Alternator Modifications (2016)
 

So, building off this thread -> http://ecomodder.com/forum/showthrea...ons-18045.html

I have gone ahead and purchased all of the required components.

http://ecomodder.com/forum/attachmen...1&d=1465413810

I also purchased a used alternator for a 4.7L Dodge engine, and took out the rectifier assembly. I was somewhat surprised to discover that the alternator apparently has two complete sets of Y windings. I verified this by taking a multimeter to the winding leads, and by examining the rectifier assembly. The rectifier assembly has a total of 12 diodes, and one of the common planes is split electrically. To me, that says "two sets of Y windings."

The MOSFETs are probably overkill, as they are rated for 260 amps apiece. However, they have an on-resistance of 0.00195 ohms, and calculations show an 84% drop in dissipated power, compared to a standard silicon rectifier diode.

Hm... Just did some research into the press-in rectifier diodes used in the rectifier assembly. Apparently, they have a forward bias of around 1 VDC, instead of 0.6 VDC.

SG-9LCNR datasheet [pdf] SG-9LCNR datasheets manu [SANKEN] pdf datenblatt ,view PDF - - WWW.ICPDF.COM

That really changes things. Re-running the numbers, using both this new information, and the fact that my alternator uses 12 diodes to begin with, gives a more stark difference in thrown away power.

At a current draw of only 35 A, my alternator currently wastes about 210 W through the diodes, or a bit over .28 HP. At a current draw that I figure to be about 90 A for what I would normally run, my alternator wastes 540 W through the diodes, or 0.7 HP. At the rated output of 136 A, my alternator would waste 816 W through the diodes, or 1.09 HP.

With installing this modification (and presuming it works), I would see those numbers drop dramatically. At 35 A, my alternator would waste a total of 7 W (0.01 HP) through the MOSFET circuits. At 90 A, my alternator would waste 47 W (0.06 HP) through the MOSFET circuits. At the rated 136 A, my alternator would waste 108 W (0.14 HP) through the MOSFET circuits.

All I can say is... wow.

Eager to see how this turns out. How does packaging look? (as in, how much more space do the MOSFETs take up?)

And yea, they are usually 1V drop not 0.7. I figured that out when searching through electronic component catalogs and found the high amperage diodes having higher voltage drop. It's kind of insane that car companies use something that wasteful to save a hundred bucks...but then again, belt driven power steering and water pumps are even worse, and barely save any money as well.

It's also crazy that without the 2V total voltage drop, the alternator pulley ratio or engine speed could be lowered while maintaining enough electrical power, or kept the same and avoiding flickering lights.

I really want to do this mod to an alternator so I can use an underdrive pulley without lowered voltage.

Going to have to turn back the controllers to DigiKey - apparently, I did not do enough research.

Texas Instruments also puts out a controller similar to the LM74610-Q1, which is the LM74670-Q1. However, the LM74670 is described in the datasheet as a rectifier diode controller specifically for car alternator rectifiers. There doesn't appear to be a whole lot of difference between the two chips, otherwise. Just in case, though, I ordered the LM74670s, and should get them about a day after I get the LM74610s.

Well, the MOSFETs are in a TO-220 package, so they should be able to fit between the two common rectifier plates. I'm going to screw the MOSFETs onto the rectifier plates, to use the plates as heatsinks. The B+ plate won't need any isolation, but the ground plates will. I'll probably also press out the existing rectifier diodes, since the MOSFETs themselves can also act as rectifier diodes if their controllers somehow fail.

The controllers themselves are about 1 mm thick, so maybe I can epoxy them directly onto the MOSFETs.

It's engineering-by-accountant. If you can chop off a dollar for every vehicle you make, and you make 300,000 vehicles a year, you just saved $300,000. That's the thing that pretty much killed Chrysler after Mercedes bought it. Chrysler's 4th gen minivans are about the most visible example of this - rustbuckets that literally start to fall apart after 150K miles.

If we can lower the alternator speed, we could realize even further gains. Alternators are said to be about 55% efficient. They're more in the 70% range at low speeds (say, at engine speeds from 700 to 1200 RPM)

http://ecomodder.com/forum/attachmen...1&d=1465569250

this is pretty much true across the board?

Pretty much. Alternators themselves pretty much are the same in function across different makes and models of cars. The only differences are how each automaker chooses to build them.

Yes. Eddy current losses increase with the square of speed (although field strength decreases, so it's a linear increase in losses), and cooling fan losses increase with the cube of speed. Lower speed is nearly always better.

If you look at a similar amperage and kV permanent magnet motor to an alternator you'll notice that they need 1kW just to spin at full speed. Since the field current of an alternator is reduced at high speed it's more like a few hundred watts, but that's still a lot.

I hope you guys are paying attention to the frequency some of these components can support. In the LM74670 data sheet it says it only supports a max of 300Hz.

You bring up a good point, with 3 pole pairs that would correspond to a max alternator rpm of 6000, which is not very high.

So, above 6000 alternator rpm, or above 2000 engine rpm, the very worst case scenario is that the MOSFETs would act like normal rectifier diodes. That might not be good for people who normally cruise at an engine speed above 2000 rpm.

I went by the description of the duty cycle given by the data sheet, because while TI gave the graphs to show why they came to 300 Hz, they neglected to give their test circuit. V(gs-threshold) appears to a heavy factor in the arrival of that figure.

It would appear that TI arrived at its 300 Hz figure to say that 300 Hz was the highest frequency at which their rectification circuit would deliver an output comparable in waveform to an equivalent diode rectifier circuit. Beyond that, the charge time would start to eat into the upswing of the AC waveform, distorting it significantly. Again, though, the MOSFET would act like a diode duting this time, so rectification would still take place. I can live with that.

I'll post a few pictures illustrating this, later on.