The cell becomes damaged if it’s allowed to drop too low. Combine that with the fact that Framework will send a new one for free & there isn’t any point in messing around with it, just replace.
Since this is now in a coolermaster case, I’m installing a much larger ML2020 with solder legs. There should be room to install at least 2 of them actually to provide about 90mah. Same voltage, same battery chemistry, and there is a spot right next to the holder that should hold 2 of these stacked, and dead simple wiring right over to the solder legs on the holder, without even buggering up the holder. Could still install a normal cell, although I think I won’t mix different size cells connected at the same time, either just the normal cell, or one or more of the bigger cells.
I already ordered the ones from amazon before finding out that FW would supposedly send a free replacement so no sense even asking now. I don’t need this type for anything but this one 11th gen board. And now I’m probably not even going to use those since I’m using one or two 2020’s instead now.
There is room for even more, like at least 4, but the board gets a little warm a little further to the right, (20mm further away from the holder), and I don’t want to put cells right over that spot. But even just a single 2020 takes the capacity from 17mah up to 45 mah, so that’s already plenty. And the way the case is so tall, I think there’s room to put 2 2020’s on top of each other, so now 90mah.
There are a few different chemistries that are all rechargeable coin cells with lithium, and even the same nominal output voltage, but they all take different charge voltage, so it’s critical to only use ML series not VL series which look the same, or anything else.
1 Like
Whether the cell being damaged or not and whether the mainboard failed to recognize a discharged coin cell are not the same thing.
The point is to figure out which one is the possible cause of the cell “not taking a charge”
Sorry for nitpicking, if that annoys you or other users I’ll shut up…
I think that messing with old worn down devices is a legitimate hobby.
The board is putting 3.17v into the cell.
That’s probably a damaged cell, wait for replacement
BTW You haven’t said anything that makes me want to say shut up. I think it’s equally valid to say this isn’t worth wasting time on, or to say it’s as perfectly interesting and valid to spend time on as any other interest.
You never figure out anything if you don’t just poke at something for a while some times. Every minute can’t be either sleeping or directly productive billable hours or taking a selfie at one of the wonders of the world.
I don’t even have a use for this stupid thing. By the numbers it made no sense to buy the coolermaster case (or even print one).
I upgraded the motherboard and didn’t have to replace anything else at the time. That’s good. But now so far I have spent:
$40 coolermaster case
$80 1T TN470 nvme
$40 2 x 8G ddr4-3200
$25 intel AX200 wifi-6/bt-5 card WITH 2 antennas (and 2 desktop/server pcie brackets)
$6 2 x ML2020
Plus tax & shipping
…on the left-over motherboard, and no real use for it.
And that’s not counting
- $20 the ML1220’s from amazon which got obsoleted by the ML2020’s
- 45 minute drive each way drive out to MicroCenter (hd, ram, and wifi were all right off the shelf there. But I have a convertible and the weather was nice and I like audiobooks, so eh
- $120 chargeasap 100W power supply
- $10 100w usbc power cable
- $250 ASUS MB16AC screen (back in 2018 when I got it)
- $100 HP LifeStyle 1000 wireless keyboard
- $75 Microsoft Arc Mouse
- $9 usba module for the keyboard receiver (everything else is rawdogged right into the motherboard ports)
I already had these for a long time, but, I still only have them because I bought them, just, previously, and now they are occupied for this instead of for something else)
It’s ultimately just a much less gainly laptop, or much less useful desktop. I already have fully working whole laptops which are better for most situations where I could use a 2nd computer for some reason like if I need to use Windows or FreeBSD and not in a vm and don’t want to boot my main machine into it. Most likely is maybe I’ll mount it on the back of a tv at one of my work benches to use for looking up documentation or running eprom or fpga programming software etc.
The most sensible thing to do was probably to just try to get $50 for it on ebay or something.
But not the most interesting.
As far as I’m concerned, pick this battery issue all the way apart.
sorry I was replying MJ1.
Totally agree
I don’t think you want to be doing that as you will double the voltage. This means they probably will not charge as the battery voltage is higher than the charge voltage, and also the RTC/CMOS backup chip will be receiving too much voltage and probably fry.
That would depend on how the batteries are connected to each other. If they’re connected in parallel, the total amperage will increase, but not the voltage. Charging might require more amps at that point, though, which could be a problem.
Yeah, but I wouldn’t want to directly connect them in parallel as they have a nasty habit of unequally sharing, and the talk was of ‘stacking’ them on top of each other, suggesting the two cells would be in series, thereby doubling the voltage.
Nothing about the word stacking implies connecting in series. Or in parrallel. These are individually insulated cells with solder legs welded on. Not stacked bare cells like in a flashlight. Multi-cell packs of all chemistries are routinely spot-welded in whatever combination of parallel & serial the application requires. 2, 3, or 4 cells directly hard-wired in parrallel, or series if you wanted the voltage to add up, are perfectly fine. You just can’t mix chemistries or capacities or ages. Individual cell management is only needed in high current or especially “dynamic” chemistries like lipo. These only require a particular voltage and current for charging, a pretty narrow range but otherwise no especially exotic management. The motherboard is producing the correct voltage, and the current will be less than the battery could tolerate, making the full charge time take longer and making it merely easier on the cells.
One possible problem is the parallel cells could possibly try to draw more current from the source, if the source were not current-controlled, but it is so there is no chance of drawing any more current even if there were a dead short wire in place of a battery. There is no danger of hurting the charging circuit, but there is a possibility that the charging IC senses the less-resistive load and treats it as some kind of short in the battery and decides it better stop providing power. So maybe it fails to charge simply because the charging circuit says “something funny here, I’m going to be safe and shut down”. In that case I just remove the extra cell, or add a resistor.
Yes, in this case, obviously the connection is parallel. Also the + goes to the + and the - goes to the -, and all the bits are insulated with kapton or other insulator instead of allowed to short all over the place.
Yes, at some point one of the cells will be the first to start behaving a little different from the rest enough to make the whole pack no longer viable. That day is no sooner than with a single cell.
In other words, this is not an issue.
Yes, that’s true, I didn’t consider that. Connecting them in parallel would require a more complex setup than stacking them in series.
No it’s not complex. It’s not like the main battery that needs each individual pouch individually monitored, nor like a car or house or even a power tool with high current and heat etc. You just solder the pins together.
Quite the opposite, for most battery chemistries parallel is significantly easier than series
Wandered off the original topic about the main battery, but just to wrap up the talk about the cmos battery, I did end up getting 2 ML2020’s from digikey and soldered them right to the terminals on the battery holder with wire-wrap wire.
It went fine, easy to do. Seems to be working fine and I see no reason to worry electrically.
So the end effect will be that these cells are getting charged a slower rate than a single one would have, and less current doesn’t hurt anything. (the battery manager ic or circuit will be limiting current since that’s 90% of the definition of it’s job, so a single larger cell will take longer to charge to full than a single smaller cell, and 2 larger cells will take even longer, but it doesn’t change how much charge you get in say an hour. Say on the original single small cell you get a week of standby time from 2 hours of charging time (I don’t know the actual number, just use this as example), the much larger cell will take longer to get full, but you will still get the same standby hours per charge hour, just you will get more total of both. So the board should be able to sleep with no power a lot longer.
I got cells with solder legs and soldered them together in parallel, seperated by 2 layers of kapton tape and bundled together by kapton tape.
In the CoolerMaster case (not the laptop!) there is a ton of space to the right of the ML1220 holder. You could actually fit at least 4 cells, maybe even 6 in 2 stacks of 2 or 3, but the aera further to the right is a bit shallower and also that area gets warm, while the space directly next to the normal holder has more room and doesn’t get warm, so it’s a better place to put cells, and even a single ML2020 is a big increase over a ML1220.
If you do this, remember, the chemistry is critical, you must use MLxxxx not anything else.
The exact cells I used are: https://www.digikey.com/en/products/detail/panasonic-bsg/ML-2020-V1AN/431504
1 Like
@Brian_White
Just out of curiosity, why do you need a larger RTC battery?
How long do you need the RTC clock to last while the laptop is switched off?
The original battery is undersized. Any motherboard should be able to hold cmos settings for 5 to 10 years without power.
The original Framework dies in as little as a week.
Even with power, the original battery dies and stops taking working at all after just 2-3 years, as we have all now started to discover the hard way.
Mine died and would lose it’s settings instantly when turned off, and no amount of charging time worked. The cell no longer took a charge. This is especially annoying when the motherboard is not in a laptop any more and no longer hooked up to the default video and keyboard. I need it to retain it’s settings.
There is no specific time that it must last without power, but a week or even a month is too little. There is no excuse for it not to last at least 5 years. That is not some entitled demand, that is the norm since the 80’s.
The problem is a combination of 3 things: using a rechargeable battery instead of non-rechargeable, using such a small one, and the motherbaord draws too much current while asleep.
We can’t do anything about the motherboard drawing too much current, and putting in the right kind of battery (a CR2032, which in this case would also require adding a diode to prevent charging) will still drain too fast because of the motherboard as well as the voltage drop across the diode.
The easiest thing we can do is just keep buying new ML1220’s every couple of years and keep the machine plugged in at all times so it never goes a month without power.
The next easiest thing is just install a larger capacity of the same type of battery that the charging circuit is designed for. (which is what I just did.)
https://community.frame.work/t/laptop-wont-power-on-unless-i-plug-in-ac-power/
https://community.frame.work/t/viability-of-an-ml-1220-rechargable-battery-for-rtc-cmos-11th-gen/
Framework themselves proposed this fix which I think is a bit ridiculous as a solution, but shows that they recognize it’s a problem, something they did wrong and need to come up with some kind of fix for:
https://community.frame.work/t/rework-instructions-for-11th-gen-mainboards-to-enable-powering-the-rtc-circuit-from-the-main-battery/
I see now. The RTC circuit is just badly/wrongly designed.
I have seen other computers keep their RTC settings for 4 hours without a battery at all, and with the ML1220 battery it lasts months(no one has measured the time).
As a comparison, ARM pc’s work just fine without using RTC or storing anything in CMOS. It’s just X86 PCs and their overly complex BIOS that need it. All the settings in X86 BIOS could have been stored as config in the OS, skipping the need for any bios screens at all.
1 Like
The bios settings control things that can’t be managed by the os, they are hardware settings that need to be in effect when there is no OS.
You can’t store hardware settings in the os when the settings determine how to even find the os, what video device to use, whether to insist on a password, etc.
Some platforms do this with flash or eeprom instead of battery-backed sram, but a pc needs to maintain a running clock anyway, and it’s handy to have a guaranteed way to erase all settings to factory default to clear any error by just removing power rather than requiring some more complicated and error prone active process to rewrite eeprom or flash.
2 Likes
Did the battery spend most of its time at 100% full, or did you use BIOS settings to keep it at, for example, 70%, or was it regularly cycled fairly deeply by often using the laptop for a significant time without AC power?