I just had an odd power problem, resolved now, so FYI.
I used my Reform battery powered (charger unplugged) then forgot to plug it in; when I noticed battery was at 7%. I plugged the supplied charger 2.1mm into Reform; the battery continued to drain, 6%, then 5%. The blue LED on the charger was pulsing bright ~once/sec.
I unplugged the charger again, did shutdown -h now (see below…), OLED was on, so I did circle-0 to power off.
I waited a few minutes, the blue LED is on steady, plugged charger 2.1mm in, circle-B, now it’s charging. The blue LED is on steady. It’s now at 25% and rising.
To hazard a guess, Reform loaded the charger too much, which folded back. Shutdown reduced current draw, battery came up and all’s well.
Upon reboot, fsck fixed a lot of errors in one of the drives; I missed which drive and which error… I have an external SSD plugged in for rotating backups. Drive is OK now.
Before I issued shutdown I ritually ran sync a couple times (what can I say; I’m old) so the fsck surprised me.,
If you have motherboard 2.5, then it can happen that the reform tries to draw more power than the supplied stock AC adapter is able to provide. This is more likely to happen the more empty your battery cell are because the cells are charged faster the more empty they are (and slower the fuller they are). There is a patch for the LPC firmware which works around this issue.
@minute did you add your patch to the original firmware? I am unable to find it in the git log.
This is good to know. I’ve had this happen to me a bunch with the LS1028A firmware, where the notebook would not start charging if I plugged it in while it was running. But when it’s shut off it works reliably.
I haven’t had the time to investigate this as of yet, but this explanation sure seems to make sense.
These are 18650 cells, variously 1200 to 2000 mAh… the recommended maximum charge rate is C/10, so max would be 120 mA to 200 mA. Faster than that shortens cell life… if all eight are in series then over 2.5A draw implies that those batteries are getting cooked. Have I got this wrong?
Can the firmware set a configurable max charging current?
OK I loaded Reform2 in kicad and had a look at the schematics. The LTC4020 has 0.01 ohm sense resistors, which should be 0.2 amps max current. NEVER MIND my foolishness.
I have looked into this a bit more. Not with any specific goal in mind, just for fun.
I have a batter bank that shows input and output. It’s probably not terribly accurate, but still fun to watch.
The battery bank has DC input / output, allowing me to use the MeanWell power adapter shipped with the MNT Reform to charge the battery. According to that, the power adapter, although labelled to max power out of 60W, is able to deliver up to 68W.
When attaching the laptop, I don’t use the DC output since I don’t have an matching cable. I use a Type-C to barrel adapter instead.
I have discharged the laptop battery down to 25% and then started running stress-ng to create load and further increase power consumption of the device. I am doing this on the LS1028A, since that one is more power hungry than the stock CPU module.
In this state, when I plug in the MeanWell power adapter, the laptop does not charge. Even if I stop stress-ng it still won’t.
When plugging in my battery bank, the power draw spikes up to 93W for a short moment. It then settles down at just below 80W. So still a good bit more than the MeanWell can deliver, but should be easily handled by a 100W USB-C power brick.
I did not take a photo of it but the IRC log also shows me having observed a power draw of 88 W which is close to what you saw.
So the delayed charging patch is nice but it does not fix the situation where the overall power draw is far above what the meanwell adapter is specified to deliver. I have successfully used this 100 W USB-C adapter without any problems since January: https://www.reichelt.de/de/en/-p344620.html
Buying an additional power supply is not a solution. Is this fixable via firmware? There haven’t been any updates to the LPC firmware or to the mainboard CAD files as far as I can see, which really bums me out as I have an MNT reform on order. It’s especially concerning to see that any other product (MNT next, MNT desktop module) appears to have priority over standing issues.
Wouldn’t a 0.01 ohm sense resistor between CSN and CSP be a 5A max charge current instead? I still have the older schematic of my earlier 2 motherboard which has a 0.02ohm resistor across that or 2.5A theoretical max. I’m sure there’s other limiting factors there too as when I plug in my reform the current display it shows it’s charging at 1.47A. Perhaps a 0.03 ohm resistor might be more suited to limit at 1.6, though at least in mine I’m only seeing it charge around 0.8C anyways
I’m not sure where you got the C/10 figure from though, LiFePO4 cells typically have a max charging current of 1C (and the datasheet for the JGNE 1800mAh cells confirms this as 1.8A). And even if they were plain Li-ion most of those have a max of 0.5C recommended. Though C/10 is a normal stopping point for charging though when charging at 1C (The 4020 data sheet is specifying C to be the charging current when mentioning this stopping point, ie. 5mV across the resistor).
I’ve personally not had any issues with the power supply itself at least on my 2.0 board with the provided 24v 2.5A meanwell supply, you should be able to lower the max charge current by changing the current sense resistor to a slightly larger value like 0.03ohms (1.66A) or 0.04ohms (1.25A) (I = 50mV / R).
Alternatively a ~10k resistor could be wired across pins 2 and 3 of Q4 which would raise the voltage on RND/SS to 0.5v when the mosfet is not on (normal charge mode), and thus would limit charge current to 50% of max.
I recently tried to use an old power supply I still had lying around, which has a maximum output power of 78W at 13V, and it indeed does not show the charging issue. So, I guess the overall power draw is just slightly above the 60W that the stock power supply can deliver.
However, I am reluctant to use that 13V power supply, as the Reform gets quite warm while charging with it. Probably because there is more load on the voltage regulator circuitry.
I have great respect for your work, but I find it rather irritating to sugar coat the issue and not be honest about it. If this is a hardware issue that requires me to get out my soldering iron on day one to fix a 1400€ device, I’d consider cancelling my order.
Yes, for charging via USB-C. I thought you were talking about the stock adapter?
What am I supposed to check out in this post?
I indeed missed that post. I have added an addendum to my last post. As far as discussions on IRC are concerned I was under the impression that the patch fixed the issue for good. I think that is also what @minute said recently on IRC.
Please do not accuse me of dishonesty. I simply did not read the post by @soulsource or I did read it but forgot about it. I did not intentionally lie or try to sugar coat. I was one of the first people who received motherboard 2.5 and please believe me, it was very frustrating to run into the charging issue and try to figure out what was even going on. This is not something I’d try to sugar coat nor am I suggesting that you buy another PSU or hand-solder your motherboard to fix it.
You are probably taking about a “barrel jack to USB 3 for power delivery” cable, but why does it matter that type of connection sits on the other side? How would the computer differentiate between a regular power supply and your USB wall wart and draw different amounts of current? It carries no data at the end.
Yes, let me expand on that. The MeanWell GST60A24-P1J that comes with the Reform is rated for 60 W. Accordingly, when I started using USB-C charging (with a cheap barrel jack adapter) so that I can share a power supply with my partner, I had bought a USB-C adapter that was rated for 60 W. This worked fine with motherboard 2.0. This setup worked without any issues until I received motherboard 2.5. With that setup, my 60 W USB-C adapter was not able to charge my reform when it was somewhere below 80% battery level. I was the only one with these issues, so I thought maybe my USB-C adapter has a problem. At that point I did not own any USB-C measurement hardware (I only bought that later, you can see it in that video). I rarely have my Reform non-stationary, so my batteries rarely go below 50%. When they did, I used the MeanWell adapter to get them back up to 100% and at that point I was able to use the 60 W USB-C adapter. Using the MeanWell adapter was also not without problems. Sometimes it took some un-plugging and re-plugging until the MeanWell adapter worked but it did work. In the end though, I became very annoyed with the situation so I bought another USB-C adapter (100 W just to be sure) and a USB-C power meter that you can see in my video. With that setup I measured 75+ W power draw that I meantioned in my earlier post. This makes it understandable that my 60 W USB-C adapter was not sufficient to deliver. It was around that time that @minute got aware of the issue and provided the charge-delay-patch. How does the MeanWell power supply still work even though more than what it is rated for is drawn from it? I have no idea. You have to ask somebody who knows electronic stuff. Maybe the components can do more but it’s only safe at 60 W? Be it as it may, it turns out that the MeanWell is able to deliver that power if it is not drawn all at once but if it first powers the system and only after a short delay draws some additional power to charge the batteries.