Issues with NiMH cells in configurations popular in diving applications.
The following information was compiled into an article for the DIR Explorers library at the request of CHG, mainly from replies to a post on X-scooters, with some additional external information.
Credit is owed for the majority of the technical information to TobinGeorge who has considerable experience in designing, manufacturing and selling rechargeable battery packs. Reference should be made to the original thread which contains further information on scooter applications, Li-Ion cells and other related topics, as well as other lists where such issues have been discussed before.
It could be said that the use of NiMH batteries with a scooter might give concerns over reliable and predictable burn times with no way to guarantee full charge and no easy way to check. The same could be said to apply to lighting (perhaps less critically), with a more predictable alternative being sealed lead acid (SLA) batteries. So, to what extent is this true or of real concern?
It is acknowledged that NiMH batteries are much fussier to charge, but the known problems may not necessarily stem from the chemistry of the cells themselves but their use in parallel/series combination packs, such as those used in higher capacity canister lights. This may have created the perception that NiMH batteries are unpredictable in terms of charging and known capacity. However, if a pack is properly designed, and the charger is properly implemented, the capacities are predictable. If one is still concerned, a “watt-hour” meter can be used in series with the battery when charging to log what capacity was returned to the pack.
Comparisons with lead acid batteries
No meaningful information about the state of charge of NiMH cells can be directly obtained by measuring the voltage, which is a reliable (and easy to test) method when dealing with SLAs. However whilst for SLAs either open circuit or loaded voltage is a meaningful measure of state of charge, this cannot always be extrapolated to useable capacity. A SLA battery that has lost 20-30% of it’s original capacity may well exhibit the same open circuit voltage and loaded voltage when fully charged as a battery with a higher percentage of original capacity still available. Here again a “watt-hour” logger is the best way to know how much energy is stored in the pack.
Also in a multi-cell pack if one cell fails then the whole series is wiped out. The same situation exists with SLAs (though most series are made up of only two rather than ten cells - e.g. two 12V 7Ah SLAs in a Pro 14 vs. ten 1.2V 4.5Ah NiMH cells per series in a multi-series nickel pack (two groups for a 9.0Ah pack, three for a 13.5Ah design).
What’s the issue with NiMH cells in parallel?
The flaw is the series-parallel design. SLA batteries can be charged and discharged in parallel, Ni-Cads and NiMH cannot (though they can be discharged in parallel - see below). This is no great secret, any applications guide from any cell manufacturer makes this point clear.
But why should 20 cells in one configuration (say series) be worse than those in another (2 sets, each of 10 cells in series, joined in parallel)?
Lets take a simple example, just too cells. First configuration is series, i.e. the positive terminal of the first cell is connected to the negative terminal of the second. When charged in this series configuration, all the current has to flow through both batteries, as it has no place else to go. Now consider a parallel configuration, with both negative terminals connected together and both positive terminals together. In this configuration you can see at least the possibility that two paths exist for the charge current, i.e. all through one cell or all through the other.
Think about paralleled resistors, with only a minor difference in resistance, the current flows will be greatly unbalanced. For Nickel based chemistries the tendency is for any cell impedance imbalance that exists at the start of charging to be magnified by the charging itself. The net result is one cell’s (or string of cells’) impedance drops and almost all the charge current flows through it whilst the other paralleled cells (or strings of cells) remain uncharged.
Other chemistries do not behave this way. Lead Acid and Lithium-Ion for example can be charged in parallel without a problem.
Lets now look at a 9.0Ah x 12 volt pack made up of 20 cells. Configured as two series strings of 10 cells each, with the most negative, and most positive cells in each string connected in parallel. This is a very common type of pack for can lights (and 13.5Ah is just one more series string).
When the cells are new and hopefully fairly close in original capacity (graded cells) there may not be enough impedance imbalance to cause charging problems. However in time, well before the pack has lost any significant useable capacity, enough imbalance is likely to develop (remember this is an imperfect electrochemical system) to make parallel charging problematic.
How does the charger detect a full charge?
Lets look at the typical charger used for NiMH. The chargers sense the change in voltage relative to time (delta V / delta T). For NiMH they are attempting to detect when the voltage stops changing (zero delta V) which is pretty hard to do, and requires a fairly sensitive circuit. When charging a parallel series pack the first string (the one that’s actually accepting current) reaches this zero delta V condition, and the charge terminates.
In most chargers that use thermistor feedback (e.g. the green wire on Halcyon packs) the thermistor is an over temp gauge, i.e. a failsafe, used to stop charging if the pack temporarily exceeds a certain temperature. They are not the primary “full charge” sensor.
Now lets look at some of the typical trouble reports from the perspective of bad pack design.
“My burn time was only half what it should be” Really? no kidding?
“I had to restart my charge x times!” The chargers typically turn off the zero delta V sensing for the first 30 minutes or so because batteries at a low state of charge can give a false reading. Repeated restarting is one (bad) way to fool the charger.
“I had to charge in the refrigerator!” One series string, taking all the charge that was intended for 2 or 3 strings, overheated tripping the thermistor (cooling by the way can limit the cell impedance variation).
Do these problems seem predictable now?
Why did this happen?
Why did this happen in the industry? One guess is that the designers were incorrectly applying their previous experience with Lead Acid (though see industry response, below). Again it’s worth repeating that what the prohibitions against parallel recharging of Ni-Cads and NiMH cells is not buried in some NASA research paper, it’s as plain as the ass on a goat to anybody reading application guidelines from any of the cell makers.
What’s the answer?
What is the correct way to use Nickel batteries? For greater capacity use pure parallel strings. If you need greater amp-hours, use bigger cells. If you absolutely need to use parallel series pack provide a means to charge them string by string separately, as parallel discharge is no problem.
Consider for a moment if your 9.0Ah pack was really two 4.5Ah packs that had a harness that would allow parallel connection for discharge, and could be unplugged from each other for series only charging. All for the cost of a few connectors.
Slow vs. fast charging
It can be counter intuitive, but for NiMH fast can be better. Why? Because is makes the “full charge detection” at zero delta V more reliable.
One is looking at the change in voltage against time. If one charges slowly then the change in voltage vs. time can be extremely subtle and very very hard to detect, it’s really a signal to noise ratio problem.
If one fast charges (~4ºC) then the transition from a positive slope to a zero slope becomes more distinct. In the real world this is the preferred technique.
Can a reliable NiMH pack be as reliable as a SLA? The long proven use of SLA certainly argues in their favour, and the same data pool simply does not yet exist for NiMH. Having said that, if one knew the source and quality of the NiMH cells, that they were well graded, and one had confidence in the charger used, there should be no more concerns about NiMH than SLA.
To put the above into some perspective, the following is an extract from a post by Barry Miller to thedecostop on a thread regarding similar issues.
“[it is right to say that] parallel is not the best way to go. However it has been working well with the right charger. NiMH batteries have proven to last longer than their lead acid counterparts. It was not uncommon to see lead acid batteries fail or have reduced burn times in short order. The 9 amp packs as I build them today and for the last 5 years have proven to be reliable and give very few problems. The choice for using the 4/3A battery was due to reliability of the cell. 5 years ago there were not that many D cells that had a 9 amp rating and the ones that were out there not always available and not that reliable. The D cells today may be much better. The 4/3A battery also allows for a smaller canister. The most difficult issue was getting a charger that would charge all 3 battery packs, 4.5, 9 and 13.5. Using -dv does not work well with batteries in parallel and finding a charger that primary termination is -dv was difficult. The charger I am using today does a very good job and although made by the same company the H is, I have made some changes to it that seem to have addressed issues that the H charger has or had. I am not sure what they use today. In general, the 9 amp NIMH battery as built by myself and H has been very reliable over the years and has given very few problems compared to lead acid. This is not to say some people have not had burn time issues, but with thousands of the 9 amp batteries built in the last years have resulted in lower returns than the lead acid batteries could ever dream of.”
The following are extracts from posts by George Irvine to the gavinscooters list in response to questions over reliability of nickel packs used in scooters, following GI3 stating that he was using the Gavin Mini G (which utilises NiMH) in caves in México.
“I am talking Mexico caves that are shallow and you can swim out of, and I am talking using this instead of the Mako scooters that I have been using there for 15 years, and I am talking repeat diving, not going long on one scooter.
What I have been doing is leaving Makos there. I take two Makos each for me and my dive buddy, one set of doubles each, and three stage bottles each. This is enough gas and scooters to dive all day and come back with full doubles.
The problem with my Makos is that they are now getting old, and I do not want to take my big scooters down there. The scooters I use here for cave diving are all Magnums and they never go anywhere but north Florida.
To answer your question, I would not use nickel packs for any dive that is at all scooter dependent, nor would I use only one scooter for a dive that is scooter dependent. For instance, Casey and I have done dives in Manatee where we ran Magnums 200 minutes, but then I broke the blades off my scooter in there one day and swam the scooter out using exactly the same gas as it took me to scooter in (due to the flow). By way of comparison, we use five Magnums each to do a six hour dive in Wakulla.
I also would not do a no flow dive even in shallow water over reasonable swimming distance with only one scooter, and as anyone who ever dove with me can tell you, I do not use my backgas at all - I always use stages - even where it takes only a part of stage to do the dive as in many Mexico caves. I don’t trust any scooter, to tell you the absolute truth, or any other piece of gear by itself.”
By way of further explanation:
“[A]ssuming everything is done right, the issue still becomes the uncertainty of the burn time RELATIVE to the certainty of burn time with lead acid. Again, this is assuming you have done the best job possible of making the other issues go away. You can expect the same kind of action you get with your nickel lights. This is the tradeoff for lightweight with scooters and for ridiculous burn times with lights.
Obviously, using more than one Mini reduces the risk. Again, when we use two scooters, we are not running one the whole time and towing one as a bailout, we are running each one a little bit.
I have split the back into two packs to lower the current of each pack and the heat, I have them welded and pressed together, and I have the wires so they can not pull on the connections, and I have a splitter in the nose that allows the packs to be charged separately. I am just being upfront about my experience with celled packs and my concerns about them, and making it clear that any scooter with any batteries can fail for any number of reasons, so don’t get overconfident. If you knew the routine I go through before and after every dive with my personal scooters for diving here in Florida, you would see how seriously I take them. Some people just jump in and ride - I just can’t do that.”
What about lights in critical applications.
Of related interest, the following is an excerpt from a post to DIRQuest by Reinhard Buchaly regarding technologies used in the EKPP.
“In my opinion lead acid is hard to beat. It gives a sufficient burntime, at least with WA based HIDs, that means you have a solid six hours with a 14 amp lead acid pack. It is easy to charge, easy to maintain, easy and cheap to replace. I am still using lead acid battery packs as workhorse for usual dives. But for travelling it is a heavy pack.
Whenever it comes to needing very long burntimes, NiMH is hard to beat as well. It gives a lot of capacity in a smaller volume than lead-acid. In older days we were running 20 Ah NiCad for this, but nowadays NiMH is the way to go. It is more expensive, you have to take care of a quality way of charging and checking them. And it has a nice weight when using them for travelling, at least when you are looking at the normal sized NiMH sets.
I mentioned these in the context that using a highly efficient NiMH set does not make sense when the light itself is not as effective as well. Of course these burntimes may be only relevant for a relative small number of people in the community. Even our NiMH backup battery packs for the HID is giving two hours, which we carry in the pocket on long dives, add to this another two hours if you use the pack from your dive partner as well. This backup burntime gives us four hours, which may be sufficient in case of a primary light failure, to return with full light before we may have to go to the normal backup lights. And it makes a real difference when being on HID compared to the usual backup when it comes to scootering speed and signalling ability.”