marty
10 MW
If someone was to build a Low Voltage Cutoff switch, would it be better to measure the voltage from one cell, or the total from all the cells? Example 8 cells at Low Voltage Cutoff of 3.2V x 8 = 25.6Vtostino said:
Battery Building Competition
- Thread starter marty
- Start date
It would be more idiot proof to measure each cell individually, but quite a bit more complicated (and possibly not necessary).
A CellLog8 would certainly be an easy way to do it. The CellLog does both HVC and LVC and is fully programmable. Can't beat it for $18. The CellLog would still need some kind of interface circuit to control the cutoffs.
A CellLog8 would certainly be an easy way to do it. The CellLog does both HVC and LVC and is fully programmable. Can't beat it for $18. The CellLog would still need some kind of interface circuit to control the cutoffs.
marty
10 MW
John in CR
100 TW
marty said:
The problem with LVC's is that they aren't smart enough for these cells. Konion cells sag a good bit under load, and in the bottom half of capacity they sage significantly more. For example, on my ride this morning the return trip included a long hill. I still had over 40% capacity remaining, but the pack voltage was sagging to 63V or 3.15V per cell. A 3.2V avg LVC or individual cell LVC would have left me walking the bike up that hill despite having plenty of capacity, and at the peak I would have been able to get back on and finish the ride because the battery load was light the remainder of the ride. If I was riding around on generally flat terrain then I better be pretty close to home if my voltage was sagging to the same 63V, so under light load conditions the 3.2V/cell LVC (64V in my case) would be appropriate. If you have no hills to contend with, that helps with the issue, but tripping the LVC under acceleration will be common, a safety issue IMHO. I would also argue that running under light load down to an LVC cut at 3.2V/cell is too low and will significantly reduce the life of the pack.
A much better solution would be an LCC (Low Capacity Cutoff) with first just an alarm that trips 5% higher than the actual cutoff, but you guessed it, it doesn't exist. As someone stated before a usage meter could work, but you still have to cutoff manually, and you'd have to start at fully charged. A better answer is a simple volt meter on the pack. Since the discharge curve of these cells isn't nearly as flat as something like LiFePo4 cells, voltage works just fine as a fuel gauge. It can be simplified even further to a multi-LED dummy gauge as long as it's tuned for the pack.
If you want to be happy using your EV, just be familiar with your range and always put it on the charger upon return. If it's used to go to work, then charge it there too for more ride flexibility and lower DOD cycles. Even at the 12p pack I'm recommending range will be fairly limited, because it's a small pack. Double that and then you're getting into truly useful territory.
I think the LVC will work OK if you just set it for a lower voltage to accomodate the sag. Some cells could sag more than others toward end of capacity and the cells can certainly be damaged if one goes all the way to zero or gets reversed.
Not perfect, but about as good as it can get without resorting to some really sophisticated system that also measures the current.
In a pack made from series-parallel combinations of cells, there's always a chance that you could lose a parallel connection in a sub pack that would drastically reduce the capacity of that sub pack. In this case, individual cell monitoring will prevent damage to the remaining cells.
Not perfect, but about as good as it can get without resorting to some really sophisticated system that also measures the current.
In a pack made from series-parallel combinations of cells, there's always a chance that you could lose a parallel connection in a sub pack that would drastically reduce the capacity of that sub pack. In this case, individual cell monitoring will prevent damage to the remaining cells.
John in CR
100 TW
If you tie all the parallel cells together, which is required by a cell monitoring system, then you open up a whole can of worms from my point of view, the most likely of which is to create imbalances in the pack due to IR differences between the series levels. If you go that route, then you also have to go to a sophisticated matching scheme like Doc does for the pack to stay balanced, which requires cell testing for both capacity and IR, and I seem to be the only one mentioning cell testing at all, but it's the most critical and involved part of the build. In such a small pack you may not even be able to achieve the total pack matching required of paralleling all levels without breaking up the blocks of cells, because it requires matching both parallel and series for a well balanced pack. Then you give up the great physical structure from the factory.
I go a different route than Doc with no parallel ties between the series strings at the end. I believe it helps keep the cells in balance through the capacity match of the cells in each string, and the inability of the currents to take a meandering route through the width of the pack. It's worked so far with good balance maintained thru hundreds of cycles of each of my packs containing a total of well over 1000 cells without a single cell failure, since the breadth of the pack ensures no cells are stressed. One of my oldest packs is even a 10p7s that would slide right into Marty's scooter, though I'd like to do a 12p7s pack for him.
Marty doesn't want to build a pack, so he certainly isn't going to repair one. That means a cell failure is a pack failure, so what's the point of identifying where it is? The cells don't need balance charging. Any strategy that includes a bunch of extra cell soldering, lots of small wires, extra circuit boards that can themselves fail, etc. is more likely to kill cells, just like the circuit boards in the Makita packs apparently often kill that end cell to which they are connected, and just like many have had their BMS's kill cells they were supposed to protect. These cells have a tendency toward self balancing that negates the need for all that stuff, which can be verified by all who have properly built packs using these used cells.
I go a different route than Doc with no parallel ties between the series strings at the end. I believe it helps keep the cells in balance through the capacity match of the cells in each string, and the inability of the currents to take a meandering route through the width of the pack. It's worked so far with good balance maintained thru hundreds of cycles of each of my packs containing a total of well over 1000 cells without a single cell failure, since the breadth of the pack ensures no cells are stressed. One of my oldest packs is even a 10p7s that would slide right into Marty's scooter, though I'd like to do a 12p7s pack for him.
Marty doesn't want to build a pack, so he certainly isn't going to repair one. That means a cell failure is a pack failure, so what's the point of identifying where it is? The cells don't need balance charging. Any strategy that includes a bunch of extra cell soldering, lots of small wires, extra circuit boards that can themselves fail, etc. is more likely to kill cells, just like the circuit boards in the Makita packs apparently often kill that end cell to which they are connected, and just like many have had their BMS's kill cells they were supposed to protect. These cells have a tendency toward self balancing that negates the need for all that stuff, which can be verified by all who have properly built packs using these used cells.
kfong
100 kW
Woohoo! Sometimes I work best under pressure. Here is a picture of the smart switch. At first I tried etching a circuit board, no go. Forgot how much I hated doing that. Luckily, I had sections from unrelated designs I could piece together and cut up. Things had to be changed from my original design, some mistakes and some rethinking. I originally had it setup with a DC/DC power section, but didn’t like the cost and realize it can be done with linears and resistors. This is actually how a lot of the ebike controllers do it to cut cost. A couple of revisions later and some test programs. We now have working switch that is controlled by a processor that can monitor the voltage of the pack. The advantage of this is any voltage sag can be placed on a moving average and filtered out. So short bursts of energy can be ignored and only if the battery stays at the low voltage set point for a determined time will the switch totally shut down and totally remove the battery from the rest of the circuit. There is a reset switch that you push to start it up again. I would have taken a video, but with no moving parts, it wouldn’t have been very exciting. Gotta love the 4110’s they are cool to the touch; it’s the regulators that get warm. I will be testing it on the bike next, but need to get the programming cleaned up and finished.
Noticed it's rather small too!, eventually I will get pcb’s made, but need to go through the testing/torture phase. Preliminary results look good, not much I want to change.
Marty, another idea is to take a cycle analyst and one of the smaller bike controllers and just do what I’ve done with my BMC build, you still have to remember to switch it off though.
John, paralleling packs shouldn’t be an issue with the konions. I haven’t had problems with my setup.
Ignore the big resistor at bottom center, I didn't have a 2 watt at the time.
Noticed it's rather small too!, eventually I will get pcb’s made, but need to go through the testing/torture phase. Preliminary results look good, not much I want to change.
Marty, another idea is to take a cycle analyst and one of the smaller bike controllers and just do what I’ve done with my BMC build, you still have to remember to switch it off though.
John, paralleling packs shouldn’t be an issue with the konions. I haven’t had problems with my setup.
Ignore the big resistor at bottom center, I didn't have a 2 watt at the time.
kfong
100 kW
Zero current when off at least not measurable if there is any parasitic through fets that are in off state, it's actually pretty cool how I did it, but can’t give any details until I start producing them. It runs off full pack voltage. Current consumption when on isn’t much, only the wasted heat generated by the linear regulator. I could have used switching supplies but they are pricey compared to a resistor. The design is scalable. Lots of software possibilities on this one as well.
From concept to design, it went rather fast. Even though I had most of the things already worked out in my head. I just needed some motivation to get it done. Now if I can only get some of my other projects finished this fast. I have a cool bike display in the works, but I need to get out and play.
From concept to design, it went rather fast. Even though I had most of the things already worked out in my head. I just needed some motivation to get it done. Now if I can only get some of my other projects finished this fast. I have a cool bike display in the works, but I need to get out and play.
KTP said:
I have a couple of ways in mind to do the total shutdown also (well, except for the reverse leakage current in the body diode of the mosfet, as you mentioned). One way purely analog, the other way with a microcontroller. Probably nowadays, the microcontroller way is cheaper/fewer parts...it is actually cheaper now to use a 8 pin micro than a simple 555 
I guess you are driving the largish gate capacitance of the mosfets with something other than the micro pins to keep it firmly in the linear region of operation?
I guess you are driving the largish gate capacitance of the mosfets with something other than the micro pins to keep it firmly in the linear region of operation?
marty
10 MW
Nice job on the Smart Switch! No idea how it works? But if it works? All is good. Your reset button reminds me of a circuit breaker reset button. Other Low Voltage Cutoffs reset when you charge battery up full again. Your reset switch will work fine for me.kfong said:
I prefer that my scooter have your Smart Switch rather then a Cycle Analyst and one of the smaller bike controllers. Cycle Analyst tells me way more then I want to know. Here is how the scooter will work. Ride it till the LVC shuts it off. Walk back home. After I determine how far it can go, I will keep all trips under the maximum distance that the scooter can go.
John in CR
100 TW
So Marty is going to do some testing after all, and then after testing use the pack smartly. It sounds like we have a moving target. Marty's moving in the right direction, so that's good. A voltmeter is the useful piece of electronics for these batteries, which can take the form of a tiny little display on the handlebars, especially since you range can very quite significantly. The point I gauge everything on is the nominal voltage, which is supposed to be the half a tank point. The motor already has a last resort LVC to protect the cells, but they can't deliver the much current when they get low, so the batteries themselves already have a soft cushy LVC that speaks to you quite clearly. You feel the power and speed fall off, and that's the point you want to avoid repetitively getting to.
Kfong, cool switch. I hope it works out. If it can cut the current, can you figure out a way for it to replace contactors and be able to handle higher currents? We don't need LVC's, since they're built into the controllers, but something that is small and can act as a switch has an instant market for those wanting to do Delta/WYE switching, or series/parallel switching. Regarding your pack, how many cycles, and how many full discharge cycles?
Kfong, cool switch. I hope it works out. If it can cut the current, can you figure out a way for it to replace contactors and be able to handle higher currents? We don't need LVC's, since they're built into the controllers, but something that is small and can act as a switch has an instant market for those wanting to do Delta/WYE switching, or series/parallel switching. Regarding your pack, how many cycles, and how many full discharge cycles?
kfong
100 kW
John, I can easily add a push button switch that can shut the system off. This will allow you to plug batteries without any sparks, but you will need to remember to push the button before disconnect since you need to tell the processor. Nice, I hadn’t thought about that. Further, button can be used to disconnect to charge, so all that would be needed is a cable hookup to the batteries. No need to disconnect anything but the charging cable.
These are IRF411B’s ganged, we are talking about. High current is already there. Just a matter of how many I want the final version to have. I don’t think it could be use in wye/delta, not actually a relay. I’ve been using my packs all summer, lots of serious crashes too. No problems with them. I plan to warranty the pack as long as Marty owns the scooter. I’m pretty confident on my builds. Too many years in the auto industry.
These are IRF411B’s ganged, we are talking about. High current is already there. Just a matter of how many I want the final version to have. I don’t think it could be use in wye/delta, not actually a relay. I’ve been using my packs all summer, lots of serious crashes too. No problems with them. I plan to warranty the pack as long as Marty owns the scooter. I’m pretty confident on my builds. Too many years in the auto industry.
John in CR said:
kfong
100 kW
Just did a quick field test, used one 10S A123 pack, same one on the bench. I wanted to see if I needed any heat sink. Not a problem even had the unit secured with packing foam. Any heat would have showed with a melted indent. Used my Cyclone/GT build, ran it past 35 amps at times and over 500 watts, did easily 450watts continuously. I rode it up to Subway and back for lunch. It’s plenty good for a scooter, not bad for an average ebike as well. I have more software to do so I need it working for now. No serious torture test yet, but at least I know it can handle the current loads.
kfong
100 kW
More like 5 
More testing, gotta love it when thing work out as planned. The LVC is working as expected. I’m only using one A123 pack and the voltage sag is significant. I can trip it when I want to just by putting it high gear. Just reset and away I go. Now I have to work on the software and give it some smarts. I have several ideas to try so it won’t cut off until the batteries are truly depleted. Fun stuff.
I wish I could say more about it, but I want to keep it to myself till I start producing them. I can say that heat isn't a problem and I can put it in an enclosed box.
More testing, gotta love it when thing work out as planned. The LVC is working as expected. I’m only using one A123 pack and the voltage sag is significant. I can trip it when I want to just by putting it high gear. Just reset and away I go. Now I have to work on the software and give it some smarts. I have several ideas to try so it won’t cut off until the batteries are truly depleted. Fun stuff.
I wish I could say more about it, but I want to keep it to myself till I start producing them. I can say that heat isn't a problem and I can put it in an enclosed box.
KTP said:kfong said:
Obviously it could handle the current loads, you have like 4 power mosfets paralleled and they are not even being switched continuously
How fast are the 4 mosfets switching on or switching off? (ie, what are you driving their gate with?)
Randomly designed a LVC cutoff circuit much like that a while back. It was mated to the LVC only boards we have.
I usually figure about 20 amps per IRFB4110 to keep the heating below the point where a heat sink is required.
Note the diode across the load connection. If the load is dropped at high current, there could be a very large inductive voltage spike created. The diode protects the FETs from the spike.
Since the FET is only going to switch once, the switching speed doesn't need to be super fast. It won't have time to heat up during switching. Heating due to resistance will be more of an issue.
I usually figure about 20 amps per IRFB4110 to keep the heating below the point where a heat sink is required.
Note the diode across the load connection. If the load is dropped at high current, there could be a very large inductive voltage spike created. The diode protects the FETs from the spike.
Since the FET is only going to switch once, the switching speed doesn't need to be super fast. It won't have time to heat up during switching. Heating due to resistance will be more of an issue.
John in CR
100 TW
kfong said:
I understand it's not a relay, but can you ramp up the power handling so it can take the place of a multi-kw relay up to 100V, cheap and compact? If so, I personally need about 50.
kfong
100 kW
Yep, but you better get me the specs you need before I send out the boards. There won't be multiple versions so I plan to try to cover as much of the ebike needs as possible. I won't be sending them out for a few weeks. The IRF4110 are already rated for 100 volts so it can already do this. The only question is how many fets do I want to gang up in the small space I plan to put it in. Since I don't have a 100volt setup I can't test it but I'm pretty sure it won't be a problem. I too bought those kilovac relays, but thought them too bulky for a bike and used them for something else. This is a very cool solution. I should have had something like this done already but it’s hard to do anything in the summer and other priorities keep getting in the way like kiteboarding
John in CR said:
John in CR
100 TW
Sounds great. We should wait to make sure Doc Bass isn't already doing the same with the series/parallel pcb he's working on, and get feedback on his results. No use reinventing the wheel if it's already being done. Maybe your design or portions of it is better, with a meeting of the minds leading to the best approach. The lack of gearing is the biggest strike against hub motors, and if series/parallel switching gives low turn count motors better hill climbing and take off acceleration without the need of extreme current, then it's potentially a huge deal.
John
John
kfong
100 kW
John, I picked up a 10x6 hub this summer, so I might look into that. I've seen some of Doc’s work, but haven't kept up with it. Now that I have a torquey hub motor, I might want to speed it up too. I'm curious if it is possible to use fets in this situation. I know he has been using relays, but that to me takes up space and there is also the current limit and possibility of a fused contact at such high currents. As far as I know he is still using relays and it might still be the only option for WYE/DELTA switching.
My smart switch would come into play if you were to use different battery packs, but I wonder if they would have problems when they interact with each other. I would need to do some experimenting when I have multiples of them.
My smart switch would come into play if you were to use different battery packs, but I wonder if they would have problems when they interact with each other. I would need to do some experimenting when I have multiples of them.
John in CR said:
John in CR
100 TW
All of our current already goes through FETs and is switched on and off, so what's a few more, unless they don't like being on for long periods. I'll leave that stuff to you e-guru's, but as soon as you mentioned going that route for a battery switch I thought whoa, what about over here. They're not being too forthcoming about plans involving a PCB and series/parallel switching in that thread at the top of the ebike Technical forum, but I don't see contactors mentioned.
John
John
kfong
100 kW
John, send me the link and I will read through it.
More testing, I have the cutoff switch tailored to my liking, but I might need to make two software versions. One that is more aggressive for guys like you John. I’m running with the 80/20 rule. Doctorbass told me to cut off at 3.3v per cell if I want the longest possible life with my Konions, and I agree after some research. At 3.3V, my cutoff with 10s is at 33volts. Cycle analyst tapers the power down so I don’t really experience the voltage sag you do with your setup John. I know I have a couple more ahrs left but I’m willing to do that so none of my cells can get taken down. These are all used cells and you don’t have any idea of how many cycles are on them even if you match the cells. I get 9.5 to 10 ahrs from a 12ahr pack. So there is 2 ahrs in reserve providing the cells actually have that much left in them. But like I said, I rather not push them hard.
With the A123 pack this morning, it’s a beat up pack I bought from some Ezone member. After receiving and seeing how crappy a soldering job he did. I decide not to use it for anything but as a beater pack. I was pushing 1200 amps this morning with it with some serious voltage sag. Happy to say smart switch stays cool. Remember these cells can push out 70amps. So at 40 to 50 amps the cells will hold at 2.5volts and bounce back happily. The Konions are rated at 15amps and even this I think is optimistic. With the Konions I would not even attempt this with a single pack since I’m already drawing 13-15amps just riding normally, and that doesn’t take account of startup. I have an 8S konion pack left, but will have to do the testing on the bench with it.
John, how low do you bring your packs down to and how much voltage sag do you push. Your earlier post suggests to me you are pushing your packs way too hard to me. So how long do you keep it under the LVD point? With my conservative setup, I’m only allowing 20 seconds of voltage recovery. This should be fine for startup and quick bursts of energy. With a full charge, LVD should even happen on a system with enough capacity, only when you are running low, but do you really want to push your pack that hard?
More testing, I have the cutoff switch tailored to my liking, but I might need to make two software versions. One that is more aggressive for guys like you John. I’m running with the 80/20 rule. Doctorbass told me to cut off at 3.3v per cell if I want the longest possible life with my Konions, and I agree after some research. At 3.3V, my cutoff with 10s is at 33volts. Cycle analyst tapers the power down so I don’t really experience the voltage sag you do with your setup John. I know I have a couple more ahrs left but I’m willing to do that so none of my cells can get taken down. These are all used cells and you don’t have any idea of how many cycles are on them even if you match the cells. I get 9.5 to 10 ahrs from a 12ahr pack. So there is 2 ahrs in reserve providing the cells actually have that much left in them. But like I said, I rather not push them hard.
With the A123 pack this morning, it’s a beat up pack I bought from some Ezone member. After receiving and seeing how crappy a soldering job he did. I decide not to use it for anything but as a beater pack. I was pushing 1200 amps this morning with it with some serious voltage sag. Happy to say smart switch stays cool. Remember these cells can push out 70amps. So at 40 to 50 amps the cells will hold at 2.5volts and bounce back happily. The Konions are rated at 15amps and even this I think is optimistic. With the Konions I would not even attempt this with a single pack since I’m already drawing 13-15amps just riding normally, and that doesn’t take account of startup. I have an 8S konion pack left, but will have to do the testing on the bench with it.
John, how low do you bring your packs down to and how much voltage sag do you push. Your earlier post suggests to me you are pushing your packs way too hard to me. So how long do you keep it under the LVD point? With my conservative setup, I’m only allowing 20 seconds of voltage recovery. This should be fine for startup and quick bursts of energy. With a full charge, LVD should even happen on a system with enough capacity, only when you are running low, but do you really want to push your pack that hard?
marty
10 MW
Ladies and gentleman, I am pleased to announce the winners of the Battery Building Competition.
First place goes to KFONG. Kfong please email me a ship to address and I will ship batteries to you. Think it would be best if I also buy this charger - [Tenergy Intelligent 3.5A Charger for 29.6V 8S Li-ion / Li-Polymer Battery Pack] and have it shipped to you so you can test scooter battery with the charger that I will be using.
Second place goes to JOHN in CR. John if I am able to get any more Makita battery packs, I will contact you first before I sell them in the For Sale section of this forum. Please get your debit / credit card or PayPal account all loaded up with $.
Third place goes to GRINDZ145. Grindz since you live right down the road from me, I would like to invite you to come to Buffalo. Will buy you a dinner in a nice restaurant. Bring family and friends if you like.
Thanks to everyone else for sharing your knowledge.
First place goes to KFONG. Kfong please email me a ship to address and I will ship batteries to you. Think it would be best if I also buy this charger - [Tenergy Intelligent 3.5A Charger for 29.6V 8S Li-ion / Li-Polymer Battery Pack] and have it shipped to you so you can test scooter battery with the charger that I will be using.
Second place goes to JOHN in CR. John if I am able to get any more Makita battery packs, I will contact you first before I sell them in the For Sale section of this forum. Please get your debit / credit card or PayPal account all loaded up with $.
Third place goes to GRINDZ145. Grindz since you live right down the road from me, I would like to invite you to come to Buffalo. Will buy you a dinner in a nice restaurant. Bring family and friends if you like.
Thanks to everyone else for sharing your knowledge.
Similar threads
