Electronics > Power/Renewable Energy/EV's

Active Power Factor Correction Gods - a challenge.

(1/2) > >>



Some background, please skip if it bores you!...

I don't even know if this is possible, but here goes. With an inductive generator (a wind turbine) at a given frequency one can phase correct using capacitance and present a varying load which will force the input voltage to change, offering maximum power point tracking potential.

The problem is that the generator (fixed current, single phase, voltage related to frequency) changes speed and so the capacitance is no longer correct. Switching capacitance in/out on zero-cross is possible but not ideal.


To implement APFC for a low voltage system, ideally without using a chip as the input and output parameters are known.

Imagine there's a large li-ion battery at the output side, currently 3.5V, so you know the voltage and it's slow to charge. The generator roughly generates Vac=Freq/1.5. Frequency varies but voltage is clamped by triac at 30V AC. 6V AC is common.

My attempts revolved around using a boost to correct the phase and then a buck, which knowing the output voltage and if suitable inductance is used one can assume CCM and derive a duty that would present an input voltage. At this point I become confused as to how PFC corrected boosted voltage could be utilized correctly to ensure power factor is correct. It was time to call in the experts!

I've created a very basic LTSpice simulation with a blank space in the middle. Is it possible to solutionise this problem?


Li-ion battery @ initial 3.5V.
Single phase generator, 0-255Hz, fixed current of approx 0.5A, voltage = 0-30V AC (overspeed voltage capped when no load). V = F/1.5 (no load).

Hmm.  Is there something particular about this wind turbine that we should know / might be useful..?

For example, commercial ones are variable pitch, so the rotor RPM is a free variable with respect to wind speed, within reason.  Perhaps this costs some efficiency, I don't know.  A gearbox (transmission) can also be used to set rotor and generator RPM independently.  And a variable field generator (alternator type) can adjust phase without capacitors or inductors.

I guess this is a much smaller scale question, given the single-cell battery.

You can buck-boost in two stages combined, or SEPIC or etc. to get a free ratio of input to output.

Will MPPT actually be important with respect to the battery's capacity (charging curve)?  Is the turbine strictly less powerful than the battery, or will they overlap?  (Trick question, really: charging can't go on forever, so something has to either stop the power flow, or dump the excess.)


There will be other smarts for stopping charging etc, the challenge isn't to produce a product but to solve the PF for an inductive source.

Here's my work so far. I haven't been able to shift phase using a boost. It could be that the boost characteristics are completely wrong, or some other mistake.

Here the duty is increased as well as the resistance of the load.

Updated LTSpice for the pic also attached.

Ok, so just to make a reactive PFC.

Reactance necessitates energy storage.

A plain boost cannot store energy, any more than the inductor that is, and a non-synchronous one can only draw current.

Even a synchronous boost can only sink or source from voltage of a single polarity, so a FWB is needed, but that forces current to be unidirectional again.  We might consider a synchronous FWB, but this is starting to sound like extra steps.

The canonical solution is a full bridge, with current controlled proportionally to voltage, and a variable phase shift introduced.

The phase shift might be a time delay for convenience, but a distorted source complicates that; in general, you should adjust the phase for each harmonic.

We can then test the source's impedance by dithering the operating point and seeing how it responds; the dither effects a change in load current, and we measure the change in voltage.  The ratio gives the impedance; these are all cycle-averaged values (by convolving with sin and cos kernels) so the complex result drops out.  We then set the V/I of the inverter to the conjugate, and we're done.  Repeat as needed.

Note the similarity to an MPPT strategy, where we dither and take the maximum each time.  We can further simplify the above to say: dither some in amplitude and some in phase (or in I and Q), and pick the best of each.  MPPT might be a four-step process, dithering each (I +/-, Q +/-) in turn; the order of which probably doesn't matter.  We of course need two axes to solve for two degrees of freedom.

Or in a more analog sense, we might make some kind of orbit on the I-Q diagram (i.e. plotting the input phasor on a grid), where the circle has a small radius around the setpoint phasor and we measure the bias along that circular path i.e. which direction leads to greater input power; when the net direction goes to zero, we've arrived at the MPP.  (Note that the circle might not have uniform power all around, but it will will a net zero dipole moment.  Multipole moments average to zero.  This implies saddle points are semi-stable; which makes sense given the simplicity of the method, it's just gradient ascent.)

The full bridge needs a supply rail which equals or exceeds the maximum peak input voltage, so a buck is needed to furnish battery voltage.

Note that the bridge can be used to circulate reactive power, at net zero power flow to the battery or input (to a first order, i.e. not counting losses).  I don't know that this is at all relevant or useful here, but it does imply a pure induction generator could be used (needs a capacitive load to regenerate field; the inverter could also kickstart the field in the first place, or serve as "grid" even if presenting an arbitrary phase; though as an induction machine always has an inductive phase, so some capacitive load will give the relevant conjugate matching).

I suspect the effort required to produce this far outweighs the value it produces.  And at the scales where it would be valuable, there are better solutions (i.e. the standard commercial methods).  But if nothing else, it would make a nice bikeshed to shave ones' yak in. :)



[0] Message Index

[#] Next page

There was an error while thanking
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod