DC - DC / efficiency

[dark]

Hi,

I need advice on what topology to adopt for 40 K W (forty kilo watt) Ac - Dc converter . Input from 120 V AC to 420 V AC , and output about 140 v dc . What topology is adopted in order to get 95 percent or more efficiency.


Thanks

 

[ronv]

Probably a full bridge forward converter. But what oh what are you doing?

 

[dr pepper]

Maybe a supply for an industrial servo motor?
You'd be able to get variable voltage if you used scr's as the bridge, you can get single and 3 phase isotop modules made for the purpose.

 

[ronsimpson]

Probably a full bridge forward converter. But what oh what are you doing?

I agree, look at quasi resonant or resonant.
40kW....You should look at power-factor correction.
Vin=120 to 420? That is a big range. 3.5:1 That will hurt you. The silicon (MOSFETs) must handle the 420V and the 400A. If the supply was 420 only then the MOSFETs will see 420V and 100A.

 

[dark]

It is a windmill generator three phase o/p ( freq is varying 50 to 2K) , could you advice more in how to go ahead with power factor correction. Thanks

 

[ronsimpson]

I don't know if you need PF correction if you are getting power from a windmill. The power company likes PF correction.

What is power factor? Some loads are a resistor. (heating element) The current and voltage waveform are the same shape. Some loads look some what like a inductor or capacitor where there is phase shift. (the current is shifted from the voltage) Some loads are like a computer power supply where the current is very non linear. (not sign wave) The power company hates these. The wire loss in the power line is high.

I have done PFC at 50/60hz. I think it should work 50 to 2k but have never thought about it. If the copper loss in the generator and short wires are low then don't add the extra effort.

edit:---------------------------
Now that I know what you are working on; probably low power is (120V, 50hz, 20A) and high power is (420V, 2khz, 100A) input. In those cases the output is 140V 17A to 140V 280A.

1) If you output voltage is above the input voltage the problem is more complicated! Using a simple buck PWM you can only reduce the voltage. If the input was 130 to 420 and the output was 129 a simple buck will work. -------hold on------- The 120VAC is probably rms or average so its peak is 160. We can buck down 160 pk (120rms) to 140 dc just find.

 

[johansen]

It sounds like you have a permanent magnet alternator.. field weakening or no?

I would look at just using a three phase diode stack since the output of the alternator is probably trapezoidal anyway, and build a big multiphase buck converter.
285 amp output.. i would look at 9 or so phases.. each configured in constant current mode.
you'll need to interleave the phases
you can get 95% efficiency with IGBTs, but not at 120vac input if you include the diode losses.

 

[ronv]

What are you going to do with the 140 volts DC?

 

[ronsimpson]

you can get 95% efficiency with IGBTs, but not at 120vac input if you include the diode losses.

I think you are in that voltage area where IGBTs and MOSFETs will have about the same loss. It will be very close.

multiphase buck converter

I was thinking one buck. With 3 bucks the current is down to 100A. With 9 bucks the current is down to 30A. At 30A I have the parts here to do that. Little modules have a advantage in that if one fails the rest will do the work until it can be replaced.

 

[dark]

It sounds like you have a permanent magnet alternator.. field weakening or no?

I would look at just using a three phase diode stack since the output of the alternator is probably trapezoidal anyway, and build a big multiphase buck converter.
285 amp output.. i would look at 9 or so phases.. each configured in constant current mode.
you'll need to interleave the phases
you can get 95% efficiency with IGBTs, but not at 120vac input if you include the diode losses.

Hi,
I am not using field weakning, its a permanent magnet alternator windmill . I have googled a lot on PFC , but nothing meaningful comes out is it required ? . Even if we adopt it nothing is stable from frequency to voltage !!. I am not sure if not using a PFC would compromize efficiency please advice , can this be omitted as the system is already very complex.

Thanks

 

[johansen]

Power factor correction will reduce copper losses in the alternator, and to some extent iron losses as well.
but at 2Khz, it will be difficult to get power factor correction, because you have such a wide range for the input frequency.

i see no reason to use power factor correction. some improvement might be made with a choke after the three phase diode bridge.

to make the system more complicated, you might go with this topology:
**broken link removed**
note that you can add phases as well with this design.
and the mosfets can be paralleled with IGBTs

for 40KW i would still look at at least 4 phases. unless you want to go with CCM for the inductors.
to operate the inductors in discontinuous mode (high frequency) you're going to need more phases because the current ripple on those three caps (the ones connected in delta) are going to be insane.
furthermore, because the alternator puts out up to 2KHz, those capacitors have to be relatively small, otherwise they are going to present a very non negligible reactive load on the alternator.
so you might have to go with more than 4 phases so those caps can be reduced in size.
this isn't a bad thing, because the more phases, the smaller and cheaper the capacitor just after the 3 phase diode bridge needs to be.

do you need constant power over the full range 120vac to 420vac?
or does the power increase with the voltage?

420vac has to be boosted to above 600 volts to maintain discontinuous current in the boost inductors.
600 volts could be stepped down to 140vdc with a fixed ratio transformer running at a fixed 50% duty cycle. the 600 vdc could be bucked down to 140vdc more efficiently, but then you lose voltage isolation. (do you need it?)

unfortunately, shorting three phases together causes the dc output of the rectifier to have a high frequency ripple with relation to the generator.
with multiple phases this would be reduced but it won't be eliminated completely.


to keep this system as simple as possible just run the alternator into a three phase bridge, an inductor and capacitor sufficient to handle the ripple current of the multiphase buck converter and buck it down to 140 volts dc. i would run it in forced synchronous mode to keep the feedback loop simple.

 

[ronsimpson]

Input from 120 V AC to 420 V AC , and output about 140 v dc .

At first there was no mention of alternator windmill so I assumed the 120V AC to 420V AC came from the power line where PFC makes a difference.
Now that I understand the alternator; forget the PFC.

 

[ben7]

OMG dude that is a big project to take on!

Especially getting it to work at 40kW!

I do think you should use 9 or 10 phases.

Start with a 3-phase 300-amp bridge rectifier, and have some filter capacitors, say, rated at 600 volts and 10,000uF. You will need to limit the inrush current from the capacitors with a soft start circuit! You should also have a circuit breaker on the input. How many amps is the short circuit current of the wind generator?

Then make ten 5kW buck converters. The idea with using ten 5kW bucks is that if one or two fail, the rest can still handle the load. Also it is good to keep them from running all the time at full capacity, for the sake of them having less stress on the components. To get high efficiency, you will need to use a synchronous buck circuit. Here is an example: **broken link removed** - it uses two switch devices instead of one switch and a diode. The advantage is that the switches have less loss than the diode.

I agree with all that johansen said.