NiMH Thermal Balancing

[Trumpet]

Zapat,

So you have a thermal switch every how many cells for this to work well? I guess to have good thermal contact with the switch, round cells would have to be in groups of seven (this gives a hex shape cluster). Any more than this per group and the switch will too be far from some cells, no?

I think a lot comes down to personal preference and how far you want to go with a protection scheme like this. What you have said about a cluster makes sense. The heat build-up will nearly always be greatest in the middle of a pack - so that is the obvious place to put one trip. Then distributing the trips throughout the pack according to how much you want to measure localized temperature. If you are not too worried abut getting the absolute maximum charge out of all cells then a derated temperature trip makes sense - this will lessen the need to monitor so many places throughout the pack. Heat will tend to spread fairly rapidly anyway (as long as the charging currents are not too high). You could also thermally bond a trip or two to a copper strip (or similar) going through the pack - this would tend to improve the sensing of the overall pack temperature and normalise the temperature being sensed by the trip. Do the likes of Dewalt/Hilti/Bosch have multiple thermal sensors in their packs (nimh)? I think they only use the one.

While I agree that the extra resistance is not really a problem while charging, while discharging it will hit your efficiency and performance. What I don't quite understand here is how your "extra positive lead used just for charging" will manage to bypass ALL your thermal switches... Unless you bring out two leads from around each switch, and short them together when you are discharging the pack. Am I missing something or what? for a 48V pack, that's about 7 clusters per string, so 7 thermal switches, and thus about 14 or so wires to use as bypasses during discharge. And this is when you only have ONE string... imagine two or three!

As above - I think it depends on how far you want to go with optimising each cluster independently - you could have multiple trips to monitor a string with each trip in series - if one of the thermal switches cuts out that string no longer receives any more charge (until the temperature drops). Only the separate charging leads to each string (or whatever arrangement you have) contains the trips. For discharge you do need a switch mechanism or connectors acting as cross-linking keys (if you are using strings which you parallel up to make the necessary connections) to join the outgoing power (postive) leads - either that or use diodes to isolate the charging feed wires.

 

[jha07]

The klixon idea is good. Too bad there aren't more temperature selections for them.
I've been sort of doing the same thing in terms of thermal balancing. I bought 7 cell subC NiMh packs from Tenergy and put 5 in series for 42V. Then made a second 42V pack to put in parallel. I have thermistors in each pack and they are wired together in parallel. This reduces their resistance and therefore stops my charger when the pack reaches a temp of 36C instead of 52C as the charger was designed. I added a 500 ohm resistor to the thermistor to raise the temp to 38C to get a bit more charge. This setup has been working very well for months. My two packs are always balanced. Both packs are at the same temp when charging is complete. Since each pack has a thermistor, even if one pack heats up before the other, it won't get too far before the charger will stop.
Recently, I decided to add 5 more cells to make my pack 48V. I already had a bunch of really old NiCd subC so I made a 5 series x 5 parallel pack and put that in series with my 42V parallel pack. My NiMh cell are rated 3800mah but in reality are about 3000mah. So paralleled is 6000mah. My NiCd are rated 1700mah, but are over 10 years old and are about 1200mah now. So 5 paralleled is also 6000mah. I put the whole mess together and charge with a 48V charger. I still use the thermistors from the NiMh pack. The NiCd don't have any thermistors. From testing, my parallel NiMh pack definitly has less capacity then the NiCd so there's no danger of overcharging them. I've been using this setup for a month and it's all working well.
People always say don't charge packs in parallel and don't charge NiCd and NiMh together. Well it all works for me.

 

[Knuckles]

Guys, I am amazed how well it works for me. I build a cheap 30-cell C-size (36V-5AH) pack with just 1 Klixon 50C in the center. The Klixon only has a max amp rating of 5 amps at room temperature.
Again, take a look at this current vs. temperature data ... http://endless-sphere.com/forums/download/file.php?id=11613
Looking at the specs, It is clear that at a higher current the Klixon will trip (open) at an even lower temperature than 50C.
(For the life of me I can't really explain why the cells self-balance. But not one dead cell yet and the capacity of the pack has remained fairly consistent.)

So get this ... I started charging my 36V NiMH pack with a 2.5 amp 48V SLA charger (only for a very rapid charge time).
Normally I use a 1.8 amp 36V SLA charger and I can leave it plugged in (unattended) all day and night.
I had done an experiment and monitored the amp draw of the 1.8a36vSLA charger and the amps drop down to about 0.7 amps for the last 1/2-hour of charging.
This actually makes the pack warmer than the 48V SLA charger. But I would never leave the 48V charger attached without a timer.

I have plenty of pics but I was ashamed to post them as my soldering (and tape jobs) may appear rather ugly.
Also I solder button-top cells and don't bother wither solder tab cells. All of these are major no-no's but I don't care.
The radioshack project box hides all, however, and these packs are like workhorses. 30-amps from a 5AH C cell!
View attachment ugly_pack_1.jpgThe white lead is the charge (+). The other leads are the high amp (+) and (-) to the controller (or other bats).

 

[ZapPat]

Well I'll be darned... only one for 30 cells!

I'm guessing you placed the klixon in the dead center of the cells, right Knuckles?

Not too bad a performance hit I guess then, unless you have some kind of super low internal resistance cell, under 2mohm DC... unlikely to exist I believe. NiMh have fairly high DC resistance in general (someone correct me if you have measured otherwise, I would like to know!)

Well this might be good for me too, and I would likely attach the klixon to a thin heat-conductive metal plate going between each 2 rows of cell ends for each string (another user suggested this idea using copper, but I think aluminum is better because it has quite a bit less thermal mass as well as less weight)

I will definitively think about trying the klixon NiMh charging solution.

 

[Trumpet]

Knuckles,

I think it works so well because you are probably using good quality cells that are in good shape - nimh is pretty damn rugged, just susceptible to damage due to incorrect charging - their inverse voltage profile at full charge being the culprit. Whilst all the cells are in good condition then the single trip buried in the centre of the pack is sufficient - it is measuring the highest temperature spot within the pack (unless one of the outer cells goes U/S) - there is the potential problem that towards the end of the pack life one of the outer cells goes bad and gets overcharged whilst the trip in the middle is still measuring the bulk temperature. One thermistor is all I have seen being used for 18v (15 cell) power tool battery packs - so it is a well used and proven method.

I've got two expensive chargers a Robbe and a Schultz - to be left unattended I'd only be happy with the thermal cutout of the Robbe active or the Schultz with a charge limit - even that is not great becuase then you have to guesstimate how much charge your pack needs each time.

As a safety precaution and to spread the thermal monitoring I think using two or more of these trips in series would be a good idea - then you are guarding against failure of one of the trips as well becuase if the trip does go U/S then you are looking at guaranteed thermal runaway.

 

[Knuckles]

Well ... (more confession) I use Tenergy and Powerizer (w/ 8 m-ohm internal resistance). Cheap cheap cheap. These bats will get quite warm on a long ride (from the high amp draw) and I know I am pushing the life expectancy of these lil' c-size cells. My battery cages can accept 4 of these "black boxes" ... 4 x 30 = 120 c cells but I am only using 2 boxes for now for a 72V nominal system.

I abuse the hell out of these boxes too. Pull way to many amps, leave them in the rain, and more. And here is another goofy trick I do ... I charge them ... they get nice and toasty ... I then put the box in the freezer for a while and try to charge them again. All for the sake of ... heck I not even sure why I do this ... just to see what happens to the voltage of the pack maybe.

FWIW my single string of 24 a123 cells also warms up my chopper forks when I go for a "full blast" 10-minute 30 MPH dirt trail session. Bats will get warm when you pull the juice from them.

But the use of many Klixons in series throughout the pack is absolutely a good idea. I would do this if I was to ever sell a NIMH "hot" box to someone. Also the cheap sub-c cells have a 5 m-ohm internal resistance and are rated for like 40 amps. You can get these at a 4.5 - 2.8 AH rating. Just a lot of solder to build a lot of smaller packs to make a big capacity system.

The idea of using project boxes that are dimensionally similar to various SLA battery dimensions is kinda neat as you can get to 72V with a much smaller volume.

960 AAs anyone?

 

[ZapPat]

Knuckles,

I think it works so well because you are probably using good quality cells that are in good shape - nimh is pretty damn rugged, just susceptible to damage due to incorrect charging - their inverse voltage profile at full charge being the culprit. Whilst all the cells are in good condition then the single trip buried in the centre of the pack is sufficient - it is measuring the highest temperature spot within the pack (unless one of the outer cells goes U/S) - there is the potential problem that towards the end of the pack life one of the outer cells goes bad and gets overcharged whilst the trip in the middle is still measuring the bulk temperature. One thermistor is all I have seen being used for 18v (15 cell) power tool battery packs - so it is a well used and proven method.

I've got two expensive chargers a Robbe and a Schultz - to be left unattended I'd only be happy with the thermal cutout of the Robbe active or the Schultz with a charge limit - even that is not great becuase then you have to guesstimate how much charge your pack needs each time.

As a safety precaution and to spread the thermal monitoring I think using two or more of these trips in series would be a good idea - then you are guarding against failure of one of the trips as well becuase if the trip does go U/S then you are looking at guaranteed thermal runaway.

Well your efficiency and performance go down as you add these thermal switches, so add the minimum you need. My ideal of one per 6 or 7 cell clusters would turn a good low resistance NiMH pack into a mediocre one, unless you feed out shorting wires for each switch to use during discharge. These klixons are also quite expensive from what I've seen, much more so than regular thermistors.

My ideal would be to use a micro-controller to monitor regular low-power thermistors installed in every 7 cell cluster (or other good thermaly connected group). Each complete string of cell clusters would have a ground referenced MOSFET switch controlled by one uC output. There are common pics with about 12 ADC inputs (40 pin DIPs), or if one wants to monitor multiple strings with one single micro, we could easily use a multiplexing chip to read extra thermistors, or just use a seperate monitoring micro board for each string. This type of program writen in C would be fairly trivial to write, but how many people would be interested in this I wonder?

 

[Knuckles]

Well your efficiency and performance go down as you add these thermal switches, so add the minimum you need.

Oh no ... not at all. The thermal switch is independent of the positive (+) main (current out) of the battery pack. That is why there are three wires and not two.

ie. Solder all the cells together and place in a box. Then add a second lead to the positive terminal and solder that to one terminal of the Klixon. Then solder another lead to the other terminal of the Klixon and run that wire out of the of pack (the thin red wire in the 15 cell sub-c pack in the pic below). Embed the Klixon at the center of thermal mass within the pack. The red wire is the charging lead and is never used for discharging the battery. The charger (low current in) and the load (high current out) share the common ground (-) of the battery pack to complete their respective circuits.

View attachment GEDC0377.JPG
btw. I leave my charger in the bat rack on the bike all the time. There is no reverse current when the charger is unplugged but still connected to the battery charging lead. A true PHEV. Very simple to construct too.

cheers

 

[ZapPat]

But the use of many Klixons in series throughout the pack is absolutely a good idea. I would do this if I was to ever sell a NIMH "hot" box to someone. Also the cheap sub-c cells have a 5 m-ohm internal resistance and are rated for like 40 amps. You can get these at a 4.5 - 2.8 AH rating. Just a lot of solder to build a lot of smaller packs to make a big capacity system.

While I don't at all dispute the fact that the klixons would be doing a good enough job of protecting the NiMh cells if they are enough of them at the right spots, I think that in your setup you may be reducing your battery pack performance quite a bit. If you use 5mohm cells as you state here, so 7 cells would be 35 mOhms, and you would be adding 20mOhm to this. This is a 57% increase in resistance, and thus 57% more voltage sag too. Higher internal resistance cells will be comparitively less affected by this of course.

My micro/thermistor/mosfet battery monitor only has the added resistance of a mosfet, so from 4 to 10 mohms for the whole string, or about 1 or 2 mOhms added per 7 cell cluster. Only 5 to 10% of the klixons for equal protection.

I could also use the mosfet's internal resistance to use as a current sense if needed, if someone would want to add over-current protection to their battery string too with little added complexity.
Heck, with just a couple resistors more we could check also for over-voltage or under-voltage of the string too! Adding these features costs us extra time to program initialy, but little extra cost and parts are needed. I'm even starting to convince myself, as I sit here looking at the many kWh's of NiMhs I could use on a motorcycle or something...

 

[ZapPat]

Well your efficiency and performance go down as you add these thermal switches, so add the minimum you need.

Oh no ... not at all. The thermal switch is independent of the positive (+) main of the battery. That is why there are three wires and not two.

ie. Solder all the cells together and place in box. Then add a second lead to the positive terminal and solder that to one terminal of the Klixon. Then solder another lead to the other terminal of the Klixon and run that wire out of the of pack (the red wire in the 15 cell sub-c pack in the pic below). Embed the Klixon at the center of thermal mass within the pack. The red wire is the charging lead and is never used for discharging the battery. The charger and the controller (or next bat pack) share a common ground (-).



cheers

I understand this when you are using only one protector per pack, but some might feel a bit nervous about this setup, and would prefer clusters in which each cell directly touches a thermal sensor. But hey, if we aren't charging too fast I might well try to get away with only one switch too. Your setup is simple for sure, and seems to do the job, so good luck and let us know how many miles you run with them over their lives, I want to know! Cheers back at ya, Knuckles!

 

[Knuckles]

Building packs is easy. I like 12V nominal strings of cells.
View attachment Bat_String_1.jpgAnd the strings are flexible too! View attachment Bat_String_2.jpgTape them up and connect together. And be HAPPY!
View attachment 1Powerizer and Tenergy C-cells are both rated (ha ha yea right) 5000mAh.
But these lil babies kick serious e bike ass! I pull 30 amps from them! I so bad! :lol: 8)

 

[jha07]

Hey Knuckles. Have you measured how much capacity your Tenergy or Powerizer cells have?
My Tenergy subC's are way under what they are rated. Just wondering if C's are any better.

 

[Knuckles]

Hey Knuckles. Have you measured how much capacity your Tenergy or Powerizer cells have?

Never bothered checking. I assume the 5AH rating is BS.

But I can do 30 MPH on a fresh 60-cell charge.
(ha ha ... as based on local radar sign at 2:00 AM - no cars around, middle of the road and head held high to reflect the radar beam).

I'd say 20 mile range on 60 ... 5-FAH C-cells (FAH = fake amp hours) with "real" pedaling at 18 MPH. Cut that range in half at 25 MPH and "fake" peddling.
(over 20 MPH my highest gear is useless so I fake pedal). It's a NY thing ... I am a master of "fake" peddling. Keeps the EVIL spirits away.

Heck ... at $0.60/Whr (using BS rating) it's OK. These are "poor man's" bats. But I do dig the burst of amps as I make many short and speedy trips.
Juice is juice. I take what I can get. And these fake bats do deliver the amps. A single string of 60 C-cells! Gotta love it!

cheers

btw ... AH ratings are measured at 1/20th the C rate. So I would need to draw 5/20 = 250 mA for 20 hours to test the rating.
Meanwhile I am pulling 30,000 mA out of these suckers during my "sortie’s" into the urban jungle!

 

[docnjoj]

Just get on your bke and see how far U go! A Wattmeter would be easier, though! But U can use distance and a cell phone (to call 4 help!)
otherDoc

 

[Knuckles]

But U can use distance and a cell phone (to call 4 help!) otherDoc

Hello! I have peddles! It is still a bicycle. Don't need no cell phone. JUST LEGS!

OTOH ... Flat tires do SUCK!

MOMMY!

 

[Trumpet]

Knuckles,

Another type of thermostat that might be of interest - rod thermostats used in water heaters/immersions - will easily take the current, are light and the metal rod will measure an averaged temperature throughout the pack and thety are variable trip temperature and pretty cheap. Min length is about 7" though.

http://www.seered.co.uk/sunvic_low_res_catalogue.pdf

 

[ebinary]

Love the thread - especially when my virgin DeWalt battery packs - drawing through the internal BMS - have now had 2 failures in two months of daily riding. It doesn't bode well for the longevity of the unmodified packs (although the cells may last for many years).

 

[ebinary]

-

Just bought these from Overstock.com (40 cells at 4,500mah):
Low Self Discharge NiMH C (2 batteries/pack) Ground 20 $239.80


As far as I know, its the only LSD NimH C cells available, and I got em for about regular NiMH price (thanks to a great starting price and a discount code at overtock.com). Unfortunately, I cleaned them out. More are available here @ $13.95/2:

http://store.batteryspecialists.com/ape45002.html

I need 10 - 20 more packs for the system I'm considering, but I'll test out these cells before buying more.

Did I mention - low self discharge!!

BTW - someone has tested these at 3C and gotten the full 4500mAH, so they can provide at least some decent current

 

[amberwolf]

Ultra Thread Necromancy: Secret of NiMH the sequel: NiMH Rats Unite!

Knuckles, in the year and a half between then and now, how did your NiMH hold up to charging this way?

I ask because I am right now working with a 36V 9Ah and a 24V 13Ah set of packs, to be used separately on DayGlo Avenger, and am currently manually monitoring charging like a somewhat nearsighted half-blind hawk. I'd rather just plug and go.

Since the only thing I can replace these packs with is the SLAs I have, if I toast them, and DGA is not really capable of carrying those at the moment (not plus cargo), then I'd like to be sure that your charging method doesn't long-term damage the packs.

Now, the 24V pack is already known to be thermally damaged, so it sags at about 0.5C, but the 36V pack is probably ok (I've not really tested yet at full charge, still working on getting it fully charged via micromanagement).

I've got a few thermal switches from various packs, including several laptop batteries. Unfortunately most of the ones from the NiMH laptop packs are 70C. Some of the packs haven't been opened up yet so I don't know their values, but I suspect the same in them. A pair I have from an SLA wheelchair pack are 55C. So I don't yet have a thermal switch safe to use for this method, but I will probably find one in something I already have here.

 

[Knuckles]

LOL I still LOVE NiMH

Temp sensor failure did kill a pack during charging!

pepi sells 50C sensors. But I am all ears on cheap temp sensor solutions.

But ... Start a new thread about NiMH packs and I will add all I know. :lol:

(NiMH kicks ass BUT temp is the key to charging)

-K

 

[amberwolf]

Well, at the moment I'm trying to keep all the stuff I'm posting about them in the DayGlo Avenger MkII thread, since that's what they're being used on.

But it's good to know that they're still working for you. Right now I am successfully charging without getting the pack warm, simply by watching the current and lowering it even further when I see it start to drop a lot toward the end. Once it gets down to about 250mA or so (down from ~900mA), the pack will start to warm up, since it is within half a volt of full. Then I turn it down to 100mA and let it charge another couple of hours or so there, and it's basically done (the voltage drops at that point). Seems to take around 1.4-1.5x as much power to recharge the pack as I take out of it, which is what I've read as typical.