mrbill
10 kW
Last summer I acquired a small supply of non-working MAC-BMC motor cores in both the "600-watt" and "300-watt" variety. These motors were used in Currie, US-Pro Drive, Synergy, and other retrofit ebike kits over the last 10 years or so. They were originally built with internal controllers, but all of the motors I received had damaged and non-working controllers. The controllers being potted, were impossible to repair. About the best that could be done is to pound out the potting (and the controller components and board) using a a chisel, hammer, and heat gun, to make room for running new power and Hall phase wires for external control. As far as I can tell the core motor of the 600-watt version is functionally identical to the Powerpack motors sold by Tim O'Brien.
I have now had the opportunity to test these motors' efficiency at different supply voltages, throttle settings, and different controllers (Headline, Transmagnetics, and Infineon) and with different rotors (stock, N42 16-pole, and N42SH 20-pole). Testing these motors has given some unexpected results that I thought I'd post for comment on this forum.
My efficiency testing was conducted using a calibrated CycleAnalyst to measure energy drawn from the battery compared to energy produced at the rear wheel of a bicycle with a PowerTap power-measuring rear hub. The testing I conducted includes the effects of the controller, wiring, and two stages of chain and sprocket gear reduction (ANSI 25 and 40). My tests cannot directly be compared with other stand-alone motor dynamometer tests unless the effects of controller and gear reduction are backed out, and this can only be done approximately. Nevertheless, since I used the same test rig and measurement tools for all tests (except where noted), the tests can be compared between themselves.
http://bit.ly/gvxe7q
Observations:
Hall sensor positions must be fine-tuned for most efficient operation. In all cases I positioned the Hall sensors on the stator such that the free-spin current draw was minimized and smooth spinning without artifacts such as squealing, whining, knocking, or non-linear behavior as the throttle was advanced through its range. In no case did the stock Hall position give the best behavior. Furthermore optimal CW and CCW operation required different Hall positions. These motors are not bi-directional by swapping Hall or phase wires with the controllers I used.
I found that the optimum Hall position for CW operation using the Headline and Transmagnetics controllers was at the center of the tooth immediately to the right of the stock Hall position. Optimum efficiency when using the Infineon controller required positioning the Halls a few mm to the right or to the left of the stock Halls. Efficiency with these motors is quite sensitive to Hall position; being off by more than a couple of millimeters in either direction results in measurable loss.
Perhaps a more sophisticated controller can adjust dynamically the Hall signal phase so that the stock Halls can be used, but these relatively inexpensive controllers require precise (but different!) Hall positioning for best performance. The Headline and Transmagnetics controllers are designed to drive motors that use 12 stator teeth and 8 magnetic poles. When driving these motors direction can be changed from CW to CCW by swapping the appropriate Hall or power phase wires. The MAC-BMC motors appear to use 18 stator teeth, 16 poles and a distributed-LRK winding in Delta configuration.
Another poster on this forum found some of my earlier posts on these motors and contacted me about testing some custom rotors he had made for them using N42 and N42SH (Neodymium) magnets.
The first round of testing I did on the "300-watt" core, same diameter and winding pattern (though twice the number of wire coils on each stator tooth), and half the height of the stator. I tested a 20-pole rotor using N42SH (high-temperature) magnets that fit on the inner diameter of the rotor without leaving significant gaps between the magnets. Not unexpectedly, but rather inconveniently, I discovered that the 20-pole rotor required searching for another optimum Hall position. Once found, I tested the motor and discovered unexpectedly that the peak efficiency was no better than with the stock rotor, indeed, it was slightly worse. The N42 magnets were certainly stronger than the stock magnets. I had just enough strength in my hands to remove the rotor from the stator, while the stock rotor was easy to pull off with one hand. Using the N42 rotor I found that I could get more power out of the motor with good efficiency, but the peak efficiency was never as high as it was with the stock rotor. For an ungeared or under-geared motor, stronger magnets may help, but for a bike with gears, the stock rotor would actually give better efficiency provided the operator kept the motor running in its efficient range.
Compare these:
http://bit.ly/eIkQWM - stock 16-pole rotor
http://bit.ly/g50vOM - N42SH/20-pole rotor
Later I compared a 16-pole N42 rotor to the stock rotor on the larger, 600-watt motor:
http://bit.ly/hEhfza - using Headline controller
http://bit.ly/hElhjh - using Infineon controller
In both cases, the stock rotor exhibited higher efficiency over most or all of the usable power range at all three supply voltages: 24, 36, and 48. In the case of the 16-pole N42 rotor, I speculated that perhaps the gap between the magnets on the rotor gave rise to the efficiency loss. But, since there was still reduced peak efficiency with the more tightly-packed 20-pole rotor on the smaller motor, I'm not sure this explains all of the loss.
Questions:
1) Why don't the N42 magnet rotors give higher peak efficiency?
2) Are there controllers that can electrically offset Hall phasing so that optimal efficiency can be obtained with the stock Hall sensors in this and other brushless DC motors?
I have now had the opportunity to test these motors' efficiency at different supply voltages, throttle settings, and different controllers (Headline, Transmagnetics, and Infineon) and with different rotors (stock, N42 16-pole, and N42SH 20-pole). Testing these motors has given some unexpected results that I thought I'd post for comment on this forum.
My efficiency testing was conducted using a calibrated CycleAnalyst to measure energy drawn from the battery compared to energy produced at the rear wheel of a bicycle with a PowerTap power-measuring rear hub. The testing I conducted includes the effects of the controller, wiring, and two stages of chain and sprocket gear reduction (ANSI 25 and 40). My tests cannot directly be compared with other stand-alone motor dynamometer tests unless the effects of controller and gear reduction are backed out, and this can only be done approximately. Nevertheless, since I used the same test rig and measurement tools for all tests (except where noted), the tests can be compared between themselves.
http://bit.ly/gvxe7q
Observations:
Hall sensor positions must be fine-tuned for most efficient operation. In all cases I positioned the Hall sensors on the stator such that the free-spin current draw was minimized and smooth spinning without artifacts such as squealing, whining, knocking, or non-linear behavior as the throttle was advanced through its range. In no case did the stock Hall position give the best behavior. Furthermore optimal CW and CCW operation required different Hall positions. These motors are not bi-directional by swapping Hall or phase wires with the controllers I used.
I found that the optimum Hall position for CW operation using the Headline and Transmagnetics controllers was at the center of the tooth immediately to the right of the stock Hall position. Optimum efficiency when using the Infineon controller required positioning the Halls a few mm to the right or to the left of the stock Halls. Efficiency with these motors is quite sensitive to Hall position; being off by more than a couple of millimeters in either direction results in measurable loss.
Perhaps a more sophisticated controller can adjust dynamically the Hall signal phase so that the stock Halls can be used, but these relatively inexpensive controllers require precise (but different!) Hall positioning for best performance. The Headline and Transmagnetics controllers are designed to drive motors that use 12 stator teeth and 8 magnetic poles. When driving these motors direction can be changed from CW to CCW by swapping the appropriate Hall or power phase wires. The MAC-BMC motors appear to use 18 stator teeth, 16 poles and a distributed-LRK winding in Delta configuration.
Another poster on this forum found some of my earlier posts on these motors and contacted me about testing some custom rotors he had made for them using N42 and N42SH (Neodymium) magnets.
The first round of testing I did on the "300-watt" core, same diameter and winding pattern (though twice the number of wire coils on each stator tooth), and half the height of the stator. I tested a 20-pole rotor using N42SH (high-temperature) magnets that fit on the inner diameter of the rotor without leaving significant gaps between the magnets. Not unexpectedly, but rather inconveniently, I discovered that the 20-pole rotor required searching for another optimum Hall position. Once found, I tested the motor and discovered unexpectedly that the peak efficiency was no better than with the stock rotor, indeed, it was slightly worse. The N42 magnets were certainly stronger than the stock magnets. I had just enough strength in my hands to remove the rotor from the stator, while the stock rotor was easy to pull off with one hand. Using the N42 rotor I found that I could get more power out of the motor with good efficiency, but the peak efficiency was never as high as it was with the stock rotor. For an ungeared or under-geared motor, stronger magnets may help, but for a bike with gears, the stock rotor would actually give better efficiency provided the operator kept the motor running in its efficient range.
Compare these:
http://bit.ly/eIkQWM - stock 16-pole rotor
http://bit.ly/g50vOM - N42SH/20-pole rotor
Later I compared a 16-pole N42 rotor to the stock rotor on the larger, 600-watt motor:
http://bit.ly/hEhfza - using Headline controller
http://bit.ly/hElhjh - using Infineon controller
In both cases, the stock rotor exhibited higher efficiency over most or all of the usable power range at all three supply voltages: 24, 36, and 48. In the case of the 16-pole N42 rotor, I speculated that perhaps the gap between the magnets on the rotor gave rise to the efficiency loss. But, since there was still reduced peak efficiency with the more tightly-packed 20-pole rotor on the smaller motor, I'm not sure this explains all of the loss.
Questions:
1) Why don't the N42 magnet rotors give higher peak efficiency?
2) Are there controllers that can electrically offset Hall phasing so that optimal efficiency can be obtained with the stock Hall sensors in this and other brushless DC motors?