First post, first build: Hall-effect trouble?

[Wisco]

Hi folks,

This is my first post, though I've been referring back to ES throughout the build process and am emmensly grateful for this great resource and community.

So I got a 48v 1000 watt rear hub motor through Aosom on Amazon and installed it on my touring bike (Nashbar aluminum frame, touring triple chainring, drop bars). For the battery I soldered up a 14s5p pack of 18650s (full voltage around 58v). Besides the challenge of mounting the throttle and LCD on drop bars, and soldering protected cells (ugh) the build went smoothly and the bike is nothing short of amazing. Incredible acceleration, flatland top speed with no pedaling of 30-35mph. With a good breakfast and full battery I think I'll be able to hit 40.

But.. It has one quirk that I'm worried about. After turning it on, the speedometer remains at 0 for 30sec to a couple minutes (either pedaling or motored). I also got an error code "03" which is a Hall-effect sensor malfunction today. I checked the wires. Red to black has a steady 4.36v. Black to yellow, green or blue will drift around -100mv for a couple minutes, then start acting normal (alternate 5v and 0 as the wheel turns). The speedometer has a longer delay after I read normal behavior in the sensors. But once it starts working it keeps doing so until I turn it off again.

Is it normal to have a warmup period? I've only known Hall sensors to work or not.
Could the problem lie elsewhere? Since the red-black voltage is good I can't see blaming the controller or LCD/computer.
This pack is a bit higher voltage than most "48v", could that scramble anything?
If I want to get it exchanged I need to decide soon. I really don't want to strip all the gear off the bike now that I've tasted the sweet thrill of an ebike.

Thanks for any input!

 

[dogman dan]

In general, it's always just the plug to the halls, plug to the throttle, etc. Look hard at those connections, particularly for contacts backed out of the housing.

You can test the halls in the motor, by giving red and black 3-5v, and spinning the wheel. As it spins, look for voltage to turn on and off on green blue and yellow.

 

[Wisco]

Thanks for the input.

I'm going to get more systematic about my diagnosis and try to sort this out. Or at least report back here. I'll pull the Hall connectors (except red and black power) and see if there is still this weird warmup period.

I rode it to work today, 23 miles out into the suburbs. My goal is to bike it in under an hour. Today I got turned around by closed trails and unfamiliar roads and added a few miles. I ended up trying to make up time by going 30mph for several miles and my battery overheated and conked out. Not surprising, but really frustrating. I ended up pedaling the last 10 miles and getting into work late.

I'll have to increase my battery capacity and improve cooling, or find a route where I can maintain a more reasonable speed.

 

[DAND214]

14s LiPo is a normal voltage for a 48v system. Most to all 48v LiFePo4 packs are 16s and peak close to 60 volts. So voltage is not the problem.
Does the bike run when first turned on or do you wait? If it runs right away I would blame it on the display and or the controller and not the wheel.

Your 14s5p battery is what AH? If it's only 10 to 12 AH I would up the capacity if you plan on 20+ mile runs. You really don't want to run it flat each run. Which 18650s did you use? I see about 2 miles per AH if I keep it under 20mph and even more if slower.

Dan

 

[Wisco]

The motor runs great immediately, even if the display shows 0 mph. I agree that sounds more like a display/controller issue but I'm still scrambling to grasp the electronics involved. I haven't had a chance to do more difinitive testing, just got off a 12hr shift (at a bike shop, appropriately enough).

I haven't done a capacity test on my cells. They are Ultrafires, and the packaging has the outrageous claim of 5800mah on it, which would be 29ah. Realistically I've been figuring about half that, in the 15-20ah range. I think more battery might be needed, although if I used it more conservatively it might make the trip. My kit has a pedal assist attachment, which I originally dismissed in favor of a throttle only setup. But I expect it would improve my mileage. I also used protected cells, which were very affordable and I thought would offer some protection of I did something stupid. Although I really doubt the circuits could clamp down a short circuit when wired in series. I have wondered though, if a pack of protected cells would have a lower effective capacity because any cells that shut down from low voltage will take the string with them and cause a cascade.

I hope to get more concrete data tomorrow evening, and I have Wednesday off of work. I've alerted the seller that I'm having trouble, and it seems they will do right if I need parts/exchange etc. I'm not surprised, as the kit has almost entirely 5 star reviews.

 

[Wisco]

So... I think my bike is fine.

I retested the Hall-effect sensors and they are fine. I just had to give the wheel a spin before I got sensible results.

And the speedometer warmup period is still confusing, but I actually timed it and found out that it will only begin to register speed after 17 seconds at at least 1/3 throttle. At full throttle it still takes 17 seconds, from being unplugged from the battery it still takes 17 seconds, and at less than 1/3 throttle or under pedal-power it will not register.

Why 17 seconds, why 1/3 throttle? I have no idea, but the regularity of it makes me think it's a software quirk, not a hardware problem. I'd still love to hear from anyone if they think this is worth worrying about. Otherwise I'm going to keep an eye on it and chalk it up to new ebike owner jitters. I can either ride off and wait for the speedometer to kick in, or spin the wheel while up on the kickstand (double legged).

I still need to sort out my battery needs for a 23 mile commute, but that shouldn't be too hard with all the ES resources.

Thanks for the feedback folks, I'll toss up the model numbers for my components in case anyone else encounters this issue.