Programming Infineon Controller for Powerpack motor

[mrbill]

I've read through a bunch of threads on programming the Infineon controller, but most of the contributors are using hub motors. Does anyone have a set of recommended parameters for the Powerpack motor used with the Matex gears, Lashout gearbox, and 18mm one-way clutch as supplied by EVDeals? The controller is an EB312-AS-Z from Lyen.

In particular, I'd like to get good starting values for Block Time, a ratio for Phase Current to Rated Current, and advice on setting one of the three speeds to 120% or some value >100%. I have the three-speed switch.

I'm looking keep the torque low enough that it doesn't damage the one-way clutch, keep efficiency high in speed range #1 and #2, yet allow for a high-speed setting in #3.

Thanks.

[rojitor]

Probably the most suited person for that question is lyen himself,he solved many of my doubts, did you ask him?

[mrbill]

Probably the most suited person for that question is lyen himself,he solved many of my doubts, did you ask him?

Yes, I did. And he's been very helpful. While the default .asv file he sent me for my controller works well, he added that it was not tuned specifically for my motor. I'm wondering if anyone has spent time tweaking the programmable parameters for this controller (Infineon EB312-AS-Z) to the Powerpack motor, in particular.

[Ham549]

BUMP
I need to know this as well since my controller caught on fire and I need one for my Powerpack motor.

[mrbill]

BUMP
I need to know this as well since my controller caught on fire and I need one for my Powerpack motor.

Which controller are you using that caught fire?

Lyen and I are still working to get this motor running smoothly at all reasonable loads, voltages, and throttle settings. it's running really well--smooth as butta, even at very low throttle and very high load--at 24 volts nominal supply (22-29v actual), and mostly OK (with some roughness in a narrow region of low throttle and very high load) at 36 volts nominal (35-42v actual). Lyen added a third shunt wire (like the EB318, 18-FET boards), and this improves performance with the Powerpack. But, I am still getting some rough running at 48 volts supply, over a fairly wide range of operating conditions that are likely to be encountered: 20-50% throttle, and moderate to high load.

Decreasing shunt resistance while keeping the programmed current limit the same by adjusting the values accordingly, by using the EB318 template or multiplying the desired current limit by 2/3 in the programming interface seems to be the needed strategy to get these controllers to work nicely with the Powerpack motor. The trick is to get it to drive the motor smoothly at high supply voltage (e.g 36 or 48 volts), low throttle setting, and high load. I am still waiting for Lyen to get back to me if it's O.K. to decrease the shunt resistance further so that we can get smooth operation at 48 volt supply over all throttle settings and loads, and get the 36 volt performance as smooth as the 24 volt.

I recommend using a current limit of 50-55A with this motor as the efficiency starts to drop off around 35A, and really drops quickly after 55A. I'm using 55A in the controller itself (50A on the CycleAnalyst) and 140A for the maximum phase current.

With the smaller BMC "300w" (single stack) motor, the wire is thinner gauge and will get hot enough to actually burn off the enamel at 60A. Efficiency on that one starts to decrease at 25A and really drops after 40A.

[mrbill]

With Lyen's blessing--(Please check with him before doing this to your own controller!)--I decreased the controller's shunt resistance further by adding solder blobs to each end of the original two-wire shunt. I now get smooth running of the motor across the performance envelope (all throttle settings and loads) at 36 volts supply. At 48 volts supply I get usable behavior at partial throttle, but at very low throttle settings and high loads, I still get some roughness (a surging and pulling back at roughly 2-3Hz frequency).

Here's the history:

Original shunt measured approximately 2.6 mOhms, as measured on the CycleAnalyst.
Original shunt with additional shunt wire added to backside of circuit board measured 1.616 mOhms on the CycleAnalyst.

http://mrbill.homeip.net/albums/infineo ... age_8.html

MrBill's solder blobs added to each side of the original shunt further reduced the resistance to 1.198 mOhms on the CycleAnalyst.

http://mrbill.homeip.net/albums/infineo ... age_7.html

If I were to run frequently on 48-volt batteries I would probably go in and increase the size of the solder blob to reduce the resistance further. I'm thinking that about 1 mOhm on the shunt would probably be sufficient.

After each of these modifications I discovered the RShunt setting on the CA by running a known current through the shunt and comparing with a calibrated DVM, adjusting the RShunt value until the currents on both instruments matched. During programming I tested the controller with motor under load to make sure that the programmed current limit produced my intended actual current limit.

For example, after my modification I found that to see an actual current limit of 55 Amps on the CycleAnalyst I had to program in a current limit of 33A using the EB318 (18-FET) template. This was found iteratively by placing an increasing load on the motor while noting the maximum sustained current observed on the CA. If the observed current was higher than 55A, then the programmed limit was reduced. If the observed current was lower than 55A, then the programmed current limit was increased. This took about three iterations.

Note that my controller has an EB312 (12-FET) board. If I had been using the EB312 template I would probably use a programmed current limit in the low-20A range to see an actual 55A limit. (I didn't use the EB312 template, so I don't have a precise figure.) Phase current limit was also reduced accordingly.

One question I still have is whether this hardware modification places the controller at increased risk of failure after many hours of use, or if there is some unlikely but unavoidable operating condition that could lead to failure.

[NeilP]

I know nothing about these motors, but I do know that increasing block time above a second or so is bad for the controller.
Good for performance in the short term, but kills controllers.

Block time is the length of time before the current limit kicks in . So if you set block time to say 5 seconds, then the controller is running at an unlimited current limit for 5 seconds, so if the motor demands it, you could draw, 100 200 amps or more in those 5 seconds.
This is motor dependant of course, so if your motor setup is not a current hog like the Xlyte 53xx series motors, then a high Block time maybe OK, and give you good acceleration. But for the good of the controller, I'd say at least leave block time low.

[mrbill]

I know nothing about these motors, but I do know that increasing block time above a second or so is bad for the controller.
Good for performance in the short term, but kills controllers.

Block time is the length of time before the current limit kicks in . So if you set block time to say 5 seconds, then the controller is running at an unlimited current limit for 5 seconds, so if the motor demands it, you could draw, 100 200 amps or more in those 5 seconds.
This is motor dependant of course, so if your motor setup is not a current hog like the Xlyte 53xx series motors, then a high Block time maybe OK, and give you good acceleration. But for the good of the controller, I'd say at least leave block time low.

I didn't mention block time in my postings as it has nothing to do with the problems people have running the Powerpack (aka BMC scooter or non-hub) motors or (I am told) the geared BMC hub motors that use the same internal winding pattern (dLRK, 16 poles, 18 teeth).

Block time is the time during which the controller current is NOT limited, the current limit being "blocked".

Having a short block time of a few seconds can give a single-speed hub motor (DD or geared) that little extra kick of acceleration until the motor gets up to speed. On a mid-drive that powers the bike through the gears as one would have when using a Powerpack motor, you can set block time to its minimum value without noticeably affecting performance. I set mine to 0.1 seconds, the smallest value I can select in the programming interface.

But, thanks for adding the caution that setting a finite and significant block time can risk damaging the controller, especially if the operator likes neck-snapping starts.

[NeilP]

Yes you did, mention Block Time.. your first post.

In particular, I'd like to get good starting values for Block Time, a ratio for Phase Current to Rated Current, and advice on setting one of the three speeds to 120% or some value >100%. I have the three-speed switch.

.

I am fully aware of the fun large block times can give on DD hubs too

[mrbill]

Yes you did, mention Block Time.. your first post.

In particular, I'd like to get good starting values for Block Time, a ratio for Phase Current to Rated Current, and advice on setting one of the three speeds to 120% or some value >100%. I have the three-speed switch.

.

I should have written "last posting". I discovered since my first posting that adjusting Block Time had no effect on how smoothly the Powerpack motor ran.

But, your reposting my initial query reminded me that I have come to a conclusion about these throttle limits.

I'm not exactly sure what the controller is doing to achieve a >99% throttle, but I can observe its effects.

I tried setting a throttle limit >99% and discovered that doing so leads to unstable performance, at least with the Powerpack motor. If the limit for a particular switch setting was above 99%, advancing the throttle through the 99/100% transition would cause a sudden change in motor behavior. At no load the motor RPM would suddenly increase, about 50% by the tone. But, the freespin current would triple.

At higher load the motor RPM would increase but the difference would be less as the load was increased. At very high load, I could coax about 5% more usable power out of the motor at 115% throttle.

The behavior difference at the threshold under no load is less for higher supply voltages. E.g. at 24 volts system voltage the discontinuity is sudden and jarring. At 48 volts system voltage, the shift to >99% throttle is barely noticeable.

I did encounter one problem, a condition that led to failure. While I was running a test to characterize motor/controller efficiency and had the throttle set to 115%, I used the cruise control to hold the throttle. Then, putting the motor under high load for roughly 20-30 seconds, the controller suddenly stopped working. After much debugging (and sending my controller and cruise control back to Lyen) we discovered that the cruise control was fried. Fortunately, the controller was O.K.

So, I have concluded that using >99% throttle is not safe for the controller or its components as it enables a failure-prone operating condition. I also don't like the discontinuous behavior between 99% and 100% throttle as I run my systems at 24 volts.

Because it does allow one to get a little more power out of a motor a maximum throttle, sort of like a very mild afterburner, I can see why someone might like to use it. But, I would recommend having the cruise control NOT engaged when flipping the speed switch to the setting that gives this extra throttle limit.

Lastly, I saw no difference between having a max throttle limit of 100, 105, 110, or 115%. They all seemed to give the same behavior at maximum throttle. If you don't want this "afterburner" setting, then set your throttle maximums no higher than 99%, not 100%.

[John in CR]

Bill,

From my understanding people have had trouble over the years matching BMC motors with controllers. It's been speculated that the issue is one of advanced timing placement of the hall sensors. You've called it a powerpack motor, but what motor is it exactly, because I see some different MAC motors at EVdeals? You should contact EVdeals, which seems to know some stuff since he mods motors, and maybe the manufacturer too. Hopefully the motor is neutrally timed, so the normal ebike controllers will work with it, since I don't think you're getting beyond the electrical rpms these controllers can handle. If the motors have some kind of advanced time in the halls, then I'm sorry to say but you may need Burtie's timing adjustment board to get the motor running optimally, or it could require a proprietary controller from the manufacturer.

I'm confused as to why you modded the shunt. Were you trying to get beyond the program settings available? Typically shunt modification is only done when you don't have a programmable controller, or if you need to fool the controller about how much current is allowed like I've done to increase regen braking force.

I'm also confused by your statements about a CA measuring the new shunt resistance, because a CA can't measure shunt resistance. In fact, unless you have the CA model with it's own shunt, you have to set the shunt resistance of the controller in the CA for it to give valid data about current. If you mean that you recalibrated the CA after you modified the shunt then all is good. Note that if you calibrate the CA by measuring actual no-load current from the battery and adjust the shunt resistance input until the CA reports the same as measured current, then to get it exact you have to account for the small current used to power the controller, which is a fixed amount, not variable like what runs through the shunt.

FWIW you divide the number of magnets by 2 to get the number of poles, and the tooth count is commonly referred to as the slots, so your motor is an 8 pole 18 slot motor. I bring this up in case you want to dig into some research, because the manufacturer may have done some special things with the windings that are possible with 8 pole 18 slot motors.

Have you tried running the motor with no accessories connected like the CA, cruise control, etc? It's possible they could be creating your issues. Personally I don't use the speed or current limit functions available in the CA, so I don't connect the CA wire for the throttle. I use the CA only as a measurement tool and don't want it ever overriding my own input. The CA does this only through the throttle wire.

You also mention efficiency. How are you determining that? To know efficiency you need to know output, but our tools like the CA only give us input along with speed and distance results. Without a dyno efficiency comparisons can only be made over the long haul by comparing consumption at different settings, and creating identical loads is virtually impossible due to outside influences unless you're using some kind of pony brake in lieu of a dyno.

You should read MWKeefer's posts here http://endless-sphere.com/forums/viewtopic.php?f=2&t=18675#p272438. The motor inside the geared hubbies he runs is a lot like the motor you have (assuming yours is neutrally timed). Your phase/battery current limit ratio seems out of whack. I think you should start over on your program settings and do like he does with a new motor and start with a 1:1 ratio and go from there. Especially now that you've modded the shunt so much you'll want to be careful with Block Time to make sure your current overshoot isn't out of hand by going too high for the controller and/or motor, but it's a valid tool otherwise in tuning your controller for optimum performance.

Regarding the speed switch settings, I'm not sure what you're trying to accomplish. I don't consider any above 100% to be valid. While they may increase top speed for some motor/controller combos, there's definitely a cost in the form of decreased efficiency and more heat in the motor and controller. There are better ways to increase max speed. I only use the speed limits for 2 things, since below 100% all it does is apply that ratio to the throttle voltage. I use it to limit the max speed for people to test ride my bikes, since they're too fast for inexperienced riders. I also use a low % like 50-60% to tame a twitchy throttle response on high powered bikes.

You asked about long term durability. Once you get it running smooth in all circumstances (jerks or pulses of power or other symptoms you mention are unacceptable), then as long as the controller isn't getting hot it should be durable in the long-term. What you'll want to be most careful of is partial throttle or modulating your throttle going up hills. That's how phase currents can get out of hand and fry a controller. This is another instance where a much lower ratio of phase current limit to battery current limit will prove beneficial for you, and since you have lots of hills keeping Block Time at the lowest setting is probably a good idea.

The last thing I want to bring up is regen. Since you use brakes so much that you're worried about rim temps, then you really need regen. If you ride your ebike always under power and not sometimes pedal only, then you will absolutely love regen braking. It's a tremendous safety factor coming down hills, because brake fade is non-existent so your mechanical brakes are always fresh and cool and ready for emergency stops. It also greatly reduces brake maintenance. Start with the lowest setting which is 0, and if the braking forces is still too strong you'll need to modify your shunt in the other direction to lower the actual regen current limit. On my extreme power ebikes (one at 12kw and the other now at 20kw), I need all the regen force I can take to help get these bikes stopped, but on my moderate power ebikes I like a nice gentle regen force. It takes a bit of time to learn how far ahead of a turn or stop to activate regen, but it's silky smooth, and the light force works great coming down hills to keep speeds down without using the regular brakes.

Whatever you learn about your motor, please share, since they seem to be good quality motors, but controller matching has been an issue that keeps many away from using them.

John

[mrbill]

From my understanding people have had trouble over the years matching BMC motors with controllers. It's been speculated that the issue is one of advanced timing placement of the hall sensors. You've called it a powerpack motor, but what motor is it exactly, because I see some different MAC motors at EVdeals? You should contact EVdeals, which seems to know some stuff since he mods motors, and maybe the manufacturer too. Hopefully the motor is neutrally timed, so the normal ebike controllers will work with it, since I don't think you're getting beyond the electrical rpms these controllers can handle. If the motors have some kind of advanced time in the halls, then I'm sorry to say but you may need Burtie's timing adjustment board to get the motor running optimally, or it could require a proprietary controller from the manufacturer.

John:

Thanks for your detailed posting.

I get these "BMC"'/Powerpack motors as discards from a local bike shop that has a service arrangement with another shop that rents out ebikes with Currie/US Pro Drive/Synergy/etc. drive systems. The rental bikes are frequently abused, so there is a steady stream of burned-out controllers where the motor is in fine shape and only the controller is fried. These motors have a built-in controller that I remove and convert to a standalone motor. It is my understanding that the "600w" version of these BMC motors uses the same motor core as the Powerpack motor (Tim O'Brien's trade name). I have also played around with the smaller "300w" BMC motor that is identical to the "600w" motor in form factor and faceplate mounting but is half the height and on each tooth uses 15 turns of thinner wire instead of the 9 turns of thicker wire on the "600w" motor.

After cleaning out the non-working internal controller, I install my own Hall sensors positioned on the stator to give minimum current draw when the motor is spinning under no load at maximum throttle, or about 2 Amps at 24 volts supply. Most of my motors are set up for CW spinning and for the Infineon family of controllers. One of my motors is set up with two sets of Halls, one for the Infineon, the other for the Headline controller that I was also using with a Headline/Cyclone motor for a while.

You can see all of this in the photos at the links below.

Photos of "600w" BMC motor:
http://mrbill.homeip.net/albums/mac_bmc ... index.html

Photos of "300w" BMC motor:
http://mrbill.homeip.net/albums/mac_bmc ... index.html

You might notice that I position my Hall sensors slightly offset from the stock Hall sensors that are found in the slots between the stator teeth. The stock Hall sensor position did not result in the most efficient operation with either Headline or Infineon controllers and may be one reason why people have a hard time running this motor well. I recommend running your own Halls if you really want to tune it up.

I'm confused as to why you modded the shunt. Were you trying to get beyond the program settings available? Typically shunt modification is only done when you don't have a programmable controller, or if you need to fool the controller about how much current is allowed like I've done to increase regen braking force.

I believe that the BMC motor presents a tougher peak load than most motors, requiring higher peak current under partial-throttle, high-load conditions. I was not able to achieve smooth operation with the Infineon controller with these motors at partial-throttle, moderate/high load with system voltages higher than 24 volts nominal with the stock shunt on a 12-FET (EB312 board) controller without physically modifying the on-board shunt by decreasing its resistance.

The controller clearly has some hardware peak current limit that depends only on the resistance of the shunt not on its programming, and believe me, I tried modifying all of the programming parameters (Block Time, Phase Current limit, overall limit). Only by reducing phase current limit significantly was I able to reduce or eliminate the rough running, but it also made the motor quite gut-less when running at partial throttle. It would bog down very quickly at partial throttle.

After making this hardware modification it was then necessary to adjust the programmed current limits proportionally so that the original desired current limits were retained.

I made these changes to the shunt with Lyen's blessing--he had already added a third shunt wire to the board--after he consulted with his source for these controllers.

I'm also confused by your statements about a CA measuring the new shunt resistance, because a CA can't measure shunt resistance. In fact, unless you have the CA model with it's own shunt, you have to set the shunt resistance of the controller in the CA for it to give valid data about current. If you mean that you recalibrated the CA after you modified the shunt then all is good. Note that if you calibrate the CA by measuring actual no-load current from the battery and adjust the shunt resistance input until the CA reports the same as measured current, then to get it exact you have to account for the small current used to power the controller, which is a fixed amount, not variable like what runs through the shunt.

I have a power supply that into a dummy resistive load can supply a constant current, say 8 Amps. (It takes a minute or two for the large power resistors to settle down to a constant temperature before the current is stable.) I can measure this current with relatively high accuracy with an in-line DVM (not a clamp meter) set to measure DC current. I can also observe the current displayed on the CA and then adjust RShunt on the CA until the current displayed on the CA matches that of the DVM. It should not be necessary to account for the quiescent current of the controller (about 80mA) as that is included in the reading on the DVM. I want the CA to read the total current being drawn from the supply, including the residual load of the controller circuitry and the CA itself.

FWIW you divide the number of magnets by 2 to get the number of poles, and the tooth count is commonly referred to as the slots, so your motor is an 8 pole 18 slot motor. I bring this up in case you want to dig into some research, because the manufacturer may have done some special things with the windings that are possible with 8 pole 18 slot motors.

I'm using nomenclature consistent with this chart:
http://forumrc.alexba.eu/nutpol_e.htm

from this web site:

http://www.bavaria-direct.co.za/models/motor_info.htm

The box on the chart corresponding to "16 poles" and "18 teeth" gives the exact winding of the BMC "600w" and "300w"motor.

Have you tried running the motor with no accessories connected like the CA, cruise control, etc? It's possible they could be creating your issues. Personally I don't use the speed or current limit functions available in the CA, so I don't connect the CA wire for the throttle. I use the CA only as a measurement tool and don't want it ever overriding my own input. The CA does this only through the throttle wire.

Yes, I have tested the controller/motor without the cruise control attached. The problem still exists (or existed, as I think I've fixed it now).

I set the controller current limit to a figure that is just above what I feel is reasonable with these motors. Trying to push more than 55-60A through these motors results in more heat than motion, so I set the controller limit to 55A. On the CA I set the limit to 50A as I feel that is enough for my enjoyment, and I don't want to be enticed into running wastefully. My recommended limits are lower for the smaller "300w" motor. I listed them in a prior message.

You also mention efficiency. How are you determining that? To know efficiency you need to know output, but our tools like the CA only give us input along with speed and distance results. Without a dyno efficiency comparisons can only be made over the long haul by comparing consumption at different settings, and creating identical loads is virtually impossible due to outside influences unless you're using some kind of pony brake in lieu of a dyno.

To measure power output at the rear wheel I use a PowerTap rear hub with a rim brake (my "pony brake") to generate the torque. It works reasonably well.

http://www.cycleops.com/products/power-meters.html

I calculate efficiency by measuring power "out" with a PowerTap rear hub and dividing by power "in" with a calibrated CycleAnalyst. Absolute accuracy is around 3-5%, but relative accuracy is better. So using the same equipment and set-up I can easily discover changes that improve efficiency and changes that do not.

Unfortunately, I have been unable to test the efficiency after the most recent shunt modifications because I'm getting too much RFI from the motor/controller that is interfering with the PowerTap signal. But, I don't expect the full-throttle efficiency to change nor do I expect much change for the partial-throttle tests.

I have tested my BMC "600w" and "300w" motors here:

http://mrbill.homeip.net/albums/mac_bmc ... ta_Inf.pdf
http://mrbill.homeip.net/albums/mac_bmc ... lta_HL.pdf

and even using a Wye connection for the motor phases:

http://mrbill.homeip.net/albums/mac_bmc ... ye_Inf.pdf

Other motors tested:

http://mrbill.homeip.net/hybridBike.php ... encyCurves

You should read MWKeefer's posts here http://endless-sphere.com/forums/viewtopic.php?f=2&t=18675#p272438. The motor inside the geared hubbies he runs is a lot like the motor you have (assuming yours is neutrally timed). Your phase/battery current limit ratio seems out of whack. I think you should start over on your program settings and do like he does with a new motor and start with a 1:1 ratio and go from there. Especially now that you've modded the shunt so much you'll want to be careful with Block Time to make sure your current overshoot isn't out of hand by going too high for the controller and/or motor, but it's a valid tool otherwise in tuning your controller for optimum performance.

Thanks. I did read that thread some time ago. Most or all of the advice is about tuning the controller for hub motors rather than mid-drive or crank-drive motors, but upon re-reading I think I might experiment with a lower phase current to overall current ratio. Right now I'm using 2.5x for phase current limit (about 140A), and that seems to provide plenty of pull at partial throttle. My memory of reducing the phase current was that it reduced the torque too much at partial throttle, leading to a gut-less feel when, say, riding up a hill with a varying grade. As the hill steepened momentarily, the motor would bog down more easily with a lower phase current. At 2.5x, the motor power would increase to compensate for the increased load, keeping the bike speed constant, or nearly so. Still, I might try experimenting with it again.

I use the shortest Block Time available, 0.1 seconds. Since I'm running through the gears, I don't ever want to have the motor pulling with "infinite" torque. I'd choose 0 seconds if I could. I occasionally and accidentally "slam" the throttle when there's backlash in the drivetrain, and I'd rather not abuse my components this way.

Regarding the speed switch settings, I'm not sure what you're trying to accomplish. I don't consider any above 100% to be valid. While they may increase top speed for some motor/controller combos, there's definitely a cost in the form of decreased efficiency and more heat in the motor and controller. There are better ways to increase max speed. I only use the speed limits for 2 things, since below 100% all it does is apply that ratio to the throttle voltage. I use it to limit the max speed for people to test ride my bikes, since they're too fast for inexperienced riders. I also use a low % like 50-60% to tame a twitchy throttle response on high powered bikes.

My original thought was to see how I could use throttle limits >99%. Maybe a could have a "turbo" option for those rare occasions when I need a quicker boost off the line. In the end I decided it wasn't worth the trouble or risk.

I agree with you re: >99% speed limit settings. i.e. Set limits no higher than 99%. I use the switch to set the limits lower so that I can ride more easily with un-assisted bikers or if I hook up a higher voltage battery. I normally run at 24 volts nominal system voltage.

I did find that at 115% throttle I was able to coax a little more power out of the motor before it stalled. I also measured a slight efficiency improvement at high output power (top 1/3 of its power range), but there is a price. I didn't published the results on my website because I don't believe this operating region is safe for this controller and motor combination. The controller shuts down after a number of seconds just below the stall load--gave me a scare the first time, and while I was running a test under this condition it blew out my cruise control. I had been using the cruise control to "hold" the throttle, freeing my hands for other tasks.

You asked about long term durability. Once you get it running smooth in all circumstances (jerks or pulses of power or other symptoms you mention are unacceptable), then as long as the controller isn't getting hot it should be durable in the long-term. What you'll want to be most careful of is partial throttle or modulating your throttle going up hills. That's how phase currents can get out of hand and fry a controller. This is another instance where a much lower ratio of phase current limit to battery current limit will prove beneficial for you, and since you have lots of hills keeping Block Time at the lowest setting is probably a good idea.

I agree that smooth running without hiccups or untoward noises is a good sign. I've already got Block Time as low as it will go. I will take your suggestion and see if I can reduce the phase current so that performance is still acceptable, especially if doing so reduces the likelihood of damaging the controller under unusual operating conditions.

I find that the motor gets warm in proportion the power it is called upon to produce (or transform). The controller gets warm the higher the power as well, but it also gets warmer if power is held constant while the throttle is reduced. In practice when there's a little breeze on it from forward motion neither the motor nor controller get more than warm. Even after slowly climbing a steep hill on a hot day, the motor and controller are not so hot that one can't grasp either firmly with a hand. Bench testing is always hardest on heat-generating parts because there's no air movement around the parts that get hot. Sometimes when I'm testing it helps to set up a fan to blow on the motor and controller.

The last thing I want to bring up is regen. Since you use brakes so much that you're worried about rim temps, then you really need regen. If you ride your ebike always under power and not sometimes pedal only, then you will absolutely love regen braking. It's a tremendous safety factor coming down hills, because brake fade is non-existent so your mechanical brakes are always fresh and cool and ready for emergency stops. It also greatly reduces brake maintenance. Start with the lowest setting which is 0, and if the braking forces is still too strong you'll need to modify your shunt in the other direction to lower the actual regen current limit. On my extreme power ebikes (one at 12kw and the other now at 20kw), I need all the regen force I can take to help get these bikes stopped, but on my moderate power ebikes I like a nice gentle regen force. It takes a bit of time to learn how far ahead of a turn or stop to activate regen, but it's silky smooth, and the light force works great coming down hills to keep speeds down without using the regular brakes.

I've looked into regen a number of times. It isn't practical on my bikes for a number of reasons.

1) I'm using a through-gears mid-drive or crank-drive, so I can't put braking force through the drivetrain. It would require braking tension be transferred through the "return" chain run, through a derailleur cage that isn't designed to handle a tense chain. It would also require an unavailable locking freehub that can be locked remotely by a cable connected to a lever on my handlebar so that my rear wheel cluster doesn't freewheel when braking is engaged. Too complicated.

2) I could add a second motor, say a DD hub motor the front wheel and a second controller. But, then I have to live with the added drag when I'm not braking and the added weight of the additional hardware. My batteries would also need to be able to handle the substantial reverse current. I often ride with more than one battery in parallel connected through Schottkys, so that would block any regen current. In the end it's not practical.

If braking and not recapturing energy is the objective, I'd probably do better with an aerodynamic brake, a parachute or drag flaps, but that also requires additional hardware, levers for deployment and complexity.

I've settled on using a large disk brake on the rear of one of my bikes, and on the other a rim brake with a sturdy, wire-bead and well-fitted tire that has no bead damage from too many installations/removals. An easy and relatively cheap. I haven't had any blow-offs since I started this protocol. When I replace my rims I will choose wider, heavier rims with more metal, preferably with an "aero" profile as these dissipate heat better than box cross-section rims.

Whatever you learn about your motor, please share, since they seem to be good quality motors, but controller matching has been an issue that keeps many away from using them.

These are excellent motors for the price if you're willing to get your hands dirty and clean them up for your use. So far they're the most efficient motors that I've tested, with a conveniently broad efficiency peak. They were, after all, designed for single-speed systems (Currie, US-ProDrive, etc.).

One advantage as I see it is they do not use rare earth magnets, just a composite magnet ring held closely to the stator teeth, so there is less danger of overheating the motor and destroying the magnets. I did overheat one of the smaller 300w motors once by pumping 60A through the coils, and the result was cooked enamel on the wire. The motor still worked fine afterward, but the wire looked burnt.

The only thing that could be improved is the ability of the motor to dissipate heat better. As it's an out-runner all heat generated in the coils is transferred through the central core to the face plate. That may be one reason the internal controllers fried easily on these motors as they are mounted and potted on the inside of the faceplate.

The heatsink fins on the cover rarely get hot enough to do any good as the cover has little metal in contact with the faceplate. Fortunately, if it's set up right the motor runs efficiently, so it doesn't generate too much heat in the first place.

[John in CR]

Bill,

Wow, it sounds like you know more about these motors than anyone except the manufacturer. You might want to play around with the ebikes.ca simulator at http://ebikes.ca/simulator/. They have some BMC motors listed, though I don't know if they're the same as yours, but they're probably similar, and the simulator reflects the lower efficiency of the non-Neo magnets. I played around with the V2 motor a bit, and the simulator shows that at 100% throttle the power and efficiency curves look great at various voltages and current limits, but below about 80% throttle the curves go to crap.

Something you might be interested in trying is to give yourself variable current limits on the fly, so you can spend more time at full throttle but change the speed range with different current limits. Take a look at this thread http://endless-sphere.com/forums/viewtopic.php?f=2&t=37068&hilit=controlled+shunt+modification and BigMoose's posts here http://endless-sphere.com/forums/viewtopic.php?f=2&t=31643&p=458366&hilit=controlled+shunt+modification#p494749.

I understand you have a lot invested in these motors, but for me motors with efficiency in the 70's don't cut it anymore. While most of my hubmotors are the common design that has peak efficiency around 85%, a newer one I have has peak efficiency in high speed mode of 93%, and a bit lower in low speed. The speed switching is accomplished via a switch of half of the strands in each phase's windings. For high they are connected in parallel, and in low they are connected in series thus doubling the turn count. The result is the unique combination of a billygoat mountain climber and a high speed motor for flatter terrain. Unfortunately the motor is designed for scooters, and too heavy and powerful for your needs.

Take care,

John

[mrbill]

Wow, it sounds like you know more about these motors than anyone except the manufacturer. You might want to play around with the ebikes.ca simulator at http://ebikes.ca/simulator/. They have some BMC motors listed, though I don't know if they're the same as yours, but they're probably similar, and the simulator reflects the lower efficiency of the non-Neo magnets. I played around with the V2 motor a bit, and the simulator shows that at 100% throttle the power and efficiency curves look great at various voltages and current limits, but below about 80% throttle the curves go to crap.

I have played around with the simulator when evaluating hub motors. But, the Powerpack motor (or equivalent re-worked "600w" or "300w" motors) isn't included in their list. I suppose someone with time on their hands could take my data and come up with good estimates for the motor constants and plug them into the simulator.

I did test this motor with rotors where the composite magnet ring had been replaced by Neo magnets. What I found was that efficiency was not improved. At best the curve was shifted slightly to the right, meaning that the peak efficiency occurred at a slightly higher output power. Kv of the motor was reduced slightly, too.

Because I use the motor in a geared setup I did not feel the tradeoff was worthwhile as I could always gear the motor to be spinning fast enough to keep it at its most efficient. With the N42 magnets I had to use a press to install and remove the rotor--those magnets are strong! And, there was increased risk of overheating the motor. But, if I had a single-speed setup that had me pulling heavy loads from a stop frequently, then the Neo magnets might help overall.

http://mrbill.homeip.net/albums/mac_bmc ... ta_Inf.pdf
http://mrbill.homeip.net/albums/mac_bmc ... lta_HL.pdf

also

http://mrbill.homeip.net/albums/mac_bmc ... lta_HL.pdf - stock magnets
http://mrbill.homeip.net/albums/mac_bmc ... HL_N42.pdf - N42 magnets

I don't think one can as a rule swap out for stronger magnets in a motor and expect an automatic overall improvement. The motor is designed for a certain strength magnet.

A more useful substitution may be to rewind the stator with thicker wire with the same number of turns. Doing so would require positioning the Halls at the base so the leads don't need to be threaded up between the coils. This would require pulling the stator off the core mandrel, then pressing it back on again afterward. I've thought of trying this at some point, but I'm pretty happy with the motor as is.

I understand you have a lot invested in these motors, but for me motors with efficiency in the 70's don't cut it anymore. While most of my hubmotors are the common design that has peak efficiency around 85%, a newer one I have has peak efficiency in high speed mode of 93%, and a bit lower in low speed. The speed switching is accomplished via a switch of half of the strands in each phase's windings. For high they are connected in parallel, and in low they are connected in series thus doubling the turn count. The result is the unique combination of a billygoat mountain climber and a high speed motor for flatter terrain. Unfortunately the motor is designed for scooters, and too heavy and powerful for your needs.

Which motor do you have that gives 93% efficiency? Geared or direct-drive? The only motors I've seen that claim >90% efficiency have a very high Kv.

The problem is if you want high power, low RPM, and high efficiency you need a big motor. The BMC "Powerpack" style motor fits nicely on a bike and isn't too heavy. I'm sure I could do 5-10% better on efficiency with something weighing 3-8x as much, but then the bike would be that much heavier. And, I rarely use more than 1hp out on my bikes. In any case regularly pulling more than 1hp puts too much stress on the drivetrain and frame.

Be careful to read the comments at the bottom of my graphs. My efficiency measurements include the effect of all components from controller power input to the rear wheel, which includes the drive train. My drive train includes two chain & sprocket links.

So, if we're looking at just the motor, then we have to back out the loss of each chain & sprocket link (3%) and the controller itself (2-4%). So, if we look at an efficiency of 75% on the chart, we need to divide by roughly (0.97)^3 to get the motor efficiency itself. That would be 0.75/(0.97)^3 = 82%.

Just for reference and comparison to hub motors, I did test a StokeMonkey kit. This one uses a 400-series Crystalyte DD hub motor. Using the same instruments I observed peak efficiency in the low-70% range, suggesting that were I to back out the losses of controller and drivetrain I'd see about 79-80% for the motor alone.

http://mrbill.homeip.net/albums/stokemo ... _stock.pdf

[John in CR]

The 93% peak efficiency motors I have are direct drive of Japanese design for a 2 speed scooter that didn't enjoy broad market success as it got left behind in the chase for much higher power for scooter. Plus it was on a 300lb lead sled, so when people forgot to take off in low with the even heavier load of a passenger aboard or up an incline, then they had heat problems, because the 2500W just wasn't enough power to get out of low efficiency operation quickly with a the load of 300lb of scooter and 200-300lb of people.

It's the best example I've run across of the Chinese struggling with thinking out of the box. They've got the motor all but mothballed and along comes me who could puts it on a 110lb ebike with a slightly higher voltage and double the current that the lower load enables and it's the best thing to come along for ebikes since lithium batteries. High speed capability and billy goat mountain climbing with half the heat generated of common ebike hubbies makes it the best thing available. It's only issue is that a Kv in high of 17rpm/volt requires a wheel smaller than 20". The 6.25" diameter once you strip the in-wheel mounting flange makes it great for mid-drive mounting, and my mid-drive bike is almost road-ready to prove it. Too bad I've been distracted with playing with Hubmonster and 20kw, which is just too fun not to be distracted. Mini-Monster will be available soon, I promise.

John