OK the title is somewhat clickbaity but it's true, I have one of the ubiquitous devices spinning at 100 RPM. That is, of course, with nothing more that a popsicle stick attached, used as a pointer to indicate start and stop positions. And converted to bipolar and running at 15+V using a H-bridge (8833) "driver".
Just how much of a load, what level of torque can it provide is a question I will answer in a future post as I've not yet completed that test set. I have done a lot of other testing and the purpose of this post, and a few to follow in the coming days, is to publish those results for people to use and comment on.
To start let me say that the motors I got from China (where else) seem to be fairly routine, run of the mill 28BYJ's except for 1 parameter ... they appear to have a true 64:1 gear reduction, not the 63.68+ discovered and posted in another thread in this forum. I say this after doing a lot of testing using 4076 halfsteps as "the number" to get a full 360 deg rotation of the output shaft. What I noticed was that the indicator (popsicle stick) would stop just shy of the starting mark (for a single 360 rotation) but yet reversing the motor would bring it back exactly (to my eye) the starting mark. This was repeatable. I then ran the MUT (motor under test) for 4 rotations before stopping. This time the shortfall was distinctly noticable. I thought that just perhaps the batch of 28BYJ I got mine from had gone back (?) to the 64:1 gears and so I changed the code to run 4096 halfsteps per rotation. Sure enough the indicator now stops "exactly" on the starting mark after 4 rotations, and does so repeatably. Where it the usual 63+:1 gearing it should have stopped ~7 deg short. I can think of no other explanation for the above other than gearing, can anyone ?
Otherwise the MUT has the usual 50 ohm coils (measured 43.6, 43.3 ohms) given the 7% tolerance on the spec I've seen. They came with the usual ULN2003 Darlington board, perhaps a bit nicer than usual as these had LEDs on them (some don't). I used one of these board during all my unipolar testing, which started with a 4AA cell battery pack (my bench supply was down). I then went on to characterize the torque vs speed at various supply voltages, from 5V to 9V applied. Note I say applied voltage as the ULN "driver" will claim about 1V of drop at these current levels.
My test rig was a printed drum, 20 mm in diameter, cradled by 2 supports, one of which also held the MUT. The rig was clamped to a table with the drum hanging over the side. Below it I dangled a thread attached to the drum and then to a water bucket, that I could fill with varying amounts of water. The testing would add/subtract water in 10CC increments until I found (by ear or eye) that the MUT was either missing steps ... or not. That later was considered a pass, and the weight of the water plus bucket (etc) was weighed with a postal scale. This directly translates into the sustainable torque (g-cm) at that speed and applied voltage. Does this sound like a reasonable test methodology to you ? After all I was trying to get a ballpark idea of the torque curve not a uber rigorous measurement. That said I think my results compare to other measured results if not the spec (which is ambiguous by itself, not even counting the 6+ variations of 28BYJs there seem to be).
The attachment is a plot of the MUT torque curve vs speed in PPS (pulse per second) when powered by a 4 cell AA pack. The pack voltage varied from 5.5V to 5.3V during the test period and this slight difference could be seen to make noticeably different results. I use PPS (terminology) to make a distinction from steps, which might be full or half or some other microstepping method. For all my tests I used the preferred halfstepping method, even though full stepping might have produced more torque. Thus a pulse per second equals a halfstep of 5.625deg/64 per second on the output shaft. I note that the spec lists 1000 Hz (?steps?) as the max running frequency and I (and others) have achieved a bit more than that at the usual voltage. The spec also calls for 34.3 mN-m of intraction torque (at 120 Hz, no stepping method given), which is what I think I've measured ... the max torque output at speed allowing for a ramp up (acceleration) to that speed. This is different from the max torque the motor can put out starting from no speed and instantly trying to get to speed. That 34.3 number translates into ~350 g-cm. I measured much more than that, as you can see, but inline with what Solar Robotics lists for their similar motor. So who is right ?
In any case I have more data for the unipolar case (vs voltage) and then again some more (incomplete) for the bipolar case (converted MUT to bipolar winding). I'll post those later after I've had some time to review them. I'll also post the code I used to do the tests as I couldn't use the otherwise excellent AccelStepper library to do some of the bipolar tests (per it's author it's sketchy past 4000 PPS).
Is anyone interested in this data or should I not bother ? I hope I have the right forum for the topic, if not perhaps a Mod can move it as deemed appropriate.
That said corrections and improvements are welcome.
