(11-13-2019, 09:55 PM)VoxUnius Wrote: (11-13-2019, 09:09 PM)MrTester Wrote: What exactly do we need tracked?
I am happy to throw the PBP on a DAQ at work with AC recording abilities, or open up and probe other values over a long term test.
Please let me know, I am happy to help collect data.
Well, it's been proven that the PBP needs more than 3A at high load (in my case, it was compiling a kernel with -j6). The question is whether a more powerful DC adapter would work with it. Is there any limitation in the charging circuitry? Can it take more than 3A?
As I said, I should be able to answer this question next Monday.
Hi,
Any update mate? Does the 5V 4A barrel? adapter work?..
If the laptop is taking 3.5A on full load with -j6, I think it might be better to test it with -j2 to limit the current to 3A.
(11-21-2019, 12:13 AM)aaditya Wrote: (11-13-2019, 09:55 PM)VoxUnius Wrote: (11-13-2019, 09:09 PM)MrTester Wrote: What exactly do we need tracked?
I am happy to throw the PBP on a DAQ at work with AC recording abilities, or open up and probe other values over a long term test.
Please let me know, I am happy to help collect data.
Well, it's been proven that the PBP needs more than 3A at high load (in my case, it was compiling a kernel with -j6). The question is whether a more powerful DC adapter would work with it. Is there any limitation in the charging circuitry? Can it take more than 3A?
As I said, I should be able to answer this question next Monday.
Hi,
Any update mate? Does the 5V 4A barrel? adapter work?..
If the laptop is taking 3.5A on full load with -j6, I think it might be better to test it with -j2 to limit the current to 3A.
Hi,
I haven't had time to test it properly. The only thing I noticed was that it still discharged the battery with -j6.
The combination that worked for me with the 3A adapter was limiting cores 5-6 to 1800MHz and putting the rest of them to powersave. That way it didn't drain the battery. And using -j2, yes.
I doubt using anything more than a 3A power supply will increase charge times (unless your 3A power supply isn't really giving you 3A), as the lithium charger IC is limited to 3A input current to protect the power supply. The charge management IC is capable of a maximum of 4A charge current, although I'm not quite sure what charge rate it is set to - I'd suspect a just a little below 3A since the calculation looks basically the same, and the configuration resistor values are only marginally different.
(11-13-2019, 09:55 PM)VoxUnius Wrote: Well, it's been proven that the PBP needs more than 3A at high load (in my case, it was compiling a kernel with -j6).
There my be an issue with your unit or power supply.
I ran a stress test on mine using kernel compilation -j1 thru -j6 WITH an external USB 3.0 hard drive attached. The standard DC power supply provided with the unit was used. I am running the factory installed Debian desktop on eMMC. I measured voltage_now throughout each test.
The results of my test did not confirm what you are seeing with your unit. I did not observe the battery discharge issue you described.
You may want to open a support ticket.
11-21-2019, 06:43 AM
(This post was last modified: 11-21-2019, 06:45 AM by VoxUnius.)
(11-21-2019, 05:58 AM)hmuller Wrote: (11-13-2019, 09:55 PM)VoxUnius Wrote: Well, it's been proven that the PBP needs more than 3A at high load (in my case, it was compiling a kernel with -j6).
There my be an issue with your unit or power supply.
I ran a stress test on mine using kernel compilation -j1 thru -j6 WITH an external USB 3.0 hard drive attached. The standard DC power supply provided with the unit was used. I am running the factory installed Debian desktop on eMMC. I measured voltage_now throughout each test.
The results of my test did not confirm what you are seeing with your unit. I did not observe the battery discharge issue you described.
You may want to open a support ticket.
Both power supplies? I had this problem with both 3A and 4A units. Although, I haven't had a chance to test the 4A unit thoroughly. I'm stuck with my PBP being unable to boot. Still struggling to figure out why. It doesn't work with Manjaro kernel anymore, for whatever reason.
Since you said you didn't observe this problem with Debian, it could probably be the mainline kernel issue. I haven't tested it in Debian, honestly. I should be able to provide more info once I've made it bootable again.
(11-21-2019, 05:58 AM)hmuller Wrote: I measured voltage_now throughout each test.
Wouldn't rk-bat attribute "charge" be more appropriate?
(11-21-2019, 10:22 AM)Der Geist der Maschine Wrote: Wouldn't rk-bat attribute "charge" be more appropriate?
That attribute (charge) is not present in /sys/class/power_supply/rk-bat/ on the factory installed Debian desktop on my unit. Even so, I would have to read the source code to understand the intent of an attribute named "charge" if it were present.
voltage_now seemed self-explanatory, and that attribute changes over time so it was a likely candidate for monitoring. If the battery does discharge while attached to DC power supply then it is reasonable to expect to see the voltage drop on voltage_now. This was not observed under test conditions I previously mentioned.
I'm not near my laptop and can't pull the plug to confirm what values change on a normal discharge.
I would expect the voltage to be constant perhaps with some small swings. Perhaps you observed small swings?
What I wrote is not what I meant ;-/ I meant "capacity". "current_now" might be also worth monitoring.
11-21-2019, 09:42 PM
(This post was last modified: 11-21-2019, 09:44 PM by pfeerick.)
(11-21-2019, 03:07 PM)Der Geist der Maschine Wrote: What I wrote is not what I meant ;-/ I meant "capacity". "current_now" might be also worth monitoring.
Any of those three would give you some idea of what's happening... `voltage_now` constantly dropping would tell you either the unit isn't plugged into external power, or the power supply can't keep up with the load. `current-now` would be the best of the three to monitor, as it will respond a lot faster than `capacity`, and will go positive/negative depending on if the unit is charging or discharging.
Depending on what how the paths have changed (if any), this script I used for the original pinebook with A64 processor might be handy to use now... logs to screen and file for later analysis. 
https://github.com/pfeerick/pine64-scrip...yLogger.sh
11-21-2019, 10:00 PM
(This post was last modified: 11-22-2019, 12:54 AM by Der Geist der Maschine.)
(11-21-2019, 09:42 PM)pfeerick Wrote: (11-21-2019, 03:07 PM)Der Geist der Maschine Wrote: What I wrote is not what I meant ;-/ I meant "capacity". "current_now" might be also worth monitoring.
Any of those three would give you some idea of what's happening... `voltage_now` constantly dropping would tell you either the unit isn't plugged into external power, or the power supply can't keep up with the load. `current-now` would be the best of the three to monitor, as it will respond a lot faster than `capacity`, and will go positive/negative depending on if the unit is charging or discharging.
Depending on what how the paths have changed (if any), this script I used for the original pinebook with A64 processor might be handy to use now... logs to screen and file for later analysis.
https://github.com/pfeerick/pine64-scrip...yLogger.sh
Earlier this day, I measured all three with a 1 liner 
make -j5.
Copy/paste removed the tabs.
time capacity current voltage
18:58:13 100 0 4357000
18:58:14 100 0 4357000
18:58:15 100 0 4357000
18:58:16 100 0 4357000
...
19:31:55 100 0 4203000
19:31:56 100 0 4203000
19:31:57 100 0 4203000
19:31:58 91 0 3971000
19:31:59 91 0 3971000
19:32:00 91 0 3971000
19:32:01 91 0 3971000
19:32:02 91 0 4202000
Capacity significantly went down within a second. I assume during that second, there was a spike in current as well.
|
