As promised, pics (and video!) of the Cieling Fan motor rewired as BLDC.
First up is the raw pieces, showing the brown Hampton Bay motor from 1986 in a partial state of disassembly. It looked much like the other cieling fan motor in the background of the first image, which is a 1995 motor from China, by an unknown manufacturer, for Hampton Bay. I chose the 1986 Hampton Bay unit to start experiments with primarily because it was easier to get apart, having been apparently designed to be serviced, but also because it was marked as made in the USA (though the windings could have been done by a 3-year-old for all I can tell by their messiness, especially compared to the Chinese motor :-) ). I also picked it because it is a rounded case, with thicker metal and no vent holes for water and dirt to enter (unlike the other that has plenty of finger-sized holes), and I might as well go for any advantages it might have being used on the bike. It does have a lot of threaded mounting holes, but that's an advantage, since any I don't use can still have screws threaded into them to seal them up, or silicone if I'm feeling chintzy.
This motor has 9 teeth on the stator, which is a common number, as is 12, for RC hobbyists to pick to rewind into BLDC's. I'd rather have had a 12-tooth version, as there is a variation on winding called LRK that gives better efficiency, but I'll have to make do with what I have in hand. I don't like the open bearings it's got--they have no cover plates to keep dirt and dust out, and I have no way of pulling them off and replacing them with different ones (even if I had others to do this with right now).
Here's an image from during the unwinding of the outer coils, showing the old RadioShack wire spool I was putting it on. It barely fit; by the time I was done it was bulging over a bit, since I didn't wind it tightly (didn't want to risk stretching the wire; it's pretty thin).
The Chinese motor is below, though I haven't done much to it besides open it up to look at it. That was tough, as they didn't make it to be serviced, and I was afraid I'd have to damage the casing to open it. Fortunately, being very patient and gently pulling the case halves apart with two wide-blade chisels by rotating and rocking them as levers on opposite sides of the case, over about 30-40 minutes, eventually opened it without damage.
As you can see here, with it on my "test rig", it's got a different way of winding than the 1986 motor--every gap has been used to wind on, where the 1986 motor only used every other one, since the smaller gaps on 1986 were only so the inner circle of windings could be installed. I have no real idea why it's wound differently from the other one, but since it is still an AC motor with those windings, it's irrelevant, since I must unwind and rewind it differently to use it as BLDC anyway.
Something else you can see here if you look carefully is that one of the bearing case covers (which the 1986 doesn't have) is dented. That's probably why it was in the junk box of fans in the first place, since it causes it to rub and make a bit of noise every rotation--probably got annoying in the middle of the night to whoever had them originally.
The test rig itself is just a couple blocks of wood with two of the fan blade mounts screwed to it with some of the fan case screws, and a long thick nail inserted thru the shaft hole (it's hollow for wires to run down to the lamp and switches, remember?) to hold it up, to allow the rotor bell to freely spin without me having to hold the thing up all the time (it's a bit heavy, perhaps 6 pounds or more). Since it doesn't clamp the motor shaft down, both the rotor and the stator can spin, which is a problem as you'll see in the video below. But it did let me test a theory, and it was very quick to throw together out of parts already laying there from the fans. Only thing I had to step away for was the block of wood, just around the side of the house outside. The two-minute test stand. :-)
For this test, I did not yet rewind the 1985 motor, I simply cut the loose wire between each of the inner windings, so I could temporarily change the order they're hooked up in (which for AC was just series circumferential), to put them in the Wye configuration. I didn't do anything fancy, just lap-solder some wires between the appropriate winding ends, to give me ABCABCABC for the pattern.
Then, I used 6 harddisk magnets placed all in the same polarity around the circle (you'll see some appear backwards, but they're not--they're not polarized front-to-back, but rather end-to-end on the same face, which makes them poor candidates for motor magnets). I'll get better magnets before I actually make this a motor for the bike, but it's all I have right now. They're of course just close to where they should be, but since they're all different strengths and sizes as well, it's irrelevant how I place them, since I only need to see if it will work *at all* for this test.
I haven't had time to build a controller yet, but as I was digging thru parts bins, I found both a dead 5.25" floppy drive and a mirror-scanner assembly from an HP laserjet. Both of them have controllers built in to them, as single-chip types. Of course, they're nowhere near the wattage I need to run the motor I will make out of all this, but they will both run this test-stand version long enough to verify the idea. Both can run on 12vdc, with just an enable line held low on the 4 wires coming off of each one. I think the 4th wire is a "motor is at speed ok" signal, but I don't care, since I only want to see it spin at all. In any finished controller for this, I would not have a set speed anyway, it would change all the time so such a feedback would be nearly useless to me (unless I had the whole bike computer-controlled, which might happen some day long from now).
Since I couldn't get the FDD motor off to access it's hall sensors (they're under a metal plate that's held down with some really soft and tiny screws, probably locktited in place, which I stripped out thoroughly trying to undo them), I went with the easily-disassembled HP controller board. The HP board has a different and lower-wattage chip on it, but it is very simple to wire to the motor, once I carefully removed the motor that was on it. This also left the Hall sensors exposed so I could simply hold the board up in the magnet path, rather than having to unsolder the sensors, glue them in place on the stator teeth in a position similar to the ones on the board, and run 8 wires from the three sensors' 12 pins total back to the controller (4 pins are simply chained together between the sensors). Big time saver for the test, though of course it can't really be used properly like this for any real work.
Now that you've suffered thru my rambling, here's the video of it actually working, only a bit over a minute. I was mumbling to myself, really, not expecting the camera to be able to pick me up over the heater fan in the room (the central heat failed just as it was getting dark today, and even though this is Phoenix, it still gets mighty cold mighty fast in this old house). The camera itself is the same one I take the stills with, a Sony DSC-50 I got really cheap at the CompUSA closeout when they shut down here back in May, because they had it marked as defective. It wasn't defective, just missing it's charger (which I found later in a bin of assorted adapters and junk, which they were gracious enough to give me for free).
Saturday, December 29, 2007
As promised, pics (and video!) of the Cieling Fan motor rewired as BLDC.