Search all of my sites with Google

Saturday, July 12, 2008

Torqued Out

I've been experimenting and thinking on how to use the ex-treadmill-motor on the bike most efficiently, which includes a stall current test at various voltages.
@12V it can be held with visegrips but it is difficult. For that I just used a battery across the motor supply wires, since the controller won't run at that low a voltage.
@24V I cannot hold it myself with the visegrips. They are pulled out of my hand as the speed increases, perhaps at 1/2 throttle, using the controller already on the bike. If the motor wasn't bolted to the worktable, I might've been able to hold onto the visegrips but it would have then tried to spin around the shaft.
@36V I felt it safer to just put the visegrips on in such a way that they'd be pinned to the table surface by the shaft's trying to turn, so I wouldn't risk any possible injury, and I'm glad I did. As I increased past about 1/2 throttle, it suddenly started squeaking as the visegrips lost their hold on the shaft, then they somehow moved parallel to the shaft and it began to spin and flipped them thru the air.

The shaft got pretty hot from the current the motor had sucked up in the second or so of doing the last two tests, but cooled quickly once it was rotating. The controller's MOSFETs didn't even hardly warm up from that, since they're on that giant aluminum plate. BTW, the stall current is about 21Amps. I posted the label saying 12Amps before, but I somehow got dyslexic on that figure.

That makes my previous list of calculations off by quite a lot:
@12V (one battery) it could run as 252 Watts max
@24V (two batteries) it could run as 525 Watts max
@36V (three batteries) it could run as 777 Watts max
So that is why the shaft heated up so quickly--at stall it was putting a heck of a lot of power thru the motor for that second-or-so. But not as much as I would've thought....

I discovered with the help of a new multimeter that can measure frequency & duty cycle (I don't have an oscilloscope yet, and need to build the interface to let my PC's soundcard be one) that the controller is never going higher than 75% for some reason. So I never get more than 27 volts into the motor system right now, no matter which motors I use on it. That's 567 Watts on this new motor at full stall load, assuming full throttle is putting out 27 volts instead of 36. No idea why it's happening yet, but will troubleshoot whatever day off I have next week, I guess.

So with these power levels, it's pretty likely that I could use just 12V to power the system, and use all three batteries in parallel for more current capability, or just one battery for times I need it to be lighter (but I will not get much range out of it). Currently I have to use all three in series just to get enough power out of the ex-fan-motors to do anything useful. Part of that might be because of the 75% duty cycle problem, but it might also just be because the fan motors are not designed for torque, just for speed (unlike the treadmill motor).

Thursday, July 10, 2008

From My Lens To Your Screen

Here are most of the promised pics of the updated Friction Drive 2.0, now at 2.0.2. Something wierd is going on with Blogger at the moment, as it won't let me upload anything else (it just sits there waiting forever after submitting), so I'll add more later to a new post if it starts working better later. I did get the motor test video to upload; it's at the end of the post.

These are the left and right views of 2.0.2, with not much visible change from 2.0.

In the left view (showing the left side of the bike), you can see some new heatsink fins at the rear (righthand in pic) edge of the motor panel. That's where the 2 MOSFETs, 2 driver BJTs, and 1 dual-diode pack are now mounted to. The heatsink is out of an old active-air-cooled car amp, used here mainly because it fit between the space available between the rear triangle legs and had enough spots already built into it's clamping bar to hold all the parts I needed to keep cool.

In the middle view, you can sort of see the controller itself, to the lower left of the top motor, but it's mostly behind the tire in that pic. What you *can* see here is the clamping bolts and plate added to the roller skate wheels on the motors, which now clamp those wheels firmly to the motor hub. Unfortunately, without a drill press, I was unable to keep all the holes perfectly lined up; this results in a small but noticeable wobble of each drive wheel, which is magnified many times at higher speeds of the rear wheel, causing the rear wheel itself to vibrate (in combination with it's imperfect centering during my wheel relacing & rebuild, which I need to redo).

In the right view is a closeup of the righthand side of the bike, looking thru the rear wheel into the space between the PWM controller and the heatsink. The controller itself is partly visible towards the upper middle of the pic. The two electrolytic caps partially visible behind the rim come quite close to that rim, within a few millimeters--they're unfortunately the shortest I could find in my salvage that have sufficient capacitance at a high enough voltage rating to deal with spikes, etc, and still physically fit in the space on the existing salvaged PCB. A single much larger and much shorter cap will also fit there, but it gets significantly warmer than two of them, due to the current flow inside it as it tries to stabilize the voltage across the motor during PWM cycles. There are caps made that will do what's needed and fit and be shorter for better clearance, but I'd have to purchase them new, and I'm trying for as much recycling as possible.

Some parts I already *had* to use new (but didn't have to buy), like the 100v / 123a MOSFETs, which can easily handle the power levels involved in this ebike version. I simply haven't found anything remotely powerful enough in the salvaged components I've collected so far, even if I were to parallel quite a number of them--I don't have enough of them with the same characteristics to make a difference, and mixing them could result in unpredictable behavior, as some might begin to conduct before others, or stop conducting later when turned off, derate differently across the power and temperature ranges, etc. Potentially could cause catastrophic failure, and that would probably be at the worst possible moment, when I most need that extra power suddenly. That's why I went with the new MOSFETs. The driver transistors are salvaged, though (they were the MOSFET drivers for that audio amp--the MOSFETs themselves were physically destroyed, but the drivers were miraculously fine).

Some closeups of the top motor drive wheel, 2.0.2 version.

Bottom drive wheel.

The bottom wheel is more precise, and wobbles much less--it was also the second one I did, so when I make some spare wheels for this version (as it works well enough to keep around as a secondary bike's drive system, even after I get a better system going on this bike), they'll probably be even better than this one, even with just hand tools and a power drill.

The hole in the center of the clamping plate on the wheels is to be able to install/remove the nut securing the wheelhub to the motor shaft, as that's most of the way inside the wheel (these motors have *very* short shafts).

The black stripe on the skate wheels actually appears to be some tar/asphalt picked up by the bike tire bit by bit and transferred to the rollers; they're sort of "gooey" when very warm, and pickup stuff like that the way Silly Putty picked up newsprint ink. :-) The tire itself does show some wear from friction, but mostly that seems to be from the deliberate loading tests I did, such as braking with motor power at full throttle (to make sure I *could* stop in the event of controller failure in that mode, and yes, I can, albeit about 4-5 feet longer stopping time).

Closeups of the new heatsink.

It's attached via 3 rackmount screws threaded directly into the aluminum plate of the motor panel (I just drilled holes 1 size smaller than the threads of the screws). Plain old white heatsink compound (from an old CPU heatsink kit I'd used Arctic Silver III on instead) fills airgaps and scratches between the motor panel and the heatsink, as well as between the components and the heatsink.

This assembly is placed in such a way that airflow from the spokes going past cools the MOSFETs, drivers, and diode pack significantly--about 17F cooler than if I put a plastic barrier sheet between the spokes and the controller/heatsink. Unfortunately sometimes I have to sit still for longish periods (3-5 minutes) at some intersections for either left or right turns, during high-traffic times. That keeps the already-built-up heat from having much of a place to go besides spreading thru the entire bolted-together aluminum assembly while I'm stopped, so I might add a temp-controlled small fan (have a bunch, need to test best characteristics

Some of the parts used to create the drive wheels:

The aluminum framework is the bottom of the camera stand, so you can just ignore it.
The two blue ones are unaltered skate wheels, to be converted as spares. The black one is the nylon-ish core of the skate wheel that disintegrated during 2.0's first test run. There are two of the bearing sets next to the black wheel; I don't currently have a plan for using those on this version, but I might need them for a jackshaft-wheel or idler on another thought I have pondered.

These are pics of the brakehandles I'm rebuilding to contain a pair of N.O. reed switches for braking. One will be used to switch the brakelights on, one will be used to switch the PWM controller off or into standby/braking mode. It would be simpler to use only one switch per brake handle, but right now the lighting and the motor systems are on two separate battery supplies, one at 12v and one at 36v. That's sufficiently different that any cross-connection of the system is going to destroy *something*, if done at the braking point, due to the way the PWM controller was designed.

And now the 45-seconds you've all been waiting for, the test video. Keep in mind that since the bike's baskets are being held up by two empty rubbermaid dog food containers, which in turn holds the rear wheel off the ground, there's no load on the rear wheel, so it accellerates faster and decelerates slower tthan miight be expected.

The wobbling of the wheel (which shakes the bike) is partly from my rebuild of those baskets, but really that shaking isn't that bad or noticeable when riding on the road.

Maybe more later--I'm literally dozing off at the keyboard....

More Power, Igor!

Apologies to the few readers here that I have yet to post pics of the fixed skate wheel solution. Laziness and being busy rebuilding the PWM controller with upgraded heatsinks and MOSFETs (100V/123A capable!) has kept me from sitting down and taking pics like I need to remember to do. I should have taken pics of *that* work, too, but I forgot. Have to do it when I go back to it and fix up some rough edges tomorrow evening, I hope.

Though it's been working ok with the radiator fan motors and skate wheels, I'm always looking to improve it. Preferably with slightly less crazy contraptions to drive it. :)

Pursuing that goal, I got a 10-year-old used Proforma 730si Treadmill today off of Freecycle, posted up as broken because it doesn't run, though it turns on, and the motor tested ok by the poster. Since all I *really* wanted was the motor (which I would never be able to buy even used, because they usually cost almost a couple hundred dollars!), that's fine by me. I had misunderstood exactly how big it was, and had biked out to the location to pick it up (stopping to pickup other Freecycle offers held for me along the way), only to discover it was far too big to fit in my trailer safely without completely disassembling it, which there simply was no way to do at the location (I had the tools, but would have had to do it in the street). So I gave in and called a friend with a large truck, and a couple days later when he was available, we picked it up and took it back to my place, where I took it apart to store the rest of it away for later salvage (or re-Freecycling), and pulled the 12lb+ motor.

FWIW, it looks like the AC-input rectifier bridge for the motor's PWM-style power supply has one branch cooked, with only 3 working diodes in it, which would explain why it's not getting enough power to run correctly--might've been an AC-line power surge that caused it. Doesn't matter for what I want to do with it, but satisfies my curiosity as to why it was not working (so I don't have to worry about the motor down the line).

It's a 120V *DC* motor, very powerful at that voltage (2.5HP continuous duty, 12Amp), but can be run at lower voltages--it just won't spin at the 7000RPM it was designed to run the treadmill with, which is fine by me. I tested it at 24VDC on my bike batteries, and I can't stop the motor shaft even with vise-grips. That ought to work out fine to drive the bike wheel, then. Probably not with skate wheels, though. :)

I do have to find a way to fix the shaft in place so I can unscrew the cast-metal (iron? soft steel? I can nick it with a file easily and it's highly magnetically responsive) flywheel from it--that may be half the weight (it's got to be at least 1/3). It used a belt drive with multiple small v-channels in parallel to drive the walking belt in the treadmill; I might be able to use that to drive my bike wheel, or at least as the first stage in the reduction from speed to torque that I might require. If so, I'll have to cut the drive sprocket for that belt off the flywheel after removing it from the motor, since it's not a separate piece, but is all cast as one part.

The matching wheel on the walking belt roller was easy to slide off, and is just cast plastic (dense, but light), with a core hole just about 1/2" larger in diameter than the bike's rearwheel hub bearing-holder section, inside the circle of spoke mount-holes, and about 1" wide. It'll just barely fit within the gap between the spoke-mount flange of the hub and the inside of the rear triangle frame, which means if I can bolt it to the hub/spokes I could drive the rear wheel by belt from the motor. Unfortunately that would require mounting the motor itself horizontally out from the left side of the bike in such a way and place as to be sticking out streetward it's entire length, about 10". That's not practical, both because it could hit or be hit by something, especially during a leaning-left-turn, and because it's difficult to support hanging out like that.

If I can find a longer belt than the very short one the treadmill uses, I could extend that out to where the motor could parallel the rearwheel axle but outside the diameter of the wheel itself, allowing it to be mounted across the centerline of the bike--for instance, under the seat, above or below the point the seat tube crosses the rear triangle frame. The lower, the better, as the center of gravity is best as low as possible with all that extra motor mass.

Now, since it's a 2.5CHP motor, technically it would be too much power to be legal for the bike. However, I'm not running it at even 1/4 of that power, so not even 1HP, which should be fine. It can take 12Amps of current, so a rough guesstimate is 12A x 24V = 288Watts. If I use 36V like my current motors do (which were designed for only 12V, and get to 153F after extended use at their axles!), then it's roughly 432Watts, continuous. Still well below the 750W legal limit (Federal, I think). And I could go up to as high as 72V and ensure it doesn't run it at 100% duty cycle, and it'd still be under that limit.

I'll know the actual power usages and drive capabilities in the next few days, as I test different schemes to drive the standing-test bike wheel, before I try to apply it to my (now working) Electricle™. In the meantime, I'll try to get pics of the various upgrades and fixes to the radiator-fan / skate wheel solution up here.