gearless / geared brushless motors.

[Burgerman]

http://www.recumbents.com/WISIL/shumaker/edrive.htm

In the absence of decent suitable brushless powerchair motors, these types of reduction motors/gearboxes could be used. This shown will give more power and MUCH greater efficiency than existing powerchair motors.

[jim4472]

That looks really good John the question is how do we control a pair in our chairs ?

Is there a nearly solution, reversing included ?

I think The mech engineering wouldn't be too bad perhaps using old F55 Gearbox with internals removed just using the axle side of things.. just thinking aloud really.

Jim

[JoeC]

That particular reduction, if it's the one I've seen elsewhere, uses an Astro Flight 3220 motor, which costs about $700 before the reduction stage, and doesn't include the sensors that a controller would need for high torque at low speed.

An alternative that I have been investigating is the Golden BLT-800 motor. I have contacted the manufacturer for a custom winding, which should cost *far* less than the Astroflight. It won't be as light, and might only be 80% efficient instead of +90% efficient, but if it can be made to work with a single stage timing belt, then it will still beat the pants off of existing powerchair motors for overall efficiency.

Golden does sell a 50 amp 36 volt integrated wheelchair controller, which could be useful for some people, but I have my eye on the Roboteq 75 amp 50 volt controller.

[ex-Gooserider]

Looks neat, but IMHO one of the other big advantages of a reduction drive, which this seems to miss out on, is that it gets the motor out from between the battery box and the wheel... Assuming that the drive and the wheel mounting hardware are narrower than the motor, then you gain extra room to possibly put in some sort of suspension system for the back wheels. I'm assuming that a suspension would be fairly stiff since the tires already soak up the small bumps, what would be needed is something to take care of the big stuff, like when BM does his curb jumping tricks..

ex-Gooserider

[JoeC]

I think that's exactly right. It's not outrageous to mount the motor near (or at) the pivot of a longer arm that supports the wheel, with just a chain or belt traveling through the arm. It could keep the moving mass for a suspension low enough (and the travel long enough) that it might start to make a real difference for big bumps.

In my case the soft suspension and narrow width work against what I'm trying to do, but a chair with an arrangement similar to the Bounder could have a really attractive arrangement with a few changes. If the Bounder had a lower speed (less than 8 mph) and a sufficiently efficient motor, then their huge battery box with group 27 batteries could shrink, the chair could get narrower, and the rear end could shrink. It could be an improvement to their narrow option.

[Burgerman]

That looks really good John the question is how do we control a pair in our chairs ?

Roboteq

www.roboteq.com

[Williamclark77]

An alternative that I have been investigating is the Golden BLT-800 motor. I have contacted the manufacturer for a custom winding, which should cost *far* less than the Astroflight. It won't be as light, and might only be 80% efficient instead of +90% efficient, but if it can be made to work with a single stage timing belt, then it will still beat the pants off of existing powerchair motors for overall efficiency.

I have been looking at the same motor for my chair build, preferably the 650 though due to space. The main hold up on sending the CAD files to the machine shop to be made is the drive system I choose. I can't make up my mind. I would MUCH rather go brushless (I'm big into large scale rc and it makes me sad how those little brushless motors can compare to my 30cc gas engines). However, I'm not sure on how to compare these motors to our brushed ones to get an idea of the power/torque at equal voltages and rpm. If the blt-650 would give equal torque as my Permobil c350 motors with greater speed and efficiency using a 3 to 3.5:1 reduction from a stop without stalling it would be an easy choice. The battery I'm making is a 56 cell pack at 14s using Headway 12ah cells. I could go 70 cell, but space gets tight. I'm also not too sure how long 24v brushed motors would last under my VERY demanding usage at 48v. I'm barely 140 lbs, but my chairs are pushed very hard.

I had an Invacare with the huge brushless motors. I was not impressed. Speed and efficiency were fantastic. Torque from a stop was not. If stopped midway up a steep hill they would not go, just stall. The brushed motors on the chairs I've had would either go or flip me over backwards. Granted, the Invacares are outdated technology now.

What about brakes?

[JoeC]

For brakes, I think the best bet is to use go-kart parts and rig something up with go-kart brakes on the wheel axles and a solenoid to release the brakes when you drive. I'm talking to another company about motors that would include brakes. I'll report back if anything interesting comes of it.

The BLT-650 will really shine at 48V, but not as much at lower voltages. My research has indicated that its resistance is in the neighborhood of 0.4 ohms, so it's not going to do well at low voltage. If you can give it 48 volts and a higher reduction ratio, say 5:1, then it ought to have decent torque. My estimate is roughly 160 pounds of thrust per motor if you do a 5:1 ratio at 75 amps and 14 inch wheels. My trouble with this is that I want to be able to get *somewhat* OK performance at 24V, but a 0.4 ohm motor will only draw 60 amps at stall at 24V. That's why I'm talking to them about a custom winding.

Another thing worth noting is that for sport use we use 10" tires. They're cheap, they tend not to ride over the ball, and they leave more space on the chair that's not devoted to a moving wheel. They also allow the motor to run at a higher RPM since there's a better translation of torque into thrust with a smaller diameter wheel. We don't have a lot of bumps or curbs on the soccer court, so it's a tradeoff that people are willing to make. I recognize that a 14" or larger wheel is much more practical for everyday use.

[Burgerman]

Why not use roboteq and just go 48v? Still possible to charge at 12v, parallel, and so on. There will be some efficiency gain too. And the roboteq allows up to 55v before it complains. It means better acceleration all the time not just at start up/stall.

[Burgerman]

The battery I'm making is a 56 cell pack at 14s using Headway 12ah cells. I could go 70 cell, but space gets tight. I'm also not too sure how long 24v brushed motors would last under my VERY demanding usage at 48v. I'm barely 140 lbs, but my chairs are pushed very hard.

So are mine and I weigh a lot more! What controller?

I chose the Roboteq and 13s (78 cells) because it stays just below the Roboteq maximum 50V (55V gives shutdown error). The 13s at full 3.600v charge sits for a while at 47 Volts. That's 3 below the maximum allowed. And in the first few hundred yards falls to 45V and drops to 42V during the full discharge cycle.

As for how will motors survive, well that depends.
The 24v motors I use draw over 120 Amps at stall at 24v. But the R-Net controller they are used with limits this to 120 Amps MAX. It does this by reducing pulsewidth (so maybe they will draw 120 Amps at say 18V averaged volts) If I use a 45V battery (my lithium) and the Roboteq controller, set to 120 Amps MAX then the motors will STILL see only 18 Volts at stall averaged regardless of battery volts. (shorter pulsewidth x motor impedance = 18v at 120 amps limit). So at stall and at low speed the motors do not know the difference between a 24V battery and a 45Volt one... So stalled they do not care.

At max speed, the RPM is doubled (almost) over 24v. But the amps are very low as speed goes up. So if they do not fly apart at 9000 RPM (and they don't, tested at 60 Volts on my bench and no problems!) then theres no real issue. UNLESS, you drive say at high speed or full power on sand or up a hill for a while. At this point it is possible to overheat a motor. Because its possible to produce much more torque at high rotor RPMs. I intend doing 2 things about this.

1. Watching temperature carefully at first, and if motors get hotter than I like I will fit water cooling and a radiator. (PC water cooling technology).
2. REDUCE the setting on the roboteq called POWER. This controls the max watts. (Motor Amps x Motor Volts - max value) Set this to say 50 percent to control heat, it will then not get hotter than a stock 24v chair.

[JoeC]

Why not use roboteq and just go 48v? Still possible to charge at 12v, parallel, and so on. There will be some efficiency gain too. And the roboteq allows up to 55v before it complains. It means better acceleration all the time not just at start up/stall.

For me? I'd definitely do Roboteq. I'm satisfied that for a brushless controller receiving CAN commands from the joystick, it can be made every bit as safe (and probably nearly as reliable) as a medical controller.

48V would be OK, but I don't think I'm going to be comfortable putting lithium into every chair I work on for quite a while. That means that I need to find high quality AGM batteries that aren't extremely heavy, bulky, or expensive in a group of four. It's not me sitting in the chair, and although some of the folks I build things for are engineers or are at least technically minded enough to fully understand and accept the risks that are involved, some are not.

I'm aiming to find a lead acid 36 or 48V arrangement that will work for cases where lithium isn't the right option, but I want to be able to do a bit better than what's already out there if I get stuck on 24V.

For anyone else out there building his or her own chair from scratch, I think that a 48V Headway or prismatic lithium pack with the Roboteq could be an excellent way to go.

[Burgerman]

Three 42Ah odyssey batteries on end, are the same exact size as two group 34 Odysseys...

So an option for 36v lead. But the more I learn about lead v lithium, the more I see it as very dead and a crappy solution. Lithium prismatic may be OK just about for replacing 24v, but don't really have the C rate for going to 48 (or 45) Volts - especially in sport use.

[Williamclark77]

I was actually researching the Robotech controllers when I came across this site a few weeks ago. I plan to use either of their suitable controllers once I figure out which motors. FYI - I read the entire Robotech thread sticky after I found it while researching how to program it for a powerchair application. Good read.

I designed the battery box around the seat base I already have and dimensions for the locking system I use in my Chevy Uplander I drive from. There are only so many inches between where the front locks and bolt locks. Also, my bolt is retractable and must be added in as more room to the battery. I am sticking to something that will fit without modifying those since they work with my current chair. 60 cells work out to fit perfect and 14S divides right at 56. 13S and 65 cells would also fit. 14S may fit with 70 cells, but it'll be close. I'll have to measure very closely.

I'm positive the blt-650 motors can easily move me. It's the torque to start it moving. What worries me, as with the older Invacare brushless motors, stop on a steep hill or ramp or trying to pull something and they would not move, just stall. Once they started turning though, such as go along the hill a few feet before trying to go back up, and they'd continue on and pop wheelies. Brushed motors do not need the slight head start. This is what I need to avoid. I do not know if the blt motors would behave this way. I do not believe the Invacare controller was hard programmed to act that way. Hold on the joystick for 5 or so seconds with the motors stalled and fuses would blow.

3:1 reduction with a 14.5" tire on a 1000 rpm motor should get tire rpm to roughly suitable for a 14ish mph top speed. I'd prefer not to go much lower on the gear reduction.

Anyway, at around 3:1 reduction with 14.5" tires on 14S (possibly 13) using the Robotech controller and blt650 motors I'm HOPING it starts from a dead stop well.

[JoeC]

It will probably work out. I'd suggest leaving yourself the option for a greater reduction, although I understand why you wouldn't want to (larger sprocket, worse wrap angles, etc), because if the torque isn't sufficient then your only other option will be to use two of the 150 amp single channel controllers, or to go through the pain of figuring out some of the Kelly brushless controllers, or....?

[Burgerman]

You cannot use 14s with the roboteq.

14 cells fully charged is 14 x 3.600v or 50.4 volts. Now it MAY not complain but you have to consider the regeneration spikes above this point when you decelerate. That may cause it to see over volts and shut down. Its really designed for 36v lead systems. They are about 40v or so fully charged. The extra is headroom. This is the reason I went with 13s. And 78 cells. or 3k watt hours. That's 47v charged. Leaves a few volts spare!

For what its worth efficiency increases if you gear down and up the volts.

For a SMOOTH brushless you need about 25 poles on the motor or more for direct drive. So about 8 poles on a 3 to one gear ratio. So it may be better to use 6 to one, and use a lower impedance motor. They all hesitate at a start, because until they move they don't know which direction they are going to go. The more poles the better, and it must be sensored.

[Williamclark77]

It will probably work out. I'd suggest leaving yourself the option for a greater reduction, although I understand why you wouldn't want to (larger sprocket, worse wrap angles, etc), because if the torque isn't sufficient then your only other option will be to use two of the 150 amp single channel controllers, or to go through the pain of figuring out some of the Kelly brushless controllers, or....?

Yup, the motors' and tires' centerline will be pretty close together. Much more reduction and few teeth will make contact. Another reason why these motors looked promising. No room for reduction if a 2 or 3k rpm motor was used and I do not want the added complexity of a double reduction. Of all the controllers I looked at, Robotech seemed to be the easiest to set up. I came across the sticky thread above where so much of the code has already been figured out and it's definitely the obvious choice. I hope to benefit from yaw's labor. Code is not my forte. I've built dozens of computers including this 6 core watercooled winders 7 64 one I'm on now, but can't read a line of code.

I'm hoping to start having parts machined about the middle of next month. That's my goal anyway.

[Williamclark77]

Sorry. Didn't see your reply yesterday.

You cannot use 14s with the roboteq.

14 cells fully charged is 14 x 3.600v or 50.4 volts. Now it MAY not complain but you have to consider the regeneration spikes above this point when you decelerate. That may cause it to see over volts and shut down. Its really designed for 36v lead systems. They are about 40v or so fully charged. The extra is headroom. This is the reason I went with 13s. And 78 cells. or 3k watt hours. That's 47v charged. Leaves a few volts spare!.

Thank you. I was going to email Robotech to ask if 14S would be too much or if it would be ok once the voltage settled down a bit to ~48v after running a few minutes. I doubt there will be enough "seat of the pants feel" performance difference between 13S and 14S to justify pushing the safety limits.

For a SMOOTH brushless you need about 25 poles on the motor or more for direct drive. So about 8 poles on a 3 to one gear ratio. So it may be better to use 6 to one, and use a lower impedance motor. They all hesitate at a start, because until they move they don't know which direction they are going to go. The more poles the better, and it must be sensored.

Thanks again. That's the information I was looking for. The actual physical design that determines their behavior. I'm still bench racing the electronics. It'll still be quite some time before wires are cut.