Steel-geared hub motor?

So total for just shipping and export documentation cost for the $100 dollar motor from vlad to australia came to $140 dollars, so it looks like wading through alibaba supplier "negotiations" time, even vlad said it was a better idea then buying from him (which is nice of him)

If I can get any decent supply information ill post it here, or if I have to buy like 5 I'll offer the extras up for sale at cost
 
OK, so I have received my motors as well. One is below:

rC94dPH.jpg

lZDtiFn.jpg


I figured out that I have a nasty habit: no pictures of actual thing, only internals...

What I can say so far:
Bad:
1. Liquid oil is only "for smell", as there is only like half of teaspoon of it in entire motor.
2. Gears are lubricated with thick grease (a bit silly to do that in the same bath as liquid oil)
3. Sun shafts are lubricated with third kind of grease (green). Also in same batch with liquid oil.
4. Quality of metal is as always. Shit.
5. All bearings are of "rubber sealed" type, even though they are in liquid oil bath and have additional glands on shaft. All these greases will wash out and make a mess eventually.
6. Both ends of motor shafts are not connected one to another, so possibly will break under heavier loads (drops, potholes, etc.)
7. Even though motor was sealed with silicone, some oil has leaked. Possibly during delivery via air, when atmospheric pressure dropped. That raises some questions about sealing the thing for various operating temperatures.

Good:
1. Suns use needle bearings - wow.
2. Ran from first try on Kelly KBS controller, all wire colors are correct. Was quite slow at 30V. No-load current ~1.2A (not sure if this is correct).
3. Gear runs quietly, almost no gear noise at no load. However there was some rattling. Possibly will go away with load.
4. Due to shaft dimensions, thicker phase wires are not possible. Probably will have to machine new, thicker shafts, to fit 4x thicker phase wires.
5. Even number of stator poles allows KV increase of 2x or 4x. Probably will make it 4x. Pictures above show this in process.


crossbreak said:
around 420rpm loaded @48V, so you'd need 90V xD I'll really measure KV tomorrow when i have my CA alternative connected
OK, so at 42V (loaded 12S LiPO) it will be 22km/h with 12.5" wheel...
So I have a dilemma now. I can increase KV by 2x or 4x. 44km/h is a bit too slooow... So... 88km/h? Probably that is too crazy for 12" wheels, right?
 
6. Both ends of motor shafts are not connected one to another, so possibly will break under heavier loads (drops, potholes, etc.)
this is the case for all geared inrunner hubs. If it was no inrunner, it could not be oil cooled. I think it can be as stiff and robust as a through axle, just a question of design


thanks for your input+pics. I managed to get 2.5mm² cables into that axle..even with additional temp probe wires. Cant say this was easy. The axle is 14mm so this is the largest common for these hubs..bafang have 12mm and they work

didnt really check on KV yet. think you should do that before you proceed.. have no chance to check that tonight..sry..anyway you could try with 2times the speed.. think this should fit as I do 12s as well in a wheel that is double thes size..BUT maybe induction will kill top speed.. no idea yet.. what induction did you measure?

My calc still shows horrible efficiency for such a high inductance motor.. somehow reasonable.. but efficiency >50% can't be true..it's too low temp for that. Maybe my values are way off and the 1000µH i measured are just BS
 

Attachments

  • MotSim_Alpha_7_steel_geared.xls
    102 KB · Views: 145
Earlier you said it heated up significantly slower then the Bafang SWXH, if the inductance is this high and efficiency is only 50%, compared to 80% you would think that no matter how much more efficient at shedding heat this motor is that it would still heat up significantly faster then the SWXH, 30% is a lot of heat!
 
yeah.. as said that can't be true. Let's wait for circuit's measurement. Maybe i have access to a dyno in the next weeks so i can find out more

If inductance is really 1000µH then i should ask myself what is wrong with the MotSim :p
 
OK, I said it runs fine, but after taking my time to listen better, I noticed strong cogging noise. Hooked up my scope...

Half throttle:
hjho0dj.png


Full throttle:
ynoxfpa.png


CH1 is hall signal, CH2-CH4 are phases.

Measured at 35V. So it looks like 60 rpm/V (actual motor, before reduction).
However I don't think this can be trusted, because I very much don't like what I see in those waveforms. It looks like very advanced or very retarded, can't decide which one.
Also, motor spins, better or worse, with any hall combination. And that signals a problem with phase angles.

Controller is Kelly KBS4805.

EDIT: My chinese LCR shows 1170-1380 μH :shock: Measurement was very unstable and I would not rely on it. But it is very similar to what crossbreak saw.
Total resistance is 380mΩ :shock: (38mΩ cables alone).

Wow, just wow. How can that be so bad?
 
that is a good question :shock: Maybe they use alloy wire :lol:

had similar problems with the halls. all my drive test where sensorless...they kelly didn't like either combination but always showed "hall sensor error" when applying serious load. Maybe we can fix hall position? I did not open the motor itself yet.. maybe you could post a pic of that hall locations?
 
I've emailed the vendor and he states
Vladimir said:
We never measured inductance, but motor efficiency is about 80-85 %, so something is wrong measured, or concluded.

I've also asked him if he would like to make a comment/add his expertise to the thread about any of the hall issues/inductance and he seemed to take a bit of offence :( . He does seem very set on the 48volt 25 amp figure too.

Also looking at the picture you've posted the halls are mechanically spaced 30 degrees apart (360 degrees/12 teeth), so there shouldnt be an issue there, 8 magnets /2 = 4 locations for same output, 360/4 = 90 degrees
90 / 3 = 30 degree spacing, so thats all correct, 120 degree hall spacing should be correct (I think :? )

It could be that the halls are too sensitive, causing them to advance timing slightly due to being triggered earlier.

Also i've traced your scopes output and measured from a phase fall to the next phases rise, between phase 3's fall and phase 2's rise the spacing seems to be longer, 64ish degrees compared to 54ish for the other 2 phase combos
LTTAUDq.png


Although keep in mind this is just traced so it wont be the most accurate results, but it does seem like their could be something weird going on, I can't see why it would be though :?
 
Vladimir said:
We never measured inductance, but motor efficiency is about 80-85 %, so something is wrong measured, or concluded.
Well, If we have 48V battery and 25A phase current, realistic power peak would be at around 70% speed. So:
Input power = 70%*48V*25A = 840W
RI losses: 25A*25A*0.38Ω = 237W

So η = 72%. RI losses alone, plus magnetics, gears, etc.
With lower phase currents, yes, 80% efficiency looks plausible.


Vladimir, if you are reading this, I would like to strongly say that we are not doing any conclusions, yet. It is very possible that our controllers are having issues, etc.
If our conclusion(s) will be negative, still, from personal practice I can say that even bad review is still better than no review at all.

EDIT:
was looking at the waveforms again.. To me it looks like phases are advanced by ~20 degrees. This may be reasonable with such insanely huge inductance.
Timing inconsistency between phases is another issue and I have seen it on many, many motors. Will give it another look in couple of days.
 
how ist it possible that inductance+resistance are so high?

I mean:
- there seems to be enough copper in it
- KV seems also reasonable
-number of turns seems to be in the 10-12T range from what i saw

Where does all of the inductance/resistance come from?
 
Number of turns is 26 per tooth, so 104 per phase. Windings are made of 4 strands, each 0.5mm in diameter, so total winding cross section is only 0.8mm².
Motors need that many windings when they have very weak magnets inside. However these magnets do look like neodymium and are quite thick, so I can not understand how did we end up with such numbers.
 
ok weak magnets could be some explanation.

I was at the dyno test bench today. Sadly it has no real/precise torque measurement and only does around 300Watt/ 20Nm or so ATM..now i have to redesign the brake/torque transducer. Hopefully this will take only some weeks, i get a lot of support there. This motor will definitely be the first i'll bench test :D

circuit said:
Windings are made of 4 strands, each 0.5mm in diameter, so total winding cross section is only 0.8mm²

Still, the one strand (~AWG20 ??) should have a resistance of around 36mOhm/meter. So four have 9mOhm/meter. With ~342mOhm winding resistance (WYE? ) there must be 19 meters of wire per phase :shock: 57m total
 
Short update...
95QevRO.jpg


But there is a problem. There was very little room left between stator's windings and outer shell and with additional wires it does not fit. Will have to find additional 2 mm somewhere. Such as like this:
atu8SMO.jpg


1.5mm off, will have to machine additional 0.5-1.0mm off the shell. Tomorrow maybe... My benchtop lathe is resonating very badly with such "large" parts and time here is 10pm now, so will have to continue some other day. Not in mood to deal with angry neighbors or police today.

BTW that stator was some work... Would not like to repeat that any time soon.
 
update #2.

Found few more hours to play with this motor today:


So it DOES spin now. Could not find a good phase combination for right direction, so had to swap two hall wires and repeat a search for good phase combo. I have found a combo, when motor spins fine, more or less:
Pgqdn9n.png


Signal is still retarded, but not by very much. Maybe due to huge hysteresis of hall sensors? Don't know why exactly. Will have to investigate.
Notice that hall frequency is 432 Hz, which is 6475 rpm at 37.5V => kv=173 RPM/V.

So in my case (12S LG 18650) - 7785 RPM => 66.7 km/h with 12.5" wheel, at no load. So probably ~55km/h in real world.
Keep in mind these figures are with slightly retarded phase angle, so actual kv will be higher by probably 10%?


Since inductance has decreased drastically after this mod, motor runs a lot smoother. Also I hate one more thing about this motor:
v42MX8s.png


Wow, have not ever seen such ugly BEMF. Should be quite lossy with trapezoidal control.
 
OK, another update. I compared hall signal to differential phase BEMF and here is what I get:

4Fl3TbS.png


Not only there are sector length variations on each period (due to uneven magnet alignment), but also there is physical delay between phase BEMF and hall signals. I compared all three halls - same thing. Actual delay is 14°. That is a lot.

This measurement was done during free spinning, so no influence from controller. Tried to spin it both ways - delay still the same and still positive. This rules out bad alignment of hall sensors, but proves that problem is not related to controller.

Maybe just too large hysteresis on halls?
Had a few lying around, will try to replace them.

EDIT: new halls (Allegro A1203) decreased delay to 10 degrees. Could be better... :(
I'm thinking that weak magnets make hall hysteresis more apparent.

Ideas?
 
Thanks for all the information so far. I find this fascinating. Partly since I was on the verge of ordering that motor. I live in Europe and am having a hard time finding parts in a reasonably price range (and imports from outside EU become expensive).
 
:) I'll wait for you guys verdict. Not looking for the perfect motor and I like to root for the underdog. BTW, very impressed by your battery project.
 
I am surprised that more geared motors don't use helical gears. These gears are metal, but are reported to run surprisingly quietly. The other perplexing thing is the shockingly poor performance of the motor part...I would have guessed the motor part would be very well understood by even the most low-priced suppliers...even if they just copied a successful motor of the same size with no real understanding of how the thing works!
 
Yes, to me it is surprising as well. There are many small details that could be improved, such as skewing the magnets or stator to get near-zero cogging. And using outrunner to get a one-piece axle.
Oddly, those who make above things rigt, use some crappy gearing.. I mean, nylon gears are quite good, but they are still quite loud with straight teeth.

Search for perfect motor continues.. It's close to taking the beating and making one myself ...
 
Helical gears are not used more often because they create drag and a loss off efficiency compared to strait cut gears. And generally they need a larger volume gearbox to get the same gear reduction which also adds to drag. And with Ebikes its not like their is much spare battery capacity where efficiency is not important. For a ICE engine where you have heaps more range then a small amount of efficiency loss is irrelevant.

The ideal gearbox on a brushless motor would be a strait cut compact planetary gearbox, encased in Vacuum chamber, or sand or liquid damper box to stop most of the noise. That way you would have the maximum efficiency and low noise as well.
 
People have been asking what is the ideal motor so I gave my opinion. Just because you don't like it you can't call it BS. prove it wrong otherwise you are speaking BS.

which part of what I said is BS ?

1) That helical gears are less efficient then strait cut gears ?

2) Or that ebikes/ev 's are severely range limited by battery technology hence efficiency is very important unlike ICE engines ?

3) or that a vacume box or sandbox cover would significantly reduce the gearbox noise ?

4) And if you did reduce the noise with these then you have the highest efficiency and low noise.

Everyone of those are undeniable facts prove them wrong otherwise you are the one speaking BS.
 
Back
Top