X1203 UPS Voltage Drops | Geekworm

X1203 UPS Voltage Drops

r

razmichael
9/26/24, 12:24 AM

X1203 with two x 2 battery packs on Pi5. I'm seeing voltage drops to the Pi under various loads that will drop to 4.9 or so and often trigger the Pi Low Voltage threshold warning. This will occur both with and without the battery packs connected. Without the UPS connected the Pi reported input voltage is very steady (5.12 to 5.14) despite running various stress tests. There are no heavy loads attached to the Pi (a self powered external HDD and a system NVME board) and I'm using a authorized pi5 power supply (that works fine directly into the Pi). I'm using vcgencmd pmic_read_adc EXT5V_V to measure the input voltage seen by the Pi. As soon as I start a stress test with the UPS, voltage drops to 5 or less - does not happen without the UPS

Hopeing for a solution or cause as I do not like getting low voltage warnings!

1 likes 0 likes

8
Posts

101
Views
DW

Douglas Wyman
10/7/24, 8:08 AM

Same problem. 12.2 volts to the UPS and same low voltage warning Pi5

0 likes 0 likes
r

razmichael
10/7/24, 9:40 AM

I switched out the approved power supply for an 18v 6a to try this and I still get some (although I don't think as many) low voltage warnings. I've sent a direct support message to GeekWorm and also contacted SupTronics but no one has chosen to respond. Not really very happy with this product or support at this stage! I realize the cause may be related to the way the power is being fed 9through the pins) and not actually a low voltage but i have no way to know this one way or the other.

1 likes 0 likes
DW

Douglas Wyman
10/7/24, 10:37 AM

I'll have to check the pogo pins on mine. A very possible cause.

0 likes 0 likes
r

razmichael
10/8/24, 12:37 AM

I was thinking more a way the Pi measures the voltage source and, since it is not coming through the USBC connector, the measurement is not valid - but this is just wishfull thinking I admit. Seriously thinking of trying a different brand although I like the "remote" battery pack as it makes it easier for other boards to be added.

0 likes 0 likes
HH

Harry Huang Staff member
10/10/24, 6:12 PM

Sorry for the late reply.

Today, we tested X1203 and the results are as follows:

1: This is a full picture of the test.

Please excuse my messy desktop.

2. There is no power supply to charge the X1203, and the output voltage is 5.17V.

Using the Geekworm 2-cell 18650 battery holder.

3. We ran stress-ng --cpu 4 to simulate a high load.

You can download this image to see it more clearly.

I also did not configure the option PSU_MAX_CURRENT=5000.

From above picture, it can be seen that there is no low voltage warning.

4. We use PD27W power supply to charge X1203 through Type-C on x1203. Please pay attention to the TYPE-C port.

5. We use Geekworm DC12V@5A power supply to charge X1203, please pay attention to the DC Jack.

PS: Throughout the test, we ran stress-ng --cpu 4 to maintain a high load.

I am not sure how you use it, I suggest you provide detailed pictures to describe it. In addition, the 18650 battery I use is also a very common battery, not a big brand battery.

Thanks.

0 likes 0 likes
r

razmichael
10/13/24, 9:04 PM

Thanks for the response. I want to clarify what I saw (and still see although less often).

When I first posted my Pi was on my desktop as I was reconfiguring it so it was only conected to a monitor with the 1203 and an NVME base (no external devices). The Pi was powered by an approved 27w PS through the usbc port of the 1203. When I report low voltage this is based on the desktop 'lightning bolt' and a check pf vcgencmd get_throttled. This occured fairly often, not just when running a stress test.

When I removed the 1203 I was not able to reproduce the issue but it returned when I put it back into the system - this occured with or without the batteries connected.

Since posting, my Pi is now back controlling a few relays but all external devices are self powered so the power demand should be low. I am providing power to the 1203 with a 18v 6amp power supply and see the low voltage warning much less often although this morning it have it and a check on the throttled state shows:

Status: 0x50000
Undervolted:
Now: NO
Run: YES
Throttled:
Now: NO
Run: YES

Running a script to monitor the voltage every second shows some drops to 4.9 or so (still within range) but this is unlikely to capture instant drops below the low voltage level (4.62 I think??). I also believe (could be wrong) that the Pi monitors GPIO 2 for the voltage rather than the input.

I do not want to pull it all apart to do more testing so I have a question - while the 1203 is attached, can I bypass it by removing the batteries and power and powering the Pi directly? This would allow for more testing where I could compare using the normal 27w PS through the 1203 or the Pi.

The photo of the current setup is NOT similar to the original testing described above!!

0 likes 0 likes
r

razmichael
10/17/24, 2:01 AM
Follow up:

I wrote a script to check voltage every second and then log every minute the results for that cycle including minimum voltage, max voltage and average voltage over the 60 seconds along with the Under Voltage/Throttled status directly from the system. Along with that, a separate log entry would be added every time the voltage was below 4.9v and, if the system detected an under voltage situation, this would also be logged (I believe the Pi considers 4.63v as under voltage but not sure on the exact value). Every third 60s cycle a stress test would run at the same time.

I ran this test 3 times, each about 8 hours (and while the Pi was continuing with it's normal operations, with the following hardware/power setup:
1. Using the 1203 powered through the jack with an 18v 6a power supply (with 4 batteries, fully charged);
2. Using the 1203 powered with the standard 27w Raspberry Pi5 USB C power supply(with 4 batteries, fully charged; and
3. Without the 1203 and the 27w Raspberry Pi5 adapter directly into the Pi.
Test one and 2 had similar results, although 1 included a system Under Voltage alarm while test 3 (without the UPS) showed no drops below 4.9v registered (keeping in mind this is a spot check once per second) and what I would propose is a more steady power supply while the 1203 is definitely pulling power at times causing more drops and potentially resulting in system under voltage. Summary below (logs availabe if wanted). At this stage I’m going to just continue to use the 1203 and monitor for any more under voltage alarms although it does look like it is using more power at times than I would expect.

Test 1:
Average Minimum over each 60 cycles: 5.03
Average Maximum over each 60 cycles: 5.16
Average Average over each 60 cycles: 5.11
Number entries for under 4.9: 15
System Under Voltage Alarm: 1

Test 2:
Average Minimum over each 60 cycles: 5.03
Average Maximum over each 60 cycles: 5.16
Average Average over each 60 cycles: 5.11
Number entries for under 4.9: 9
System Under Voltage Alarm: 0

Test 3:
Average Minimum over each 60 cycles: 5.12
Average Maximum over each 60 cycles: 5.15
Average Average over each 60 cycles: 5.13
Number entries for under 4.9: 0
System Under Voltage Alarm: 0
0 likes 0 likes