Most reliable ebike controller for >1.5 KW @ 48v+?

[mayhaveburnt]

I've bounced off of no-name ebike controllers, most of which have burnt out or mysteriously failed within a month. I've tried a sabvoton 7245 and it outright didn't work on arrival, same with a Flipsky 75100. I gave the Flipsky another go since VESC is FOSS-ish and available on linux, it worked for about 2.5 weeks @ 40 amps batt, 110 amps peak phase. It's also dead now, so that doesn't fill me with confidence.

I've heard good things about Far-Driver controllers, along with (very expensive) Curtis controllers, but I'm not building a race bike here, I just want something reliable and powerful enough to carry 10 kmh up a fairly steep mountain with a considerable load. My BMS is DALY, and it hasn't exploded or failed, so evidently not everything cheap is crap, and I'm hoping someone here knows a controller with a good price-power ratio. 220 USD is about as high as I can go, if price is the limiting factor.

I have my eye on an ND72200 or 72350, with the plan being to derate it to 150 peak phase amps, 70 peak batt amps. I know about the PhaseRunners that GRIN makes, but they cost a pretty penny. I'm just so lost in the lack of definititive testing and review for these controllers that I really don't know what I can trust.

 

[amberwolf]

I've bounced off of no-name ebike controllers, most of which have burnt out or mysteriously failed within a month.

@ 40 amps batt, 110 amps peak phase

I just want something reliable and powerful enough to carry 10 kmh up a fairly steep mountain with a considerable load.

What is your motor? How steep a slope? how heavy? How long does it have to sustain this? You can use this info to go to ebikes.ca 's motor simulator and get guesstimates of the actual power required to do the job, and even experiment to find a range of more optimal solutions to do it (not just changing the controller, but possibly the motor, or gearing, etc).

If the motor isn't matched to the speeds and loads, it may be demanding more current than the controller can really handle, or some other issue causing the controller failure.

Could also be a timing issue--if it's a sensored setup, if the halls are not neutrally timed, or are not equally spaced, vs the stator teeth, etc., a controller that is trying to push current based on each halls' timing per phase could be mistiming the phase currents and overloading one or more of the phase bridges.

Do you happen to know what the specific failure mode was for each dead controller? That might help point to the cause, which might help fix it so the next controller is less likely to fail.

Was there any commonality to the point in the trip that each failed? There might be some specific circumstance or collection of them that exacerbates the problem and makes it more likley to kill hte controller.

 

[mayhaveburnt]

Quite heavy, around 150 kg, with bike, gear, batteries and all. The motor is a knockoff G062 6T, I rewound it with 220C wire to handle all that extra load. I get pretty tired of pedaling up the hill every 20 minutes, so 20 minutes on, 5 minutes off, repeat. Batteries are two 15 AH generic li-ions in parallel, with a hefty diode to prevent cross charging. At points the road is up to 70% grade, at which I hop off the bike and walk while it pushes itself.

All the controllers seemed to fail randomly, the VESC failed after a pothole shook it, the no-names just went poof without apparent cause, and the longest lasting one (7 months) failed due to poor circuit design. The only controller that I could do an autopsy on had a failed microprocessor, everything load-bearing was pristine, FETs, shunts and all. I checked my phases on the motor for shorts, which there was none. ~300 milliOhms from all phases to each other. I rarely go up that mountain, and all the controllers have failed on the flatland, where only about 15 amps are drawn while cruising.

I have run my motor sensorless due to the low quality halls failing repeatedly, so the sensors can be safely ignored. The VESC showed overcurrent errors at first, because I set the phase amps too high relative to the battery amps, but after fixing that I had no errors. The only repeatable error I could think of is when I flamed out the low quality original windings on the motor after a long trip, but even that didn't cause a controller failure immediately.

 

[E-HP]

I don’t think it’s a quality issue. Where are you mounting the controllers? If you’re not overdriving the units, then nearly all failures are due to heat. A bigger controller might help but you need good airflow even if you go bigger. There are some cheap controllers with huge heat sinks.

 

[mayhaveburnt]

I mount my controllers in a void space in the frame, exposed to open flowing air. When I was monitoring the VESC's FETs, they never went above 55C. The only controllers that got hot were the stock ones that came with the bike, which used terrible FETS with very high R-on values. Temperatures when I'm out and about are often just above freezing, so everything is quite cool.

Despite my signature, I do consider thermal limits, and so far I've not found any controller-side thermal problems, only motor overheating issues common with the G062. I'm decently sure I just have bad luck, because even *way* over-rated controllers have met their doom with me. Even a Sabvoton, which was rated 2x above my intended amp target, didn't even boot when I got it. I've gotten a bunch of DOA controllers for some reason, it's highly discouraging.

How would I overdrive the units? Do you mean ask them to put out higher current than they are rated for? I try my best to stay within the advertised parameters, if that's the concern. The VESC 75100 is rated for 100 amps continuous, but the lackluster thermals don't allow that, so I let it run at 40 batt amps max. If it's voltage, than I should be well within the Vmax, since I'm not topping 55V.

I'm just wondering what controllers won't arrive DOA, or if going with a higher rated Flipsky Vesc would be a better idea.

 

[amberwolf]

Some thoughts below; dunno if any of it is helpful.

Quite heavy, around 150 kg, with bike, gear, batteries and all. The motor is a knockoff G062 6T, I rewound it with 220C wire to handle all that extra load. I get pretty tired of pedaling up the hill every 20 minutes, so 20 minutes on, 5 minutes off, repeat.

I am not sure I understand--you go down the hill then right back up the same hill immediately? At what speed? (the speed, weight, and slope are all needed to find out how much power it takes to do the job).

Batteries are two 15 AH generic li-ions in parallel, with a hefty diode to prevent cross charging.

Note that if there are diodes between the controller and the battery, the battery cannot act as a damper for any kind of voltage spikes that might occur.

There are various things that can cause the motor to generate voltage spikes; the most common is spinning as a generator (which shouldn't happen with a geared hub that has a clutch).

For instance,if you ever ride *down* that hill at a faster speed than the system could power you on a flat road, and the motor gets spun up (such as if the planetary's clutch sticks or if it doesn't have one), the voltage produced by that could be higher than the controller can take, and can damage the parts of the controller that are on the battery bus. It might not be catastrophic damage, but it could be cumulative until it fails.

As long as the freewheel/clutch inside the motor prevents the motor from being spun up that shouldn't be an issue.

but it can also happen from being switched in certain ways, if ther'es enough inductance the collapse of the current flowing thru the coils could generate quite a voltage spike, which has nothing to damp it without the battery being directly attached.

So...if it's possible I'd recommend directly parallelnign the batteries without the diodes. If they are identical and at the same state of charge that won't be an issue. If they're not identical then "it depends" on the differences whether there are any potential issues. (usually, there aren't). There *are* differeces in BMS designs that might mean you need to disconnect charger input paralleling while riding and disconnect discharge output paralelling while charging.

At points the road is up to 70% grade, at which I hop off the bike and walk while it pushes itself.

Hmm.
If I simulate even a 50% grade with a 48v phaserunner (closest to VESC), and a different bafang geared hub (didnt' see yours in the list), I get essentially no progress uphill at all, and rapid overheat of the system. I'm pretty sure the numbers below the graph are wrong, "blown" by the situation placed upon the components. Motor Simulator - Tools


Trying to use the sim with "unlimited" controller, battery, and the biggest motor there to find out how much power it would actually take to go up that kind of slope with that load, I get numbers in the 7-10kw range for power used, and only speeds of a couple MPH, which makes everything worse. I don't think any of the parts you have in the system could handle that....even if the numbers are off by half, I have a feeling that the system is being overloaded by the hill(s), and that overload may be damaging things that cause later failures.

Might not have anything to do with the issues, but something to look into.

All the controllers seemed to fail randomly, the VESC failed after a pothole shook it, the no-names just went poof without apparent cause, and the longest lasting one (7 months) failed due to poor circuit design. The only controller that I could do an autopsy on had a failed microprocessor,

Failed MPU usually means that something sent a voltage into it to kill it, and that usually means either a phase/hall short that sent phase (battery) voltage into a hall signal line and thus into the MCU, or the battery-to-low-voltage converters failed and passed a high voltage into the MCU. The latter failure can happen if voltage spikes occur that aren't absorbed by the battery (which it can't do if ther'es a diode between it and the controller).

everything load-bearing was pristine, FETs, shunts and all. I checked my phases on the motor for shorts, which there was none. ~300 milliOhms from all phases to each other.

Just curious: What kind of meter was used to do the phase-to-phase check? (a regular multimeter doesn't go low enough to do this; I have a DE5000 that can do it with some motors)

And were the phases disconnected from each other inside the mtoor at the wye connection point? (if not, that is an alternate resistance path and can affect the readings).

I have run my motor sensorless due to the low quality halls failing repeatedly,

Even the low quality clone sensors don't normally fail repeatedly. If this is happening, there is probalby something that is causing them to fail--overheating, voltage spikes (either direct on the supply line or the signal lines, or induced voltages on these from high currents on the phase wires that run in the same cable with the hall wires), or something else.

Note that geared hubs can easily overheat because there are multiple airgaps between the heat-generating windings and the outside air, and it can take a long time for that heat to get out of the motor. Under high load the motor can overheat before the outside even feels warm.

 

[lnanek]

Yeah, I have a Grin Baserunner on a geared motor with clutch and it still managed to off itself and blink the over current error message on the LED if I go downhill very fast with a battery blender or DC-DC converter between it and the battery. Some controllers are just designed to need to flow back into the battery when needed, unfortunately.

That said, I think any hub motor would die if so loaded and going up such a step slope at such a slow speed except maybe the Grin wheelbarrow ones. If that's really needed, I'd go with a mid drive and help it spin fast enough despite the bike going so slow by using the bicycle gears.

 

[E-HP]

If I simulate even a 50% grade with a 48v phaserunner (closest to VESC)

ya the 70% part was when I decided to not invest in a lot of typing.

 

[mayhaveburnt]

I understand that I probably sound like an idiot, but this is a mountain. You don't just go and climb it in one singular run, you take breaks, unless you want to be hurting yourself.

To check the phases I used a regular Fluke 75 DMM. Yes, I kind of forgot that all the phases were tied and would affect the reading, sorry. If voltage spikes are the problem, than a simple set of capacitors should help.

So, essentially, the hills I'm trying to climb here are just too damn steep unless I want to be using dual qs273 and a monster battery, with a ridiculous controller. If so, I'll just walk the bike, no use in challenging physics to a fistfight.

I would hope that the designers of these controllers would have redundant voltage regulators and such, but that's not something we can control. It sounds like the overall "solution" to this issue would be to save up for a Surron or Talaria, with their high wheel torque and all.

If I magically made a G062 with a chain drive 10:1 ratio like in a dirtbike, the results would be more reasonable, some simple venting would help keep temperatures below melting for the insulation.

 

[E-HP]

I understand that I probably sound like an idiot, but this is a mountain. You don't just go and climb it in one singular run, you take breaks, unless you want to be hurting yourself.

Send a google maps link to the coordinates of the road. Maybe you’re using a different definition of road, but that usually means paved. The equipment used to pave roads can’t perform that work staring at around a 30% grade, so your road must a trail? I don’t think you could push a bike up 70% unless you have spikes on the shoes for traction.

 

[neptronix]

Relevant read:
Successor to the infineon clone controllers in the 1000-2000w power range?

 

[mayhaveburnt]

The equipment used to pave roads can’t perform that work staring at around a 30% grade, so your road must a trail? I don’t think you could push a bike up 70% unless you have spikes on the shoes for traction.

I slip and fall a bunch of times, trust me I wish I had spikes, the climb is brutal in the cold.
The "road" is like much of the area I live around: slimy, clay-gravel mix. I correct myself, it wasn't 70% grade, it was a 62% max grade, with an average of 20-40% for the rest. The road looks like it's "stepped" according to the raw geo data, but the grade is fairly smooth IRL.

Here's a Google link to the route: 30298-30288 Lake Cavanaugh Rd to 35399-35393 N Shore Dr
And a FacilMap Link, for those avoiding Google : FacilMap

 

[amberwolf]

To check the phases I used a regular Fluke 75 DMM.

I have the 77-III, and it can't read anything like low enough to actually measure phase coil resistance with it's ohms function. I had to buy the DE5000 to begin to be able to measure stuff like motor windings and shunts.

I haven't used the 75, so maybe it can read much lower resistances than the 77? You'd have to check it's manual.

If voltage spikes are the problem, than a simple set of capacitors should help.

Perhaps, if they have a low enough ESR and can respond to the frequencies involved, and are placed close enough to the FETs so that PCB trace inductance doesn't cause problems.

You can look at the various controller design threads for info about this sort of thing.

Seems easier to directly parallel the batteries without the diodes, unless there's a specific reason they're required. (though it's possible extra caps might help even in this case).


But....realistically, that size and type of motor is going to have a hard time with such steep slopes, and probalby so is the battery (and it seems apparent so will the controllers, though I suspect that has more to do with the diode-battery issue; that would require experimentation to determine the relevance of).

I would recommend at least checking out the simulator and using your actual riding conditions, including speed, etc., to see what kind of power it should take to do what you're after. If you find that it is well within the ability of the parts you have to deal with without stressing them at all, you can start looking for other causes. If you find that it shows it would exceed the parts' capabilities, or would be stressing them, then you can begin looking for replacements that can handle the power required to do the job under those specific conditions.


Alternately, if you don't already have a wattmeter, I'd recommend installing one between your battery output and your controller input, preferably one that can at minimum retain the peaks and minimums of the various values. Something like the Cycle Analyst and a datalogging device to record the serial output of the CA would be even better, as you can then use the Grin web tools to analyze the trip, show you a chart, etc. Either way you can use the data for each segment of your ride(s) to see what the worst-case parts are, and use the data to see if there is or isn't a potential for stress or damage from the power required. Unfortunatley this method requires a working system, so you can only do it after replacing the controller if it's failed.

 

[amberwolf]

I slip and fall a bunch of times, trust me I wish I had spikes

You can isntall them, if you don't want to buy studded tires and change them out for the ride. If you are using knobbies it's relatively easy to install, say, drywall screws, or machine screws, thru each knob from the inside of the tire. You'd probably want to also install a tire-liner made from one or two old innertubes, with their valve stems removed, and slit circumferentially then slipped over your actual innertube, to proetect it from the screw heads.

The disadvantage of doing this is you need two sets of tires and swap them when changing from the mountain to paved roads, etc. Or carry two wheel sets, one with these tires and one with regualr tires. Neither one of these is convenient, but it might help your climb.



BTW, it's also psosible that if the wheel spins in slick conditions, and hten suddenly grabs, there can be quite a current spike in a contrtoller that can damage things (even if they don't fail till later).

It can also damage the gearing or clutch in a geared hubmotor.

 

[E-HP]

I slip and fall a bunch of times, trust me I wish I had spikes, the climb is brutal in the cold.
The "road" is like much of the area I live around: slimy, clay-gravel mix. I correct myself, it wasn't 70% grade, it was a 62% max grade, with an average of 20-40% for the rest. The road looks like it's "stepped" according to the raw geo data, but the grade is fairly smooth IRL.

Here's a Google link to the route: 30298-30288 Lake Cavanaugh Rd to 35399-35393 N Shore Dr
And a FacilMap Link, for those avoiding Google : FacilMap

I use this tool. It's always been reasonably accurate. The steepest section I found on the route was 20.5%. I pasted the coordinates of the 4 steepest sections below. Maybe if you have the exact coordinates of the 62% grade, that would help. The rest of the climb was pretty tame.

HeyWhatsThat Path Profiler

Compute elevation profile for an arbitrary path

www.heywhatsthat.com

 

[mayhaveburnt]

Maybe if you have the exact coordinates of the 62% grade, that would help. The rest of the climb was pretty tame.

Huh... Odd, I can't because apparent it doesn't exist. It isn't common for OSMAnd to lie to me like that but the GPX logs check out right along with E-HP's utility. Absolute maximum grade was ~30%, with an average of about 5-15% via +/- 2m gps. Nothing to sneeze at but also not the ridiculous 62% my map was giving me. I don't think 15% is "tame" even with a motor to help, being that most biking magazines I've seen list consistant 6% grades longer than 20km as extreme. Here's some coord's for cross-reference: 48.272308, -121.934189 || 48.285444, -121.943942
28.4% grade via GPS: 48.284594, -121.937672

We'll just assume it's about 12%, that seems like a good baseline for most of the steeper mountains and hills around me.

But there remains the problem of the controllers, which regardless of hills is failing consistantly. If any of you think it's a phase short, hall short, or other low impedance fault, I'll get some 24AWG 200C wire ordered. Nonetheless, even when the motor was brand new 6 months ago, controllers were failing. This controller failed within 20 minutes of a normal ride, it was the first controller I used after I ordered a replacement motor. ~0% grade, 30kmh speed, cool temperatures, single (unrectfied) battery.

Failure was common with that controller, according to reviews. The VESC had lower failure rates, but still was considerably shaky. What controllers would survive the current spikes and back-EMF spikes that are common with dilapidated "roads"? I looked in the reference thread that was provided, and it essentially retold my entire research experience, but unfourtunately didn't help.

I'm looking for the kind of reliability that can survive the unforgiving weather, overly ambitious rider, and terrible roads all at once. My theory is QC in lower grade FSESC's is lacking, while the bigger, better ones get checked better. The first VESC I got literally had a sharpied "X" on the back, marking it as bad. Lo and behold, no working controller.

Do you think the more expensive FSESCS (like the 75200 pro V2) would handle the unforgiving electrical characteristics of my bike? Or should I throw in the towel and take out a loan on a ND72450 and derate the crap out of it?

 

[doublepower]

i have plenty of hills around where i live that say its like a 6% grade but if you put a digital level on it at certain parts of the hill it is more than 20 degrees keep in mind a 100% grade is a 45 degree angle

 

[scianiac]

I mean a 75200 pro V2 would be more current and overheating tolerant for sure but that's about it, it's still a bottom tier VESC and honestly having used all low end VESCs up to now I don't think it's worth it. They work but the QC and design is half assed at best. If I were you I would get a mid tier VESC, a high end tier would be better but those can be pricey. For some things you do get what you pay for. So something like a Spintend or MakerX (not makerbase those are as crap as flipsky).

If the VESC was poorly setup, if you were getting overcurrent errors and the max current was set to something like 1.5 or so phase current then it was not performing correctly and lowering the battery current was only a workaround to a poor tune or poor controller. Which is part of the problem with cheap VESCs they have terrible curcuit design without enough filtering so run unstabily in sensorless and that gets dramatically worse with higher currents. So it's possible a bigger cheap VESC will last longer but it may still be unstable and hard to tune, my 75200 pro V2 certainly is not exactly impressive and while I did get it to push some serious current it was a lot of work to get it stable and even now it has some instability, and I'm pretty sure it's all down a few missing ceramic capaictiors and lazy curcuit design.

 

[E-HP]

I don't think 15% is "tame" even with a motor to help, being that most biking magazines I've seen list consistant 6% grades longer than 20km as extreme. Here's some coord's for cross-reference: 48.272308, -121.934189 || 48.285444, -121.943942
28.4% grade via GPS: 48.284594, -121.937672

It's a good climb, but by tame I mean there's enough of the ride that allow the motor and controller to cool before another steep part. It's not a straight 1000 ft climb.

i have plenty of hills around where i live that say its like a 6% grade but if you put a digital level on it at certain parts of the hill it is more than 20 degrees keep in mind a 100% grade is a 45 degree angle

20 degrees is 36% so pretty big difference from 6%. These tools do have limitations and can produce different results when adjusting the points being used. The hill I used for temperature and voltage sag testing averages around 18% for the last third, but there are several cross streets where the road levels briefly, so its like a stair step. So while the slope is close to 22% if you take data points on the sloped portion between intersections, the tool averages in the flatter intersections when computing the grade. The hill is still good for my testing purposes and climbs a little more than 700 ft in just over a mile with a gradual increase in grade.

 

[mayhaveburnt]

Hmm, well, I am still within the return window for the VESC, and I'm a little more knowledgeable than I started out with, if anything the sucky VESC was a decent experiment, and I didn't make any edits to the harnesses so I'll probably get my money back.

I mean a 75200 pro V2 would be more current and overheating tolerant for sure but that's about it, it's still a bottom tier VESC and honestly having used all low end VESCs up to now I don't think it's worth it. They work but the QC and design is half assed at best.

The VESC blew me away honestly, I'm so used to absolute bottom-of-the-landfill electronics that the ability to actually change the way the controller works was a big plus. I could tell the VESC wasn't gonna last more than a year, it's too dang simple, I can count the number of components under the Alu box. I had hoped it would last at least a few months, but nope. Not a single backup zener diode I could spot, that would at least provide emergency voltage regulation for the STM32. As one may guess, I've got little to no money to be using on ebike controllers, so bad is all I can get. The best helmet is the one you'll wear, and the best controller is the one you can buy.

That being said, the reviews and reports of the 75200 pro V2 have only gotten better over the years. The lil' 75100 (ebike box version) doesn't have any phase filtering, barely can pull 40kHz switching, has the thermals of a sauna in the desert, and is a borderline smoke machine. The 75200 prov2 visibly has more hardware protections (still not much), and has much more surface area to dissipate heat, not to mention 2x the raw capacitive filtering. Cheaper Trampa and even MakerX VESCs are also known for failing (not like FS, but still), all VESCs it seems are hit and miss until you reach like 350 USD? Here's a teardown of a 75200 prov2, it's not as bad as I thought but it's also not great either: Teardown.

my 75200 pro V2 certainly is not exactly impressive and while I did get it to push some serious current it was a lot of work to get it stable and even now it has some instability, and I'm pretty sure it's all down a few missing ceramic capaictiors and lazy curcuit design.

Yeah, I can get that. I'm not going to be pushing more than 150 max phase amps, only about 70 consistent. The 75100 I had just up and died, if your 75200 lasted long enought to find all that out, than It'll work for me. Extravagance or safety is of little concern, I can handle that, I just need a base that won't die after a pothole.

On the other hand, the FarDrivers I've seen are bulky, heavy, and ugly, but have been said to be extremely reliable as compared to the competition. from 1kw-100kw is a huge range, but none of the controllers have good documentation or programming guides, and warranty is outright misleading. the cheapest FarDrivers I've seen are 140 USD, and that one (ND72200) has just as many bad reviews and reports as the FSESC 75200 prov2 AFAIS.

It's not an easy decision, but my dedication to open source and accessibility has me leaning toward the VESC, even with crappy hardware. Nobody but dedicated DIYers with an appreciable income would think to buy a 300 USD controller, even if it is GRIN or BAC or VESC, that's almost enough to build a whole other battery. I think I'll take a chance at the higher VESC, after I repair the existing stuff that's on my bike. I hope I'm not entirely disappointed, but I wouldn't be suprised if I was.

 

[scianiac]

Yeah I mean my original 75200 is several years old now, I haven't put a million hours on it but it's running at 130pa. Even to get it to be stable at that level took a lot of work and it won't go any higher even though the motor can take 180-200 due to instability. My 75200 Pro V2 I have far less hours on and with a in theory much easier to tune motor, a DD hub vs an IPM inrunner. It's running at 200A but was a nightmare to get it to that point and still runs through instability regions that cause a little shudder but not enough to trigger a OC fault. Yes they have improved but still seem to be pretty bad in some regards, if I recall my 75200 Pro V2 still has zero ceramic bus caps and a pretty roundabout board layout. I think that is why both are so noisy, well the original also probably has the stupid opamp running on the 5V rail which is a known design flaw with these. Also the 75200 pro v2 is running the wrong firmware, as in they just hacked it together to use the trampa firmware instead of making a correct branch, yes they are that shitty. In practice this isn't a huge issue but there are some settings which it will allow you to turn on because the firmware thinks it's configured one way and it's not that cause very bad things to happen. Now I didn't blow up the controller with them but I probably got pretty close.

Oh and I also have a flipsky 75100 dual running two bafang G062s and that has run them perfect with zero effort at 80PA, maybe I have the rear motor higher than that I can't recall. Was that luck, is the G062 just really easy to tune, would it missbehave if I turned it up any higher, I have no idea.

My point though and I have yet to pony up and buy a higher end controller to test it but it seems the time you waste in tuning them (which admittedly depends a lot on the motor, how hard you are pushing it, and luck) is worth spending more. If you are careful VESC is pretty resilient to blowing up even in unstable events during tuning and what have you but lower quality means you have to be more careful.

I think VESC in general though suffers from more failures for the exact reason you stated, you can change so many settings, and a lot of those settings, like a lot a lot will have the effect of somewhere between some instability and instant cookoff. So you give people a lot more dangerous settings and you're likely to increase the number of blown controllers vs a controller where you can't change much, at least that's my theory.

I mean Vedder has some videos of running a massive super low inductance outrunner on a Trampa and he is just triggering the OC over and over through lots of instablility and he's just messing settings and seeing what works and it's just fine. As in he wasn't being that careful or maybe it just looks that way.

 

[amberwolf]

it's too dang simple, I can count the number of components under the Alu box. I had hoped it would last at least a few months, but nope. Not a single backup zener diode I could spot, that would at least provide emergency voltage regulation for the STM32.

I haven't seen redundancy of, well, any type, in any controller of any level. I haven't taken the Sevcon apart yet, but even in the takeaparts of some fairly expensive controllers around here I don't recall seeing anything like that. (not that it wouldn't be a good idea...just that they don't seem to do it).

There are a lot of very simple and pretty cheap ways to "failure proof" (not really, but...) assorted ebike electronics that simply aren't done (probably because they cost money, and even pennies per unit add up to money they could pocket). See this other post I made today for possible fixes for another common and preventable issue:BBSHD won't spin after replacing controller & updating FW to Nilsson v1.5

As one may guess, I've got little to no money to be using on ebike controllers, so bad is all I can get. The best helmet is the one you'll wear, and the best controller is the one you can buy.

I'm familiar with this problem....the only good stuff I have is stuff people have sent me or sold me used really cheap...(ok, so most of the crappy stuff I have also comes from the same sources :lol: but if i had to buy everything on my own "at retail" I wouldn't have much beyond the cheapest kits like Yescom.

(and the stuff I find at goodwill / freecycle / etc...just check any of my build threads :lol: ).

I've blown up and broken my fair share of that stuff, some of it from not checking to see what kind of loading there would be vs what it could handle, or doing any math (we don't get along) to precheck suitabilty of things, etc. Some of it from the stuff just being junk.