Two small transformers versus One large transformer

Thread Starter

SiCEngineer

Joined May 22, 2019
442
Hello,

I am designing a power supply for increased power density in a large step-up, high voltage application that requires isolation.

Since the transformer is typically the largest and heaviest part, and high voltage transformers bring with it increased considerations of transformer parasitics (namely capacitance) - I am looking at ways to reduce transformer size and improve power density. In my application, I have multiple secondary outputs that are stacked together by connecting in series the capacitors after a voltage multiplier. This is one way to reduce the ratings of the secondary components and the transformer. However, I am looking at more creative ways also.

I have recently seen some topologies that use two smaller isolation transformers rather than one large one. My question is, what exactly is the benefit of this? The input current will be shared between the two transformers, reducing their VA rating (and possibly size?), but the voltages seen by the transformer do not change - therefore I am unsure if it brings benefits to me. Is there other benefits to using multiple, smaller transformers than the one I have outlined for purposes of increasing power density, size, weight etc?

My idea was instead of using one transformer with, say, 8 sec. windings connected in series, to use 2 transformers each with 4 sec. windings connected in series to supply the load. I'm a bit of a newbie to transformer design but I am looking at creative ways to improve the design!

Thanks in advance.
 

Thread Starter

SiCEngineer

Joined May 22, 2019
442
Using two transformers is not going to reduce the footprint or weight of the power supply.
Max.
Hi Max,

Thanks for your reply. If so, then what is the benefit realised by implementing two transformers? Why for example may someone choose to use two half bridges each feeding a transformer as opposed to a full bridge feeding a single transformer?

Thanks
 

Thread Starter

SiCEngineer

Joined May 22, 2019
442
What are the voltage and current requirements, input and output?
What efficiency level are you trying to achieve?
What is your level of power supply design experience?
Hello Charles,

The input voltage varies between 240-300V, and the outputs are -3kVDC and -6kVDC (DC/DC converter), whereas the output currents are in the mA range. I would be looking at efficiencies above 85% ideally. I have only about 8 months of power supply design have plenty of time to work on the project.
 

Janis59

Joined Aug 21, 2017
1,834
RE: Max. Generally yes, but not always. Exactly with similar idea as Your`s in mind I made a just deadly mistake year ago. It was one 120 kW 36 kV SMPS where I designed and installed for all H-bridge and logics feeding just most ordinary old-good classical trafo. Say straight, rather impressive in size, as the total power was near to 1/2 kW, and the count of different floating windings against gnd 5V, 12 V, 19V and so and so was strong over dozen. After the mounting was ended and first couple of times the igbt was beaten out, correspondingly all the Gretz diodes was need to check even at this small servo-PS, became clear that dozen small trafos would be far more handy, nicer, less proned to beat out the diodes, more handy for screw off and screw on, etc etc etc. At last that heavyweight machinery fell off the table some may be 40 cm high, and guess what was the major damage - my trafo was crushed the screw-places and cut the wires Thus, next time for sure I shall make a number of smaller trafos.
 

Janis59

Joined Aug 21, 2017
1,834
RE:SicEng
If no much more than 10-20 mA, my advice is to forget the HV trafo. Use ordinary isoleringtransformer in entrance and after organize the Cocroft Walton circuit. If size and weight is not critical, nothing more is needed. Note, that C-W circuit allows to exploit the Graetz style diode bricks. That is way to reduce the number of needed solderings about 4-fold, thus the time of knitting around. For such construction, yet, capacitors will be in range of 10...20...40 uF. If any size and weight is critical, then use a Royer circuit to chip the grid, and chipped frequency lead to small ferrite isolation transformer after what stays tiny high frequency Cocroft-Walton circuit. Then capacitors will be 1n...10n...100n range however diodes will be demandy because of higher speed. At least the 1-cent 1n4007 will not be applicable for sure.
Why to listen me? If any trafo already stays there why to not wind the HV coil straight on it? Because, if any trafo have more than 1 kV the isolation becomes serious, but if it cares until 3 kV the insulation is the main thing of all construction, but if there are even more - like some 12-50-120 kV, whole the construction design is centered to this damn detail. Still after years and years I am not able to buy ANYWHERE on the planet a 5 meter long cable piece for our lasers repair. Just 10 identical lasers are resting in purgatory shelves, because all have identical defect - 50 kV trafo have beaten-through this coaxial cable insulation and all power is going be lost there. Last year I fould one American company what wanted a just nightmare-like 250 USD per meter of such cabe, and we was willing (surprize) for it, but then they started to ask question what military authorization of USA we have, so the deal was cancelled by existing rules. And our lasers still stand. Let never make a transformers over 2...3 kV except that few-turns RF type. Or at least up to 20-30 kV is rather effective measure to fill in the epoxides by vacuum. Technology is rather prost, take the plastic sack from the nearest food store, place the transformer in, hermetize by scotch (I mean tape not a drink), and cut two small holes. One hole hermetize by the same tape for the pipe leading to forevacuum pump, and other pipe immerse in the tin with already mixed epoxydes. As less coloured are those resin, as better the HV security.
 
Last edited:

Thread Starter

SiCEngineer

Joined May 22, 2019
442
The currents drawn by the load are in the range of *hundreds of mA, not tens of mA - my apologies for the misunderstanding there. Size and weight are some of the main design criteria and isolation is pretty much essential, I have looked at different ways to achieve isolation but none really are applicable here. I plan on using multipliers but only through connections to the transformer. I don't think much can be done in topology to improve upon previous designs but some modifications to the transformer may prove fruitful in improving the solution density.

At least one benefit of multiple smaller transformers then, is less possibility of having one fall on your head :)!

But yes, I don't think removing the HV transformer is a practical consideration in this case.

Thanks for your input anyway!
 

MrAl

Joined Jun 17, 2014
11,389
Hello,

I am designing a power supply for increased power density in a large step-up, high voltage application that requires isolation.

Since the transformer is typically the largest and heaviest part, and high voltage transformers bring with it increased considerations of transformer parasitics (namely capacitance) - I am looking at ways to reduce transformer size and improve power density. In my application, I have multiple secondary outputs that are stacked together by connecting in series the capacitors after a voltage multiplier. This is one way to reduce the ratings of the secondary components and the transformer. However, I am looking at more creative ways also.

I have recently seen some topologies that use two smaller isolation transformers rather than one large one. My question is, what exactly is the benefit of this? The input current will be shared between the two transformers, reducing their VA rating (and possibly size?), but the voltages seen by the transformer do not change - therefore I am unsure if it brings benefits to me. Is there other benefits to using multiple, smaller transformers than the one I have outlined for purposes of increasing power density, size, weight etc?

My idea was instead of using one transformer with, say, 8 sec. windings connected in series, to use 2 transformers each with 4 sec. windings connected in series to supply the load. I'm a bit of a newbie to transformer design but I am looking at creative ways to improve the design!

Thanks in advance.

Hello,

In theory the width of the stack could be halved if the voltage was halved. That would be for two transformers with primaries in series. The secondaries would probably also be connected in series.
The window area could be reduced by about 80 percent if the current was halved. That would mean the primaries connect in parallel.
 

crutschow

Joined Mar 14, 2008
34,280
I would think that, all else being equal, two transformers with 1/2 the equivalent VA rating of a single transformer would cost more, weigh more, and have less efficiency.
 

MrAl

Joined Jun 17, 2014
11,389
I would think that, all else being equal, two transformers with 1/2 the equivalent VA rating of a single transformer would cost more, weigh more, and have less efficiency.
Hi,

Ok well then let's start with a single transformer with 100 turns on the primary and a 1 inch stack.
Now remove all the turns, then separate the stack into two cores each with a 1/2 inch stack.
Now wind 50 turns on one core, the other 50 turns on the other core.
Now connect the two primaries in series.
We havent added anything to the mix so the weight is the same.
Now for the secondary do the same.
The total weight must still be the same because we havent added anything.
Flux density the same also.
 

crutschow

Joined Mar 14, 2008
34,280
Hi,

Ok well then let's start with a single transformer with 100 turns on the primary and a 1 inch stack.
Now remove all the turns, then separate the stack into two cores each with a 1/2 inch stack.
Now wind 50 turns on one core, the other 50 turns on the other core.
Now connect the two primaries in series.
We havent added anything to the mix so the weight is the same.
Now for the secondary do the same.
The total weight must still be the same because we havent added anything.
Flux density the same also.
That's a hypothetical where you are assuming that the two split units will have the same characteristics as the single transformer.
That doesn't prove they would be equal in the real world where winding area, core area, etc. do not scale linearly but as the square or cube of the dimensions.
 

MrAl

Joined Jun 17, 2014
11,389
That's a hypothetical where you are assuming that the two split units will have the same characteristics as the single transformer.
That doesn't prove they would be equal in the real world where winding area, core area, etc. do not scale linearly but as the square or cube of the dimensions.
Hello,

Not sure what you are trying to say because everything i mentioned scales linearly.
When you halve the number of turns, you can halve the stack, and the flux density stays the same. The number of turns vs area scales linearly.
For example, if you have 100 primary turns on a core and the flux density is 5000G, if you halve the number of turns to 50 and halve the core stack the flux density is still 5000G.

It was always hypothetical, this is theoretical and not a bad assumption because the flux density stays the same.
 

oz93666

Joined Sep 7, 2010
739
Two smaller transformers has to improve efficiency since removing heat is one of the main issues , two smaller ones spaced apart are more effective at dissipating heat to the environment.

four or eight would be even better , assuming no other issues negate the benefits
 

Thread Starter

SiCEngineer

Joined May 22, 2019
442
Two smaller transformers has to improve efficiency since removing heat is one of the main issues , two smaller ones spaced apart are more effective at dissipating heat to the environment.

four or eight would be even better , assuming no other issues negate the benefits
That’s interesting, and what I thought might be the case. My application kind of places emphasis on power density, so making the overall solution as small as possible. My logic was also that if splitting one transformer into two smaller ones could improve the dissipation of heat, then the size of the heat dissipating equipment would reduce - so even if two transformers took up More board space, the overall size might see a decrease.
 

BobTPH

Joined Jun 5, 2013
8,805
Hi,

Ok well then let's start with a single transformer with 100 turns on the primary and a 1 inch stack.
Now remove all the turns, then separate the stack into two cores each with a 1/2 inch stack.
Now wind 50 turns on one core, the other 50 turns on the other core.
Now connect the two primaries in series.
We havent added anything to the mix so the weight is the same.
Now for the secondary do the same.
The total weight must still be the same because we havent added anything.
Flux density the same also.
The two cores will have a lower saturation current and must carry the same current as the original. Unless the saturation current of the orignial was larger than needed by a factor of 2, you lose.

Bob
 

Thread Starter

SiCEngineer

Joined May 22, 2019
442
The two cores will have a lower saturation current and must carry the same current as the original. Unless the saturation current of the orignial was larger than needed by a factor of 2, you lose.

Bob
Hello Bob,

That’s true, didn’t think of that. Although the same current will flow through the primaries of the two transformers, does the halving of the voltage not have an effect? What about the series connection of the secondaries, or parallel connection or the primaries?
 

Thread Starter

SiCEngineer

Joined May 22, 2019
442
It is only current and number of turns that determine the field.

Bob
I understand, but what I mean is field aside, are there no benefits in this type of application where multiple transformers in a certain configuration? Such as parallel input series output, or series input/series output? Or would it just be easier and more effective to just implement one single transformer rated for the entire VA?
 
Top