'lightest.bike' 1.7kg 1000w mid drive

[neptronix]

Update from here. Been too busy with business until now.

The gears have been waiting for a cleaning since i drilled the unit out and ended up with a ton of shavings on the gears.
If i could do it again, i'd put a wet napkin in front of the gears. but you live, you learn!



I wanted to avoid disassembling the entire unit and ending up with less than perfect factory perfect fit, so i kept the gear stack in.

1 hour of manual labor, 5 ear swabs doused in isopropyl alcohol and a super tiny but long screw driver later..
I have the gears free of all debris particles that can be seen with the naked eye, and no visible binding or added friction in the gears. Had to scoop up some extra lube off the sides and throw it on the gears afterwards.

Close enough for government work!



I threw some extra solder on to make sure these wires aren't going anywhere..



and then i ghetto rigged a way to tension the wires together so that as the bike is shaking, the wire insulation isn't getting beat up and eventually shorting.



Will give it a spin on the VESC and report back in some days from now.

 

[neptronix]

Latest update from the indiegogo:



Looks like they are still having problems getting product out of the door.

I'll see how mine goes at 25A batt, 26A phase ( x 52v = ~1300W ) in sensorless mode and base whether to buy another unit as a backup ( assuming i won't get it in a realistic timeframe ).

 

[rhauff]

and then i ghetto rigged a way to tension the wires together so that as the bike is shaking, the wire insulation isn't getting beat up and eventually shorting.

View attachment 370639

Looks like you need a P-clip right there..

Nice work, looking forward to the results! Still haven't received mine.

 

[SCBiker]

Newbie here, thanks for all the prior experience with this kit. I ordered my kit about 2 years ago and recently received it and finally got the first short ride in the woods today. This is my first ebike kit install, but I'm generally comfortable wrenching on bikes. A few thoughts/experience in case it helps anyone else.

I used the pressfit adapter which caused me the most issues as the loctite product fully grabbed before the insert was in place and I couldn't move it - ended up cutting it out and ordering a second through a more local bike shop (I'm in the US). For the second one, I waited to apply the loctite until the insert was almost in place, no issues.

The standard mount (originally recommended since my bike has a straight downtime) was not close to working with a 36 tooth chainring. The medium mount (I actually ordered the hybrid mount, but only used the medium mount from it) works, but not overly generous clearance from the chain to the top of the chainring.. but no problems on the short ride today.

I ordered the custom shaft for the BB.. no issues, good length, no noticeable wobble. You do not need any of the spacers for use with the custom shaft, but they are still in the install kit.

I have the 700Wh battery.. I wished that I would have gone smaller currently. The 700 (or actually 720) is too large for in my bike frame, so I have it on a rear rack. I extended the cable with 12ga, silicone covered wire to XT60 connectors.

First ride was muddy and slippery. The motor takes a pedal stroke to respond and surged some on me (on and off), so I have tuning to try to improve that. I have the 750W motor, superhero kit. Plenty of power for me as I wanted something for when I'm taking tools into the woods to work on the trails. The 36 tooth chainring is a limitation for top speed that will likely bother me down the road. Have connected the app through an old android phone - a somewhat newer android phone did not work with the app.

 

[neptronix]

Howdy.
You should try a medium mount or hybrid mount, on most bikes, this will give you the ability to run a 42T or larger.
You'd be the first person to mention no wobble in the cranks, i think. Wondering if they finally fixed it.

Best place ( for weight balance and handling ) to mount a battery that doesn't fit in the triangle is on the top tube and forward. Look into a grin technology bottle bob for a ready to go mounting solution.

Curious to see if you can get the torque sensing tuned to work right. Or if it remains pulse-y no matter what.

 

[SCBiker]

Thanks - I hadn't seen that offering from Grin technology before. I'm not anxious to give away my standover clearance, but will keep that one on the table as I look for options.

I hadn't seen any other notes (maybe missed them) on any one else trying the custom BB spindle? But maybe the machining for that or the supplier is different from the standard offering which has the wobble?

I have the medium mount currently installed (tried the standard mount initially, but it was a no go). I may try the hybrid if I encounter any chain slap hitting the chainring.

Yeah, don't have a game plan on tuning the torque sensing yet.. I think mud and slippery roots yesterday wasn't helping, but will share if I find any improvements that work for me.

 

[neptronix]

I would recommend it towards the front of the bike if you still need standover clearance.
There exist shun 170mm cranks you could use on this motor to reduce the standover problem ( now, seat can go down 5mm )

The wobble was in the cranks, not the bb thankfully. Haven't heard of a custom BB spindle experience on here.

You have the medium mount but can only fit a 36T? that's strange, can you post a picture of the bike with motor attached?

 

[SCBiker]

Here's a pic, 2 up from the highest gear.

 

[neptronix]

Dang.. this bike has an unfortunate shape.. i think you might get to 40T with the hybrid mount if you are lucky.
The dual chainring mode would be the way to go if you need substantially better than that.

 

[rhauff]

I ordered the custom shaft for the BB.. no issues, good length, no noticeable wobble. You do not need any of the spacers for use with the custom shaft, but they are still in the install kit.

Is your custom BB shaft ISIS or square taper?

Just curious, as the kit I have been waiting years for, is an ISIS custom shaft for fatbike. Wondering if that has been the holdup.

 

[SCBiker]

Mine is ISIS. I asked if I removed the custom shaft from my order if that would allow it to ship faster and they never pointed to that as the hold-up.

 

[neptronix]

FYI. This is what the VESC detects when plugged in sensorless. I used the 'medium outunner' setting when running autodetect.

 

[neptronix]

I get about 496rpm at 55.3v unloaded or 8.9kv when unloaded.

This sounds about right compared to the listed spec:


VSS drive mode does not work and just produces a crackling sound when the throttle is pushed. Same for all other modes.
Sensorless drive mode works, except for i can tell that the motor chokes when you are moving the throttle anything more than slowly.

 

[neptronix]

Had a bit of a twist in the plot.
Smelled something, then the motor kept acting funny. VESC tool wouldn't detect it, gave me ABS current fault and also failed to detect flux linkage many times.

I open the motor, the smell hits me, and i fear for the worst, that maybe my wiring or something odd happened.

Well.. turns out that the motor phase contact still touching the original controller caused a small capacitor to explode inside and there's probably an electrical short somewhere.

So i machined off bits of the board until only one phase was contacting



Plug it in.. no problems with VSS mode. Somehow the RPM of the motor changed to 470rpm from 495rpm, so about 8.5kv

The motor consumes about 1.1a at 55.1v. a 60 watt no load is super impressive for a motor with 2 gear reductions on it and 1000w capable!

 

[neptronix]

Applied the observer gain cut in half trick and in VSS mode i'm no longer getting the occasional stutter on initial throttle-in, seems to work the same as on a hub.. at least in the air
Trick details: VESC Tuning Advice, Poor Acceleration compared to other controllers

Ok... time to hook it up to the bike

Edit: Damn, i'm a quick link short of victory. You should hear more about this external controller quest tomorrow.

 

[neptronix]

Motor is together and i got the chainline within 2-3mm of perfect.
A little over 81F by the time i was done getting greasy.

Temps outside should be decent tomorrow morn. Will give it a spin in the next 1-2 days.

 

[limiky]

Motor is together and i got the chainline within 2-3mm of perfect.
A little over 81F by the time i was done getting greasy.

Temps outside should be decent tomorrow morn. Will give it a spin in the next 1-2 days.

Curious!!!!!

 

[neptronix]

Sigh. The controller won't fit in it's pocket. I don't have long enough wires, and there is not any room to make a kind of shielding for the mass of VESC wires.



This controller would also be very difficult to re-case.

Can't seem to find a VESC online that would fit into the pocket. all of them want to have wires along the wrong sides.
With the exception of this one that has oddball connectors and what looks like zero matching accessories.



I guess i'll make some extenders to pull the controller into the triangle and i'll take another run at this tomorrow.

 

[neptronix]

I finally figured out how to get the controller mounted in an almost sane way. Man, i really need that smaller VESC!



The VESC power button seemed to have broken on me. Some isopropyl sprays and picking at things later, i finally got it unstuck and gave it a short ride up my test hill course ( 2 miles of very flat, then very steep )

I programmed in 20A batt / 20A phase with 25A max overcurrent. I also added a 1 second gradual ramp up on throttle.
I could tell the motor was VERY underpowered, as it would make warbling sounds when getting started.
Noise level was basically identical to the previous controller.

Maximum power draw was about 600w on a hill climb, and i kept tripping the overcurrent limit if i hit the throttle and wasn't pedaling hard.
But this is my fault because it's supposed to be 2x the phase amp value, my setting was 1.25x.

Other than these things, it seems like the motor is driven well by the VESC, and behavior is similar to driving a hub motor.
Temperature of the case was basically just ambient temp.

My original stock 1000w controller had a lot more push at low RPM. So i think a 1:1 phase to battery ratio is incorrect for this motor. It may be limiting my maximum power at higher RPMs too.

So I'm going to give it another run in the morning tomorrow on:
25A phase
20A batt
35A max overcurrent limit

..and hope that this is a better imitation of the stock tune. And once i get to baseline, i'll start adding amps from there.

 

[neptronix]

So here's the imitation stock tune:



Raising the absolute maximum current to almost 2x eliminated the cut-outs, to a point where the motor will usually tolerate a worse case scenario where you're getting start in a very wrong gear and not pedaling. I call this good enough, considering that sensorless drive will never be perfect.

The motor still sounds wobbly, but less so.
What's strange is that i get the same power draw... 900w max.. but i should have a 1350w power draw at least.


I theorize that the lack of phase amps is the reason why we can never draw the rated power.
I also think the motor can handle higher phase amps because the sensorless drive and slow start wimping out the initial torque.

So i went to try an inverse proportion:



Now i can draw up to 22a peak, but this only happens in a very short part of the powerband, usually we're talking 14-19.5A.

Heat wise, it's 34C/94F out here and my tires are a little bit hotter than the motor despite the fact that i went full throttle up a 5% grade w/o pedaling.

Haven't even started touching this motor's thermal capabilities yet.

 

[limiky]

Awsome testing!

 

[neptronix]

I ramped it way up, knowing that the sensorless mode is a good safety buffer against too much initial torque.



It briefly hits 1050w on a 52v nominal battery @ 40% SOC and draws 24.7A max, but usually hangs out in the 900w range while climbing.

I went up and down the steep hill on full throttle and in the wrong gear for the job 2 times and it's about 95F at this point outside so i figured, i should have produced some heat by now.

..but my high score for temperature is only 55.5c on the hottest part of the case.



What's unfortunate is that as you add amps, you get more power in the mid band, but no extra down low or up high. That's because sensorless mode is dramatically clipping the power in low range.

I will fiddle with the sensorless control settings to see if i can get a little more low to mid range torque, so that the extra power is actually doing something.

 

[neptronix]

I keep tuning in tiny bits of more low end torque.. and the drive keeps getting better..
With dual 44T chainrings, i can now hit about 32mph with pedaling on a 80% charged battery, and i see a peak of 1100w on the screen now.

Cadence is around 90rpm at 80% charged, and that's good because it's still gonna be in the 80's or upper 70's as we get closer to 0% charge.



Not noticing any additional heat as a result.. this motor is awesome and has a big efficiency band.

Will continue working on it until i can hit 35mph.

 

[neptronix]

Update on their patreon page.
Good to hear for people still waiting.

 

[neptronix]

Running this thing on 36A phase / 30A batt and i'm seeing 1150w peaks now.

Efficiency is still impressive!
Rode it up the 1.5% 6 mile grade course while pedaling and in 1 gear that's non-optimal in an attempt to build some heat.
The last part of the course is the steepest. This was in 75F weather.



I managed to get the initial torque to be a bit too strong by dialing up the observer gain. Started letting out a little kick, lol. So that needs to be turned down a little.

Love that the VESC allows you to tune different parts of the power band. Makes it possible to dial in a clutch/gear preserving tune by emphasizing the power band in the mid to high range - almost like a gas motor

Next, i will try 30A batt, 40A phase.