Kelly KLS & Revolt RV-160E watercooled : cogging noise and overheating

[arnlej]

Hello,

Since I upgraded my boat to brushless I added acoustic noise & vibrations to my thermal issues.

I am obviously disappointed since the cost (about 2000€) and all my efforts in this migration.

I hope to find here some expertise to help me understand my mistakes and options.

My 6m displacement boat is parallel hybrid, with a 13hp Volvo Penta 2002 diesel and an electric motor on a 10kWh 53V LiFeYPO4 pack on the same shaft (25 mm, 2m, bronze propeller).

I can use the Volvo or the electric, and charge the pack with the Volvo (regen).

Peak power is 1000 rpm, approx 90 N.m. Cruise is around 600-800 rpm, 2 to 5kW.

My previous setup was a brushed Perm PMG 132 with a belt transmission to the shaft (a Bellman product, quite well designed).

It was inaudible, i could only hear the water… a pure pleasure !

But I had power limitations (due to the belt) and cooling issues. And the belt pylon was taking a lot of space.

The motor eventually died from overheating.

I decided to replace it with a compact water-cooled solution :

- Kelly KLS-72701-8080H sinuziodal controller on a big cool plate.

- Revolt RV-160E kv26 with water-cooling, directly on the shaft between the Volvo and the propeller.

- 17l/min @ 24V Flojet water pump from the river (14 to 25°C) through a filter, targeted to use at 12V (quieter).

I am very happy with the power performance and size, but I have 2 major issues : noise and thermal.

1) the motor is noisy

Even if i turn it by hand, i can feel and hear the cogging resonating on the shaft and boat hull, despite the silent-blocks.

I though cogging was a problem at low speed but i feel that noise amplifies with speed. When cruising serval hours it ruins my pleasure.

I am now hesitating to replace the 6 mm 10x10cm aluminium plate that hold the motor stator to its mount with a 15mm rubber plate. I have done no calculation, not sure it will hold the torque, and not sure it will absorbe the vibrations anyway, because they are also transmitted to the shaft.

I was not familiar with the cogging before. I now read it is a tradeoff with efficiency, noise, size…

I imagine it is not as crucial for a bike with other noises (tire/street) and shorter trips.

If I have to replace the motor to get a silent cruise, how to be sure there will be less cogging, ie for example does Golden Motor have less cogging than Revolt ?

Unless I go for a slotless motor, which does not seem available for those rpm & torque.

2) despite water-cooling the motor gets way too hot

For example :

- at 2000W the motor's temperature stabilises at 80°C after 30 minutes.

- at 5000W (100A) it goes > 130°C in 12 minutes (while the controller stays < 30°C).

As Revolt told me, it is possible I am not using the motor on its optimum working point : the perfect matching would have been kv22 or kv24 instead of kv26. Still it is far from expected : RV160E is marketed for continuous operations at 150°C, 15kW continuous power.

Last, the efficiency might be bad : my BMS displays a current on the battery pack about 30% superior of motor current from the controller.

Revolt confirms those temperature are not normal, and tells me to adjust the timing in the controller. I am not sure we can do that in the Kelly KLS. do you think i should get the « Timing Adjustment Tool » ?

Sorry it was long, i wanted to give you enough details.

Thank you so much for your ideas,

Arnaud

 

[fechter]

One possibility is you don't have the hall/phase wire combination correct for that motor. Just because you match the colors on the motor wires to the controller doesn't mean it is correct. It may be hard to determine the correct phasing with the load attached. If you have a way to disconnect the motor (remove belt?), you can more easily determine if you have the right wiring. There are several topics on how to determine the correct wiring combination. Basically you do trial and error on different combinations of the wiring and look at the no-load current as you give a little power to the motor.

If you actually do have the correct wiring combination and it's still noisy, you may want to find a different controller that has a sine wave output. This greatly reduces the noise and improves efficiency. Not sure if that model of Kelly is sine wave or not.

 

[arnlej]

Thank you Fechter,

the Kelly KLS controller has a sine wave output and has a learning mode that automatically tries and find the combinaison, moving steps by steps.

Since the controller has completed that process and the motor is running (from rpm 50 to approx 900, neutral and reverse without stalling) I assume it did find the correct combinaison.

Do you suggest it might not be the best one ?

 

[larsb]

Sad to see another Revolt owner bite the dust.

Revolts comment on your kV is not true, they are only out to protect themselves. Main problem with their earlier motors has been that Revolt hall sensor placement kills motor performance, this can be improved by placing new sensors or going to an encoder. I have tried both and it works well.
I don’t know what the halls look like for the 160E motor, it’s new, could be the same issue also on the 160e since Revolt don’t know what they’re doing. Their comment about phase angle points to sensor timing issues.

Second thing is that if motor runs at only 900 rpm and 90Nm on a regular basis then this is way too high torque for this motor. You should gear the motor with a belt, this’ll help with the heating issues but at your low kV it will be too low rpm to gear down i guess. A 1:2 or 1:3 gearing with HTD-8M pulleys doesn’t need to take a lot of space and will manage the torque.

The noise will be mostly the same no matter what you do, but a rubber spacer between motor flange and mount should help with the noise transfer, could be worth a first try. to get a really silent setup you need to switch motor. outrunners of low quality won’t do :wink:

How much power do you need? A QS inrunner motor would probably be good for your application. They are really good and cheap.

 

[arnlej]

Thanks Lars, i read your posts about rewinding the Revolt and improving the cooling, what a journey !

On the E version Revolt put the sensors on a PCB to improve their placement. Do you think it will still benefits from new sensors & encoder ?

The problem I have with a belt is it takes way more space than directly driving the shaft, so the motor does not fit bellow the deck of the boat. It is why I was happy to remove the previous belt system.
I now understand the belt would be great for a brushless as it would absorb part of the cogging vibrations.

old setup : pulley & belt, the motor was on a pylon, above the deck, requiring an extra cabinet and ventilation.


new setup : direct drive & watercooling, the motor is bellow the deck (old picture, work in progress)


I though I could go direct drive with a low KV. at KV=26 and 50V it makes 1300 rpm with no load, so about 1000rpm at full load, close to what I get.
Maybe I made a design mistake : It would be more efficient to select the motor not for full load (10kW, 1000rpm) but for cruise (5kW, 850rpm), so this is why KV 20 to 24 might have been better. But I wanted a reserve of power to go full throttle for few minutes, useful for emergency situations.

Thanks for QS, I did not know them. I dont see watercooling but maybe i should ask them, as well as low KV and shaft on both sides.

I am also starting to think maybe I should go back to a DC brushed motor, in my experience, contrary to what I read everywhere, the noise of the brush is way more silent than the vibrations caused by cogging. I would be more than happy to be wrong on that !

 

[larsb]

The kV part is a misunderstanding.

1) Every kV has the same torque capability if the winding is done to same copper fill since torque relates to the amount of steel in the stator.
kV dictates what voltage and current is needed for the correct rpm and also what phase current you need for full torque so it has to be balanced with your controller and battery.

2) motor peak power is always at high rpm since P=M*n/9.55
This means that a 15kW peak motor with a peak rpm at 10krpm is only a 1.5kW motor at 1krpm. This is why it is beneficial to gear it down from full rpm.

Cogging is not related to brush/brushless as long as both have slots in the stator. Torque ripple from commutation is also present on both motor types

 

[larsb]

Beware also that revolts are known for poor windings. If you are up for rewinding the motor then this’ll probably help but i doubt it’ll be enough.

5kW at 800 rpm is 60Nm and it’s needed continuously. Not many direct drive motors that can do that.

The brushed motor diameter looked huge. What gearing was it for this?

I thought about your case and the major improvement for both heat and noise would be to gear. A 20:60 thooth htd8M pulley gearing is only 150mm in diameter for the larger pulley. Should fit with some thoughtful packaging and is not so expensive.

 

[arnlej]

Thanks a lot Lars for the clarifications

The brushed was a pancake motor, larger and thinner. Its reduction ratio was 1.47:1, with a nominal 2300 rpm under 48V.
Ideally that ratio would have been 1.7:1 due to the propeller, which I could not change because it matches the Volvo motor.

In addition to the size of the system, the poor air cooling, I had also power limitations due to the poly-V belt : I could not change the ratio without big modifications (unmount everything to change both pulleys, switch to a track belt). So I was limited to 2.5kW.

When the motor died, direct-drive seemed the perfect solution and I based my kV choice on the Revolt graph :

http://www.revolt.org.il/images/companies/1/RV-160E%20graphs/RV-160E%2026kv%2048V.pdf

I though : if the motor has enough torque for full speed (1000 rpm), It could easily drive at lower speed.

- I am extremely happy with the power, speed. It is a very compact, easy to integrate motor at a good price.

- Very disappointed about the vibrations and very limited water cooling

- Concerned about the efficiency

About the cooling Revolt told me that it requires air-cooling in addition to watercooling.

Maybe that motor is too small, it would have needed thicker wire.

I should have taken pictures of the winding before installing the motor, it is difficult to see now but i dont feel there is a lot of free space :


Actually I wonder if the winding has been damaged by high temperature (which never exceeded 130°C according to the temp sensor, but maybe the temp sensor was off, it died by the way)

Does this look normal to you ? (is it glue or melted wire ?) Maybe i should check for phase to phase resistance ?



thank you,
Arnaud

 

[fechter]

The windings look overheated.

 

[larsb]

Yes, that winding is cooked. Time to either rewind or switch motor before a winding short circuit takes out your controller also. Don’t use it anymore!

 

[arnlej]

Thank you both,

you were damn right !

It actually went very bad just after I posted my last message, I tried to fix weird controller throttle issues, the motor stalled at startup and burnt in awful toxic fumes in few seconds. It was medium rare, it is now well cooked.

I just hope the controller is not dead. I could reflash it, it seems alive. I ordered a random cheap 36V sensored BLDC to test.

So here i am.

Long term solution is to switch to a less cogging and more efficient - or a least better cooled - motor.
As absolute silence is my priority I start to think an axial flux, with multiple stack and proper watercooling could be great. If only it was affordable.

Short term (coming days) is to fix the motor, with the boat in the water, to enjoy the rest of the boating season despite covid rain and cold weather
3 options :
- rewind myself (+ change the temp sensor)
- send it back to Revolt (expensive)
- buy a new stator (fastest)

Revolt proposed me those 3 options, and offered a reasonable price for the new stator, but is not sure it is the best solution since it might brake the PCB when putting the rotor back in place.

Lars, how easy was it to open your motor ?
I read you built a special tool, jig screw, would you have a picture ?

Thanks,
Arnaud

 

[_GonZo_]

I made some videos about how to set-up Kelly controllers:
I think if you check them out, you will be able to find different solutions for your problems.
As there is many parameters that can help you with the noyse issue and with overheating, etc.
If you can not find solutions, please just ask and I will try to help you.

Here is the first video:

[youtube]A_si4IfU5so[/youtube]

 

[larsb]

It needs a puller to open, not hard really, just some screws and a pulling bar with holes matching the revolt

Rewinding or switching stator will not solve your issues with noise and heating though, gearing will help you with both.

The major improvement for the noise will be decoupling the motor from the shaft and adding some rubber dampers and the major improvement for the heat will be gearing it down. 1:3 reduction will lower the needed current with 3x and the heating 9x as copper losses are squared relative to current.

 

[arnlej]

Thank you,

I think I would need a complex setup to fit in the tiny space, maybe a 90° right angle gear + a pulley & belt. this is too much modifications for right now, I will look at that option next winter.

Do you think it is feasible to rewind just removing the cover of the motor (without removing the back black part of the rotor) ?
This would allow to leave the boat in the water... but maybe i am too optimistic.


I asked Emrax for the quote of their motor 208 with watercooling. They just told me that I could not plan to use the hall sensors together with high torque because the magnetic field from the high currents (>165Arms) will disrupt the hall sensors. It needs an encoder.
I wonder if there is the same problem with the Revolt, or if it is radial flux specific.

 

[larsb]

This problem is not related to axial or radial flux, it is a generic problem for hall sensor installations and revolts have been especially poor in this regard, at least before the hall sensor pcb:s.
Problem possibly exists also now on the RV160E, can be verified with an oscilloscope.

I have RLS encoder partnumber in my threads if there is a need for it for the revolt.

 

[larsb]

Thank you,


Do you think it is feasible to rewind just removing the cover of the motor (without removing the back black part of the rotor) ?
This would allow to leave the boat in the water... but maybe i am too optimistic.
revolt_open.jpg

It will be impossible to do a rewind with good results without removing motor.

 

[arnlej]

Thank you,

I did remove the motor from the boat, was not easy, and just opened the motor.
Well... everything that could burn did burn.


The reason : i had to move the boat (9 hours) to go back to its port.
Since the Revolt had stalled and partially burnt the week before, it shorted some windings.
Having the Volvo powering the shaft made the magnets spin, generating big currents and high temperature, a lot of smoke during the first hour, and at the end of the trip big vibrations.
This is a limit of my parallel hybrid direct drive setup : if the electric motor is dead, I can't remove it, it is in the way between the diesel and the propeller.

For now I will ask Revolt for a new stator winding as it is fast and quite reasonably priced.
I will have to replace the bearings too, as they do not spin correctly anymore, and do a lot of cleaning.

For the future, I wonder if i could address my two problems, noise and overheating, by stacking 2 stators & rotors (custom shaft / custom bushing), and shifting their angular positions ... could that make a 24N28P instead of a 12N14P ?
but maybe overlapping teeth would kill performance.

 

[fechter]

Holy crap, that is some serious toast. The magnets in the rotor are most certainly fried too.

So when it was "working", was the motor also turning the diesel engine?

 

[larsb]

I must say this: switch motor instead of throwing more money at revolt motors. Fechter is probably right that the magnets will be damaged now, that’s one of the most overcooked motors i’ve seen here.

The QS 138 90H motor would probably work for you in this drive application, also the qs138 70 motor with integrated gear (https://www.aliexpress.com/item/1005001515841817.html?spm=5261.ProductManageOnline.0.0.7c004edfcrfahE&fbclid=IwAR3B3RrTTphuAqVwaOXSwbdE2qZs8qNfVWZv2WvSkNgXZraNbUad7bRfvkQ)

No matter what, gearing will probably be the best solution since it’s a continuous high torque/low rpm drive.