Methods for charging 7S+ lithium-ion battery packs from DC sources

[derGoldstein]

I apologize for the messy post, there are really a couple of different questions here but I didn't want to just leave a terse question without posting what (I think) I found out while trying to find the answers.

I've been looking for ways to charge 7S and 8S battery packs from a DC source, mainly 12v and 24v. They all have built-in BMSs, so balancing is done internally. The first part is just getting the voltage up using a boost converter, that's not a problem. However, when looking into charging ICs, monolithic designs up to 4S are very common, but beyond that I'm having trouble finding any (it may very well be that I'm misunderstanding the search parameters).

Right now the only way I can charge them is using a charger that plugs into an AC, very similar to this one:
http://www.batterysupports.com/24v-294a-5a-lithium-ion-battery-charger-7s-7x-36v-lion-lipo-p-165.html

I opened it up and, beyond the circuitry required for the SMPS (based on a TL494), it only has a TL431, a dual op-amp and a few discrete transistors. There's no dedicated charging IC or an MCU. Does this mean that it's only doing the CC/CV stages?

It looks like quite a few chargers are just buck converters with current regulation. Here's an example:
http://www.powerstream.com/Product3.htm

This looks very similar to the countless CC/CV boards on E-bay, which are usually based on the LM2596 or XL4015, where you adjust the voltage to 4.2v per-cell and limit the maximum current to 1C (or less). Is this a reliable way of charging li-ion packs?

Finally, if a CC/CV circuit is good enough for charging li-ion packs, could I just use a boost converter with current limiting to charge the packs? This would let me just have one circuit that I could connect to a 12v or 24v source, maybe based on an LTC3780.

Sorry again for the mess, any advice would be appreciated.

[NiHaoMike]

Just charge them to 4.1V/cell. That voltage is OK to leave connected at all times.

[jt]

Finally, if a CC/CV circuit is good enough for charging li-ion packs, could I just use a boost converter with current limiting to charge the packs? This would let me just have one circuit that I could connect to a 12v or 24v source, maybe based on an LTC3780.

Yes you can.  As NiHaoMike indicated, just set the max voltage of the boost regulator to 4.1V*the number of series cells.  If you have other requirements such as maximizing cell life, maximizing charge capacity ect. then the more complicated charge profiles are needed. 

[Siwastaja]

There are no "complicated profiles", and no "algorithms" at all. You just follow the datasheet limits, and they are only limits: Maximum current, Maximum voltage, Minimum charging temperature, maximum charging temperature. You enter the current and voltage to a CCCV supply and go. There is nothing else. Never has been. Myths originate from NiMH and lead acid, which are different and need complicated algorithms.

If the pack already has a BMS, that BMS is exactly the thing that does all thinking. It shuts down the charger when any one of the cells hits 4.2V or so. Connect the BMS outputs (it at least has a "charger enable" output, otherwise it is not a BMS) to your CC-CV charger and make sure the current is set below the specified maximum charging current.

When you were looking at those multiple-cell chips, those were the BMS's! You don't need two.

The first thing to check is if the "BMS" indeed is a working BMS or a scam. Look at the imbalances using a multimeter when first charging. All cells should rise fairly close together near 4.2V. If the BMS is set to 4.2V and your charger to 4.1V*num_of_cells as suggested above, you may never reach balancing state, which may still be fine. It is usually not needed. Balancing in a BMS is a completely irrelevant feature, although it's nice to have, it's just a bonus.

Checklist for minimum working li-ion BMS:
- Charge cutoff signal when any cell hits high limit (e.g., 4.20V)
- Discharge cutoff signal when any cell hits low limit (e.g., 2.70V)
- Charge cutoff signal when temperature is below low limit (0 deg C) or over high limit (e.g., 50 deg C)
- Discharge cutoff signal when temperature is over high limit (e.g., 60 deg C)

If you manually measure that the cells are in balance, the BMS is optional for charging. It is more important to stop discharging when any single cell is at low voltage limit.
 
TLDR; Limit current, limit voltage, make sure the BMS is real, and obey the BMS "stop charging" signal.

Edit: If you are unsure about the maximum charging current, use 0.5C max. Even if it's specified at 1C, charging at 0.5C increases battery life.

[NiHaoMike]

The simple BMS boards I have seen cut off at something like 4.3V/cell. Enough to prevent a catastrophic failure, but the cells will not last very long at that voltage.

[jt]

There are no "complicated profiles", and no "algorithms" at all. You just follow the datasheet limits, and they are only limits: Maximum current, Maximum voltage, Minimum charging temperature, maximum charging temperature. You enter the current and voltage to a CCCV supply and go. There is nothing else. Never has been. Myths originate from NiMH and lead acid, which are different and need complicated algorithms.

I agree there is a lot of misinformation out there.  Someone in a review board once asked my team what protection we had against our 18650s spontaneously reversing polarity; he insisted it was a well known issue.  But is there truly nothing to be gained by adjusting the charge profile of a Li-ion?  Seems like the constant current - constant voltage profile is too widely adopted to be based on a myth (to put myself at risk of committing an argument fallacy).  I've also commonly heard that Li-ion capacity is dependent on charge rate (slower the better); is this not true? 

[nctnico]

Charging Li-ion is straightforward: start with constant current up to the maximum voltage (typically max cell voltage * number of cells) and then stay at the maximum voltage until the current drops below a certain value (say 0.05 C). Just make sure the charger never outputs a voltage above the maximum voltage because that may kick the BMS into charge overvoltage protection or in extreme cases can damage the BMS. The charge current depends on the number of cells in parallel and their recommended charge current.

[HKJ]

But is there truly nothing to be gained by adjusting the charge profile of a Li-ion?  Seems like the constant current - constant voltage profile is too widely adopted to be based on a myth (to put myself at risk of committing an argument fallacy). 

CC/CV is the method all LiIon datasheets lists as charging algorithm. There are modifications to it for faster charging, the idea is that you can use higher current for the first part of the charge and then finish the charge with CC/CV.

I've also commonly heard that Li-ion capacity is dependent on charge rate (slower the better); is this not true?

The termination current will affect capacity. Actual charger current is not very important for capacity, but will affect lifetime of the cell, i.e. charging at high current will reduce the capacity of the cell faster, than charging at low current. This is also voltage depend, i.e. at low charge state high current will not cost lifetime.

[derGoldstein]

Thanks everyone, this clears up a lot.

Does anyone know if certain current limiting circuits are too "noisy" for charging purposes? Linear regulators produce the cleanest output in exchange for high energy losses, but how noisy is too noisy to charge with? Low-quality buck/boost converters will have spikes that exceed the 4.2v per cell (even if you do set them to 50mv lower), but only for microseconds at a time.

I'm trying to get something small enough that I can place the circuit on the battery pack itself and be able to charge it from 12v wherever it's available, so I'm looking to see how small of a circuit I can get away with.

[Siwastaja]

But is there truly nothing to be gained by adjusting the charge profile of a Li-ion?

Oh, there are benefits! Combination of high charge rate and high voltage causes more lithium plating, which decreases the life. This issue is sidestepped by specifying something like: "normal charge at 0.5C; or fast charge at 1C*  *not for maximum cycle life". But you can get good life and fast charging if you charge at 1C to 4.0V and then decrease the current. I have researched this on my own to confirm the idea. On a modern LG cell, 0.7C was really the max near full, but you could go to 1.3C near the mid no problem. But this is fine-tuning; it's easiest to just charge at 0.5C (or whatever specified for long-term use, depends on product.)

I've also commonly heard that Li-ion capacity is dependent on charge rate (slower the better); is this not true?

Charge rate doesn't affect the amount of charge going in (as long as you have the CV phase with low stopping value), but too high of a charge rate causes more permanent damage, dropping capacity over time faster than expected.

Discharge rate doesn't affect the capacity either. It's just that you reach the low-voltage cutoff point earlier due to resistive drop using high currents. If you add a CV phase to the discharge cycle, or lower the low-voltage cutoff below the specs (not recommended), you'll get the same mAh out at 0.1C or 2C.

BMS is a problematic term because anyone can use it for any meaning. It's totally meaningless. I would never call those fixed 4.3V "basic" protection circuitries "BMS". BMS is a management system. It needs to do all the basic functions required to, well, manage the battery! Which I listed above. Not much to do, but fixed 4.3V limit is just not enough. This should be obvious, but for starters, you need to use the actual limits given by the cell manufacturer 

[Siwastaja]

Does anyone know if certain current limiting circuits are too "noisy" for charging purposes?

Modest ripple doesn't do any harm at all. To be certain, avoid extremes. And make sure that your switcher feedback loop is stable when you have a battery as the load. Any PSU that is stable will do just fine.

Low-quality buck/boost converters will have spikes that exceed the 4.2v per cell (even if you do set them to 50mv lower), but only for microseconds at a time.

Not a problem. Still, only charge to 4.15V or so to maximize life and get some leeway against errors. Unless you really need 100% of the rated capacity.

I'm trying to get something small enough that I can place the circuit on the battery pack itself and be able to charge it from 12v wherever it's available, so I'm looking to see how small of a circuit I can get away with.

Charger should have current as a primary feedback loop; voltage feedback loop can be horribly slow, like in the 1Hz range. This helps with stability.

Or, if you are charging at rather low rates, or don't need to get exactly 100% full, just skip the CV phase and build a current-mode only charger with a simple latch circuit which turns it off after 4.2V is reached. You'll end up at over 90% if you are charging at currents that are low enough to maximize the lifetime anyway.

[derGoldstein]

I've been charging several battery types using CC/CV buck modules and it works fine so far.

Looking at what current limiting dedicated charging use, many of them still use linear regulation (is this just to keep them cheap?). Does it make sense to attempt to smooth out a boost converter's output using an LC filter?

I've been looking at CC/CV LED driving circuits like this:
http://www.ebay.com/itm/261907091027?_trksid=p2060353.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT

It has current limiting, but is there such a thing as the "wrong kind" of current limiting? I know this is more of a basic electronics question, but I don't know how else to phrase it... Most of these circuits use an op-amp and a sense resistor, rather than a function that's built into the driving IC (which is sometimes monolithic and sometimes drives external switches).
Is it plausible to take a boost converter without current regulation and "add" current regulation to the output? There appear to be ICs whole sole purpose is to do that:
http://www.ti.com/lit/ds/symlink/lmp8646.pdf

Feel free to tell me that I'm asking the wrong questions or using incorrect terminology. I'd rather understand how the circuit I'm using works than depend on an off-the-shelf black box solution.
(edited because there were a bunch of newlines after the post...)

[Siwastaja]

LED driver and li-ion charger is essentially the same thing, namely an adjustable constant-current supply. Some chips may list both as typical applications. Charger requires a more precise voltage limit than LED, however.

Linear charge regulators are sometimes used exactly for low cost, but they are low cost (or sane at all) only on very low power (< 5W input), or when the input is already relatively well regulated and happens to be a good match. 0.5A 5V USB connector is a great example of both situations. For higher power, or unsuitable input voltages, thermal management becomes an expensive problem, and you need to go to switch mode.

There is no wrong kind of current regulation, as long as it does what the name says: regulates current. It's ok to keep the average within spec; moderate ripple does not do harm as long as you avoid extremes.

Boost converter already has an output capacitor. No need to add extra LC filters. The power is just fine for charging li-ion batteries. Boost converter is good, btw, as it's rather easy to implement a current feedback in the control loop. Voltage feedback can be slow and secondary, or a latching type that only turns the thing off when the limit is reached.