Lithium auxiliary battery and cold weather charging
- Thread starter jimbob394
- Start date
shade
Well-known member
There we go.
As a nod to your unconstructive criticism, I'll mention that some reactors are cooled by liquid (well, molten) sodium. Seems appropriate for this application, too. What's the worst that could happen?
shade
Well-known member
Good point. Molten sodium isn't as heavy.
hour
Observer
Yeah, I was worried about that too. I even cut the silicone mat to not be full width, to better allow heat to rise up directly off the aluminum on the sides. Then I promptly ruined that idea by packing the sides. So I'll find comfort in luthj's comment assuming he's correct
Yeah, so here's the thing. I could base my PID value to the MOSFET on the temperature sensors of the BMS (its internal one and its remote probe) but that would cause the heat pad and plate to get extremely warm. That's why my reading is coming directly from the plate even though it's
Overnight the temperature of the plate stayed consistent to .2 degrees and was sipping on 6 watts when I woke up, maintaining ~78.5*F. I didn't/don't really have any idea how long the soak takes, or how accurate the thermistor is, but I'd think 7+ hours sealed up would be enough. The BMS read a combined temp of 74.5*F, identical on both points of measurement. This has never happened before - the probe (wedged in a crack between cells about half way down) has always been many degrees different. That makes me think this is a solid solution, low - slow - thorough. Oh yea, BMV shows 75*F. So that's consistent reading at bms, probe, and BMV's sensor on the positive terminal. I'm liking that.
Made some adjustments assuming about a 4 degree difference in plate temperature and overall temperature if left to run, so we'll see how that does over the course of the day. Then I need to test at my actual target temperature which will probably be somewhere between 47-50*F. Ya see, I have no clue if the floor size, thickness, exposed area or lack there of, insulating materials, etc.. if maintained at 50*F, is enough mass to influence the entire box to keep 50*F too, when 10*F air is surrounding it.
I had a legit idea on maintaining the temperature with liquid but didn't want to lose the real estate even if it represented thermal mass. Was going to build the sump out of plexiglass (ay, I've built 60 gallon sumps using plexi and the proper solvents and joining techniques, no leaks 12 years!) but finding a way to heat it on 12v wasn't productive. Not like they make junior sized aquarium heaters in that voltage. Doing that would have just been adding complexity for the hell of it though and I'm ready for this project to wrap up
Last edited:
Even ambient air inside the box is enough.
Just model how long the box temp needs to be held, in order to be sure internal cell temp is at target.
If starting point is 20below might be 24hours.
If 0 maybe six is enough.
That way the point where the pad temp is closest to (not touching) the cells can be moderated not to go too hot, that will be a damage / wear point.
I think a 1" space between pad and cells bottom would be best, cut a rectangle of that plastic greenhouse bench grid or something, if you melt that you're running way too hot.
Just model how long the box temp needs to be held, in order to be sure internal cell temp is at target.
If starting point is 20below might be 24hours.
If 0 maybe six is enough.
That way the point where the pad temp is closest to (not touching) the cells can be moderated not to go too hot, that will be a damage / wear point.
I think a 1" space between pad and cells bottom would be best, cut a rectangle of that plastic greenhouse bench grid or something, if you melt that you're running way too hot.
shade
Well-known member
Seems prudent. A little sculpting would also allow some airflow through the grid.
shade
Well-known member
I must be misunderstanding you, since you mentioned that "air inside the box" is enough.
If the heat source is separated from the battery, and even heating is the goal, it seems that encouraging convection of the internal air space would be a good idea, even if it isn't the most efficient transfer medium.
luthj
Engineer In Residence
Convection does transfer heat decently over modest differences. You can easily figure out the temp differential from top to bottom of the box at steady state. Put the whole pack in the fridge. Let the heater run for 5-6 hours until the temp inside stabilizes. Measure the different from top to bottom. If its within 20F, you should be good. If its greater, you may need to adjust your layout some. With decent insulation the difference between the top and bottom will be minimal, as the heat transfer within the box will be much higher rate than transfer between the box and environment.
Air warming the cells is fine, just very slow. The key is for all cell **interiors** to be pretty even, basically at the box ambient air temp.
Circulating air between each cell is IMO prohibitively difficult given the compression requirement.
So just saying flat aluminum plates between each cell would make for much faster heat transfer, long as the plate temp is not allowed to get too high.
Without that, the cells are just one block thermally, the inside cells very small surface area of edges exposed, will take a long time to get up to "box ambient".
This is why EV TMS use circulating fluid, which IMO is too complex for DIY.
Circulating air between each cell is IMO prohibitively difficult given the compression requirement.
So just saying flat aluminum plates between each cell would make for much faster heat transfer, long as the plate temp is not allowed to get too high.
Without that, the cells are just one block thermally, the inside cells very small surface area of edges exposed, will take a long time to get up to "box ambient".
This is why EV TMS use circulating fluid, which IMO is too complex for DIY.
shade
Well-known member
Ah, I see. Using aluminum to conduct heat from the horizontal plate to vertical plates between the cells makes sense. Not a heat pipe, but still should be effective.
luthj
Engineer In Residence
In situations like this its all about the heat transfer rates between the two semi-isolated systems. 1/2" of polysio is probably in the same ballpark as 4-5" of battery stack. So the transfer within the box top/bottom will be similar to the loss through the box to the exterior. Rough math indicates that the delta-T between top/bottom of the box will be the same as or smaller than the delta T between the box and environment . If necessary you can add some more insulation to the top to balance things out.
With a small air gap where warm air from the heating plate can rise from. It will collect at the top of the box, balancing things out somewhat.
Also note that charge rate restriction in lithium refer to the average cell temperature. So if the cell ranges from 30-70F (top to bottom), you can typically charge at a rate specified for 50F.
With a small air gap where warm air from the heating plate can rise from. It will collect at the top of the box, balancing things out somewhat.
Also note that charge rate restriction in lithium refer to the average cell temperature. So if the cell ranges from 30-70F (top to bottom), you can typically charge at a rate specified for 50F.
Last edited:
hour
Observer
That's pretty legit. A perfectly sized box up to the shoulders of the cells made of aluminum, with an additional 0.75" interior space. Then slide a 0.25" aluminum plate between each cell. It'd be a pain in the pecker to make though as you'd want everything absolutely flush and perfect so dividing plates make acceptable contact with the bottom and the walls., i'd imagine. Then put that box in another box.
Don't think I need to elevate them for fear of melting for the following reasons. 1) no relay to potentially get stuck 'on', 2) by default the voltage going to the pad is zero, if the controller fails or fuse blows to controller, the pad does nothing.
3) The value I'm writing to control the voltage going to the pad is between 0-255 and I can set a cap on that. Here's my line of code from last night getting figures using a 12.33v power supply, and which caps the pad at 50w. I'll have to do the figures again from the battery, as a kill-a-watt and dc brick with conversion losses gave me:
int MAX_PWM = 90; //140 @ 12.33v is 92w or something... 90 is 50w@12.33 26=14.5w 22=10.9w 15=3.6w 17=6.0w 16=4.9w
4) If the thermistor married to the pad provides garbage readings (it reads every 10 milliseconds... so a crap ton) they're thrown out, and I wait until I've collected 12 samples before adding them all up and dividing by 12 to get the average resistance, and then calculating the temperature from that and running the PID routine. Haven't had a single bunk reading though in 11 or whatever hours.
5) if the BMS' two temp measurement points are above 45 I'll send 0 volts to the pad. And I'm installing another sensor directly on top of the plate too, which will send 0 volts if it's say... 3 degrees over what the thermistor is reporting. These things being at the top of the code and causing a bail out from the rest of the routine.
At least, that's how I hope things will work. Other mildly interesting things, the perfect calculated value to maintain the target temp is issued at the end of 12 samples averaged together. With 10ms between samples that means the value is written every 120 milliseconds. I'm finding that with my current P I D values I overshoot my target temperature by 1.5*F in initial ramp up at thermistor. It then reduces voltage until it finds its target and maintains to 0.2*F. It usually writes the same value for minutes on end before it deviates by maybe 1-2 (again on a scale of 0-255, so peanuts) and then returns to the original value it had been writing. I put a frozen water bottle on the plate during testing and it seems responsive enough, though I can't think of any setup that would require immediate response as temp would never change that quickly in the wild.
Depending on how cold it gets outside, I'd do 2-3" as a minimum, with Great Stuff filling in the edge joins gaps and corners.
If you're shooting for a rate over 0.3C, target temp should be 25C°.
For over 0.5C 30°, even hotter if approaching 1C.
Assuming a goal to maximize longevity
If you're shooting for a rate over 0.3C, target temp should be 25C°.
For over 0.5C 30°, even hotter if approaching 1C.
Assuming a goal to maximize longevity