Mains isolation

[kellys_eye]

I'm preparing to construct an isolating, variable AC supply unit for testing switched mode power supplies and other equipment rated up to 200W.

The basic design uses a variac powering a 240:28V (10A) transformer which itself feeds another transformer (wired in reverse) to step the supply back to 240V but add the isolation that is required.

Whilst I can source another 240:28V transformer to do the 'up-converting' I already have an ex-UPS transformer rated at 240:24V that I'm considering using.

This would, of course, mean a final AC output that varies to GREATER than 240V but since the output is fully metered (V and I, with variable current tripping) this won't pose a problem and could also allow testing to the actual input limits of many SMPS's devices which are in excess of 240V anyway.

Question is, will the ex-UPS transformer be ok to use with its secondary not 'matching' the secondary of the first (240:28) transformer? Or should I go with the 240:28V version as the second one?

This isolation device is strictly for intermittent use only (10-15 minutes at a time)

[G6Tanuki]

It should work OK: indeed the slight mismatch could be just what you need to make up for the inevitable lossiness of the double-transformation.

[A 240-28 transformer followed by a 28-240 transformer will not give 240V at the output!]

I've used ex-UPS transformers in the past but only in the 'forward' direction [mains to lower-voltage - that's the way my current 50V-at-4A linear-amp power supply works] and though *in theory* transformers are symmetrical you may find a UPS transformer embodies some strange 'tricks' because it needs to be able to run in the reverse direction when the UPS batteries are providing the power. I've seen some where the secondary is designed to be resonant with a big capacitor in order to make the output closer-to-a-sinewave when the invertor is running.

[GrimJosef]

I don't know the details of the testing you're planning to carry out, obviously, but if tight control of your isolator's AC output voltage is required then you'll have to deal somehow with the fact that the UK's mains voltage can wander around pretty badly on quite short timescales (a few minutes or even tens of seconds).

The transformer losses will also mean that the output voltage will be load-dependent. If you're running your SMPS into a constant load then this shouldn't be a problem. But if the load will be varying then so will the voltage that is powering the SMPS.

For both these reasons you may find that you're spending a lot of your testing time twiddling the Variac to try to keep the supply voltage constant.

If resources allow, the neatest solution to these problems is to use a UPS to do your testing. Its voltage will be much more constant than the mains and if it's a decent sine-wave one then its feedback control system will also keep that voltage constant if the load varies. The output waveform is also much less mangled than the mains one. I bought a second-hand 1500VA unit from eBay for about £35 if I remember rightly, added some extra FETs (the pcb was pre-drilled for them), beefed up the cooling and connected a couple of high-capacity 12V leisure batteries and it will now deliver more than 1kVA of clean, stable, isolated mains for as long as the battery charge holds up.

Cheers,

GJ

AN advantage of 240/28 to 24/240 is that you can earth the 28/24 bit and ensure the output will not go to mains.

[Ambientnoise]

Will putting more than 28v into a 24v winding cause saturation ? Isn't it equivalent to putting over 280v into a 240v winding ? I lack the knowledge but hopefully one of our transfomer experts will comment.

Ken

[kellys_eye]

Originally Posted by GrimJosef

I don't know the details of the testing you're planning to carry out,

In the main it will be simply powering up from 'zero' and noting the current consumption as the voltage increases - damage prevention/limitation - and measuring as the various aspects of the SMP as supplies kick into action. Isolation is a valuable 'extra' for the obvious safety reasons.

TBH I haven't noticed any deviations of the mains supplies that might cause concern and testing at full load would only be momentary anyway - once I'm happy the SMP is working correctly and over a decent range of input voltages the equipment would be returned to 'normal' AC supplies for soak testing.

[G6Tanuki]

Be aware that plenty of SMPSes handle low-input-voltage (and restricted ability-to-source-current-from-the-supply) rather badly!

While the idea of using variacs and lamp-limiters is fine for legacy power-supplies using big-iron transformers and rectifiers, 'ramping up' the voltage to a SMPS can cause it to transition through deeply-odd states and go into some potentially-destructive inefficient oscillation modes.

[Skywave]

Your two 240/28v. transformers are providing the required mains isolation. But would it not be preferable to have that isolation between the incoming mains and the variac?

Al.

[Skywave]

Originally Posted by G6Tanuki

While the idea of using variacs and lamp-limiters is fine for legacy power-supplies using big-iron transformers and rectifiers, 'ramping up' the voltage to a SMPS can cause it to transition through deeply-odd states and go into some potentially-destructive inefficient oscillation modes.

Really? I've been using that technique of slowly 'winding up' the input a.c. voltage for the testing and repair of SMPSUs for years. Never found the problem that you have described yet.

Al.

[G6Tanuki]

Really? I've been using that technique of slowly 'winding up' the input a.c. voltage for the testing and repair of SMPSUs for years. Never found the problem that you have described yet.

Some Sony TVs/monitors in particular are prone to it. It's not so much of an issue with 'universal' SMPS that take anything from 90-260V as their supply but the type that have a 120/240V select switch can find running on low supply-voltage a problem. The SMPS osc can end up running at a frequency at which it's really inefficient meaning excess dissipation in the PSU.

[evingar]

Originally Posted by Skywave

Really? I've been using that technique of slowly 'winding up' the input a.c. voltage for the testing and repair of SMPSUs for years. Never found the problem that you have described yet.

Indeed !

I do a great deal of commercial SMPSU repair work and I have a set-up very similar to the one proposed (a commercial unit comprising an isolating transformer feeding a Variac).

While I accept there are specific SMPSU's (especially older ones) that may be problematical, the vast majority of stand alone SMPSU are fine with being run up slowly on a variac. When you think about the variability of "real world" Line voltage it would be a pretty poor show if they weren’t! Certainly anything designed with a decent controller chip (e.g. the Unitrode series) will have under voltage protection built in.

I have voltage and current meters on my set-up and I can categorically state it has saved me one hell of a lot of silicon and time over the years.

[Radio Wrangler]

Any decent SMPS has undervoltage lockout, meaning that it just sits there and looks at you until you give it enough volts to start up (at least without creating any new problems)

David

[kellys_eye]

Originally Posted by Skywave

Your two 240/28v. transformers are providing the required mains isolation. But would it not be preferable to have that isolation between the incoming mains and the variac?

I don't see what difference that would make electrically? My method is based on the current availability of parts-at-hand, 300W isolation transformers (240:240) being relatively difficult to source. (anyone?).

A secondary advantage is to use a c/o switch to take-off the first transformers secondary and deliver a variable 0-28V AC (@10A) - not at the same time of course - which may prove useful in an experimental sense. Adding a bridge rectifier and smoothing (10,000uF) and I get yet another variable (unregulated) supply.

Overall a pretty versatile solution to my isolation needs!

I'll post the finished device for further 'interrogation' after completion

[julie_m]

I doubt that you will saturate a 24:240V transformer with 28V at the input, unless it was already marginal. It most probably will just cancel out the inevitable voltage drop due to transformer winding resistances.

Even if saturation does occur with the variac turned up full, you can probably devise some sort of mechanical rotation limiter to prevent this.

Also, see Big Clive's YouTube Channel for some nice-looking panel meters that look as though they would be good to use on a variac.

[kellys_eye]

Originally Posted by julie_m

Even if saturation does occur with the variac turned up full, you can probably devise some sort of mechanical rotation limiter to prevent this.

Good idea - thanks! I'll probably implement that anyway as I'm already making a 'zero' interlock to ensure the variac is set to zero before enabling the hi-voltage output side.

[broadgage]

It would be preferable in my view, and simpler, to obtain a 240 volt to 240 volt isolating transformer.
These turn up fairly regularly on ebay. The ones intended for use of portable 240 volt tools whilst eliminating the risk of a dangerous shock to earth.
They are usually in a blue moulded GRP case, very similar in style to the well known 110 volt transformers, but blue rather than yellow.

Searching ebay or other sites for these can be a bit irksome as many vendors describe them wrongly.

[Skywave]

Originally Posted by Skywave

Your two 240/28v. transformers are providing the required mains isolation. But would it not be preferable to have that isolation between the incoming mains and the variac?

Originally Posted by kellys_eye

I don't see what difference that would make electrically.

'Electrically' it makes little difference. I was thinking primarily about electrical safety.
Plus, if in terms of V-A, you have a Variac that has a greater V-A capacity than the two transformer combo, (which sounds likely here), you could over-load the transformers at a high V-A final O/P when the Variac is set for max. volts out. Connected the 'other way around', it's the transformers that 'call the shots' in terms of max. V-A availability. However, having said all that, I think that Broadgage's suggestion of obtaining a 240 v. / 240 v. isol. transformer (of adequate V-A) is the best solution.
Finally, don't overlook the essential requirement of adding fuses: more than one will be required!

Al.

[Skywave]

Originally Posted by Skywave

Really? I've been using that technique of slowly 'winding up' the input a.c. voltage for the testing and repair of SMPSUs for years. Never found the problem that you have described yet.

Originally Posted by evingar

While I accept there are specific SMPSU's (especially older ones) that may be problematical, the vast majority of stand alone SMPSU are fine with being run up slowly on a variac.

Well, maybe. Way back in the early 1980's, I was regularly repairing SMPSUs using the 'Isol. xformer-to-Variac' for a large range of them from many different manufacturers - and without encountering any problems at all by using that technique. And some of those PSUs were really quite 'minimum cost' designs, too.

Al.

[MrBungle]

I wouldn't ramp up voltage on a variac to an SMPS myself. If it's loaded at all, it'll still draw the same amount of *power* from the source. To compensate for the reduced voltage initially, it'll draw more current to provide for the load. More current = more stress on the input side of things = more chance of it blowing up in your face. This actually happened to me I will add as the NTCs left the building running on a lower voltage DC input (yes they mostly run fine on DC too).

Myself, I use just an isolation transformer but don't assume that the isolation transformer is actually going to save you. If you end up being part of the primary circuit, you'll still kill yourself. Same rules as mains safety apply. All it takes is connecting a scope ground clip to the wrong bit of the supply, probing something and then you've got a ground loop as good as live mains through the scope. Differential scope probes are a good bet but expensive.

Oh and proper BS rated safety glasses.

Older SMPS, particularly from the 1980s and particularly Tektronix ones, seem to be complete dicks at low voltages and will oscillate on startup and make some horrible noises. I think that's where that lore comes from. Haven't seen that on the few little Meanwell ones I've been playing with recently. I hope that problem has gone away

Oh and the fortunately low voltage derived supply on my Tektronix 2225 is horridly unstable if you run it under voltage.

Got a dead Tek 2465 here for power supply repair at the moment. Should be fun seeing if that still does it

Lamp limiters is a no go unless you want some disco lights!

[wd40addict]

Back in the early 90s the place I worked bought a Tek Digital storage scope. Not long after its warranty expired so did the SMPS. Tek UK reluctantly fixed it. A year later it failed again and this time they just offered us a very low cost swap for one of the later model scopes! I don't think they could actually fix them and presumably there was no stock of replacements due to demand.

Similar story on my HP8648C the SMPS on these seem pretty unreliable, probably due to not enough load on the 5V line compared to the +/-15v. I replaced it with a Meanwell and added a step up module to generate the +38v from the 5v.