# Idiots guide to rewinding Revolt 120 pro

#### larsb

##### 1 MW
I will rewind a Revolt RV 120 pro motor to see if i can improve performance. It will be my first wind so that's where the headline comes from

Some measurements of original stator (rough figures since windings are still in place):

Approximate stator slot area: 190 mm2 (slot bottom width:6, top width:15, radial slot height:18)
Copper area in slot: 2(sides)x12(turns)x5(strands)x(single wire area)
Single wire area: 3.14*0.8^2/4 = 0.50 mm2

--> Copper area in each slot: 60 mm2

fill factor: Area copper / Area slot: 60/190= 32 %
That is low. I am looking forward to the rewind as it seems to be possible to make a big difference.

Max fill possible in ideal conditions can be viewed here:
https://www.emetor.com/blog/post/how-big-maximum-possible-slot-fill-factor-using-round-conductors/

Wire diameters between 2-3 mm seem to be optimal, although i think large wires will be really hard to wind.
I am surprised to see such a big effect of the type of insulation!

I will be aiming for at least 65% fill and plan to use 14awg square wire in 2 parallell strands for as many turns as can be fitted.

If someone has tried 14awg wire for a rewind of a motor this size then i'd like to hear your input. Is it too stiff to handle?

I made a model of the stator to see approximate theoretical area of the slots.
It is 230 mm2.

I will have to switch to larger wire or more parallell strands to keep the kV roughly the same as original winding. Since my bike is a mid drive i don't worry so much about exact kV as i can easily switch gearing.
Having that said i still want it to be <45 as this places output RPM @ 72V really well for a one stage gearing.

I don't want to go to 3 strands of smaller diameter as i think it will be a mess to keep the wires from getting crossed.
If i could source larger square wire then that would be my choice but i have only found one place to get it from:
http://www.electromechanicsonline.com/default.asp?cat=Wire%2C+Magnet%2C+Flat

Wow Lars, I will be watching this closely. Very interesting to see what can be achieved and how tricky it will be to rewind it. I have wound a bunch of small rc motors over the years and know from my limited experience that things gets challenging when thick wire is being used. Depending on your findings, I might follow with the rv120pro I have on the shelf.

Hello wheazel,
What is your experience as to what max wire diameter that can be handled without bleeding or getting wires badly placed? I think the larger the motor, the easier to wind but hey - in theory everything is easy!
I might use round wire instead of square as most likely i will bump into issues as i wind. Mistakes will be really expensive with a cost of around 50 USD per attempt for the square wire.

My first try with driveCalc indicates 1.7mm wire if i keep it 2 parallel, dLRK and delta terminated and aim for >12 turns

driveCalc indicates it would be better to do 3 parallel wires of 1.4 mm, that gives an estimate of almost 80g more copper in the motor - but i doubt it. Aligning three wires will be a lot harder than two.
I wonder if it will be that much easier to bend the thinner wires so that it still will be a better wind?

You are correct that the wire size difficulty also depends on motor size. When bending thick wire (1,2mm) is the thickest i have wound on a small motor, insulation to the core becomes risky. Easy to get shorts. Also i would recommend som stiff wooden wedges or similar to pack the winds as you go. And then patience can be the biggest challenge. personally i would go with round wire. How would a single thick wire add up?

I read up on skin and proximity effects. Common understanding i've seen so far is that it does not matter in our motors since frequency is so low, 3000rpm is 350 Hz for the revolt motor and at that frequency the skin effect is limited.

In this paper the effect of PWM on skin effect is studied: https://pdfs.semanticscholar.org/1e0f/b1f1c30169dff4038b24c4a3dbe0905ab13c.pdf

If i understand the results then 32% higher copper losses were caused by skin and proximity effect when comparing ac and dc resistance for 2.62mm wire when under PWM.

Results from the paper shows that the wire diameter should be below 1.5 mm to keep the ac losses low. I might use rectangular wire: http://www.wires.co.uk/acatalog/rt_ec_wire.html
The 3x1.5 wire looks like it can be a good choice, although it will be around 20 turns If fully wound. I think that is too big difference, kV will be 27 after the rewind.

Hey larsb,

Kudos to you taking this on and thanks for sharing your efforts.

I thought I might just mention form personal experience getting that last wind or two on the cad model is a whole lot easier than in practice :lol:

What I mean is you will probably only achieve 18 or maybe 19 turns when you wind what you plan above.

I don't mean this as a dampener at all, just thought it might help if you planned for that now rather than discover it when you are 80% wound! Hope it helps.

All the best!

D

Just skimming through this interesting thread again and rechecking your first post math.
It looks correct and I am just baffled there is so much potential for improvement.

Your theoretical fill with 1,5x3mm wire is 3times as much copper area.
Just sounds so big of a difference I ask if something is wrong with the calcs
But it seems the only wrong here is the crappy stock motor windings.

No wonder I didn't achieve good efficiency with my motor.

Also want to agree with Danny Mayes, its often very hard to get the last wind on.
They key is (if its even possible) to not "cheat" earlier in the wind. You really got to pack the winds tight and tidy from the very start.
Might be easier to get an overview with square wire. I never used square wire due to the price.

I totally agree, difference is huge - almost too large to be correct. As i measured the slots with windings still in place it can turn out to be smaller slots when i measure correctly but i think measurements are in the right ballpark.
My original windings look to be done in about 5 minutes per motor.

I picked up a new rv120 motor today (which is the one i will rewind). I got it from a fellow ebiker for 1500 SEK / 165 USD and it is brand new!

Some pics on the stator- it doesn't show but hall sensors are fastened with jammed bamboo sticks. It's the same stuff you use for barbeque spits, hope the guys at Revolt had some good meat to go with it!

There's a good amount of air instead of copper in the slots and there are many wires loose outside of the stator bundles

I have planned to add cooling fans to the motor so optimal winding is a balance between air flow and max copper fill / low copper losses.

only two layers of wire would be preferable for best cooling so that one layer has contact to the stator and the other to air but that would sacrifice a lot of fill. It would give me a 12-turn winding with 80% more copper than original and a lot of free space for air. Copper fill would be lowered to 12/20, only 60% of the estimation in the winding above. Too low.

To keep a decent flow through the motor i will try to keep 2 mm of free space between the windings on opposite teeth.
That would give a quite slack winding of around 16 turns of 3x1.5mm wire.

If the rectangular wires can be routed well then i think that will be my goal, it should be possible to meet!

larsb said:
I have planned to add cooling fans to the motor so optimal winding is a balance between air flow and max copper fill / low copper losses....

Change your plan to optimum winding being to fit as much copper as possible. The best cooling starts with make less heat to disperse.

You are absolutely correct John, i am also thinking like this :
With no load amps at 6-7 and stock phase-phase resistance of .026 ohm the iron losses are larger than copper losses at nominal power (72v/90A in my case)

'Edit:'
So the copper losses are not dominant and cooling of motor might be prioritized.. Or?
I think the large stock bearing and the two additional support bearings i use on the shaft ends add upp resistance. That's a total of five bearings involved. I will try to remove one of the support bearings to see the effect.

Right now i plan on using one 3x1.5 wire to keep the winding neat and easy, it is my first try of a rewind so i want to give myself some slack. That means max copper fill will mean more turns, 20-22 turns instead of 12 and i assume that will mean even larger iron losses since i need to get more torque from motor with less external downgearing to get same speed.

My assumption is that the best for efficiency is to downgear as much as possible without going to 2-stage gearing so that motor is running lower load at high rpm.

I think more parallell wires will be tough to wind well, any tips on how to get more parallel copper area?
If i would plan for absolute max copper then i would wind it LRK so i didn't have to 'tetris' wires from adjacent teeth as they meet in the middle.

All suggestions on how to perfect the winding are welcome!

Nice project ,will keep an eye on this one as thats a nice fill improvement and a lot of motivation ... its a shame the manufacturer has dropped the ball on some ie bad soldering ect one had a can badly out of balance, how is this one.

I have just bought this motor so i don't know how it is when used but my ebike with rv-120pro vibrates to the point that my privates go numb when riding at full speed. I am not sure though that it is only the motor causing this. I might post a video of a no load test at 3000 rpm.

I will post a lot of pictures after i have opened the motor (in roughly a week.) Looking forward to it!

It is unbalanced rotor and pretty bad one. The rotor's thickness is not even.
https://endless-sphere.com/forums/viewtopic.php?f=28&t=80516&start=100
Please take a look at pictures at the rotor of my RV160 which i made for middrive. I believe all revolts have the same problem.

larsb said:
That means max copper fill will mean more turns, 20-22 turns instead of 12 and i assume that will mean even larger iron losses since i need to get more torque from motor with less external downgearing to get same speed.

My assumption is that the best for efficiency is to downgear as much as possible without going to 2-stage gearing so that motor is running lower load at high rpm.

I would say it is no good idea to go up with the RPM on this motor, because the iron losses are already quite high at 3000RPM (according to the pic of your simulation).
So it would be better to go up wit the torque a bit, and here it is important that you have low copper losses.
The iron losses square with the RPM, while the copper losses sqaure with the torque (the phase amps).

I am not going for higher RPM since then i cannot down gear enough in 1 stage. My issue is that if i get 20turns instead of 12 then i need to produce significantly more torque as well as change gearing to get same target speed, i don't think that's good.

EDIT: laminations are 0.5mm when measured

(I think the big part of the iron losses from the calculation are actually secondary losses like bearing drag as the motor has *0.2 mm laminations*. The bavaria calculator just calculates iron loss as volts*noloadA so it's quite a simple estimation.)

I found a good description of how to double the kV independent of the turn count so that i can wind with thinner wire/more turns and get same RPM. It is done by parallelling the coils. I will probably do it that way so my new goal is to get >24turns on each coil. Sweet!

I will switch to thinner wire, 2.5x1.25 should be a good fit, will try it in the CAD when i have confirmed the slot geometry.

Parallel coils means i will wind coils separately, if i get a short then only two coils must be rewound. Good for a newbie winder but it will be a lot of soldering before motor is finished.

larsb said:
I am not going for higher RPM since then i cannot down gear enough in 1 stage. My issue is that if i get 20turns instead of 12 then i need to produce significantly more torque as well as change gearing to get same target speed, i don't think that's good.

if you get 20 turns instead of 12, than you would need to increase the battery voltage by 20/12 and lower the amps by 12/20 for getting same load on the motor.

i would first decide on the battery voltage (72V is the sweet spot in power density / efficiency of current available controllers).
than decide on the maximum speed of the bike and the gear reduction (i would not go above 3000-3500rpm on that motor).
now calculate the kV you need and wind it to that.

Several buyers have issues with the revolt motors, both my motors have really poor fill factor so i think they are all done the same way.

(I already have all details settled for my system: battery volt is 72v, gearing is 9/53 in a 26" wheel. I have tried Kelly and Adaptto controllers plus a lyen one and motor gets hot in 5 minutes of nominal riding.)

I think Madin is just highlighting what is probly a typo, but it reads as... if you have a 20 turn motor it is a higher rpm motor than the 12 turn. That should be the opposite.

So will you try the 3x1,5mm rectangular wire?
(Or will the 1,25x2,5 still be enough fill?)

good idea winding the coils separately. i think it makes it easier to add more copper when doing so.
only downside is the soldering work and probably it needs more space next to the stator.

Wheazel: I will decide wire when slot size is confirmed but i will most likely go with rectangular wire of some size.
2.5x1.25 looks best right now of the wires i've been able to find:
- I need to fit at least 24 turns per tooth with delta+parallel connected coils so 3x1.5 wire is too large.
- It is smaller than 3x1.5 and will be easier to wind
- the parallell connection of the coils will still give lower resistance and less copper losses.

Madin& kdog: relationship between turns and kV are clear for me, maybe i should clarify that original winding is 45 kV and 12 turns in series delta connection. I am aiming for slightly less than 45 kV and as i wrote earlier a wind with 20 turns in series coil connection give 45*12/20=27 kV and that is too low for my liking.

It gets confusing as the coils in parallell connection double the kV. In order to get a kV of 45 then 24 turns are needed. You can read the thread at rcgroups for a deeper description, this is the first i've seen regarding coil parallell connection. There's a turn vs kV calculator linked in the rcgroups thread that should help (haven't tried it yet but i will upload it when i get to a computer)

Here is the calculator
View attachment 1
"Manuel_v" on RCgroups is the creator of this doc, but it is free for use.

This is my result:
12 turn series delta shown as well as 24 turn parallell delta

resistance is not correct as the wire length is not really known yet.

Replies
1
Views
439
Replies
12
Views
8,516
Replies
106
Views
7,116
Replies
96
Views
9,170
Replies
9
Views
4,801