Ironless axial flux Halbach project - 170mm, 36 pole, 44mm bcd disc brake mounting

[madact]

Less drag though...

I figure being able to tighten up the air gap is a good tradeoff. And the extra drag introduced by a practically unloaded ball race... has a pretty small care factor.

To be honest, I haven't actually built that much, I tend to get caught up in 'paralysis by analysis' too often... much experience designing & simulating, little experience seeing if said designs let the magic smoke out . Most of my experience with power switching circuits is from building your average buck and buck/boost topologies and rebuilding a couple of PC power supplies to deliver ~10A at 14.5V (handy trick for testing car amps, RC gear etc, but I have a 'real' off-the-shelf DC regulated supply now) - I'm still a noob at motor controllers. I'll probably do my initial testing with a sensorless RC plane motor controller, and see if it can pick up enough feedback to fire the phases in the right order, then get into the controller design once I know I inserted all the magnets the right way 'round (or once I know the space ESC won't do the job)...

Thanks for the file :2thumbsup: still working out how to use FEMM for moving coil simulation... I really want to do fixed speed, varying current on the coils (treating it as a linear motor), and integrate force, but the example scripts I've tried so far crash with lua errors (could be a wine thing) - will keep trying...

 

[safe]

In FEMM there is the ability to select areas and then get the "Force Via Weighted Stress Tensor" value from it. If you have coils in your simulation (or silicon steel cores like I do) you can select them and you will get a resultant force value. Since the coils are all identical you just multiply the force value times the coil count and can arrive at an overall force.

Torque = Force * Radius

...so the bigger the disc (rotor) the better leverage you get.

Once you know the simulation results you can go back to the actual magnetic flux number and fine tune it. In mine I'm able to maintain about 1 Tesla while having room to lot's of copper. Without the cores I'd be forced to use very tight tolerances that I don't I could achieve given my fabricating abilities. I've done a lot of things to adapt to low priced magnets and low precision instruments. A professional shop could do this so much better than I can.

In the spreadsheet:

"Torque"="Current"*$Radius*$Depth*$FieldStrength*$TurnsPerCoil*$Series*$Parallel

...in your case "Depth" is 1/2" or 200% of my 1/4".

 

[madact]

Nuts...

safe: Nothing around here that can open a .wks file can you save it as something else? A .xls would do the trick...

 

[safe]

Sorry about that... I'm on an old computer and that's an old Works format.

When I convert to Excel it loses the Charting and the Range Names. In some ways that's a good thing... just go around the spreadsheet and define names for things as you go and that way you can figure it out.

Things like RpmPerVolt once defined makes reading things easier...

 

[madact]

Hopefully that helps makes sense of things.

Yeah, I spent a bit of time the other day following the cells up down and sideways but it all makes sense now.

I did promise some preliminary results... here's a scenario with 6 16AWG wires per phase (in 2 sets of 3) and an ~0.8mm air gap. At 12A per phase, 60% duty cycle bipolar, FEMM gives me an average of 2.73 N for 2 poles (manually accumulated over 10 points)

which works out to 49.14N at an effective radius of 78.75mm, that's 3.87Nm.
This is in roughly the right range - anything over around 5N and I'll be in serious doubt of my rotor and stator integrity
Copper losses are around 25W at this current level. Makes me wish I could get square magnet wire in 16AGW...

Long story short, with a 700C/25 tyre, this is giving a delivered power of 192W at 60 kph, efficiency around 88.5% ... still some optimization to go, there...

I suspect that your 88% for maximum efficiency is about right. Everyone expects anything built with a Halbach Array to automatically jump in efficiency to 96% like the CSIRO, but that doesn't happen for free... you have to do your math.

Indeed you do - but perhaps you miss my intention... this motor is being designed as a high-speed cruising power-assist for a fully faired velomobile on the open road: 192W at 60kph should be far in excess of the total power required at that speed, if I get my aerodynamics right; this is the start of my envelope, not the end of it. At 60kph, typical motor loading would be 50-75W, and 'Long range cruising speed' will be ~ 90kph, with the motor operating at 100-125W (with the rider supplying another ~100W). At 90kph and 125W delivered power in this configuration, copper losses are only ~5W, for an efficiency of... 96% or so ... at 90kph / 200W, copper losses rise to around 13W for an efficiency of around 94%, and at 250W, maximum theoretical efficiency is still in excess of 92%... All that said, I haven't worked out how to model end effects on the inner and outer edges of the rotor yet, I'm just hoping simulation with N40 and no end-effect modelling will hold for real N42 and real end-effects...

Next phase will be getting things running from octave to automate testing of stator parameters (conductor count and size, particularly) - I've looked at an example script, it doesn't seem too hard. I'd really like to do it with one layer of copper, as construction would be greatly simplified, but I'll just have to see how it goes.

An interesting effect I noticed is that force does not vary linearly with current in the simulation... turns out I had a small offset in my force calculations with 0 current in all circuits. Looks like reducing mesh size in the airgap tightens it up nicely.

In other news, a pic of the rotors which will be used in construction of the... rotor:

Promax 180mm (DT-180C) - the brake surface has I.D ~145mm, OD ~180mm at the pointy bits, so it should be just about right... the drilling happens to have 36 holes in each ring, too - which is serendipitous for lining up a large number magnets at 5 degree increments: the poles and halbach magnets will line up with the peaks and troughs of the outer edge...

Lastly, I've realised another factor which may make tight tolerances more bearable - if the coil timing is slightly advanced under heavy load, the windings will pull towards the centre of the gap - the side closest to the rotor will be pushed away more strongly, producing a magnetic bearing effect which hopefully might keep things apart if they start flexing at high torque. Conversely, if the timing is slightly retarded, the coils will pull the stator towards the sides of the rotor. The opposite is true under braking. Methinks I shall definitely be building my own controller, with sensors - this is not the kind of thing I trust to an off-the-shelf controller to be designed for (a sensored controller with an adjustible hall effect sensor might do the trick, but I'd rather not risk it).

 

[safe]

Looks like we lost a few postings.

It's so darn hot here these days it's hard to do anything past 11 am. It was 87 degrees and about 70% humidity around that time this morning. The weather in the midwest is great for crops, but terrible for humans. (so I can't do much on mine when it's like this)

About all I can do is get an early morning ride on my other bike and then that's it.

 

[madact]

Looks like we lost a few postings.

Yeah, I updated my last post with a quote from one of yours that got nuked, and a reply... just in case you missed it...

It's so darn hot here these days it's hard to do anything past 11 am. It was 87 degrees and about 70% humidity around that time this morning.

Pfft. When I went to school, hot weather go-home-early temperature for primary schools with un-airconditioned weatherboard classrooms ('portables' were quite common until recently) was 97F. That's about the point at which we start calling it a 'hot' day - it's a heat wave if it stays like that for a week. Hottest recorded temperature was 115F in Adelaide last year, but inland it can get hotter, I used to live out in the country, and it wasn't regarded as 'hot' until it hit at least 105F...

Mind you, it's dry heat down here in Adelaide, I hear Darwin is far more humid - closest I've come to hot & wet is when I visited Seoul during a heat wave in '99, 95F/90% - it wasn't too bad after a couple of days to get used to it, but bizarre to see condensation on the windows of air-conditioned buildings, and we got caught in a bit of a deluge and it was like standing in a hot shower...

I suppose the other thing which makes it bearable here is that we measure temperature in centigrade - 42C doesn't sound as bad as 108F

Right now though it's winter, and not having a proper workshop, the rain is frustrating...

 

[pedalless]

Only in the 250W region, which is what's legal here in South Australia without a license & rego....

The e-bicycle limit in all states of Australia is 200watt not 250watt

"SOUTH AUSTRALIA

The Motor Vehicles Act 1959, defines a power assisted pedal cycle to be a pedal cycle with an auxillary power drive attatched up to a maximum of 200 Watts.

Part 2, Division 1, Section 9B, covers the exemption of registration for a power assisted pedal cycle.

Part 3, Division 1, Section 25, states that a licence is not needed for a power assisted pedal cycle.

Source: SA Motor Vehicles Act 1959.
"

I hope to see this motor running soon best of luck
with your construction.

 

[madact]

The Motor Vehicles Act 1959, defines a power assisted pedal cycle to be a pedal cycle with an auxillary power drive attatched up to a maximum of 200 Watts.

True - perhaps I was a little careless with my wording here (I was thinking in terms of motor design, not the system as a whole), I certainly don't intend to break the law. The speed controller will be tasked with keeping the power output at the rear wheel strictly within the legal limit.

The reason I said "in the 250W region" is that 250W @ 60kph is what I'm using as my 'design spec' for the limits of safe operation of the motor, and it allows a little headroom for amateur build quality and hot weather. Because of the design of the motor, this will mean that technically, the motor itself could deliver 375W at 110kph, but that's down to the controller to put a 200W limit on it. At the end of the day, there is no such thing as a "200W" motor - I can buy a motor labeled "200W" at a bike store and hook it up to a battery and controller combo that would make it do 400W (for a while, anyway ), and that would be illegal, just as I could go on ebay and buy a motor labeled "1000W", pair it with a controller which limited output to 200W and it would be legal - and likely somewhat more efficient (though I wouldn't like my chances of convincing a traffic cop of that, and going to court for no reason doesn't sound like fun - better than going to court for good reason though, I guess...).

Before you point out that 110kph seems like something you'd require an illegal motor to achieve, I fully expect the vehicle in question to be capable of that on human power alone (and in real-world conditions, one might easily hit that on a slight decline - a speedo is a must)... if only for a few minutes, at my fitness level, which isn't spectacular. I know from experience that sustaining 70 to 80 for an hour or so in a similar vehicle is well within my capability. The motor is purely for extending cruise time, not increasing top speed... This is also the main reason I'm building instead of buying - the chances of a commercially available bicycle hub motor being efficient for velomobile speeds seems slim - these would likely be optimized for speeds of 30 to 60kph, not for speeds over 60.

But thanks for your concern, and it's certainly good to see someone looking out for the interests of others.

 

[safe]

The laws alway deal with "output" power actually delivered and not "input" power introduced to the controller. So a 24 V 10 Amp controller might be allowing 240 watts of "input", but in most cases after losses in the motor you will have under 200 watts delivered to forward motion.

This is why I've argued for off street racing classes to be defined in the reverse way to street laws. For racing you want equal power input levels and unequal output levels to give an advantage to people with higher efficiency. For the street the focus is on output without regard for how many losses occured to get there. Totally different mindset.