Mongoose Blackcomb

Show off your Ebike creation here.
User avatar
APL   100 kW

100 kW
Posts: 1056
Joined: Aug 06 2018 8:44pm

Re: Mongoose Blackcomb

Post by APL » Jan 01 2022 2:19pm

Often wondered..
Push button and resistors parallel to the main switch?
Push button, fill caps, throw the main switch, .. yes, no?
(Light bulb instead of resistors?)

Barncat   100 W

100 W
Posts: 269
Joined: Oct 26 2020 2:49pm

Re: Mongoose Blackcomb

Post by Barncat » Jan 02 2022 11:13am

APL- i guess you're substituting switches for plugs in my proposed wiring layout? they'd have to be high current rated, plus you need to be able to routinely disconnect the cables to charge the battery anyway. i'm still trying to devise some elegant routing of these exposed power cables, complicated by the triple leads on the battery side of the vesc, and large unwieldy emergency battery cutoff switches... i've been a bit sidetracked thinking about improved freewheel designs for the 80T sprockets on my other builds.

User avatar
APL   100 kW

100 kW
Posts: 1056
Joined: Aug 06 2018 8:44pm

Re: Mongoose Blackcomb

Post by APL » Jan 02 2022 2:14pm

I guess I don't understand what the triple vesc leads are,.. unless they all go parallel? Anyway, I would never go without a main
switch on the battery, although, I don't know what your overall circuit looks like,.. just never know what's going to happen, (or
the batteries could do the slow drain too.)

Not an expert, but it's my understanding that e-shock comes from the higher voltage batteries trying to fill the caps too fast, like
high pressure water going through a small hole.

It's all voltage and not much current, so a small momentary push button switch with resistors across the main switch should slow-fill the caps enough to allow you to close the main switch without as much e-shock.

Hope I got that right,.. I have to do the same thing on my bike, so it would be nice to get it worked out. I haven't had any trouble with the grand slam connect yet, but moving up to 60+ volts now.

Barncat   100 W

100 W
Posts: 269
Joined: Oct 26 2020 2:49pm

Re: Mongoose Blackcomb

Post by Barncat » Jan 02 2022 6:13pm

I'll be ganging the 3 reds and 3 blacks to the respective 8ga battery pack plus and minus. I guess the 3 xt60's distribute power into the board better and allow more flexibility... mxlemming could give us a better explanation.

I'll try the setup as per my diagram and report back in a while.

Barncat   100 W

100 W
Posts: 269
Joined: Oct 26 2020 2:49pm

Re: Mongoose Blackcomb

Post by Barncat » Jan 06 2022 12:07pm

The anti spark is a pain. I've got 5x20 ohm resistors parallel on an xt30 as a précharge for 20s.
mxlemming- so that's effectively 4ohms total correct?

mxlemming   10 kW

10 kW
Posts: 866
Joined: Jul 17 2020 7:56am

Re: Mongoose Blackcomb

Post by mxlemming » Jan 07 2022 3:30pm

Barncat wrote:
Jan 06 2022 12:07pm
The anti spark is a pain. I've got 5x20 ohm resistors parallel on an xt30 as a précharge for 20s.
mxlemming- so that's effectively 4ohms total correct?
It is.
Hold my beer while I divide by zero :flame:

mxlemming   10 kW

10 kW
Posts: 866
Joined: Jul 17 2020 7:56am

Re: Mongoose Blackcomb

Post by mxlemming » Jan 07 2022 3:54pm

Barncat wrote:
Jan 02 2022 6:13pm
I'll be ganging the 3 reds and 3 blacks to the respective 8ga battery pack plus and minus. I guess the 3 xt60's distribute power into the board better and allow more flexibility... mxlemming could give us a better explanation.

I'll try the setup as per my diagram and report back in a while.
Nearly missed this.

The 3 cables are the result of a not quite misunderstanding about how inverters work, and because it's easier to fix 3 small cables than 1 huge one.

At low speeds, high phase currents, the majority of the time the current is recirculating around the PCB. Only at fairly high duty cycles (speed) does the current going mainly straight from battery to motor.

So yes, just wire them all to the battery, red to red, black to black.
Hold my beer while I divide by zero :flame:

Barncat   100 W

100 W
Posts: 269
Joined: Oct 26 2020 2:49pm

Re: Mongoose Blackcomb

Post by Barncat » Jan 18 2022 6:24pm

i finally got around to wiring the HI200 in.

i used 5.5mm bullet connectors on the phase wires and covered them of course with heat shrink. soldering three 12ga battery side wires to one 8ga wire is not easy. after tinning the ends with the help of some flux, i fashioned a copper strip to sort of crimp all the wires together then hit the whole joint with a torch. there are probably more sophisticated ways to do it... then it was wiring the throttle to the, like, 30 ga wires and plug supplied with the vesc. it worked but they are too thin to inspire any confidence. i interrupted the +5V throttle wire with a simple toggle kill switch- which works as intended.

i fired up the vesc tool and got the thing spun up without much trouble, but from a dead stop (it's sensorless) on the test stand the motor is grinding and hunting for a way to get rolling... didn't have this problem with the Flipsky vesc. out on the street even if you're barely moving at all there is no problem with applying throttle though since motor is spinning a bit. may or may not be able to sort that out.

more annoyingly, i'm hitting an upper limit at full throttle that is cutting the motor off with an unpleasant shudder. i suspect the current may be limited in the vesc tool settings but did not spend any time analyzing that after working on it all day. it didn't cut out at full throttle on the test stand with no load, so it's only evident on the street. rather difficult and unsafe to try to look down to see if the red fault light is flashing on. there are a number of advanced parameters i don't yet understand in the tool.

a bigger problem may be that this motor is just not stout enough to be really robust in this application, and it can only be mounted from one side. not sure... i can beef up the mounts to some extent. more testing tomorrow.
Attachments
20220118_174934.jpg
20220118_174934.jpg (2.97 MiB) Viewed 569 times
Last edited by Barncat on Jan 19 2022 9:39am, edited 2 times in total.

mxlemming   10 kW

10 kW
Posts: 866
Joined: Jul 17 2020 7:56am

Re: Mongoose Blackcomb

Post by mxlemming » Jan 19 2022 5:47am

You might need to set it up again.

Try measuring things multiple times in VESC tool and see if it's stable. That motor probably has 5-25mohm resistance, something like 10-20uH inductance and 4-8 mWb (might want to calculate that from like issues and kV). If you get bad parameters it will behave like this. These motors can be tricky.

Try asking T motor for the resistance and inductance.
Hold my beer while I divide by zero :flame:

Barncat   100 W

100 W
Posts: 269
Joined: Oct 26 2020 2:49pm

Re: Mongoose Blackcomb

Post by Barncat » Jan 19 2022 5:35pm

after more test riding today the vesc is still cutting the motor out by either cracking the throttle open quickly or after slowly ramping up to full throttle. played around with simple voltage and current limits to no avail.

a new wizard setup yielded motor resistance @ 10.4 mOhm, motor L @ 4.95, and flux linkage @ 3.48, and it set motor current max @ 138.76 A- which seems an odd arbitrary #... current KI is 10.39, and observer gain was 82.4, don't know what either of those terms means. mapping the ADC throttle is easy enough and that's working well otherwise.

mxlemming- trying to diagnose a problem like this remotely is basically impossible, but thanks for the help. could be anything, but probably just needs a few parameters adjusted.

my advice to anyone concerned is stay the hell away from vescs if you're not an electrical engineer. [EDIT :warn: that's not entirely true, you can fake your way through it with enough study and finding help/relevant info online. the vesc tool itself is fairly easy to use once you know your way around it. keep reading]
Last edited by Barncat on Jan 23 2022 11:02am, edited 1 time in total.

Barncat   100 W

100 W
Posts: 269
Joined: Oct 26 2020 2:49pm

Re: Mongoose Blackcomb

Post by Barncat » Jan 21 2022 11:18am

spent a few hours yesterday tweaking vesc parameters, but the system is still failing at its most important task- putting power to the ground.

some 4 pages of hastily scrawled notes attest to the debacle... in retrospect, this HI200 is doing exactly the same thing as the previous Flipsky vesc- which i attributed to that unit being slightly underrated for voltage. both vescs are brand new, and i don't suspect any issues with the Tiger Motor U15. if the motor has any problems it's not obvious, but it is one of the last explanations i can think of.

also, no fault codes have been thrown by the vesc despite the numerous cutouts under load on the street. motor runs flawlessly in the shop with wheel jacked up and no load, plus i was able to minimize the sensorless start up hunting to negligible levels by reducing the switching speed to 30khz from 40. i've measured zero resistance on all of my heavy DIY battery cabling.

this loaded motor cutout is a common problem over in the esk8 community, with dire results when that happens- getting pitched forward off your longboard. of course that whole sport was invented by ER docs and orthopedic surgeons... most of those boarders simply had their max total amps set too low. i've set mine at 475A so that's not it, in fact there doesn't seem to be much palpable torque difference when playing with motor amp settings, though if you can never get on the throttle it's hard to tell.

this is pretty damn frustrating. this bike should be a serious lightweight high performance ride, feeding on (75) brand new Molicel p42a 15s5p.

one of the last things i can think of is that the motor detection- done three times including with chain and rear wheel disconnected, all with the same results- is that the vesc is saying 14 poles. the U15 has 36 stator teeth and apparently 42 magnets, 36N42P according to spec sheet (i originally thought i counted 36 magnets but closer inspection reveals that incorrect). so 14 poles seems like a bad # to me, and i don't know whether the vesc means pole pairs, or the number of magnets (update on that- found a prior response by BV himself on vesc-project that says it means number of magnets). could this presumed error simply be causing the vesc to get out of sync with the motor- cutting it out under load????
Attachments
20220121_102732.jpg
20220121_102732.jpg (2.89 MiB) Viewed 514 times
Last edited by Barncat on Jan 22 2022 3:03pm, edited 1 time in total.

Barncat   100 W

100 W
Posts: 269
Joined: Oct 26 2020 2:49pm

Re: Mongoose Blackcomb

Post by Barncat » Jan 21 2022 1:51pm

just took it out for a test ride with the setting changed to 42 poles. no improvement.

ran it in BLDC mode too, which seemed promising, but it added unpleasant high frequency noise. seemed like it was building up to a nice top end on the street but then stumbled badly at near full throttle just like FOC.

that's all she wrote folks. i don't believe any vesc can drive a U15. i'll be selling the parts and swapping in a different motor and controller. [EDIT- i was wrong :warn: keep reading]
Last edited by Barncat on Jan 22 2022 3:03pm, edited 3 times in total.

mxlemming   10 kW

10 kW
Posts: 866
Joined: Jul 17 2020 7:56am

Re: Mongoose Blackcomb

Post by mxlemming » Jan 21 2022 9:12pm

I've written a sensorless FOC algorithm that I ride around with on my ebike in less time than I've spent trying to get a VESC I built to run reliably.

VESC has a lot of proponents but there seems to be a healthy dose of it just doesn't work for some cases.

Bad luck.

There are other motor controllers that might run that u15.

Don't waste time on the poles, it's not used in the control algorithm.
Hold my beer while I divide by zero :flame:

Barncat   100 W

100 W
Posts: 269
Joined: Oct 26 2020 2:49pm

Re: Mongoose Blackcomb

Post by Barncat » Jan 22 2022 1:22am

mx- Thanks for the advice. It was a live and learn situation. I wasn't aware there was any significant risk of this system to fail to function properly.

I just ordered another 3000w MY1020 style inrunner since I've had good results with those so far, they're available, not very expensive, and will fit nicely where the U15 is. Will try that with the vesc to see if I can put bursts of 150A or more through it and gear it a bit taller than my other bikes, since the battery is 15s not 20.

Probably any other controller that'll run the U15 will be pretty pricey, so may not go that route with no assurance of functionality.

kubark42   1 W

1 W
Posts: 59
Joined: May 15 2020 7:08am

Re: Mongoose Blackcomb

Post by kubark42 » Jan 22 2022 10:31am

Barncat wrote:
Jan 21 2022 1:51pm
i don't believe any vesc can drive a U15. i'll be selling the parts and swapping in a different motor and controller.
Here is Benjamin Vedder himself using a VESC to drive a U15:

https://vesc-project.com/comment/8346#comment-8346

The video demonstrates tuning strategies, although that was with the 5.2 firmware whereas 5.3 (just released this Sunday) includes many refinements for motor identification. In fact, based on the things we did I would say that if you weren't using the latest betas as of November last year then your old motor settings were just Wrong, capital W and all.

My v5.3 experience with a 20kW outrunner was that my VESC 100/250 worked perfectly on the first try.

Barncat   100 W

100 W
Posts: 269
Joined: Oct 26 2020 2:49pm

Re: Mongoose Blackcomb

Post by Barncat » Jan 22 2022 11:54am

kubark42- that's fantastic news! thanks, and i stand corrected. i obviously didn't know that info existed- will study very closely. and i'm using the 5.2 firmware, so will immediately update and start over. maybe i'll be as fortunate as in your case.

i have some skills with fabrication and design, but i'm NOT an EE. i didn't realize when deciding to go VESC/U15 that it was a fairly exotic combo. got a good deal on the motor, and figured vescs are tiny. i like low weight high power.

according to that thread this high pole/stator motor type saturates early and freaks out, which finally describes and confirms electronically what i've been feeling on short test rides. fortunately i don't think i've harmed any components as i don't beat on things that aren't working properly.

will report back. you may have saved the day.
Last edited by Barncat on Jan 22 2022 3:01pm, edited 1 time in total.

Barncat   100 W

100 W
Posts: 269
Joined: Oct 26 2020 2:49pm

Re: Mongoose Blackcomb

Post by Barncat » Jan 22 2022 2:45pm

kubark42- dude! you're the man, and well BV of course... that was a golden nugget of information :thumb:

i spent barely 5 minutes tweaking parameters after watching the video link in that thread and bike runs great. now i can focus on refinements and squeezing max power out of it. i didn't even bother downloading v5.3.

i'm running at 60v, and set motor amps for the moment at a conservative 150A and absolute max amps to 280A, left the motor resistance unchanged @ the detected 10.6 mOhm, changed the inductance to 7.5uH, set the flux linkage to 2.50 mWb, KP is at .005, KI is at 7.00, and set observer gain way up to 1000.00 as per the vid- up from only 82 as detected.

as is, it's only about as fast as my two other builds at 45ish mph, but i plan to boost that. the new settings also eliminated almost all of the minor "judder" i attributed to mechanical motor mounting and drive train stuff. so it's smooth and quiet, with just a touch of pleasant high frequency whine. it is however still hunting badly to start spinning from a dead stop with no load on the test stand, but even the faintest push of a foot on the street makes that irrelevant. if you know of ballpark settings to clean that up or where to find them i'm all ears. thanks!

mxlemming   10 kW

10 kW
Posts: 866
Joined: Jul 17 2020 7:56am

Re: Mongoose Blackcomb

Post by mxlemming » Jan 22 2022 9:23pm

Wow, I have to admit I've learnt a lot about VESC from that one video. I'd never have thought to change the observer gain (certainly not by a factor of>10), in fact I'd seen that it was directly linked to flux linkage so assumed it was best to leave well alone assuming the flux linkage was correct.

I'll have to go back and revisit some of my past attempts.

Do you have an oscilloscope? If so, could you capture that u15 motors back EMF (voltage on the terminals as you spin in by hand/with a drill).
Hold my beer while I divide by zero :flame:

Barncat   100 W

100 W
Posts: 269
Joined: Oct 26 2020 2:49pm

Re: Mongoose Blackcomb

Post by Barncat » Jan 23 2022 11:40am

yeah, it seems risky to be multiplying settings by 10 or 12, particularly for a layman such as myself. one main point of that thread and the video was if it's malfunctioning- stop immediately. common sense. seems to be a lot of latitude in settings otherwise, but if you don't even know what prudent ranges are it's a tad intimidating. my main rule- don't electrocute yourself. be careful with amp settings and so forth... i still need to make a little lexan cover for the board.

and btw, as you suggested earlier mx, if you lose the throttle ground the motor will spin up uncontrollably so that's bad... tested that by unplugging from the comm port. that renders my throttle kill switch worthless. don't know yet if there's a setting in the tool to address that.

i need a disconnect switch in line on the positive battery cable. is anyone aware of a proven smallish low profile switch instead of those bulky 2"x3" keyed plastic units? i'm considering making one as a last resort. the available stuff won't fit well where it needs to go and is unsightly.

anyway, electronic tuning is an interesting analog to tuning the dozens of Mikuni VM carbs i've done over the years. don't have a scope. just a DMM and clamp meter. i should download v5.3 vesc tool and compare settings now that i have a baseline. still need to clean up the sensorless start a bit, which pertains to openloop/ HFI /switching speed stuff, and the newer tool might do it automatically.

mxlemming   10 kW

10 kW
Posts: 866
Joined: Jul 17 2020 7:56am

Re: Mongoose Blackcomb

Post by mxlemming » Jan 23 2022 6:24pm

Barncat wrote:
Jan 23 2022 11:40am

and btw, as you suggested earlier mx, if you lose the throttle ground the motor will spin up uncontrollably so that's bad... tested that by unplugging from the comm port. that renders my throttle kill switch worthless. don't know yet if there's a setting in the tool to address that.
Yeah, this is why I:
1) Run a solid disc brake on the front
2) use RCPWM as my throttle signal - if it goes out of bounds/disappears, it is very easy to detect and just shut down.

Barncat wrote:
Jan 23 2022 11:40am

i need a disconnect switch in line on the positive battery cable.
Try an XT90 loopkey (male XT90 with the pins connected and female one in with one side to battery, other to motor controller)in such a place you can just rip it out.

Just take care not to plug it into a battery.


If you get a chance,meet someone with a scope... please take a shot of the back EMF. I have a hunch why the observer needs the gain that high, and it is nothing to do with the motor saturating. I am expecting it is very very not sinusoidal, and thus can appear at certain points in the cycle to the observer that the motor is pretty much going in reverse.
Hold my beer while I divide by zero :flame:

Barncat   100 W

100 W
Posts: 269
Joined: Oct 26 2020 2:49pm

Re: Mongoose Blackcomb

Post by Barncat » Jan 24 2022 12:02pm

yes, i have disc brakes front and rear, and built both wheels for that and because i needed a 24" front and 20" rear.

sometime in the next few days i'll try building a disconnect switch, a quarter turn affair with 1/8" copper strip sandwiched between clear polycarbonate. since there's no shutoff on the vesc it's an item i need routinely not just for an emergency.

who knows, maybe i'll become interested enough in electronics to pick up a used scope. it's the sort of learning that would be far easier with an in-person EE friend though.

i put about 15 miles on the bike yesterday. i bumped the motor amps up to 200 without needing to reset any other values and it ran fine other than the throttle briefly going dead after WOT a few times, which i attribute to an override from absolute total amps still being set perhaps a bit low at 280- will raise that a bit and continue test piloting. dumb question: what's the relationship between "motor amps" and phase amps?

i've limited battery amps to 225 at the moment btw, the theoretical limit for 5p of Molicel p42a. when i was designing the bike last winter, affordable vescs were limited to about 14s, so i figured 15s was the most i could get away with and that the high amp rating of the p42a would enable me gear it taller @ 5:1 and make speed with torque rather than voltage. turns out torque is overrated, and 200A vs 150A was hardly breathtaking... voltage is where it's at if you want to go really fast. not sure it bothers me enough to build another battery, and this vesc is only rated to 16s...

mxlemming   10 kW

10 kW
Posts: 866
Joined: Jul 17 2020 7:56am

Re: Mongoose Blackcomb

Post by mxlemming » Jan 25 2022 11:33am

Might be able to set a bit of field weakening for more speed.

Take care with field weakening, too much of it and a controller shut down can be fatal (for the controller).

It's always safe to run a bit because there's voltage sag and resistive losses but too much and when it crashes the motor generates enough voltage to destroy the FETs/dcdc.
Hold my beer while I divide by zero :flame:

User avatar
owhite   100 W

100 W
Posts: 232
Joined: Aug 03 2020 11:22am

Re: Mongoose Blackcomb

Post by owhite » Jan 25 2022 12:43pm

sometime in the next few days i'll try building a disconnect switch, a quarter turn affair with 1/8" copper strip sandwiched between clear polycarbonate.
post a pic I'd very interested in seeing it.

kubark42   1 W

1 W
Posts: 59
Joined: May 15 2020 7:08am

Re: Mongoose Blackcomb

Post by kubark42 » Jan 25 2022 10:35pm

Barncat wrote:
Jan 22 2022 11:54am
kubark42- that's fantastic news! thanks, and i stand corrected. i obviously didn't know that info existed- will study very closely. and i'm using the 5.2 firmware, so will immediately update and start over. maybe i'll be as fortunate as in your case.

i have some skills with fabrication and design, but i'm NOT an EE. i didn't realize when deciding to go VESC/U15 that it was a fairly exotic combo. got a good deal on the motor, and figured vescs are tiny. i like low weight high power.

according to that thread this high pole/stator motor type saturates early and freaks out, which finally describes and confirms electronically what i've been feeling on short test rides. fortunately i don't think i've harmed any components as i don't beat on things that aren't working properly.

will report back. you may have saved the day.
Awesome! Do report back, I'd love to know about your experiences with 5.3.

IACE (I am a controls engineer), and so I don't find tuning by orders of magnitude to be very frightening when I know what I should be looking for. What I really like about Vedder's video was seeing what observer instability looks like in the telemetry plots.

Barncat   100 W

100 W
Posts: 269
Joined: Oct 26 2020 2:49pm

Re: Mongoose Blackcomb

Post by Barncat » Jan 26 2022 10:37am

hey guys- i'm aware of the concept of field weakening but will need to do more study; have not quite finalized how to build the disconnect switch; and was unable to download vesc-tool v5.3 yesterday despite chipping in 10 euro- other guys were having the same problem... [update- it downloaded instantly this morning so that's cool. just a server glitch or something yesterday]

as mentioned, i'm getting brief red light vesc cutouts with the "motor amps" cranked to 200. nothing as unpleasant as when the motor settings were way off, but still not right, and i don't know exactly which value is being exceeded. i'll be spending a few hours on that today, and hope to get the new firmware downloaded.

i'll also be installing a simple thumb throttle next to the left handlebar brake as regen. should be very simple to wire into +5v, GND, and ADC2 on the comm port as a reverse throttle, and then tailor the response as desired with the mapping, current, and throttle curve graph, which i really like.

Post Reply