Jump to content

Firewheel custom battery pack


esaj

Recommended Posts

What have you done with the BMS input cable? Connected it to ground ?

The BMS and the Mainboard imho should only share the ground line. So the way to "communicate" for them would be that the cable has a pull up resistor on the mainboard and the BMS shortens it to ground or leaves the cable "open" (unconnected) to signal the two states (accu can be recharged/accu is full)

Or did you mention in your post that the wheel did not start at all when you left the cable unconnected?

The wheel did power up if the cable was disconnected but started repeating something like "low battery or <unclear> in battery" immediately, I didn't have the motor connected (except for the hall-sensors), so I don't know if it would have driven the motor, but I doubt it would have let me do anything...

The 1Rad Werkstatt guys told me to what to do with it, and then asked me not to tell it forward, not sure why, maybe they suspect it would influence their business or something. But from the little I know, it seems there was more to it than just connected/not connected, as there seem to be some extra chips around the point where the cable seems to go in the BMS:

13853045.jpg?aa53ba401723803b18825079c8d

So probably the BMS somehow did tell the mainboard when the batteries are "too full" for regenerative braking, and the firmware would then switch to some other type of braking.

Link to comment
Share on other sites

  • Replies 125
  • Created
  • Last Reply

The wheel did power up if the cable was disconnected but started repeating something like "low battery or <unclear> in battery" immediately, I didn't have the motor connected (except for the hall-sensors), so I don't know if it would have driven the motor, but I doubt it would have let me do anything...

The 1Rad Werkstatt guys told me to what to do with it, and then asked me not to tell it forward, not sure why, maybe they suspect it would influence their business or something. But from the little I know, it seems there was more to it than just connected/not connected, as there seem to be some extra chips around the point where the cable seems to go in the BMS:

...

So probably the BMS somehow did tell the mainboard when the batteries are "too full" for regenerative braking, and the firmware would then switch to some other type of braking.

So my next guess would be that disconnected (Vcc on the mainboard by the pull up resistor) means that the battery is empty (0%) and/or has problems. Connected to ground (if that was the secret ) means battery absolutely full. Values inbetween could show the different capacity states. These could be "encoded" by resistance, voltage or frequency/PWM or something similiar...

Should be quite easy to figure/try out. Especially since you still have the original BMS giving the signal...

BTW: The original accu pack also had only two power lines to the main board and no wires for charging?

Edit: Sounds a little bit unlogic what i wrote above ;) Signalling battery full to the mainboard and still charging the battery... ;(

Link to comment
Share on other sites

So my next guess would be that disconnected (Vcc on the mainboard by the pull up resistor) means that the battery is empty (0%) and/or has problems. Connected to ground (if that was the secret ) means battery absolutely full. Values inbetween could show the different capacity states. These could be "encoded" by resistance, voltage or frequency/PWM or something similiar...

Should be quite easy to figure/try out. Especially since you still have the original BMS giving the signal...

I did try to measure it with my cheapo-multimeter, but found no voltage between the battery minus and the wire, there was exactly 5 ohms of resistance between the wire and the battery minus (ground), and full voltage between the wire and battery plus and very high or  infinite resistance/no connection (if I recall correctly, I could of course measure it again). Also there didn't appear to be any measurable voltage between the mainboard input and the wire, but as I had no clamps, trying to keep the multimeter-needles on the connector pins wasn't that easy, and it could actually be that they just didn't get good connection... :rolleyes:  But as you suggest, I do suspect that there is some form of resistance/voltage/pwm/magic fairy dust used to encode the status. I need to figure out some load to use with the old disconnected batteries to discharge them a bit and see if the resistance changes.

I don't have an oscilloscope to dig deeper, and of course right now the working motherboard is inside the wheel (I do have one that's broken, but probably not going to help that much). I'll try to see if I can get more measurements from it with the when I tear down the wheel the next time (like I said earlier, the new mainboard is on its way, and I just got the word that the BMS for my 4th pack has entered the country, so hopefully I'll have them both next week). If the new mainboard no longer has the BMS-input, they must deal with it somehow differently nowadays, hopefully in a way that's "compatible" with the BMSs of the new batteries, at least this time I'll know to test it and not just go riding with full battery :D

Btw, hypothetically, if the BMS-input was directly connected to the ground, and it means that the battery is absolutely full, shouldn't the wheel then NOT try to charge it regeneratively? ;)  Sorry for being a smart-ass, I do appreciate your input on this matter... ;)

 

BTW: The original accu pack also had only two power lines to the main board and no wires for charging?

The original BMS has separate "C-" -pad for the charging, seen in the picture beneath the "HG-" -pad, but the regenerative charging must have occurred over the discharge-wires, as the charging wires just went to the charge port.

Link to comment
Share on other sites

...

Btw, hypothetically, if the BMS-input was directly connected to the ground, and it means that the battery is absolutely full, shouldn't the wheel then NOT try to charge it regeneratively? ;)  Sorry for being a smart-ass, I do appreciate your input on this matter... ;)

...

The original BMS has separate "C-" -pad for the charging, seen in the picture beneath the "HG-" -pad, but the regenerative charging must have occurred over the discharge-wires, as the charging wires just went to the charge port.

 I was the same smart-ass, while you already quoted my post, too. ;) 

Keep my finger's crossed, that you find something out or just don't need to with the new motherboard.

...

Maybe with the old BMS you had just overvoltage protection while charging with the charger, regenerative breaking was "destroying" the accus and the wire was only for outputing the battery status... ;)

Link to comment
Share on other sites

Maybe with the old BMS you had just overvoltage protection while charging with the charger, regenerative breaking was "destroying" the accus and the wire was only for outputing the battery status... ;)

Kind of a scary thought, if I've been riding on top of a bomb for over 1400km, if the overcharge protection would be only connected through the "C-" -pad in the original BMS and it has actually slightly overcharged the batteries each time I've gone down that hill as the very first thing on my trips  ;)   It's not that steep and not that long, but certainly enough to shut down with the current BMSs if the batteries are full... haven't dared to try it again yet, currently the wheels been turned on with lights burning in the hallway for some hours, so I could get it to discharge to "rideable" levels after my tests  :P 

Whatever the solution will be for my new packs, I'm certainly not willing to do anything to the overcharge protection of the BMS, shunting the discharge side might be somewhat risky, but overcharging the batteries certainly is very dangerous in terms of elevated fire or explosion-risk...

 

Link to comment
Share on other sites

Wrote about this before: the only adequate solution would be to design a BMS where the excess power from reg would be channeled off into a resistor bank, embedded to the control-board heat sink; but even here you'll run into problems on a long continuous descent, where heat build up > dissipation rate. Maybe fan assisted venting could help out here...  

 

Link to comment
Share on other sites

Wrote about this before: the only adequate solution would be to design a BMS where the excess power from reg would be channeled off into a resistor bank, embedded to the control-board heat sink; but even here you'll run into problems on a long continuous descent, where heat build up > dissipation rate. Maybe fan assisted venting could help out here...  

Yeah, dissipating the excess somehow would certainly be the solution... I have no idea how the original batteries did it, whether the mainboard changed the braking logics when the battery was full, or if the original BMS has something special regarding this. I've mailed the 1Rad Werkstatt to ask about the problem, hopefully they'll have some insight into this. If not, I just have to hope that the new mainboard does not suffer from the same problem, or just charge to lower voltage. Less range, but on the other hand, the battery lifetime would go up a lot  ;)

Link to comment
Share on other sites

You really deserve better luck! It might be a bit of overkill, but the guys at eBikes.ca are offering a charger where you can set the per/cell voltage to 4.05v http://www.ebikes.ca/product-info/cycle-satiator.html, gives you longer cell cycle life as well.

Thanks for the tip, guess that's what I'll have to get if it still happens even with the new mainboard... I knew there were all sorts of risks when I started this project, and although there have been numerous step-backs, at least the wheel still works and I didn't burn down the house (fingers crossed neither of those happens next) :P  

After something like 4-5 hours idling in the hallway, the voltage has dropped (only) to 65.9V, guess I'll "have" to take a ride to bring it more down (I don't want to leave it sitting with high charge).  ;)  Ok, I also want to ride it too... although, I must admit that I'm a little bit scared of it now, at least as long as the voltage remains high  :D  Grudgingly, I will take a leash and attach it to my belt, although that'll probably mean the wheel will make me fall should it cut-off and get away from me, but I don't want to take the risk of it hitting anything expensive or anyone should that happen. Good thing I have all that protective gear, right? ;)

Link to comment
Share on other sites

Window-1.thumb.png.ac5048da7d8577723193c

No idea, whether there's an equivalent option in Finland, but it looks like in Germany we can insure our EUCs now regardless of the continuing problems with vehicle regulation.

Esa, I don't want to suggest this as an alternative to your precaution of putting your Firewheel on a leash :wacko: and yourself in all the protective goodies you can find. Fingers crossed!

For German wheelies: Details are here http://forum.electricunicycle.org/topic/717-german-discussion-group/?do=findComment&comment=11340

Link to comment
Share on other sites

Just to give you some additional Firewheel experience, I never had a control board that had a plug for that one-wire connector. The original battery had one but it wasn't used, and the replacement battery had one too. When I replaced my control board and was stlll using my old battery a few months back, I started it up and immediately got "Please restart the unicycle" messages on power-up. The Firewheel guys said it was the battery and were mailing a new one.

In the meantime I experimented a bit to see if I could get the old battery to work. While riding down a hill on a pretty full charge I got my one and only BMS cutout ever, managed to bail off the wheel standing up. So it may have been overcharge for that case, or maybe it was just a messed-up battery. I still have that battery sitting around, I was planning to turn it into a backup but hey, it's hard enough to find time to ride much less build a spare batttery box. :P

Link to comment
Share on other sites

No idea, whether there's an equivalent option in Finland, but it looks like in Germany we can insure our EUCs now regardless of the continuing problems with vehicle regulation. Esa, I don't want to suggest this as an alternative to your precaution of putting your Firewheel on a leash :wacko: and yourself in all the protective goodies you can find. Fingers crossed!

If the legislation passes as it is currently proposed, no registration will be required (as the wheels will be bicycles). That also means that normal liability insurance should cover any accidents.

The pragmatic solution would be to move to a house at the bottom of the hill. But ofcause not real techno-geek would want to be pragmatic.

Sorry for the bad joke... :rolleyes:

:D Easier said than done... ;)

 

Just to give you some additional Firewheel experience, I never had a control board that had a plug for that one-wire connector. The original battery had one but it wasn't used, and the replacement battery had one too. When I replaced my control board and was stlll using my old battery a few months back, I started it up and immediately got "Please restart the unicycle" messages on power-up. The Firewheel guys said it was the battery and were mailing a new one.

In the meantime I experimented a bit to see if I could get the old battery to work. While riding down a hill on a pretty full charge I got my one and only BMS cutout ever, managed to bail off the wheel standing up. So it may have been overcharge for that case, or maybe it was just a messed-up battery. I still have that battery sitting around, I was planning to turn it into a backup but hey, it's hard enough to find time to ride much less build a spare batttery box. :P

If you have F260, don't you have two empty battery compartments? All you'd need to is connect the discharge- & charge-wires behind single connectors and you'd have 528Wh wheel (with two BMSs). Of course the secondary PCB needs to be moved to the mainboard compartment, but I already did it once during the first tear down, so it's entirely possible (might need to move the mainboard on the metal plate, if it isn't near the edge already).

 

As for the BMS-issue, Chris from 1Rad Werkstatt answered my mail promptly and told me what's going on. The bad news are that first of all, I can't again tell the details. And the solution isn't really easy or fast. Most likely I'll have to go with how the packs work now and only charge to lower voltage, and see what I'll do about it over the winter. :mellow:  This project's been one long, slow train-wreck! :D Guess somebody had to be the "pioneer", although I'm not likely anywhere near the first person to get custom packs (but probably only one who has documented the process in more detail, although maybe only as a cautionary tale ;)). But, I think I can go with these for now...

I did go for a ride for about an hour and a half, of which about the first 40 minutes was used just by going back and forth in the hiking paths, before the voltage lowered enough for me to dare to "really" ride. The combination of a wheel that can shutdown due to too high voltage (overcharge protection), too low voltage (overdischarge protection, the batteries are not shunted) and has no sealing, completely dark paths in the middle of the forest at night and a flashlight I was holding in my hand, that developed some sort of connection problem, started flickering and sometimes went totally out was actually... pretty fun, once I stopped fearing the braking and picked up speed :D  Still, with the voltage around 62V, it can jump to over 65V during stronger braking (more powerful and especially power-braking actually seems to lower the voltage, so it does suck energy from the batteries), so I better keep the voltages fairly low if I intend to ride somewhere where I need to brake lightly more. The rest voltage after coming home and letting the wheel sit for a while was 63.0V (3,9375V per cell), so apparently I really didn't use that much battery (even though it was mostly off-road riding), I'd guess total mileage was around 15km, as I rode the first part much slower, and didn't really go much above the first warning even later.

Although it's probably "only" around +50% more power (3 packs with 10A continuous/20A pulse current per pack vs. 2 packs before), it clearly has a lot more torque, and it's a lot more stable when riding over larger humps and bumps. That one big hill where vee's 14" MCM2s was pulling 2.5kW peaks was pretty easy to ride up.  My battery guy tries to get the last pack to me by the end of the week, but it could go to next week, the BMS should be with him tomorrow (hopefully, if the postal service doesn't lose it like the last time :P). Having 4 packs to suck in the regenerative energy should also help to keep the voltage raise lower during braking, not to mention still give some more power when it's needed.

 

Link to comment
Share on other sites

I realized I never posted the last part of the rebuild... so here goes.

I was stuck with the motor wire that just wouldn't go where it was supposed to. I ended up cutting a small hole in the mainboard compartment, through which I could pull the wire to the compartment and get the shell-side finally in place (at the same time I noticed that the screw-bump netx to the hole isn't actually used, just like the similar ones inside the battery compartments):

8PPgRRy.jpg

 

After that, I bolted the shell to the pedal-assembly, connected the batteries and checked that the wheel has room to turn, the whole thing turns on & works like it should (starts balancing):

zymz6DS.jpg

So far, so good, at this point I actually thought that now this is going well and I'll be finished in a few hours. Wrong.

Then comes one of the hard parts (like there aren't those much, eh?): wiring the front. The shell-halves must be placed on the floor so that the front-sides are facing each other, then the batteries must be connected & the whole shebang must be fitted into the compartment:

zmyHDPu.jpg

I also cut small groove on both shell-halves near the power button at the top (top right on the picture, not really visible) for the display-connector wires to poke out.

In the below picture, the wiring is pretty good, nothing looks like it's sticking out and (except the display connector, that's supposed to), and no screw-holes are covered. Really not as easy as it looks, most of my own wires aren't the AWG-type with silicone-cover and lots of tiny conductor-wires inside, but the usual "car-wires", between 0.75mm2 and 2.5mm2, and they don't bend that easily in small spaces. In the end, I had to repeat these steps many, many times, before I got everything to fit (and a couple of times go as far back as taking the motor & mainboard metal plate off to fiddle with the wires):

bg2epKn.jpg

Once the wires are pretty much in place, comes another hard part, closing the shells:

kYQ2Ean.jpg

At the first try, I couldn't get it to close at all, and had to cut off the small battery display holder in the opposite shell:

ZhavMMt.jpg

As you can see from the pictures, there are lots of wires, and the wires could get between thos "poles" or whatever you could call them (the white plastic pipes/cones/whatever that hold the screws connecting the shell halves), and it's hard to sometimes see if there's something in between. I found it easiest to check with a flashlight from the outside of the shell, looking down the hole where the screw goes and then poking it to a better position with a small screwdriver.

What's worse is that you can't see what's happening at the back of the front-compartment (the "wall" between the compartment and tire), and at least once I got wires squished between the shell-halves and the wire-covers were damaged a bit (don't worry, I fixed them with electric insulation tape ;)). At one time, I didn't actually notice this until I tried turning the tire, and then I could hear it scraping against a wire (but couldn't see where). Maybe at 4th try, I figured I could use the rubber mat to help here:

hFIlvsh.jpg

There, now the wires will stay nicely inside the compartment and not wander outside the back. Of course have to check that the mat itself then inserts nicely inside the compartment when closing, but that was still easier than trying to see any stray wires.

At some points, I actually got it all the way to closed & could screw the other side to the pedal-assembly and tighten the shell screws. The only problem is, that some of the parts (battery covers or such) won't start scraping against the tire until everything is tightened up (we're really talking millimeters here):

yVqYLpF.jpg

So, from this I actually had to go all the way back to removing the motor to get to tighten some screws inside. Of course it requires then doing all the steps again.

This is the last picture I took of the assembly, after 6th time I had put it together, where I was wondering again what is scraping this time...

RRWWiE9.jpg

It turned out to be the nut of the other bolt holding the shunt in place inside the mainboard compartment (not really even seen in the picture), although I had filed them down a bit. You know what that means? Yes, full teardown all the way to the mainboard compartment, unscrew shell screws, unscrew the bolts holding the shell halves on the pedal-frames, separate the halves, remove motor, open mainboard compartment, remove bolts, hot-glue, re-wire a bit, close mainboard compartment, re-attach motor, re-attach mainboard-side cover to pedal-assembly,  re-wire the front, get the halves together, check and move all the wiring as necessary to get it to close, re-attach the other shell-half, and finally the screws holding the shell-halves together. Check that nothing scrapes. I got pretty good at this, due to all the repeats. :P

On the 7th try, I finally got it all together, and you've probably seen this picture already:

1PFnao1.jpg

Ready to rock. And rock it did, while it was still charged to only 57.2V... you know the rest  ;)

Link to comment
Share on other sites

If you have F260, don't you have two empty battery compartments? All you'd need to is connect the discharge- & charge-wires behind single connectors and you'd have 528Wh wheel (with two BMSs). Of course the secondary PCB needs to be moved to the mainboard compartment, but I already did it once during the first tear down, so it's entirely possible (might need to move the mainboard on the metal plate, if it isn't near the edge already).

 

Yeah, I was thinking that there really could be a problem with the old battery but chances are that it works pretty well and if i shunt the BMS at least it wouldn't shut off. I have to open it to shunt the BMS on the new battery anyway. I could definitely parallel the two packs. Heck, I should practice by shunting the old pack anyway! 

I like the idea of cutting the extra hole near the control board, that gives you a little more ability to jiggle around the wires. It's not going to really affect the integrity of the shell and you can slap a piece of tape over it so it's sealed.

Link to comment
Share on other sites

Went for a ride again, it's actually "rideable" once the voltage gets lower. Mostly I rode around the forest paths again in the dark (as there's very little chance there'd be anyone around in the middle of the night in the dark) and the power difference is really noticeable once I'm in more "comfortable" voltage area not to worry that much about going too high during braking or too low during accelerations. The climbing ability has certainly increased a lot, as well as the stability even with larger humps and bumps. Of course that's pretty risky, as I'm tethered to the wheel all the time, so dismounting isn't really an option (the leash would probably make me fall even if I could otherwise out-run a possible fall), but luckily that hasn't (yet ;)) happened. I started at 62.9V and after about an hour or little over that of riding (and letting the wheel then sit for a while to get more comparable readings), it had gone down to 59.2V. Still no idea about the mileage, as I haven't replaced the bike computer, but my speeds were certainly higher than before (going above the first warning a lot, so above 18km/h), so I guess I could have ridden now around 30+km since I started from 65.9V yesterday.

Also took some spins in empty paved roads. The voltage meter only updates about twice a second, so it likely misses the highest and lowest (very momentary) voltages. In general it seems to go about +- 2V from the "rest" voltage (ie. when the wheel is stationary) during harder braking and acceleration, at least in these voltages, but I don't know how much higher or lower the peak and bottom values are. The BMS specs also show that it's pretty fast at cutting out if the overcharge voltage is triggered (55ms), I've seen BMSs that have trigger times above 500ms, which would be nicer in this case, as a very momentary (less than half a second) high voltage wouldn't yet trigger it ;)  The overdischarge side is even faster at 5ms. Thank you for making such a fast-reacting and safe BMS, it's going to kill me someday  :D

I've been trying to wrap my head around the DC-motor drive of our wheels for a couple of days, and how the regenerative voltage could be regulated to stay below the cut-off, and it certainly doesn't look easy. From what I've understood (and take all of this with a pinch of salt, as I could and probably do have understood many parts incorrectly), the motor also always acts as a generator, and that's what causes the back-EMF (basically a voltage) that's in the motor while it's spinning (the back-emf is proportional to the speed/rpm of the motor).

When the voltage coming from the direction of the batteries (regulated by the PWM from the mosfets-bridge) is larger than the back-emf of the motor, the motor speeds up, of course there's more to this, as the speed must be carefully controlled in our wheels, I read somewhere that the speed in some wheels can be changed/adjusted 100-200 times per second, probably relating to the frequency of the PWM and affected by voltage raise/drop times, motor resisting the change in speed and whatever? The torque is caused by the current flowing through.

If the back-emf is higher than the voltage from the batteries, the current reverses in direction and regenerative braking occurs (the motor acts as a power source). The PWM is then controlling the charging of the batteries with the current caused by the back-emf, and apparently the "other half" of the cycle when the pulse is low it causes a dynamic braking (I think that's what it was called) by shorting the motor itself for the "non-active part" of the pulse. Or maybe the the motor simply disconnected, "free-wheeling" during the other part of the pulse cycle. So the duty-cycle is either like charge - brake - charge - brake... or charge - free-wheel - charge -free-wheel... I understood that if the motor would be shorted all the time, it would slam to stop as fast as possible, possibly burning something (the other half of the mosfets?) or itself (coils?) on the way. But I still think it's charge-brake -cycle, as the free-wheeling would cause the braking to be much less effective... And it apparently needs to be cycled, as the back-emf will start to drop during the charge-phase and the current flow could again reverse if it drops below the battery voltage... Plus there's a limit how much current you can push to the cells without damaging them, so triggering the charge on and off should cause less stress.

When the voltage from the battery and the back-EMF of the motor is same (zero potential difference), the current drops to zero and the motor is "free-wheeling" (there's no torque at all), so basically it's keeping the speed steady (minus slowing down caused by friction etc). Apparently also disconnecting the motor makes it free-wheel (there can't be no voltage / potential difference if there is no closed circuit).  I've seen this with the app and vee's Gotway earlier and the with the current display, the current will drop to around zero fairly easily when riding even a slight downhill with steady speed, of course once the wheel tilts back or forth slightly (so slightly that you won't notice it, we could be talking like one tenths or hundreths of degrees), it's doing small corrections and causing small positive and negative current, either triggering a very small (regenerative) braking or acceleration to keep the wheel near the totally straight up "0"-position. That's probably also what causes the sound the wheel makes when you hold it stationary (as it's very, very quickly actually vibrating back and forth around the 0-position).

The third kind of braking is the "plugging" braking, where the battery polarity is reversed from the point of view of the motor, and the battery power is actually used to brake it by trying to drive it in the opposing direction, and I suspect this might be what's happening during power-braking, as it the effectiveness of power-braking seems to be much lower with lower batteries (at least on my 14" generic, never really noticed this with Firewheel)... I've also tested this a little now with the voltage-display, and it would indeed seem that during more powerful braking, the voltage first raises a bit and then drops, although it's hard to say for sure due to the slow update of the display (maybe I should test it more with vee's MCM2s and look at the graphs on the app...).

So, back to the regenerative braking and regulating the voltage. Digging somewhat deeper, I've found out there's (of course) a lot more to it. There are high transient voltages occurring during the braking, I've read a post where an engineer tested 9V motor with 7V drive voltage, and got a transient voltage of 110V out of it for 12µs when the regenerative braking started (btw, that's microseconds, so 12 millionth of a second, a very, very short time), so it could be easily be over a kilovolt (1000V) in our wheels. But, these aren't the problem, as they're short-lived, the current is probably low at that point and these are apparently handled by avalanche diodes either in the motor itself or in the mainboard. Reading about high-power "transient-voltage-suppression" avalanche diodes (Transils, TransZorbs and what have you), I thought that they might be the answer, if I could place one to prevent the current flowing towards the battery and instead divert it to either the motor ground through another diode to absorb the excess power when the current reverses and the voltage raises "too high", basically creating the dynamic-braking condition over some resistor(s), or dropping the voltage over a resistor towards the batteries. Of course it's not that simple, as I would need some sort of resistor on the battery wires too or mosfet controlling where the current can go and when... I've forgotten the little I knew of such circuits a long time ago :wacko:

Another problem of course is that the power generated during the braking can be in excess of 1.5kW (tested with vee's MCM2s, if the values given by the Gotway can be trusted), although for only short-lived times, and the current peaks at over 20A, so should the diodes or the resistor(s) fail, the other half of the duty cycle would certainly become free-wheeling or the connection to the battery could die. Another thing is that I have no idea how this would affect the riding, as the firmware logics might go haywire with such a setup :P Also it might simply just blow something up in the mainboard or the motor itself, as the current could get pretty large. I really haven't got much experience of circuit design, the little I did was way over ten years ago at school and it was pretty simple stuff like oscillators and basic filters from the little I remember ;) And I'm not sure if the diode preventing the current from flowing to the battery could handle the "normal" riding currents, or if it would just burn, then that'd be a certain face-plant any way ;)  I also found some active transient voltage/reverse polarity protection chips/circuits available, but the ones I looked at didn't seem like they were rated for high enough voltages or currents. And regulators and switching regulators and lots and lots of stuff that just went way over my head ;) All of these probably also carry the added danger of causing voltage ripples and frequencies and whatever in the power supply to the mainboard that could cause it to behave strangely or crash the CPU. I rarely give up easily, but in this I'm so way over my head, that I don't think I'm going to try to limit the regenerative braking voltage, unless someone can point me to a "surefire" way of doing that, probably I'd just end up breaking something  :D  Although I do have the spare mainboard on the way and actually the 4th battery was sent to me today... bittersweet, eh? ;) Guess I could just as well put it in the wheel, at least then I'll have one more pack to soak up the regenerative charge, should prevent the voltage from raising a bit more...

 

Link to comment
Share on other sites

You've done a lot of research there!  My electronics knowledge is even older than yours so I'm also stumbling around trying to understand it all. 

The third kind of braking is the "plugging" braking, where the battery polarity is reversed from the point of view of the motor, and the battery power is actually used to brake it by trying to drive it in the opposing direction, and I suspect this might be what's happening during power-braking, as it the effectiveness of power-braking seems to be much lower with lower batteries (at least on my 14" generic, never really noticed this with Firewheel)... I've also tested this a little now with the voltage-display, and it would indeed seem that during more powerful braking, the voltage first raises a bit and then drops, although it's hard to say for sure due to the slow update of the display (maybe I should test it more with vee's MCM2s and look at the graphs on the app...).

It would seem like you'd only want to do this sort of braking in the situation where either the battery was full and you didn't want to risk charging it any more, or charging alone wasn't providing enough braking power. I wonder if the "shaky" braking effect is due to this, like maybe they didn't/couldn't implement it smoothly?

Link to comment
Share on other sites

...

If the back-emf is higher than the voltage from the batteries, the current reverses in direction and regenerative braking occurs (the motor acts as a power source). The PWM is then controlling the charging of the batteries with the current caused by the back-emf, and apparently the "other half" of the cycle when the pulse is low it causes a dynamic braking (I think that's what it was called) by shorting the motor itself for the "non-active part" of the pulse. Or maybe the the motor simply disconnected, "free-wheeling" during the other part of the pulse cycle. So the duty-cycle is either like charge - brake - charge - brake... or charge - free-wheel - charge -free-wheel... I understood that if the motor would be shorted all the time, it would slam to stop as fast as possible, possibly burning something (the other half of the mosfets?) or itself (coils?) on the way. But I still think it's charge-brake -cycle, as the free-wheeling would cause the braking to be much less effective... And it apparently needs to be cycled, as the back-emf will start to drop during the charge-phase and the current flow could again reverse if it drops below the battery voltage... Plus there's a limit how much current you can push to the cells without damaging them, so triggering the charge on and off should cause less stress.

You should be careful with the word free-wheeling - these word is already "occupied" in electronics. See my comment below with the youtube link.

Regarding the PWM pulses they should be charge - "free-wheel" - charge - "free-wheel" or discharge-"free-wheel"-... I do not think that it's possible to use the MOSFETS to shorten and burn the breaking energy. You just need to look at the numbers you mentioned for "breaking power" - there is nowhere a heatsink to keep the MOSFETS cool enough to survive...

So the two only ways to decelerate is to "pumb" energy into the accu or to put the pulses timely on the motor coils, so that the motor is actively breaking instead of accelerating. Imho ;)

...

The third kind of braking is the "plugging" braking, where the battery polarity is reversed from the point of view of the motor, and the battery power is actually used to brake it by trying to drive it in the opposing direction, and I suspect this might be what's happening during power-braking, as it the effectiveness of power-braking seems to be much lower with lower batteries (at least on my 14" generic, never really noticed this with Firewheel)... I've also tested this a little now with the voltage-display, and it would indeed seem that during more powerful braking, the voltage first raises a bit and then drops, although it's hard to say for sure due to the slow update of the display (maybe I should test it more with vee's MCM2s and look at the graphs on the app...).

For changing the battery polarity you would need a full bridge for each motor coil. Imho there are only half bridges in the wheels. (6 MOSFETS-> 3 Half Bridged for each of the three coils?)

So, back to the regenerative braking and regulating the voltage. Digging somewhat deeper, I've found out there's (of course) a lot more to it. There are high transient voltages occurring during the braking, I've read a post where an engineer tested 9V motor with 7V drive voltage, and got a transient voltage of 110V out of it for 12µs when the regenerative braking started (btw, that's microseconds, so 12 millionth of a second, a very, very short time), so it could be easily be over a kilovolt (1000V) in our wheels. But, these aren't the problem, as they're short-lived, the current is probably low at that point and these are apparently handled by avalanche diodes either in the motor itself or in the mainboard.

These should hopefully be handled by active freewheeling ( here's a nice video about this topic: https://www.youtube.com/watch?v=uqzOQGiwGnE )

 

Reading about high-power "transient-voltage-suppression" avalanche diodes (Transils, TransZorbs and what have you), I thought that they might be the answer, if I could place one to prevent the current flowing towards the battery and instead divert it to either the motor ground through another diode to absorb the excess power when the current reverses and the voltage raises "too high", basically creating the dynamic-braking condition over some resistor(s), or dropping the voltage over a resistor towards the batteries. Of course it's not that simple, as I would need some sort of resistor on the battery wires too or mosfet controlling where the current can go and when... I've forgotten the little I knew of such circuits a long time ago :wacko:

Another problem of course is that the power generated during the braking can be in excess of 1.5kW (tested with vee's MCM2s, if the values given by the Gotway can be trusted), although for only short-lived times, and the current peaks at over 20A, so should the diodes or the resistor(s) fail, the other half of the duty cycle would certainly become free-wheeling or the connection to the battery could die. Another thing is that I have no idea how this would affect the riding, as the firmware logics might go haywire with such a setup :P Also it might simply just blow something up in the mainboard or the motor itself, as the current could get pretty large.

Burning this much energy is not to easy...

The charging current to the battery only comes while the active part of the PWM Impulse - so the mainboard can controll the "mean current" going to the batterys. These could also be the problem with your new BMS: that it reacts to fast to the spikes while each PWM pulse and does not consider the mean value and "stays calm"...

In combination with the information about the state of the batteries (% capacity) the mainboard could decide to not stress the batterie and change the pulses on the coils, so that the motor is ?active/dynamic? breaking (the magnetic fields of the coils decelerate the wheel). Or is this change (the stator with the coils tries to break the wheel) the generator mode where the coild produce energy and charge the battery? 

So maybe full-bridges to change the polarity on the coils would be the only solution ("plugging breaking") to break and let the accus survive?

 

Link to comment
Share on other sites

PS: 

...Another problem of course is that the power generated during the braking can be in excess of 1.5kW (tested with vee's MCM2s, if the values given by the Gotway can be trusted), although for only short-lived times, and the current peaks at over 20A, ...

For burning up to this power "actively" one could use mosfets to "connect via PWM pulses" the motor to a resistor bank. But i have just found power resistors with 17W and a measurement of 75x9x6 mm. So with these you would need 1500/17~88 of these. Could be possible to put them all under the left and right outer shell of the wheel, make the shell out of metal and use it as heatsink. Then you have also an automatic heat protection build in - once the shell gets to hot everyone will voluntarely unmount the wheel ;)

Link to comment
Share on other sites

PPS.: I assume the details of BLDC controll for eucs could be quite like described here: http://cache.freescale.com/files/32bit/doc/app_note/AN3007.pdf . Just the stator and the rotor are changed in position - which should not alter the theory behind it? There they mention under "4.4.4 Break Controller" how to handle the overvoltage when the BLDC is in generator mode. In a little bit more detail described in : http://cache.freescale.com/files/32bit/doc/app_note/AN2845.pdf

This "excess" Voltage/Power goes all (pulsed by the PWM signal) into the battery pack of our EUCs, imho... And i am afraid (like already presumed in a post before) that the overvoltage protection of the BMS is only in the charge part, which leads solely to the charge plug. And there is no overvoltage protection after the discharge wires to the motor which are used for regenerative breaking...

If the articles linked above are "state of the art", such a Break Controller (with a big resistor bank ;)) would be needed to save the batteries, once they are full...

Link to comment
Share on other sites

You should be careful with the word free-wheeling - these word is already "occupied" in electronics. See my comment below with the youtube link.

Ok, so "free-wheeling" is the current going around the motor, I was using the term to mean the situation where the motor has no torque (so when it's either disconnected or the back-EMF and battery voltage are the same)... thanks for clearing this up.

Regarding the PWM pulses they should be charge - "free-wheel" - charge - "free-wheel" or discharge-"free-wheel"-... I do not think that it's possible to use the MOSFETS to shorten and burn the breaking energy. You just need to look at the numbers you mentioned for "breaking power" - there is nowhere a heatsink to keep the MOSFETS cool enough to survive...

 

If "free-wheeling" circuit means that the current flows only "inside" the motor-circuit, then it's charge - free-wheel - charge - "free wheel"...

 

Taken from here: http://electronics.stackexchange.com/a/56187 , here the motor circuit is "closed" via low-side (so apparently this is "free-wheeling"?) to brake hard:

braking current

We now have a large current flowing in the opposite direction. Torque is proportional to current, so now instead of applying a gentle clockwise force, just enough to overcome friction, we are applying a hard counterclockwise force, and the mechanical load is rapidly decelerated. As the speed of the motor decreases, so does V1, and consequently so does the current, and the torque with it, until the load is no longer spinning.

 I had misunderstood the entire concept of "free-wheeling", looks like it's then the same as what I meant by "dynamic" braking?  The regenerative part is then done by alternating (switching) between connecting to the battery and free-wheeling:

Let's look at what's happening a bit after we have started braking, but before we have stopped:

braking mid-way

The motor has slowed significantly (back-emf is 1V), and the current has decreased with it. Now what if we switch the bridge to the high side?

battery charging

Ah ha! We are charging the battery! Of course, if we stay like this very long (again, defined by time constant L1/R1) then the current direction will reverse, and we will be uncharging our battery, and accelerating our motor, not braking it.

So don't do that. As long as we remain in this state, the current is decreasing. So, we switch back to the other state, with the bridge low, so the back-emf can build the current back up. Then we switch again, and shoot some of it into the battery. Repeat, fast.

If this sounds like what one ordinarily does for PWM motor control, it's because it is.

 

So the two only ways to decelerate is to "pumb" energy into the accu or to put the pulses timely on the motor coils, so that the motor is actively breaking instead of accelerating. Imho ;)

For changing the battery polarity you would need a full bridge for each motor coil. Imho there are only half bridges in the wheels. (6 MOSFETS-> 3 Half Bridged for each of the three coils?)

But doesn't the polarity already change when the wheel is ridden backwards? How else could the motor be run backwards? Does the mosfet on/off -pattern (EDIT: or timing) then change (the order in which the coils are energized)? The examples I've seen so far have only used 4-mosfet H-bridges, and talk only about driving the motor in one direction & braking.

These should hopefully be handled by active freewheeling ( here's a nice video about this topic: https://www.youtube.com/watch?v=uqzOQGiwGnE )

Good video, thanks. Somewhere it was mentioned that the brushless dc-motors usually have high voltage-rated avalanche diode (or two diodes for bi-directional motors) inside the motor to "swallow" the very high and short lived transient voltages, but don't know if it is the same for our wheel motors. 

Burning this much energy is not to easy...

The charging current to the battery only comes while the active part of the PWM Impulse - so the mainboard can controll the "mean current" going to the batterys. These could also be the problem with your new BMS: that it reacts to fast to the spikes while each PWM pulse and does not consider the mean value and "stays calm"...

 

According to the BMS-specs, the overcharge protection is triggered by voltage (4.20+-0.05V), so although single PWM "on"-state probably does not last 55ms, and the average voltage is lower than peaks due to PWM, it could still build up to high enough voltage... I don't really know that much about this, I've dealt with PWM exactly once, around 2004 I think, and that was just driving 12V DC motors by controlling the voltage over a mosfet (one per motor) with PWM (so very simple speed control in one direction).  ;)

 

In combination with the information about the state of the batteries (% capacity) the mainboard could decide to not stress the batterie and change the pulses on the coils, so that the motor is ?active/dynamic? breaking (the magnetic fields of the coils decelerate the wheel).

Yeah, I could imagine how that could be done with software, unfortunately I cannot change the firmware :D

Or is this change (the stator with the coils tries to break the wheel) the generator mode where the coild produce energy and charge the battery? 

From what I've understood, the back-EMF needs to build up during the 

So maybe full-bridges to change the polarity on the coils would be the only solution ("plugging breaking") to break and let the accus survive?

Yeah, that would be nice, but wouldn't this require a whole different mainboard & firmware? ;)

 

PS: 

For burning up to this power "actively" one could use mosfets to "connect via PWM pulses" the motor to a resistor bank. But i have just found power resistors with 17W and a measurement of 75x9x6 mm. So with these you would need 1500/17~88 of these. Could be possible to put them all under the left and right outer shell of the wheel, make the shell out of metal and use it as heatsink. Then you have also an automatic heat protection build in - once the shell gets to hot everyone will voluntarely unmount the wheel ;)

:D I could use that to melt the snow from the yard in the winter, just ride around with my mobile heater ;)   I actually did find some pretty high power resistors by Ohmite (up 2kW): http://www.ohmite.com/search.php?appl=high power&function=results   Of course they're pretty large.

 

PPS.: I assume the details of BLDC controll for eucs could be quite like described here: http://cache.freescale.com/files/32bit/doc/app_note/AN3007.pdf . Just the stator and the rotor are changed in position - which should not alter the theory behind it? There the mention under "4.4.4 Break Controller" how to handle the overvoltage when the BLDC is in generator mode. In a little bit more detail described in : http://cache.freescale.com/files/32bit/doc/app_note/AN2845.pdf

This "excess" Voltage/Power goes all (pulsed by the PWM signal) into the battery pack of our EUCs, imho... And i am afraid (like already presumed in a post before) that the overvoltage protection of the BMS is only in the charge part, which leads solely to the charge plug. And there is no overvoltage protection after the discharge wires to the motor which are used for regenerative breaking...

If the articles linked above are "state of the art", such a Break Controller (with a big resistor bank ;)) would be needed to save the batteries, once they are full...

Thanks for the articles, I'll try to dig into them later, right now I need to get back to work  :)

Link to comment
Share on other sites

Ok, so "free-wheeling" is the current going around the motor, I was using the term to mean the situation where the motor has no torque... thanks for clearing this up.

If "free-wheeling" circuit means that the current flows only "inside" the motor-circuit, then it's charge - free-wheel - charge - "free wheel"...

 

Taken from here: http://electronics.stackexchange.com/a/56187 , here the motor circuit is "closed" via low-side (so apparently this is "free-wheeling"?) to brake hard:

In your link they are driving a "normal" DC Motor (with brushes) and use one part of the half bridge to power the motor (or letting the energy in generator mode back to the batteries) and the other half of the bridge to shorten the motor ( breaking)

The three half bridges used to drive a BLDC are normally used like shown in my linked article. (One side of all three coils is connected internally and only the other sides come out of the motor and are connected to the three half bridges). So you cannot shorten the single coils... (Maybe shorting all three wires at once would be the same like shorting the DC motor?)

But doesn't the polarity already change when the wheel is ridden backwards? How else could the motor be run backwards? Does the mosfet on/off -pattern then change (the order in which the coils are energized)? The examples I've seen so far have only used 4-mosfet H-bridges, and talk only about driving the motor in one direction & braking.

The direction is controlled by the mosfet on/off pattern timing.

So using full bridges (like i wrote in the post above) to change polarity should be just stupid nonsense...

Good video, thanks. Somewhere it was mentioned that the brushless dc-motors usually have high voltage-rated avalanche diode (or two diodes for bi-directional motors) inside the motor to "swallow" the very high and short lived transient voltages, but don't know if it is the same for our wheel motors. 

I think this is more for EMV - so that the highest and fastet spikes get already eliminated directly at the source. These spikes would not make it "over the wires" untill the Motor Driver, where the not as high and fast spikes get eliminated (e.g. by active free-wheeling). So you can listen to the radio and watch tv while the motor is operated nearby.

There are also small capacitors used, so that these spikes just get shortened.

According to the BMS-specs, the overcharge protection is triggered by voltage (4.20+-0.05V), so although single PWM "on"-state probably does not last 55ms,

You are right. 55 ms would be 18 Hz. The PWM frequency should be choosen higher - so that vibrations caused by the PWM cannot be heard...

 

and the average voltage is lower than peaks due to PWM, it could still build up to high enough voltage... I don't really know that much about this, I've dealt with PWM exactly once,

Yes - If the BMS reacts after either 55ms or 500 ms should not change anything. If you go downhill steadily the voltage will just be there... Could be a clue more, that your original BMS had no overvoltage protection for regenerative breaking....

:D I could use that to melt the snow from the yard in the winter, just ride around with my mobile heater ;)   I actually did find some pretty high power resistors by Ohmite (up 2kW): http://www.ohmite.com/search.php?appl=high power&function=results   Of course they're pretty large.

But some couples of the smaller ones there should be easily usable for the mobile heater ;) Would be interesting how much heat could be build up by riding down a hill...

On the other side - maybe this is overdone and overcharging the battery pack by regenerative breaking is just decreasing the battery life a little. No Videos of burning or exploding EUCs where shown till now from people going downhill...

Link to comment
Share on other sites

On the other side - maybe this is overdone and overcharging the battery pack by regenerative breaking is just decreasing the battery life a little. No Videos of burning or exploding EUCs where shown till now from people going downhill...

Actually, I was wondering if this was caused by overcharge: http://forum.electricunicycle.org/topic/1166-generic-x3-copy-catches-fire-in-norway/

Link to comment
Share on other sites

PS: 

...

The three half bridges used to drive a BLDC are normally used like shown in my linked article. (One side of all three coils is connected internally and only the other sides come out of the motor and are connected to the three half bridges). So you cannot shorten the single coils... (Maybe shorting all three wires at once would be the same like shorting the DC motor?)

...

So if shortening the three coils is the way for "active breaking", this could be easily implemented with the actual mainboards with their 3 Half Bridges. If the lower three MOSFETS are on and the upper three are off (or the other way round) this is exactly this shortening! (so one could also divide the thermal power dissipation forthe mosfets by using the upper half of the bridge for one pulse, then the lower and so on...)

Could be, that this is the reason for the shaking some users here reported imho, when they went downhill with a full battery pack. Maybe this kind of breaking was then used by their EUCs to alarm the rider that the EUC is on its limits, by using this breaking method quite rough. Or this kind of breaking can only be implemented in this rough way? Or the power dissipation for the MOSFETS is way to high, so it can only be used for a short time...?

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...