APL's DIY axial-flux motor

Yea, the little 600 sq. ft. shop is getting full these days, I have to walk sideways most of the time. The only tools I'm
missing now are a Tig welder, and a metal band saw.

Thanks for the link's guys, You Tube is such a savior. :!: The most helpful one so far has been the HSS tool bit sharpening
vid. It show's exactly whats going on with the business end, and how it needs to be used. I like the idea of grinding my own.
https://www.youtube.com/watch?v=__A2xtLF0AU&t=1539s

Of course they're all helpful, and I'll be watching a lot of them for a while..

Right now I'm trying to get the lathe mounted up on the bench, and running again. After three days of moving a 300 lb+
machine around, plus a 100 lb+ motor assembly, I'm about as sore as I've ever been. :(
 
spinningmagnets said:
Start watching youtubes from "this old Tony", and the right column thumbnails will have other suggestions.

https://www.youtube.com/user/featony/videos

You are halfway to being ready for the zombie-pocalypse...

This old Tony is AMAZING. I love his videos. Unfortunately they are not invinite.
 
I like Tony, .. he's to the point and edits his videos very well. Some of those guys go on, and on...
I'll be checking out a lot more of his stuff,.. as he has plenty, especially the TIG ones.

Well, the lathe project goes on, and on too. I'm going through the belt splice thing right now. This
will be the third splice going into the old leather belt, The previous ones were masters of 'splice',
but mine is not going to be a 'looker'. I should just get a new one, but I'll give this a shot first.


View attachment 1


Plus, some fool put the on/off switch next to the chuck, so I have to risk mangling myself every time. :confused:
Thats got to change. :!:

The chuck is worn out, so that will be the next thing to buy, and the lantern style tool post leaves a bit to be
desired, so a quick change will be in order there too, along with some carbide bits. $$$!!!!
But for now, I think I can squeak by with what I have.


Bench lathe.jpg


Almost ready to twirl something,.. but now I need to do the holiday thing, and visit the family for a week.
Sometimes life gets in the way,..

Happy turkey days! :)
 
Belt splice is easily done with steel wire "staples". I used an old leather pants belt of the right width from a thrift store, with buckle and pointed ends cut off, and was lucky enough that some staples for the purpose came with the lathe. Just had to use a drill to put the holes in the belt first. IIRC I used three staples.

The switch on mine is out front, mounted along the front edge of teh bed to teh left of the chuck.
file.php
 
Very nice amberwolf, I see you have the 'Workshop' model 'C',.. I would have preferred that one, but it is what
it is. Take good care of it my friend,.. these are nice lathes, and parts are plentiful.

Well, I've made a classic noobe blunder,.. I assumed that 9" swing meant nine inches over the bed,.. all these years. :(
Makes sense to me,..but NOOO... It's the radius!! Frack me running. Why don't they just say 9" diameter?
Geez, by that logic, I can say I've got a 16" swing on my private parts... :D

Live and learn, well, it's not really 9" either, it's actually 9-3/4". :roll: This motor's stator plates are 9-1/2", so I'm just
barely under the wire. Whew!

I don't get this old world some times,.. and would it kill us to switch to metric? It's inevitable,.. so why not just do it? :!:
 
APL said:
I don't get this old world some times,.. and would it kill us to switch to metric? It's inevitable,.. so why not just do it? :!:

I've been told it is inevitable since grade school ... about half a century. ;^)
 
APL said:
I don't get this old world some times,.. and would it kill us to switch to metric? It's inevitable,.. so why not just do it? :!:
Well, you do already. Somehow. Eg. mills? What kind of anachrony is this? Thousands of an inch? So you mix imperial and metric.
Use inches, feet, miles and stuff and half it, half it again, and again and again a again until you have 1/16th of it. Really strange and difficult to multiply or even add and subtract, but maybe easier if you learned it from the very beginning of you life. I personally will never understand.
 
APL said:
Very nice amberwolf, I see you have the 'Workshop' model 'C',.. I would have preferred that one, but it is what
it is. Take good care of it my friend,.. these are nice lathes, and parts are plentiful.
All the parts I've seen that I could use are out of my price range, but it works well enough as-is for the primitive work I've needed it for so far.

I got pretty lucky with finding this one, even with the problems it has. I still need to get some decent cutting tools to use with it; what I have is just harbor fright stuff.

A few times I have had things I wished I could do with it but couldn't because it's chuck is not far enough off the bed. So I have had the idea to come up with plans for raising it off the bed, so larger diameter work pieces can be used in the chuck, as long as I can bolt something to the work piece that wil stil fit in the chuck. ;)

There's probably good reasons not to do that...but since it's something I'm going to need to do at some point, and I can't buy a bigger lathe, I'll be trying it out when that time comes. :)
 
Finally done with the holiday travels, and getting back into the saddle. The little APL shop is starting to warm
up again.

I got the lathe spinning today, and took a first pass off the axle,.. looks real nice! The tool post is a total pain,
but it works, and I guess that's all I need. At least I can move forward, and get this thing made at long last.

Next up will be the stator plates, and it should actually start looking like a motor once they get bolted on.


First pass..jpg


On the new 'Muscle Cruiser' build, I figured out how to get six 9ah drill packs in the same space as four 12ah's. So
thats looking better. It will give me 27ah overall, at 40v, and cheaper $$. The 2S-3P set up will have an awful lot
of low end grunt, and might even break something in back.
Perfect! :)

Since Milwaukee won't give an actual max watt output of the packs, I don't know the specifics yet,.. I'll have to
hook up the CA to find out. But I've ridden a 3P-2S set up, and it's VERY impressive! Definitely scary. :shock:
Can't wait!!!
 
APL said:
The 2S-3P set up will have an awful lot
of low end grunt, and might even break something in back.
Perfect! :)
I figure if it doesn't break something starting up from a stop at full throttle, it doesn't have enough torque yet. ;)


Since Milwaukee won't give an actual max watt output of the packs, I don't know the specifics yet,..
If they list the max amps out (or max amp draw of a tool that uses them), then you can multiply that by the average voltage of the pack, and get a good guess at wattage.
 
I've been looking for a while now, and they hide the max watt draw very well. They don't want to get into a
'Watt War' with other makers, by simply stating what people need to know. It's easier to be ubiquitous with ah,
and tool draw. Confusion sells more batteries. This has been going on in the automotive battery market forever.
It's easy to manipulate the different systems for determining AH, and thus, the final ratings.

It would be easy to figure it out with the actual cell outputs, but it's the BMS settings that count. I suppose I
could just buy one and test it,..but you would think that somebody has done this already. Not that I can find.
My gut feeling is that it's around 900w for the 9ah 18650 packs. In a 3P set up that would be 2700w,..which
would throw some fire into the little controller I have. :)

All speculation mind you, but that kind of power into a hub motor mid drive is something to behold.

I did find out that the 12ah packs are using the 2070 or 2170 cells, which have a grater watt output. And have
a conservative 1200w BMS setting,.. hmm.. thats If I can believe anything I read.
But at $200. apiece, their hard to justify. A 9ah is $80.

Just a brief chart I saw;

18650's-- 1500 - 3600ma.
2070's -- 3000 - 4000ma.
2170's -- 4200 - 4800ma.

Like I say,.. single cells are not my specialty, I'm a bit of a noob on that, as well as the rectangular flat packs that
I see beging used. They seem to have a much higher rating, and are a bit more 'stackable'.
Always more to learn...
 
Well, the watt rating of the battery isn't really all that important--it's the amps it can support.

As I noted before, you can measure that (with just an ammeter), running a pack with a variable load (like a bike motor at different throttle settings and loads, or a variable speed tool under different loads) until the BMS shuts it down from overcurrent, then assume it's max is a bit less than that. ;)

Or just use the max tool amps a pack is rated to be used with; it might not be on the battery, but makers love to brag about amps on tools. Whether those amps are continuous or peak is another matter; you'd have to test that to be sure.


Then measure the voltage while under that amp load, and multiply the two, and you get watts.

But since controllers are current-limited, then the amps is a more important rating, becuase the packs you use with it have to be able to sustain that current for however long you would run it at max current (usually only a few seconds for most people).

Controller watt ratings don't have much to do with anything other than a number they assign to them...just like with motors.


As for max current capability of an 18650 (or any other cell) it depends on the specific cell, not on it's capacity or other things. Some cells are made for high capacity, but have lower current output capability (c-rate). Some cells are made for high c-rates, but have less capacity. Some are in-between. Without knowing what cell is in a pack, and it's spec sheet ratings, (or having accurate ratings for the pack itself) you can't know what it's actually capable of without testing it under the conditions you need to use it at.
 
So, let's do some fake math and say your controller is rated for 20A, and absolute max of 75v due to parts used (but is actually a "52v" controller).

Let's say you use three 18v packs in series, for 54v average (3.6v cells x 5s, 21ish v fully charged), and say those packs are rated for 300w each; that'd be 900w for three in series. Then you wire up three of those strings (9 packs total) in parallel, so you could get 2700w.

Well, 300w / 18v = about 17a. That's already close to the max current a 20A controller would ever draw, so more than two pack strings in parallel is "wasted" current capability--the controller will never draw more than the 20A, even if the batteries can supply it.

If the packs are actually capable of 900w each, then just one pack of however many in series (1p, Xs) would be able to supply more current than the 20A controller will ever ask for.


However...the batteries will sag less in voltage the more of them there are in parallel, so there is still an advantage in doing so, because you get more watts total out (depending on actual voltage sag under load), and less waste heat in the packs. At some point, cost/benefit ratio is skewed and isn't worth doing any more parallel cells, of course.


And if you have a controller that will draw more current than a single pack is capable of, then you need to parallel at least however many it takes to supply that current.
 
Let's say you use three 18v packs in series, for 54v average (3.6v cells x 5s, 21ish v fully charged), and say those packs are rated for 300w each; that'd be 900w for three in series. Then you wire up three of those strings (9 packs total) in parallel, so you could get 2700w.
Well, I see what your saying,.. although theres a slight flaw in that quote; three 300w packs in series is 300w.
Three 300w packs in parallel is 900w.

However, I see it's going to come down to the controller to get the most out of the packs,.. your right of course.
Which brings me back to knowing what kind of power I actually have to draw from, so can get the right controller.
I suppose I could just overkill it with a monster controller, and be done.

I can't trust their 'tool amp draw' advertising, they can't even get their voltage rating right. (18v is actually 20v) And I don't
really want to take a brand new battery out of the box, and push it to overload.

But, I'm afraid that it's going to have to come down to something like that, in order to get to the truth of the matter.
I have a Cycle Analyst I can hook up, I assume that it's accurate enough, finding the actual method for a good variable load
is the next question.
Maybe I can find a cheap used battery to use for this.

Otherwise, I know that a 2P pack system works well with this 40A controller. (although, the controller may still be limiting.)
So a 3P pack system will probably need a larger controller to take advantage of these batteries.
 
APL said:
Well, I see what your saying,.. although theres a slight flaw in that quote; three 300w packs in series is 300w.
Three 300w packs in parallel is 900w.
No, wattages add either way, becuase they are voltage x amps. Amps (and Ah) only add in parallel, while volts add in series.

If you have a 300w pack that is 18v x 17A, then put it in series with two others, then that's now 54v * 17a, which is now 900w.
 
Well shut my mouth! First time I've heard of it. How could such a basic bit of info. like that escape me for 50 years?
I've been into electronics since high school, and don't remember ever reading that.

:bolt: "Watt's add both ways, series and parallel." :bolt:

So take the number of batteries times their watt output, and you have the total watts available at any given moment.
(depending on their charge)

In my case,.. 6 X 900w = 5400w. (except the 900w is still questionable)

Riiiiight? :?
 
APL said:
Well shut my mouth! First time I've heard of it. How could such a basic bit of info. like that escape me for 50 years?
I've been into electronics since high school, and don't remember ever reading that.

It might not have been directly stated anywhere in the books/etc you've had...but it is what happens when you calculate watts from amps and volts. (or either and resistance)

If it helps, here's a page with the math and words I lack:

https://www.allaboutcircuits.com/textbook/direct-current/chpt-5/power-calculations/



:bolt: "Watt's add both ways, series and parallel." :bolt:

So take the number of batteries times their watt output, and you have the total watts available at any given moment.
(depending on their charge)

In my case,.. 6 X 900w = 5400w. (except the 900w is still questionable)

Riiiiight? :?

Yes.

Think of it as the amps and volts instead, and it may make more sense.

Let's assume they're 5s packs, and 18v average voltage. If they're 9Ah, then let's conservatively assume a 3C max current rate, which would be 21A per pack. That's 18 * 21 = 378w per pack.

If you put two in series, that is now 36v * 21A, for 756w per series string.

If you put three of those series strings in parallel, that's now 36v * 63A, for 2268w.

That's only a *capability*, since the controller determines what actual current will be pulled from the batteries, and the voltage sag of the batteries at any particular state of charge and current draw will change the wattage available. At full charge, more watts are available (for a number of reasons including total voltage), and near empty less are available.


If the cells' C-rate is higher, or you are willing to push them harder (assuming the BMS does not current limit it, and there's no fuse in the packs), then you could get more out of them if the controller will pull it.






BTW, some thoughts on voltage "ratings" of tool packs:
I can't trust their 'tool amp draw' advertising, they can't even get their voltage rating right. (18v is actually 20v)
If the packs are 5s, then 18v is "correct", because packs are generally rated not for their full voltage, but their average, so 3.6v/cell x 5 is 18v. 3.7v is probably more correct, but that comes out to 18.5v which isn't as easy to remember, and doesn't print as large in the same space for marketing. ;)

20v doesn't fit 5s or 6s, because at 5s it'd be 4v/cell (rather than 4.1-4.2v full), and at 6s it'd be 3.34v (rather than 3.6-3.7v).

More likely they'd call a 5s pack a 21v for it's full voltage, or 18v for it's average, and a 6s pack 25v full, or 22v average.
 
There are teardowns of the Milwaukee 18V packs on YouTube and they show 5s pack configurations. So the 18v nominal rating is in keeping with standards commonly used for ebike batteries.

These batteries will run a circular saw. My Ryobi AC circular saw is rated at 14 amps. So that's 1680 watts. So maybe they can be much more clever with how they manage peak current draw with the battery powered tools, but my finger to the wind guess has me thinking it would be hard to get good performance with much less than half of those 1680 watts. So I'd have to guess that those battery packs can deliver at least 700-900 watts.
 
Sounds like the HPC 36-72v 50A controller I have is going to be a restriction. Fine for the last bike, but this
next one is a muscle cruiser, so I should at least have the option of melting my motor. (if I want.) :)

I'll have to look around,.. but, any Ideas for a good candidate? Don't need any frills, just a basic controller.
I know the motor is going to be the next weak link, and probably the rear hub, but I'd like to start out with it,
..an old HPC Thunderbolt.

Rated at 2000w, probably good for 3-4000w burst's.

I'm going to need a 'get out of jail free' card too... :confused:
 
Finally getting somewhere on the axle, only made one mistake so far when I turned a handle the wrong way. :oops:

The bearings haven't gotten here yet, so I can't do the final fit, but she's looking pretty good. Plenty of chances to
ruin it yet, and drilling the phase wire holes is going to be a good one. I have to drill from both sides and try to
connect them in the middle.


Turnes Axle.jpg


I got a full 5 Gal. bucket of chips out of the deal, and I don't think I saved any weight over a steel axle, but at least it
looks prettier. :)

I'll have to order up the stator plate aluminum, so that it will be ready to go,.. and I'll need it to get the mounting holes
matched up with the axle.
 
APL said:
Sounds like the HPC 36-72v 50A controller I have is going to be a restriction. Fine for the last bike, but this next one is a muscle cruiser, so I should at least have the option of melting my motor. (if I want.) :)
You could do a "shunt mod" to fool the controller into providing more current than it's designed for. Sometimes this works fine, sometimes it lets the magic smoke out ;) .

There's two basic ways: decreasing the resistance of the shunt, which itself has two main ways:

--adding solder or wire to the existing shunt(s), but this is not really controllable except by experimentation, and not undoable (there's no way to go back to the original resistance even if you do your best to remove all the extra solder; its' still changed at least a little, enough to make a noticeable and measurable difference in behavior).

--adding more shunts in parallel with the existing one(s). This is both controllable, and undoable. Most of the shunt wires that are the same diameter and length are the same resistance, so if you already have two shunts, and add a third (out of a dead controller, for instance), you just added half-again the amount of current capability. (parallel resistance formulas). Add two shunts and you double the current handlign (halving the resistance). If the shunts aren't the same, it's not as easy to calculate, especially if you don't knwo what any of your shunt resistances are, but youc an experiment more easily (undoably easily, anyway) by adding / removing shunts than by adding/removing wire/solder from a shunt.


A second way to fool the controller is to find the path from the non-ground end of the shunt(s) to the MCU. Somewhere along there will be a voltage divider and possibly an op-amp or transistor amplifier circuit. Changing the divider resistance or the input/feedback resistance in the amp changes the voltage the MCU sees from the shunt for any particular current thru the shunt, so you can just put a potentiometer on there and adjust it until it does what you want or explodes. ;)

The generic "30A" controller on the left side of the SBC trike (taht runs an MXUS 4504) was turned into an 80A that way--it hasn't died yet. :)


..an old HPC Thunderbolt.

AFAICR that's the Crystalyte H354x motor. If you know what the winding of the motor is you can go to the http://ebikes.ca/simulator page and see what it can do.

I have an HSR3548 on the right side of the SBC trike right now (until i fix the broken axle on the MXUS 4503), and used to use it on CrazyBike2's rear wheel, and it can manage some torque. I checked it out with Incememed's SFOC5 I have here for testing, which can put a fair bit of power thru it, and it woke up more than with the Grinfineon 40A that it usually runs on--but it also quickly (minutes) got it too hot to keep my fingers on. (but due to limitations at the time with the SFOC5's startup methods, I couldn't do much torque with it until it got above several MPH, so no real comparison available for startup oomph. Will try to retest that with the new firmware that should have better startup).
 
Thanks wtuber, for the good gestimate on batteries, 700-900w is what I would think as well,.. enough to
make a plan from anyway.

And thanks amberwolf,.. for the mod. idea, I like the idea of ripping into the controller. Admittedly,
a dark area for me. I really need to bust out the windows on the controller thing,..and see whats really
going on in there. I understand the basics of motor control, but haven't really studied a controller circuit yet.
I've been to busy learning about everything else. Plus, they stick micro chips every where these days,..
always takes the fun out of it for me.

I didn't know you can do a tweak, and squeeze more amps through them. I would think that you'd have to
oversize the Fet's at the very least. So this sounds great. I'd rather mod it till it melts, than buy a new one.
I'll learn more this way, and the moneys already spent, so, I'm ready to go.

I like the second version to fool the controller. use a pot to vary it, and then maybe replace it with a fixed resistor
once I find the sweet spot,.. or some kind of lock knob.

I can put a thermal sensor in the motor windings, to keep an eye on that.
I suppose I could put a sensor in the controller as well,..maybe even hook it to a shut down input, like the brake wires,
to keep it from letting all the magic out? :)

I'll have to look for a circuit for this controller, and see if I can get my eye on what your talking about.

And I'll look up the windings on the H354x, and check that simulator out. One good thing is that this is a mid
drive, so it spins pretty fast right away, and heat hasn't been an issue so far.

Maybe I'll lose my nerve,.. or grip, before it ever gets that hot. :wink:
 
APL said:
I didn't know you can do a tweak, and squeeze more amps through them. I would think that you'd have to
oversize the Fet's at the very least.
Sometimes you do. Members often did this up to a few years ago (Lyen pretty much made his "name" and business doing this, and a number of members were serious pioneers in it).

Nowadays even the relatively cheap controllers use FETs taht can be good enough to get a few extra amps from, and sometimes double what it was meant for, if you aren't pushing the voltage limits at the same time, or doing it continously. (like for me, it's short bursts of a few seconds of high current, then just cruising at a couple dozen amps or less, mostly),

Sometimes it's already built quite well enough to handle more current than the shunt (or programming, which may not be user-accessible) allows, so just changing what it allows is enough to get much better performance. (again, it may only work in bursts, not continuously....)

Much of the more-continous anti-smoke limits are how well the FETs are mounted to the heatsink bar, and how well taht is mounted to the case, and how well the case is cooled. If they don't get too hot, the FETs won't be likely to blow up, as long as they are not near their voltage limits.

Run a FET near it's voltage limit, and you are gambling much more than pouring more current thru it; do both at the same time and it's practically a guarantee that it'll fail at some point when conditions are just right (wrong?).



I only mod controllers that aren't critical, meaning I either have a backup for it installed (but not hooked up) on the bike, or it's a 2WD+ bike (or trike), simply because I've read of too many "oops" moments of pushing them too hard, and have experienced a number of controller failures over the years--more of them on modded ones (like the Lyen "monster" 6FET I used on DayGlo Avenger that I broke a clutch on a geared hub with, and overheated that same hub several times almost to the point of killing the halls--but the controller itself had the capacitors fail from heat generated inside it in use, durign the summer here)).
 
Seams like everything I'm interested in happened years ago. Sometimes it feels like I'm on a ancient planet,
everything has come and gone, and you guys are the remaining survivors, at 25000+ posts, you've seen it all.

Currently, I'm reading a 54 page thread from 2009, when the H35 series first came out, to get the information
on it. liveforphysics pushed it to the limits and beyond. :shock:

From what I can gather, these motors will take 5000 watts for short periods. I'm no maniac, (yet), and the voltage
is 40v, which is low on the scale, plus it's in a mid drive,.. so I might have a chance at making this work.
(famous last words)

The batteries will have an estimated 5000w to draw from.

40v x 5000w = 125 amps. This is a 50 amp controller. Think it can be bumped up that much? You say your getting
80a out of a 30a. Just a guesstimate, I know theres no guarantee,.. nobody knows where it's going to go poof.

Like I say, I'm not sure I can hang on to that anyway, mid drive is three to one, so thats equal to 15,000w, which is
aprox. 20 HP? Must have my math wrong,... can't be that much. :?:
 
Back
Top