Single versus three-phase transformers for power density

Thread Starter

SiCEngineer

Joined May 22, 2019
442
I have currently designed a powerr converter, with a single phase transformer which operates at >200kHz. The design goal was to create a converter that increases power density, by minimzing the size of the passive components, namely the transformer and the output storage capacitors, which are pulsed (so are typically oversized for this purpose). The topology is an LCC converter, frequency modulation.

My question is, in terms of power density (transformer size, transformer weight, etc). Would you recommend I investigate the possiblity of using three-phase primary windings? I have read that using three-phase primary windings can reduce input ripple current, and output capacitor ripple current. I am thinking of a three-phase primary, single-phase secondary with voltage multipliers.

Is there any benefit to this transformer topology compared to using single-phase primary windings? May I realise some benefits, or with the extra phases always add additional size and weight to the transformer?

Although I have experience with power electronics I have not really investigated the benefits of single vs. three phase windings!

Simple spec is: 270VDC in, 5kV, 2.5kV out with 12mA, 180mA respecitvely - so a high voltage, high step-up application - if that helps.

Any suggestions welcomed.
 

MisterBill2

Joined Jan 23, 2018
18,167
With such a low power level 3-phase does not make much sense. An article about the development of a higher power inverter supply did mention a number of ways to improve it, the first being run it at 60KHz, the second being use a number of secondaries, each with it's own rectifiers, which avoids the need for long diode strings, and leads to reduced leakage inductance. Also, the transformer in that design used much more core than the minimum possible core volume found by using some formula. The greatly reduced leakage inductance allowed a reduction in the number of turns required. That reduces the overall size of the supply. Also, using the very high speed GaAs transistors allowed the high power at the high frequency. So there you have a few suggestions. I hope that they help a bit.
 

Thread Starter

SiCEngineer

Joined May 22, 2019
442
With such a low power level 3-phase does not make much sense. An article about the development of a higher power inverter supply did mention a number of ways to improve it, the first being run it at 60KHz, the second being use a number of secondaries, each with it's own rectifiers, which avoids the need for long diode strings, and leads to reduced leakage inductance. Also, the transformer in that design used much more core than the minimum possible core volume found by using some formula. The greatly reduced leakage inductance allowed a reduction in the number of turns required. That reduces the overall size of the supply. Also, using the very high speed GaAs transistors allowed the high power at the high frequency. So there you have a few suggestions. I hope that they help a bit.
Yep! I currently have a number of secondaries each connected a multiplier circuit, then they are connected in series. Creates the high voltage nicely. I was mostly interested because an article I read noted that the input current ripple/output voltage ripple were dramatically reduced with a 3-phase transformer.

For the same power level, if a three-phase transformer would be used, would it be much larger than a single-phase counterpart, though? If we make everything else equal.

I was also interested because it might allow using fixed-frequency phase shift control on the primary side for regulation rather than using frequency modulation of the LCC, which brings some benefits for the application.
 

schmitt trigger

Joined Jul 12, 2010
868
By three phase, you mean interleaved converters?

Indeed, it does reduce the ripple current requirements.

There are some caveats though, the main is the load current balancing between the three phases. It is another control loop nested within the primary voltage regulation loop.
I was not personally involved with the design so I can't give you details, but an acquaintance of mine was, and they struggled with the loop compensation.
 

Thread Starter

SiCEngineer

Joined May 22, 2019
442
Yes, so having three of the same converter each connected to one of the three-phase windings of the transformer. I have indeed read what you mention in regards to the current balancing - I have yet to get that far yet, but I was just wondering what effect this would have in terms of DC link capacitance, transformer size, output pulsed capacitor / filter capacitor size in the secondary side?
 

MisterBill2

Joined Jan 23, 2018
18,167
It is more effective in reducing the ripple to raise the frequency. The article I mentioned used 60 kilohertz for a 3000 volt 1 amp supply and used a 0.1 MFD filter capacitor.
Three cores would be quite a bit larger, because core size does not scale, And more heatsinks, or a bigger heat sink, will also take up more space. Three inverter drive circuits will take up some more room than one, although the difference in absolute volume will not be very much.
And certainly full wave rectified unfiltered power has less ripple, , but for the current levels described the capacitor size would be the same, although the ratings could be different.
 

MaxHeadRoom

Joined Jul 18, 2013
28,617
My question is, in terms of power density (transformer size, transformer weight, etc). Would you recommend I investigate the possiblity of using three-phase primary windings? I have read that using three-phase primary windings can reduce input ripple current, and output capacitor ripple current. I am thinking of a three-phase primary, single-phase secondary with voltage multipliers.
How does that work? Do you mean three single phase secondaries?
Max.
 

Thread Starter

SiCEngineer

Joined May 22, 2019
442
It is more effective in reducing the ripple to raise the frequency. The article I mentioned used 60 kilohertz for a 3000 volt 1 amp supply and used a 0.1 MFD filter capacitor.
Three cores would be quite a bit larger, because core size does not scale, And more heatsinks, or a bigger heat sink, will also take up more space. Three inverter drive circuits will take up some more room than one, although the difference in absolute volume will not be very much.
And certainly full wave rectified unfiltered power has less ripple, , but for the current levels described the capacitor size would be the same, although the ratings could be different.
That is true, but for the high output voltages I am quite limited with how much I can increase the frequency. I think 300kHz is probably pushing it (?)
 

MisterBill2

Joined Jan 23, 2018
18,167
That is true, but for the high output voltages I am quite limited with how much I can increase the frequency. I think 300kHz is probably pushing it (?)
In both posts #2 and #6 I suggester 60KHz, only 20% of 300KHz. And even at 60KHz it is pushing things, which is why I suggested the fast transistors. With the much lower leakage-inductance transformer you will need the fast transistors in the inverter power stage. And you will only need small filter capacitors.
 
Top