- I just picked up one more Power Supply from the same company
a 650W type, with TWO transformators in it..

exactly same issues found and solved.

[oz2cpu]

i made a new video, about this

Elektro Automatik EA PS 7065-100 650W Power Supply

https://youtu.be/eBPt8bqEL90