| General > General Technical Chat |
| LiPo Battery on Holding/Float Charge |
| << < (3/6) > >> |
| tooki:
--- Quote from: Siwastaja on June 29, 2021, 08:38:36 am ---I asked because I didn't understand what you mean by the next sentence. --- End quote --- Yep, you’re right, I had a brain fart there. I’m right in the middle of final exams, stressed and sleep deprived, and got myself mixed up. :scared: --- Quote from: Siwastaja on June 29, 2021, 08:38:36 am ---Never heard about such recommendation from the manufacturers. Maybe you can give an example so we can see what's it all about. --- End quote --- Take a look at the datasheet of the Samsung INR18650-29E batteries my current project uses. (Its specs are still fresh in my mind since I just used the datasheet last week to calculate my charger parameters.) Page 9 says: --- Quote ---1.4 No continuous charge - Charge cut-off condition : Cut-off Current over than 1/20C(Charging method: CC-CV) - Initial recharging condition : Remaining Capacity under than 90% or voltage under than 4.10V --- End quote --- I will concede that this is contradicted by a subsequent statement on p.10 that says: --- Quote ---2.3 Charging time Continuous charging under appropriate voltage does not cause any loss of characteristics. However, the charge timer is recommended to be installed from a safety consideration, which shuts off further charging at time specified in the product specification. --- End quote --- But since it seems they won’t warrant them in that situation, there must be some downside. |
| SiliconWizard:
--- Quote from: tooki on June 29, 2021, 07:40:26 am --- --- Quote from: imk on June 28, 2021, 02:47:31 pm ---Hello, Are there any issues keeping a LiPo battery on a holding charge of 4v? Basically if i connect a modest load of 40ma or so to a LiPo and then connect the output of a Buck Convertor set at 4v to the LiPo plus the Load are there any issues. The situation is that i want to run something 24/7 but in case of power failure the device will still work from LiPo power. I know there are many Lead Acid battery emergency light systems that work like this but how do LiPo's tolerate it? imk --- End quote --- DO NOT do this. You can’t even keep them on a tiny trickle, it damages the battery. Just use a Li-Ion charger IC that will do exactly what you want. In a project I’m just finishing up, the TI bq25895 charger handles this. It supplies the load with power from the AC adapter (which can be USB) when connected, from battery when it’s not. A microcontroller programs the charging parameters via I2C. --- End quote --- Yes, but there's still something to be careful about. (Got bitten by this once.) Even if you're using a proper Li-ion charger IC, setting too low a charging current, even if that looks right for the application in question, is often not a good idea either. It's probably partly dependent on the charger IC itself. But for very low charging currents (relative to the battery's capacity), the IC may have troubles properly detecting the end-of-charge condition. I ran into this using a Microchip charger IC (do not remember the exact ref, but I could dig it up.) Now some ICs do include an internal time-out, so the charge would stop after a given amount of time even if the IC doesn't detect an EOC condition. I don't quite remember, but I think the IC I used didn't have a time-out. The reason EOC can be hard to detect for low charging currents is simple. Most charger ICs roughly use the following algorithm: 1 - If Vcell < Vmin => source a predefined constant current < charging current (to "prepare" the cell); 2 - If Vcell < Vmax (usually 4.2V for LiPo) => source constant current = charging current; 3 - When Vcell reaches Vmax => switch to constant voltage mode @Vmax, and monitor current to the cell until it goes below (charging current)/10; 4 - Stop charging => charger output in Hi-Z. The condition in step 3 is one reason EOC detection can fail for very low charging currents. Since the threshold is usually (charging current)/10, it might be below the usable range for the internal current monitoring circuitry if charging current is set too low. Something to keep in mind... And, as tooki also said, I will never emphasize enough how careful one should be with Li-ion/LiPo batteries. Better safe than sorry. They forgive very little. |
| tooki:
I know the bq25895 has a built in timer, among other things. Given how accurate its measurements are, I’m fairly confident it could measure its low currents, too. (Since I’m using 3 cells in parallel, the precharge and cutoff current is not right at the edge anyway.) |
| Siwastaja:
I have worked with Samsung cells yes, especially the 29E as it was and probably still is cheaply available from secondary grey sources, as I'm a small-scale (prototyping scale, let's say, some thousands of cells have gone through) battery pack manufacturer. Quite frankly, the public datasheet isn't very usable, IMHO. This is typical to li-ion cells. Also typical to Korean or Japanese products is the language barrier from the engineers to the English datasheet. Numbers are fine, but text around the numbers can be pretty weird. I'm sure if you are a proper (multi-million, I'm not) customer and sign the NDAs, you will be able to get the "actual" datasheet, more data, and engineering assistance from Samsung. So it's not surprising they contradict themselves, and you need to apply critical thinking. In the end, "Continuous charging under appropriate voltage does not cause any loss of characteristics" is quite clear, and matches with what I have seen from the literature and own experience. Nothing happens because current doesn't flow. Li-ion is quite close to an ideal voltage source with a series resistor. I can see the point of adding a timer to increase safety, but let's be frank about it, it doesn't help a lot. If your charger design has a tendency of failing (for example, blowing pass regulator MOSFET short circuit), it can fail during that time, and unless the timer controls a secondary, redundant MOSFET, it won't help the slightest. The problem is, normal charging times can be long, especially CV phases when fully charging per manufacturer recommendations, so the failure event is likely to happen during charging well before the timer expires saving the day. It's like if a motorboat or jet ski had a 5-hour delay in the dead man's switch. Now one point in the safety cutout timer could be that if the cell is damaged so that it leaks charge (i.e., self-discharging well beyond normal amount, heating up as a result) then the timeout finally limits the amount of excess charge put into the cell by the charger. But this has two issues, (1) A dumb timeout will still allow quite a lot of charge to be put in, for example for a C/2 charger and 5-hour timeout this would be 250%. In order to actually limit the excess charge, you'd need a lot more complex system; i.e., initial SoC assumption based on voltage triggering charge integrator reset, then integrating current over time, and emergency stop if, for example, 110% is exceeded. (2) A timeout can do nothing if there are loads connected to the very same battery while charging, a common use case in cellphones and laptop computers. How are you going to differentiate between charge lost inside the cell, or used up by the intended load? So again we would need cell current sensing, accurate charge integration, capacity tracking, integrator reset logic and overall failure detection logic. Which can't give false positives because then no one uses your product. (edit: 3 of 2): If the cell leaks charge excessively anyway, the cell voltage would drop, and then the user would just restart the charging by unplugging and plugging in the battery, again delivering full timer worth of charge into it. For such imaginary faulty cell, this would be dangerous. So then we would need a battery ID and memory to flag the battery "do not charge". This is a lot of engineering to be done right and not make things worse. There is no easy off-the-shelf IC solution for all this. Manufacturers never seem to require such levels of BMS measures against protecting from cell damage so clearly they think their passive cell safety is enough to handle this and the BMS is required just to prevent externally induced over/undervoltage events (i.e., stopping charge and discharge based on voltage), and for a few other oddities, like preventing charging of overdischarged cell, and preventing charging at cold temperatures. And finally, temperature sensing of the cell is an easy way to deal with cell problems (the 18650 cells do have the integrated PTC for this very reason; an external extra sensor connected to your micro won't cost a much extra). All in all, for a typical float/holding charge application, do implement the following * Force turn-off if cell voltage < 3.2V (assuming low load currents and no significant peak currents. Adjust this down for larger peak loads like 2.8V at C/2) * Refuse turn-on if cell voltage < 2.5V (replacement of the cell is needed) * Refuse charging if cell temperature < 0 degC * Make sure your regulator circuit doesn't fail in any way generating excess voltage (pass element short circuit being most typical). It's recommended to add another pass element (two back-to-back MOSFETs typically) controlled by a small overvoltage/undervoltage protection IC. This way, protection is actually redundant and not dependent on some massive Benchmarq/TI monster chip no one understands. * Always remember the classic ceramic fuse, nothing beats it in fire protection in case all semiconductors fail short. * Thermal fuse coupled to the cell won't hurt if you want to do this seriously but usually designers won't bother. |
| magic:
--- Quote from: SiliconWizard on June 29, 2021, 08:25:03 pm ---The reason EOC can be hard to detect for low charging currents is simple. Most charger ICs roughly use the following algorithm: 1 - If Vcell < Vmin => source a predefined constant current < charging current (to "prepare" the cell); 2 - If Vcell < Vmax (usually 4.2V for LiPo) => source constant current = charging current; 3 - When Vcell reaches Vmax => switch to constant voltage mode @Vmax, and monitor current to the cell until it goes below (charging current)/10; 4 - Stop charging => charger output in Hi-Z. The condition in step 3 is one reason EOC detection can fail for very low charging currents. Since the threshold is usually (charging current)/10, it might be below the usable range for the internal current monitoring circuitry if charging current is set too low. Something to keep in mind... --- End quote --- Perhaps it could also be that self discharge at 4.2V exceeds the CC/10 threshold if CC is set too low. |
| Navigation |
| Message Index |
| Next page |
| Previous page |