Miles' 125mm inrunner design
- Thread starter Miles
- Start date
Calculator for annular heat-sinks:
http://www.mhtl.uwaterloo.ca/NC_annular.html
Calculators for rectangular heat-sinks:
http://www.mhtl.uwaterloo.ca/NC_rect.html
http://students.cs.byu.edu/~jgoodell/heatsink1.html
http://www.novelconceptsinc.com/calculators-forced-convection-heat-sink-thermal-resistance.cgi
http://www.mhtl.uwaterloo.ca/NC_annular.html
Calculators for rectangular heat-sinks:
http://www.mhtl.uwaterloo.ca/NC_rect.html
http://students.cs.byu.edu/~jgoodell/heatsink1.html
http://www.novelconceptsinc.com/calculators-forced-convection-heat-sink-thermal-resistance.cgi
The stator OD is 130mm. The stator width is 17mm (minus copper wires).
Thanks Spinning!
It's an outrunner running inside the hub casing? Right?
Another comparison to make is that with the GNG motor.
Both are inrunners with similar stator O.D. and active length.
GNG motor specs:
Weight 2.63 kg
12 slots 8 poles
Winding factor 0.866
Cogging: 24 step per rev.
Rm 73 milliohms (phase to phase)
Kv 67 rpm/V
Kt 0.143 Nm/A
Km 0.53 Nm/√W
Specific Km 0.20 (Nm/√W)/kg
Rotor OD 79mm
Stator OD 119mm
Stator ID 79.8mm
Stator thickness 25mm
Lamination thickness 0.5mm
Magnet thickness 4.2mm
It's an outrunner running inside the hub casing? Right?
Another comparison to make is that with the GNG motor.
Both are inrunners with similar stator O.D. and active length.
GNG motor specs:
Weight 2.63 kg
12 slots 8 poles
Winding factor 0.866
Cogging: 24 step per rev.
Rm 73 milliohms (phase to phase)
Kv 67 rpm/V
Kt 0.143 Nm/A
Km 0.53 Nm/√W
Specific Km 0.20 (Nm/√W)/kg
Rotor OD 79mm
Stator OD 119mm
Stator ID 79.8mm
Stator thickness 25mm
Lamination thickness 0.5mm
Magnet thickness 4.2mm
Yes, it is an outrunner inside an outer case. That's why it has such a poor heat-shedding path, only the axle slowly moves heat away from the stator-core. Of course the lams are steel, but (at least) the stator-core is aluminum. But...even though the core is aluminum, it is only a small heat sponge, with a poor heat bleed-off ability.
If I had more time and a shop, I'd make an de-shelled BPM core into a mid-drive. Might be a fun and inexpensive experiment...
If I had more time and a shop, I'd make an de-shelled BPM core into a mid-drive. Might be a fun and inexpensive experiment...
Hi Miles, this is very good news! I have kept coming back to this design for any new updates. I have really enjoyed watching the development of the ‘Miles 90’ but to me the 1-2Kw motor niche is already saturated with much inferior, but dirt cheap competition.
I apologise for the following selfish thread theft.
I think a quality motor like this is much in need!
I was more interested in the Miles 125 (but perhaps with a longer stator), a larger version of the Miles 90 motor, for potential very high performance off road use. Something to give IC motored enduro bikes a run for their money, at least on tight forest single type tracks. If we take the approach to design a motor that can always produce enough torque to lift the front wheel in the air on demand at most speeds, then we can maximise weight transfer to the rear wheel, increasing rear grip and drive out of slippery dirt corners and have the ability to manual over logs, ruts and whoops etc., like most IC engine bikes can. And if we don’t achieve this, electric bikes will always lose.
A 40kg downhill ebike has a very high centre of gravity because it has 8-9 inches of suspension travel. With a 26” rear wheel it can only put down about 200Nm-250Nm at the back tyre before it starts to flip (the design requirement). (Maybe Toolman2 can have some input here as he has done a lot of tuning with his Kelly controller for acceleration/wheelies/wheel spin verses motor phase amps on and off road with his awesome CA120 powered bike). This sort of torque at the rear is probably "just" within the mechanical limits of a quality rear downhill wheel and spokes and tyre if we use tyre bead locks etc.
Alternative motors:
The $1200-$1800 US Joby motor is now made in China which is still OK but the last one ordered took a veeery..... long time to get here (toolman2??). They are designed for air craft use so are of large diameter (getting harder to fit), open construction and need to be sealed up to reduce gravel consumption. Hall sensors need to be installed (a biggish job) and once sealed up, could probably do with filtered force fed air to keep it all cool again and to get it close to the quoted output figures.
The Plettenburg Nova series looks real promising but at around 3 grand aussie for the water cooled version, it is starting to get too expensive and still may not actually exist in real life (does anyone have one?). Anyway I cannot afford to find out.
CA120, at only 2.2kg, is dyno proven to whack out the power (more than a 12kg cro motor at the wheel), but has some design issues: stator tooth design, sub-standard copper fill, halls needed, large heavy bell bearing, magnet heating issues as discussed elsewhere, unsealed, excessive .35mm thick stator laminations for the high frequencies (1200Hz!) that we run.
As a rough calculation if we gear for around 70km/hr off road (the last 10km/hr of acceleration always being a lot slower), the back tyre needs about 550RPM loaded. Using 14T front and the maximum available 100T(105T may be available) rear sprocket (to reduce chain tension loads), we get about 7.14:1 reduction and that means we need around 28Nm at the motor shaft and 4000rpm loaded (so needs thinner lams). This is about the maximum tension that the best quality 219” kart chain can take (O-ring chain is also available for off road longevity). There is also a very large selection of cheap off the shelf standard mount up sprockets, keeping it simple, light and cheap to maintain and fine tune for different circumstances. A motor like this would fill the need of most die-hard performance enthusiasts, and would hopefully be under 3kg.
All motorcar manufacturers building EV motors from scratch, to maximize power density, cool their motors with coolant or oil. You just need so much more air than fluid to do the equivalent job - it starts getting noisy from the airflow and fans required. And you cannot ensure IP65 type sealing. Just trying to think outside of the square, with a .7mm air gap, could I filament wrap or wind a fibre/epoxy sleeve and then cylindrically grind it on the mandrel to a .3mm thick fluid-proof sleeve pressed inside the stator and o-ring sealed into the end plates, allowing oil to be pumped across the stator core, cooling the copper and stator more effectively than the usual water jacket on the outside of the stator laminations (which increases the motor diameter and with no coolant actually touching the all important copper)? Although a standard stator type water
jacket could be made more simply.
I think that the centrifugal loads on the magnets is a bit of a non issue. I have been involved with some work for Ford Mo Co, bonding metals and composites etc. From Lab testing, we got 20 to 25 Mpa bond strengths using methacrylates and toughened epoxies(this can drop a lot with temp), or roughly 2.5kg per square mm. As an example, a 10mm x 10mm magnet (100 square mm surface area) would need approximately 250,000 grams to pull it off! That is a lot of g force for a 5 gram magnet! Toolman2 has tested some very cheap inrunner motors of similar size up to 10,000rpm and they started showing some stator rubbing.
Radially segmented magnets may prove more efficient.
Does the higher permeability of S.G type iron reduce the material thickness needed for the backing iron? On larger outrunners(hub motors) this could add up.
Thanks Zappy
Hi Zappy,
Thanks for your thoughts.
I was thinking of the initial design as a replacement for the motor on my eMoulton, so torque requirement isn't that great but, of course, we can take it in any direction. The size is close to that of the Plettenberg Nova15. My target weight is 1.6kg for the 30mm stack but I might not achieve that. I expect a 50mm stack could be kept under 3kg, though.
Regarding the rotor. I may be over engineering it for this specification but it's what might happen to the bond strength at high temperatures when run at high speeds that worries me. CNCAddict had an Astro rotor explode on him at 13,000 rpm, IIRC. It's not that much more trouble/cost to do the additional lamination pattern. Yes, there might be more to gain from segmenting circumferentially first, as it's the shortest side, but more than one circumferential segmentation can create problems. You can do as many axial segments as you need.
Thanks for your thoughts.
I was thinking of the initial design as a replacement for the motor on my eMoulton, so torque requirement isn't that great but, of course, we can take it in any direction. The size is close to that of the Plettenberg Nova15. My target weight is 1.6kg for the 30mm stack but I might not achieve that. I expect a 50mm stack could be kept under 3kg, though.
Regarding the rotor. I may be over engineering it for this specification but it's what might happen to the bond strength at high temperatures when run at high speeds that worries me. CNCAddict had an Astro rotor explode on him at 13,000 rpm, IIRC. It's not that much more trouble/cost to do the additional lamination pattern. Yes, there might be more to gain from segmenting circumferentially first, as it's the shortest side, but more than one circumferential segmentation can create problems. You can do as many axial segments as you need.
Thanks for your reply Miles, Something similar to the Nova 15 is very close to this. 24s lifepo4 or 79.2Volts seems to be a pretty good fit with the available controller fets etc. and is a fairly standard battery pack of around 1500w/hours of A123. 30Nm would be really good. What sort of lamination thickness/material is suitable for around 5000rpm? I would be very interested in this motors further development and build. I have not been very active here of late as I have been finishing our yacht and selling our home and now live on board full time (and it does have a small work shop, as well as having more spare time :wink: ). I still have access to my lathe, welders and composites etc though. The last Joby motor(9000rpm) just had some Kevlar filaments for magnet insurance, light ,cheap, and only around 20 minutes work. Making a motor from scratch you can make things like the drive sprocket, motor mounts etc correct from the start. Instead of paying a reasonable amount of cash pulling it apart and make sprockets fit onto what was meant to be a propeller mount, carefully grinding halls into you stator with out touching your expensive windings etc.
Cooling depends on use, obviously an 800w eMoulton at 96% efficient is less demanding than 8kw offroad past its peak, then a lot more aggressive cooling system will be required. I was very impressed how small the efficiency hit was on your smaller version when over powered. Very promising work.
Zappy
Cooling depends on use, obviously an 800w eMoulton at 96% efficient is less demanding than 8kw offroad past its peak, then a lot more aggressive cooling system will be required. I was very impressed how small the efficiency hit was on your smaller version when over powered. Very promising work.
Zappy
Did a quick and dirty simulation using the 7t winding at 4000 rpm using 85A. That gave 28.76Nm torque. 11,960 Watts out, 750 Watts heat (incl. 88 Watts of Iron losses), 94% eta. This was just a linear sim. but it gives a rough idea. Voltage is too high, needs something like a 4t winding for 50mm stack.
Peak efficiency is 96.6% at 4145 Watts (10Nm).
Peak efficiency is 96.6% at 4145 Watts (10Nm).
After a bit of tweaking, I've done a non-linear simulation for circa 30Nm at 5000rpm using 0.15mm Cobalt laminations
[pre]Efficiency 95.67%
Torque 29.58Nm
Power out 15380W
Heat out 690W (Incl. Iron @ 112.6W)
Torque ripple 1.639%[/pre]
Anyway, that gives us some idea...
[pre]Efficiency 95.67%
Torque 29.58Nm
Power out 15380W
Heat out 690W (Incl. Iron @ 112.6W)
Torque ripple 1.639%[/pre]
Anyway, that gives us some idea...
SolarRay
100 mW
What is the maximum torque during acceleration? (5-10 seconds)
SolarRay
100 mW
For 5 seconds, almost no cooling system does not have time take a lot of heat, so we can assume (in some approximation) that the adiabatic process. Suppose that, for example during this heating period (5 seconds) at 50 degrees сelsius.
SolarRay
100 mW
Well, let it be 60nm.
60Nm for 5 secs?SolarRay said:
There's about 2500Watts of heat being generated for 60Nm of torque.
I guess it will depend on the contact conductance across the coil and from the coil to core?
Stator core is 1.14163E5 cubic mm (797g).
I still think this would be very hard to wind. Trying to scale off the screen, 4 turns of 6.5 x .8 strip with soldered 8awg interconnects still can't match this.
I think this motor with some large thin axial fins or thin water jacket would be great(fins cut away where space is needed). Can you cut the Miles 90 stator out of the centre? and then maybe just use a steel core ? The cobalt steel sounds to expensive to waste :wink: . Miles do you have any rough idea on the price premium?
Zappy
Yes, I can :wink: I've no idea of the price premium for Cobalt over the same thickness of Silicon lamination. I'll find out soon. I was planning to get the rotor laminations cut out of M19 0.5mm stock. I was going to use area inside the 90mm stator for a 65mm inrunner stator.zappy said:
Similar threads
