Testing back to back power transformers for use as an isolation transformer

[intabits]

I want/need an isolation transformer, but they are not especially cheap. And as I have many other power transformers lying around this place, I should use two of those in a back to back configuration if it will provide a viable result. But I was concerned about voltage "sag" at the output under load, and temperature rise.

So I tested three handy pairs of transformers, using light bulbs as a dummy load, and with a variac to adjust the input voltage for a 240V output at the load. I was going to do power and efficiency calculations, but found the temperature rises to be quite small, and so just went with the "Input-voltage-for-240V-output" as the metric to compare them with.

Here are the transformer pairs used, with estimated VA ratings of 90VA, 150VA, and 2KVA, with weights of 2Kg, 4.5Kg and 22Kg.



Those very large ones served as a reference as what the very best I could come up with would be.

Here are the results of the testing:-



I have decided to make a general purpose isolation transformer box using the 150VA pair.

And I'll attach handles to the large ones, along with some connector arrangement to enable them to be quickly configured for any purpose, including as an isolation transformer, but still allowing them to be general purpose high power units.

Video:-

[Cliff Matthews]

Each trafo sags ~10% from NL to full load, so 2 in series doubles the problem. I'll try-out this cct one day..

[bob91343]

Many isolation transformers include a slight stepup to provide full output under load.  Having said that, your solution looks good to me.

Stay within ratings and you will have good results.

[David Hess]

I have a pair of 1500 watt transformers configured back to back and mounted to a 2x6 that work fine as an isolation transformer.  Since they support 120V/240V with split windings, the output side actually provides two 120 volt AC isolated outputs or isolated 240 volts AC if needed from a 120 volt AC line.

[intabits]

The secondary output of the big ones is 130V with 240V input, and the primary has 200V, 220V & 240V tappings, so I can also get 120V if I use the 220V input. (and I just found a possible need for that). (And having three of the beasts allows for all sorts of configurations)

[Electro Detective]

and the output waveform vs the input waveform with varying load levels from 10% to max on the DSO/CRO scope looks like...   

 

[intabits]

and the output waveform vs the input waveform with varying load levels from 10% to max on the DSO/CRO scope looks like...   

No idea. Never even considered looking at it.
Are you saying there is likleyhood of significant distortion? I can imagine that happening towards max load, but have no experience of it.
Assuming there may be some distortion, is that not also true of any isolation transformer?
Would this be a problem in typical applications of an isolation transformer?
You've got me curious now. The worktable is now otherwise occupied, but I'll have a look at it in the next week or two...

[soldar]

... they support 120V/240V with split windings, the output side actually provides two 120 volt AC isolated outputs or isolated 240 volts AC if needed from a 120 volt AC line.

I have a couple of similar transformers with two separate primary windings which con be configured in series for 230 V or in parallel for 115 V.

Let us call the windings of one transformer A and A' and those of the other transformer B and B'. I connect A and B in series to the 230 V mains and connect A' and B' in series and get 230 V isolated from the mains. It is more efficient than connecting the transformers back to back.

[intabits]

... It is more efficient than connecting the transformers back to back.

That's a good idea.
With the big ones, I could connect the 220V primaries in parallel to the mains, and their 120V secondaries in series.
Then I'd have the more efficient single level of isolation at around 4KVA!

[David Hess]

I have a couple of similar transformers with two separate primary windings which con be configured in series for 230 V or in parallel for 115 V.

Let us call the windings of one transformer A and A' and those of the other transformer B and B'. I connect A and B in series to the 230 V mains and connect A' and B' in series and get 230 V isolated from the mains. It is more efficient than connecting the transformers back to back.

I specifically did not do that because the isolation between primary windings is not as high as between primary and secondary windings.  With the current back-to-back configuration, I can also ground the secondaries for even more isolation.

[duak]

Series/parallel windings are wound bifilar, ie. two wires wound at the same time, to ensure they have the same number of turns and identical coupling inductance.  This means that you've only got two layers of magnet wire insulation for galvanic isolation.  Bifilar windings also have more coupling capacitance between the two windings than if the windings are separated.

As an aside, I encountered a toroidal power transformer where the primary winding count differed by one turn.  It worked fine when the windings were in series, but not in parallel because there was a circulating current flowing that would trip the line breaker.  Going back to the build sheet, I found that the designer did not specify bifilar winding.

[David Hess]

Series/parallel windings are wound bifilar, ie. two wires wound at the same time, to ensure they have the same number of turns and identical coupling inductance.  This means that you've only got two layers of magnet wire insulation for galvanic isolation.  Bifilar windings also have more coupling capacitance between the two windings than if the windings are separated.

The dual windings on the transformers I used are on separate layers; they are not bifilar wound.  But I still chose to wire them back to back for increased isolation.

[soldar]

With the big ones, I could connect the 220V primaries in parallel to the mains, and their 120V secondaries in series.

That doubles the power they can handle. Back to back they can only handle the power of the smallest one but in parallel they add up.

Be careful about the secondaries in parallel having the same voltage.  A volt or two will not really matter, especially under load, but if there is much difference you will get wasted heat.

I connect one end of the secondaries and measure voltage between the other two open ends. That way you also make sure they're in phase.

[soldar]

Series/parallel windings are wound bifilar, ie. two wires wound at the same time, to ensure they have the same number of turns and identical coupling inductance. 

Definitely not the ones I have. I guess mine were wound by someone who could count

As an aside, I encountered a toroidal power transformer where the primary winding count differed by one turn.  It worked fine when the windings were in series, but not in parallel because there was a circulating current flowing that would trip the line breaker.  Going back to the build sheet, I found that the designer did not specify bifilar winding.

I doubt that. One turn is not going to blow anything. If the transformer tripped the breaker when the secondaries were connected in parallel it was because they were connected incorrectly, not because there was a one turn difference.

[soldar]

Some years ago, for many months, my building was undergoing renovations and the electricity would we cut off randomly and without warning which was a pain when I lost my current work on the computer. Even worse, The work was still ongoing when I was leaving for some weeks travel in China and I wanted to remotely connect to my computer at home.

So, I talked to my neighbor in the next building and he had no problem letting me hook up to his electricity and we threw a cable from my window to his window. I could have just plugged my computer into his supply but that risked the hanging cable failing so I rigged something very simple.

For isolation I used those two transformers in series as described in my earlier post.  Then I put a diode bridge and connected it in parallel with the computer PSU bridge.

With my mains working most, almost all, of the power came from my side because of the resistance of the long cable to the neighbor.  When my power was cut off the computer wouldn't even notice and would continue to draw from the neighbor.

[malagas_on_fire]

Thats a neat redundant psu soldar.

I have two sets of back to back transformers but with 6VA and 12VA only, in back to back since they are reduction transformers 230 to 6V and the other are 230 to 12V . For testing triacs with Dimmable bulbs,  capacitive dropper PSU's with small loads they quite do the job.

[nctnico]

... they support 120V/240V with split windings, the output side actually provides two 120 volt AC isolated outputs or isolated 240 volts AC if needed from a 120 volt AC line.

I have a couple of similar transformers with two separate primary windings which con be configured in series for 230 V or in parallel for 115 V.

Let us call the windings of one transformer A and A' and those of the other transformer B and B'. I connect A and B in series to the 230 V mains and connect A' and B' in series and get 230 V isolated from the mains. It is more efficient than connecting the transformers back to back.

This doesn't sound safe to me because the primary windings won't have a safety isolation barrier at all. This is also true for any regular transformer when considering the primary / seconday windings unless it is rated as being double isolated.
Just don't mess with isolation transformers. Differential probes are the way to go because they much safer compared to floating the DUT and they aren't that expensive any more. A good isolation transformer probably costs more.

[soldar]

This doesn't sound safe to me because the primary windings won't have a safety isolation barrier at all. 

True but in this case I was not really looking for strong isolation as both sides were at mains voltage. I just needed some very basic isolation to prevent GPCI faults.

[bob91343]

I can see two reasons to isolate.  One is the obvious safety concern over fault currents.  The other is common sense regarding interconnecting test equipment.  The former is important in permanent installations but not so much for test setups.

I only use an isolation transformer when running tests on ac-dc radios and similar gear.  For all other purposes, it isn't important to my operation.

The only exception to this was when I was repairing one of my Boonton Q meters.  There is a constant voltage power transformer in it that failed, and I decided I didn't need regulation and so I bypassed it.  Mistake.  I sacrificed the isolation it provided.  I found out inadvertently when feeling a tingle while making some connections.  So I bit the bullet and repaired the power supply properly and no more trouble.

As with nearly everything in life, common sense is important.  Don't fix something that isn't broken, and don't reengineer something unless you really know what you are doing.

[duak]

Series/parallel windings are wound bifilar, ie. two wires wound at the same time, to ensure they have the same number of turns and identical coupling inductance. 

Definitely not the ones I have. I guess mine were wound by someone who could count

As an aside, I encountered a toroidal power transformer where the primary winding count differed by one turn.  It worked fine when the windings were in series, but not in parallel because there was a circulating current flowing that would trip the line breaker.  Going back to the build sheet, I found that the designer did not specify bifilar winding.

I doubt that. One turn is not going to blow anything. If the transformer tripped the breaker when the secondaries were connected in parallel it was because they were connected incorrectly, not because there was a one turn difference.

Soldar, the particular transformer I referred to was rated for about 2 KVA.  This was 30 years ago so I don't have the build sheet but going by memory, there were something like 80 turns for each 120 V primary winding so each each turn contributed about 1.5 V.  The breaker was a 2 pole rated for probably 8 amps and also served as the power switch.  The same breaker was used for both 120 and 240 V and was connected in series or parallel along with the transformer when the line voltage strapping was changed.   During final test of the overall unit, it would work fine when strapped for 240 V, but would trip after a short time (a minute?) when strapped for 120 V.  All voltages measured OK and the line current wasn't significantly higher than spec.  The resistance of each winding was a fraction of an ohm so the difference in the number of turns caused a circulating current in the primary windings when they were connected in parallel.  I think I used a clamp on ammeter on one of the primary windings and found an excessive current and further tests showed the primary windings had a different number of turns.

I would have thought that the manufacturer would have known how to count accurately too, but I guess by specifying the exact number makes them actually wind them and then test afterwards.  If memory serves, this transformer was from a batch of 10 or 20 and was the only one that had this problem so I chalked it up to a spec, fab and QC issue.

[intabits]

With the big ones, I could connect the 220V primaries in parallel to the mains, and their 120V secondaries in series.

That doubles the power they can handle. Back to back they can only handle the power of the smallest one but in parallel they add up.

Be careful about the secondaries in parallel having the same voltage.  A volt or two will not really matter, especially under load, but if there is much difference you will get wasted heat.

I connect one end of the secondaries and measure voltage between the other two open ends. That way you also make sure they're in phase.

I was actually talking about the secondaries in series (each 120V) to get isolated 240V.
But should I ever need 120V at high power, I would be checking as you described.

[Psi]

If you can find a dead UPS (which is usually very easy for a few $ or free) they are a good source of a single transformer with two mains windings.
(They have Mains in,  Mains out and Battery windings).