I don't want folks to think my attitude on mppt as an unnecessary addition stems from not having a fiddle with it.
The other way to do it is to wind for 22 odd miles per hour winds, and boost the lower winds as a square function (plus a bit), and we achieve near 60%-70% from the alternator at high output when the boost is overtaken, and we get to control the lower band with the same kind of tricks as buck control (but the other direction).
Buck provides for less line loss, but that is also true of boost and transformer if you wish..... beyond that it provides little extra.
The booster provides the same control at the lower levels, but the alt windings are better suited to the power curve of the wind to the alternator than with 8-12mph windings.... plus your power handling is roughly 4 times as well. With boost, the electronics sees only the lower winds, the alt takes the brunt of the high winds where things are a bit heavy for the electronics.... so we don't need clipping either.
In order to gain reliability, the non-use of computer driven pwm, makes it easy to control in real time cycle by cycle current limiting, and so input filtering can be very basic. As the waves fall towards zero, the boost will try to go very hard, but the current limiting will protect the fets in real time, not sampling time.....
Anyway here is a story I wrote some years back regarding fiddling with this technology.
=============================================================================
In 2006:
Is was one of those days where it was too windy/cold to venture outside and I decided to tidy up my boost experiment.
The rats nest was spreading over onto the floor of the shed, and it was time to dust off the big computer and do a circuit board for this converter.
After finding the disks for the Autotrax and loading them into a new windows 98 partition, it was time to relearn how to use the program again (long long time).
After struggling with making a rudimentary library of components, 3 hours later I finished up with a board like this:
Its not done by an engineer so you get what you get... time to print it out.
Luckily the program had a driver for laser 300dpi and the old brother 730 I had bought across with me had just enough toner to print a page or two. (Moving onto an island is just so much fun).
I print it out on two transparencies (o/head projector sheets I think). I use two because this printer is pretty bad, and the toner is very low. I moved the image on the transparencies to different physical locations so that the flaws in the drum appear at different places on the layout. very agricultural.
By taping the two transparencies together (on top of each other )we get a image that is mostly black where it should be in two layers. Any drum deficiencies only effects a single layer at differing places, so the theory was that tracks would be mostly covered.
This worked out as pictured here:
After rummaging around through the container outside, I found the picture frame I wanted and removed the glass from it, found the ammonium persulphate and breathed a sigh of relief..... over here you can't just go up to the shop. I've been waiting for a month for my fibreglass to turn up for a stator....still waiting.
So being lazy, I had gotten some pre-sensitised pcb boards from Wiltronics (on the net) and proceeded to sandwich the transparency between the green pre-sensitised board and the glass from the picture frame.
Well now all I need is a UV light source. A fluorescent for 20 minutes at 2" (haven't got one out in the shed) or exposure to the sun for 1 minute or so.
So we wait for the sun to come out..........not coming out.....now its's raining again.....
The nice thing about oceanic weather is the complete unreliability of the weather. Its still raining, but now the suns out at the same time. I hold the board to the sun whilst standing undercover for 2 minutes....seems a long time to stand still and all the while thinking..... gosh this is high tech.
Now to develop the board.......oh no, can't seem to find the NaOH. When this happened on the mainland we had a dishwasher, and the powder for it was caustic, and could be relied upon to act as developer....a shed in the middle of a paddock does not have a dishwasher. A quick test with the clothes washing powder yielded no result...damn people friendly wash powder.......I did the unthinkable............
Went in and had a cup of coffee and brewed over my dilemma....The good lady wife suggested oven cleaner......silly women.. oven cleaner ….
that won't work says me...
It was ten minutes later that I looked with satisfaction at my nicely developed board:
….. and sheepishly went back in to inform she who must be obeyed that she was right, and I was wrong.....be no living with her now..
Scrabbled around the shed looking for a suitable container to etch the board in. It must be corrosion-proof, disposable, tough, and I decided a 5kg plastic nail box for 4" bullet head nails looked perfect......so.... out they go.
Armed with the newly acquired "etching tank" I added about a tablespoon of ammonium persulphate to 1/2 inch of hot water, and dumped the board in.
I sloshed the board around in the mixture for ten minutes.......I've had enough, it's too slow.....add more ammonium persulphate.....little bit more.....now I can see it peeling off (small exaggeration). it's almost done:
Note the lovely yellow nail container (I mean etching tank.).... Decided that yellow makes it go faster.
Well half the day is gone and we have a board that needs drilling. It was some time later that I found the minidrill, and importantly the mini-bits that go in it, and filled the desktop with another bunch of tiny holes. I have discovered that the drill drills quickly through the fibreglass, and that the trusty bench is at a different rate and sound so I know instinctively to move on to the next hole.. Good for me ..not so flash for the bench.
Well the next hour was used to load the board as per this overlay:
There is no trick to this, I use an ic holder to take the tl494 so that if I destroy it in testing, it is easy to replace. The coil is stolen from the original one I built a few days ago (50t ETD39)
It ends up looking like this:
So yes it can be done easily from scratch... design, board design, load and test in a single day.... This makes it very easy when compared to making the turbine itself
Testing was as follows:
Tentatively hooked up the power and the load, and the poor layout of the design was immediately apparent. The waveform was ok down at lower levels, but started to wobble around a bit at higher levels. A 470n cap across the 1M resistor on the l/hand side of the overlay stopped the spurious voltage spikes, and the fets ran cool at 200W input for about two minutes. The diodes warmed up mildly, but that was all. At 300W, my battery on the receiving end was getting far too much charge and was held for only long enough to feel a slight warming of the fets.
It can be noticed that there is provision for a second coil, which I haven't tried yet, but will wind it the same as the first inductor, and this should increase our power handling to well over 600W.
Just to get a better idea of how the switching wave was, I pulled out 5 of the 6 fets to find how heating went with only one to carry the switching current. 100w warm.. 200W nice and warm after a minute or so, 300W, now it needs a heat sink. Not super hot, but you could tell that without cooling, it would not last forever.
All in all, happy with the output. For one 12v battery to charge another 12v battery.. this will do it at 20-25A or more in this config, or if the second inductor works as planned, 40-50A.
I don't have a copy of the circuit, as I made parts of it up as I drew the board. (I can hear the sounds of disgust by people who design things properly. It is very loosely based on a circuit from Oatley electronics
http://oatleyelectronics.com/pdf/k098.pdfNo circuit of this I'm afraid. Just reverse engineer the board. The output stage is roughly as Flux had described, and the current sense and voltage sense are last minute additions on the layout board.
The current reg will need a slight mod to use as a boost converter as per Flux, and I don't have the proper hall effect diodes over here, but if we were to sense the three phase by way of the single turn current sense you could achieve roughly the same thing. (one line of "before the bridge diodes" of the genny output)
This current one will suffice for a large alternator if used with two coils. Further testing will show the upper-limits of this device, but 12v-24v is easy. It should do every bit as well at higher voltage, and it's wattage will increase as well.
I tried a “what will she do mister.”... 40A input fets still held together, and started to warm .... certainly not dangerously so. The battery fizzed loudly, so stopped the test to destruction...spoiled my fun a bit.
It has one drawback as a standalone power supply. It relies on the receiving batteries as a voltage clamp. It will require a static load to keep it under control without this clamp..... its a battery charger after all..
Notes:
This experimental board is a tidy up of the small board done a few days earlier, but simply tidying up the ratsnest on the desk and putting all the junk in one place. So the circuit board holds all the stuff including the inductors.
To further investigations I designed the board with two inductors in order to see if it was as simple to scale up as I'd hoped..... It does seem to be the case.
The wave form is same as before (scope in my files). This is directly across the switch fet. With only 1 coil,efficiency drops below 90% at about 200W, with the two coils, this seems to keep the efficiency higher for much longer. (less resistance in the coils in parallel).
As a 12v battery charger from another 12 v batt it is very good and stable. 20-30 amps.. no problem)(big deal who wants to do that anyway?) It needs a 470n cap in parallel with the 1m feedback resistor in this case.
For stand alone, the deadtime control (pin 4) needs about 220K from pin 4 to Vref. This stops it going into a runaway condition when there is no battery clamp on the output. What that could be used for escapes me as well.... but it is a very powerful voltage increaser in excess of 30v at quite high current.
For the operation it was loosely set to explore.... boost on an unstable input waveform...it appeared to be terrible. I connected it to a 1/2 wave rectified 50hz 10Amp transformer as an input, to mimic a fast spinning single phase mill. The resultant output into a battery was shocking above 2A.
This was caused from two things. As I increased the output, the rms output from the source transformer dropped . This fed directly into the voltage support for the tl494 (ka7500 in this case), and it dropped in and out of operating range at 50 hz...what a hash that made of the switching waveform.......
The second problem was that being single phase input, in every cycle we dropped through zero and so there was no voltage to boost, and so the boost fet tried to drain the filter cap on the Voltage regulator......not so good. If it was on three phase this would be masked some.... so..
We could draw the chip power from the battery on the output... too easy
I decided it would be better to solve the problem from the input side so that it could be run as a power supply from a scrappy source. This required only a diode in the line from the input pad to the regulator. This allowed me to use the voltage peaks and filter and regulate them for the pwm chip and driver circuit, but the switcher fet couldn't drain this supply back through the diode...another failure... another lesson, another victory.
So that's where I left it.
The inductor was wound on the lathe from 1.8mm wire Neatly close wound for four layers....comes to about 54 turns or thereabouts. The frequency is so low (3khz) that it offers lots of flexibility.
The core material is 3C80 (the numbers you can almost see), but any thing will do at these frequencies. The two small current transformers came from some junk psu. Could easily wind your own around a core, or use the input noise filter inductor from a standard psu and rewind the coils on that. I only used two to get enough current handling through the single loop at 40-50 amps....well.....for the hell of it .
..................oztules
Edit:
This was envisaged for an 8 footer, which I had carved the blades for. They took me a long time, so I decided to make it more efficient, so I wouldn't have to carve large blades. Before this was finalised, I developed the chainsaw method for blade production, and no longer cared how big they were to be. So the efficient 8 footer never actually went up. The same tower with a 4 meter blade set cost no more.... so I never bothered. The 4m unit has run without upset, with much more power than an 8 footer could ever dream about, and the cost was the same (ok $30 different as I carved new blades, but used the same steelwork (yaw gap changed) and a few more turns of wire....
Edit 2 fixed a LOT of typing errors and diction.