Report KD 340 about rectification reviewed

[Adriaan Kragten]

Report KD 340: "Rectification of 3-phase VIRYA windmill generators" has been reviewed. A new figure 10 has been added to chapter 3.2.1. This figure gives the variation of the DC power for 3-phase star rectification. A new chapter 3.2.3 has been added to KD 340. The title of this chapter is: "Rectification of each phase with a 1-phase bridge rectifier". The advantage of this third way of rectification of a 3-phase current is that the DC power isn't fluctuating.

[kitestrings]

Hi A,

A couple of questions, and please bear with me, my theoretical knowledge is limited, dated, or both ;>].

I thought with a single-phase, full-wave bridge we would need 4 diodes, but I think in this configuration you are saying that is only six?

The trade off with this arrangement is that the losses are higher?  Do the diodes also have to be individually more robust, because if I understand you, they would be subject to full current vs. full current, 2/3 of the time, in a (star) 3-phase bridge, in a given conductor... so the RMS current is only (2/3^.5)= .82*I ?

With a 3-phase, full-wave diode bridge there are two sets of diodes which must be electrically isolated from one another.  With this method, I assume there are three?  Does this make the build more difficult from your experience?

Thx, ~ks

[Adriaan Kragten]

For star and for delta rectification of a 3-phase current, you need a 3-phase rectifier with 6 diodes. The wire diagrams of these two ways of rectification are given in figure 5 and figure 6 of KD 340. For star rectification, a coil is used only during 2/3 of the time (see figure 7) and this result in fluctuation of the power (see figure 10). For rectification of a 1-phase winding, you need a bridge rectifier with four diodes. I don't give a wire diagram of this type of rectifier but it looks almost the same, only the diodes D3 and D6 are cancelled.

For the third way of rectification of a 3-phase winding, the three windings are not connected to each other and every phase is rectified by a 1-phase bridge rectifier. So totally twelve diodes are needed. The DC terminals of the rectifiers are connected in series. The advantage of this third way of rectification is that now every coil is used all the time. This results in a somewhat higher DC voltage and a somewhat higher power.

The power of a rectified 1-phase winding is given by the sin^2-alfa function of figure 2. For three phases which make a phase angle of 120° which each other, you have to add three sin^2-alfa functions which are shifted 120° with respect to each other. This results in a constant power which is just a factor 1.5 higher than the peak power of one phase. This is an advantage if the windmill is grid connected using a 3-phase inverter because then both the ingoing DC power and the outgoing AC power have no fluctuation and this eliminates the need of capacitors.
 
The disadvantage is that the DC current is always flowing through six diodes. If the voltage drop over one diode is 0.7 V, the total voltage drop is 4.2 V. This third way of rectification should therefore not be used for a low battery voltage as then the rectifier losses are rather large. But for a high DC voltage, the rectifier losses are small and then this third way of rectification can be used.

I assume that you don't build up a 1-phase bridge rectifier from separate diodes but that a rectifier is used with four diodes in it. The required size of the 1-phase rectifiers depends on the maximum current and on the cooling. The rectifiers have to be placed on heat sinks if the maximum current is high and one should not allow a higher current than about half the nominal current if a long lifetime is wanted. You can build up a 1-phase rectifier with separated diodes but in this case, both heat sinks of all three rectifiers should be isolated from each other.

[kitestrings]

Thank you.  I follow this third method now, though I can see why it might be less common.

I've always preferred the stud-diode built rectifier over the potted types.  It seems they can be made to be pretty robust.  Also, when I worked on wind turbines more years ago, we would occasionally encounter a failed rectifier due to overheating, or a nearby lightning surge.  With the diode type, you could test and replace just the failed part and be back in service pretty quickly.

Here is the rectifier box that I built for our turbine.  There are two because we have two controllers, each with their own inputs.  The front anodized aluminum heat sinks are hinged to allow for better access.  The enclosure is repurposed, reconfigured from an older wind control box.

[SparWeb]

Nice enclosure!
Is there a shunt hiding inside?

[Adriaan Kragten]

I have added a wire diagram as figure 16 to chapter 3.2.3 of report KD 340. This wire diagram is added as an attachment.

I have also added figure 15 which gives the power fluctuation of the three phases U, V and W and the total power.

[kitestrings]

Thanks Adriaan.  This is what I pictured after your description, but makes it much easier to follow in your report.

There are no shunts.  I used the box from our Sencenbaugh controller, cut the cover, and added the second set of heat sinks to the front.  Inside are: breakers, an Omron high V limit relay, the diodes (85A) and two relays (load & dump).  On the cover is our rpm meter.

[Adriaan Kragten]

It appears that there was a mistake in the calculation of the total power. It is only allowed to find the total power by adding the power of the individual phases if every phase has its own resistor load. But this isn't the case if the three rectifiers are connected in series. It is allowed to add the voltages which results in a fluctuating DC voltage and DC current. This makes that the power for this third way of rectification is fluctuating in the same way as for star rectification. The report was modified concerning this point.

[DamonHD]

Would adding some (tuned) level of capacitance across the DC output of each stage help?

Rgds

Damon

[SparWeb]

Yes, but the *tuned* value of the capacitor is huge.  The cost of the capacitors, and the marginal effect they will have, make it not worthwhile.
For comparison, that ripple you see in Adriaans paper is a fluctuation less than 5% of the average.  Correcting 5% to make it more like 3% ripple is a small benefit indeed.

Edit: there are extremely low voltages in Adriaans paper which exaggerate the fluctuation of voltage.  If I understand the meaning correctly (jump in if you have more to clarify AK) this is just above the rectifier threshold voltage.  Then as the generator operates at full voltage above battery cut-in and then in the range of useful power, either 28V or 56V or whatever the case may be, then the fluctuations that we see (+/- 0.3 volts) have not grown in proportion, so that's when they make up only 5% of the system voltage.  Oh I just did the math and its even less than that.

[Adriaan Kragten]

Would adding some (tuned) level of capacitance across the DC output of each stage help?

Rgds

Damon

I think that for battery charging using of a capacitor to flatten the DC voltage and the DC current is useless as the battery itself has a flattening effect. But for grid connection, capacitors might be needed if a 3-phase inverter is used and if three nice sine waves are wanted. In my report KD 712, I have explained a 9-phase winding and the fluctuation of the DC voltage and DC current of such a winding is almost negligible.