none1337 Posted June 8, 2018 Posted June 8, 2018 Good afternoon, I am seeking advice on cooling my Orange Pi R1, it seems to overheat after about 6 hours of idel even with a heatsink on it. Currently i have a desk fan moving air around it and it holding around 31c with ambient room temperature around 25c. The device is just running a network bridge, pihole, and vnstat (hourly). CPU is on average 1-3% and memory use is 53%. Touching the device it is hard as it is physically hot. The USB hat is not installed. Any ideas?
guidol Posted June 9, 2018 Posted June 9, 2018 which armbian version do you use? sometimes a newer mainline version is running cooler than mainline. But as I recall from my R1 it was running quite cool at idle. its at 49 degree celsius I have installed the "big" expansion board with adapters and did power via power jack barrel. I did put it frim the shelf and updated totday to ARMBIAN 5.46.180607 nightly Debian GNU/Linux 9 (stretch) 4.14.48-sunxi I also have a litte heatsink, but no fan. here the first number of today: Time CPU load %cpu %sys %usr %nice %io %irq CPU C.St. 08:58:00: 1200MHz 0.04 1% 0% 0% 0% 0% 0% 47.2°C 0/8 08:58:05: 1200MHz 0.04 2% 1% 0% 0% 0% 0% 48.3°C 0/8 08:58:10: 1200MHz 0.03 0% 0% 0% 0% 0% 0% 48.3°C 0/8 08:58:15: 1200MHz 0.03 1% 0% 0% 0% 0% 0% 47.8°C 0/8 08:58:20: 1200MHz 0.03 0% 0% 0% 0% 0% 0% 46.9°C 0/8 08:58:25: 1200MHz 0.03 1% 0% 0% 0% 0% 0% 46.8°C 0/8 08:58:30: 1200MHz 0.02 1% 1% 0% 0% 0% 0% 48.3°C 0/8 08:58:35: 1200MHz 0.02 0% 0% 0% 0% 0% 0% 47.2°C 0/8 [EDIT] 8 hours later: Time CPU load %cpu %sys %usr %nice %io %irq CPU C.St. 14:45:18: 1200MHz 0.15 1% 1% 0% 0% 0% 0% 48.5°C 0/8 14:45:23: 1200MHz 0.14 0% 0% 0% 0% 0% 0% 46.1°C 0/8 14:45:28: 1200MHz 0.12 1% 0% 0% 0% 0% 0% 48.2°C 0/8 14:45:33: 1200MHz 0.11 0% 0% 0% 0% 0% 0% 47.2°C 0/8 14:45:38: 1200MHz 0.18 1% 0% 0% 0% 0% 0% 47.8°C 0/8 14:45:44: 1200MHz 0.17 1% 0% 0% 0% 0% 0% 47.4°C 0/8 14:45:49: 1200MHz 0.16 0% 0% 0% 0% 0% 0% 47.6°C 0/8 with ambient room temperature around 28c.
none1337 Posted June 11, 2018 Author Posted June 11, 2018 Priolr to starting the unit I cheked the OPR1 and the temp was 27c. Plugging it into the USB on my keyboard for power it seems to consume: 4.95-5.00 V (stable at 4.68V) 0.25-0.60 A This was tested using my USB Tester. After 8 mins of idle with nothing plugged in I will reach 51c. All temps checked at heatsink with a IR temp checker. Here in the on board: 11:23:54: 1200MHz 0.88 25% 11% 11% 0% 2% 0% 48.4°C 0/8 11:23:59: 1200MHz 1.05 32% 11% 19% 0% 0% 1% 50.6°C 0/8 11:24:04: 1200MHz 1.12 33% 2% 16% 0% 13% 0% 48.8°C 0/8 11:24:11: 1200MHz 1.11 41% 3% 12% 0% 25% 0% 49.7°C 0/8 11:24:16: 1200MHz 1.71 51% 7% 9% 0% 34% 0% 50.1°C 0/8 11:24:21: 1200MHz 1.65 26% 2% 23% 0% 0% 0% 50.1°C 0/8 11:24:26: 1200MHz 1.60 25% 0% 24% 0% 0% 0% 49.7°C 0/8 11:24:31: 1200MHz 1.55 25% 1% 24% 0% 0% 0% 50.9°C 0/8 11:24:36: 1200MHz 1.51 25% 0% 10% 14% 0% 0% 51.2°C 0/8 11:24:41: 1200MHz 1.47 25% 0% 0% 24% 0% 0% 51.4°C 0/8 11:24:46: 1200MHz 1.43 26% 1% 0% 25% 0% 0% 52.0°C 0/8 11:24:52: 1200MHz 1.39 25% 0% 0% 24% 0% 0% 51.5°C 0/8 11:24:57: 1200MHz 1.36 25% 0% 0% 25% 0% 0% 51.4°C 0/8 11:25:02: 1200MHz 1.49 30% 2% 1% 25% 1% 0% 51.5°C 0/8 I have placed a basic sticky on the bottom fins on the top heatsink on it from the Raspberry Pi but other than that there is no cooling air flow wise or passively. Any suggestions? The board was from Alibaba I reached out to them for support but they best they could say is use a 32GB sd card not 128GB. The best uptime I was able to hold was around 6 hours with a fan blowing right on the unit. Edit: here is the OS version Welcome to ARMBIAN 5.38 stable Ubuntu 16.04.3 LTS 4.14.14-sunxi Looks like I am a few releases behind. Any chance I can run a in place upgrade that would resolve the issue? My other thoughts are to tune down the CPU but am unsure on this.
guidol Posted June 11, 2018 Posted June 11, 2018 Today I installed armbian dev with Kernel 4.17:https://dl.armbian.com/orangepi-r1/nightly/Armbian_5.46.180611_Orangepi-r1_Ubuntu_bionic_dev_4.17.0.7z ARMBIAN 5.46.180611 nightly Ubuntu 18.04 LTS 4.17.0-sunxi Linux opi-zero-r1 4.17.0-sunxi #140 SMP Mon Jun 11 00:25:44 UTC 2018 armv7l armv7l armv7l GNU/Linux Time CPU load %cpu %sys %usr %nice %io %irq CPU C.St. 22:13:08: 1200MHz 0.28 3% 2% 0% 0% 0% 0% 37.9°C 0/8 22:13:13: 1200MHz 0.34 0% 0% 0% 0% 0% 0% 37.5°C 0/8 22:13:18: 1200MHz 0.28 0% 0% 0% 0% 0% 0% 37.6°C 0/8 22:13:23: 1200MHz 0.26 0% 0% 0% 0% 0% 0% 35.5°C 0/8 22:13:28: 1200MHz 0.24 0% 0% 0% 0% 0% 0% 36.7°C 0/8 22:13:34: 1200MHz 0.22 0% 0% 0% 0% 0% 0% 35.9°C 0/8 22:13:39: 1200MHz 0.20 0% 0% 0% 0% 0% 0% 37.8°C 0/8 22:13:44: 1200MHz 0.19 1% 0% 0% 0% 0% 0% 38.0°C 0/8 22:13:49: 1200MHz 0.17 0% 0% 0% 0% 0% 0% 36.7°C 0/8 22:13:54: 1200MHz 0.16 0% 0% 0% 0% 0% 0% 34.8°C 0/8 22:13:59: 1200MHz 0.14 3% 3% 0% 0% 0% 0% 38.2°C 0/8 22:14:04: 1200MHz 0.13 1% 0% 0% 0% 0% 0% 36.8°C 0/8 22:14:09: 1200MHz 0.12 0% 0% 0% 0% 0% 0% 37.8°C 0/8 22:14:14: 1200MHz 0.11 0% 0% 0% 0% 0% 0% 36.5°C 0/8 22:14:20: 1200MHz 0.10 0% 0% 0% 0% 0% 0% 37.1°C 0/8 I will let run the R1 idle over night and will see if it wont hang (like on kernel 4.14.x or before)....
none1337 Posted June 11, 2018 Author Posted June 11, 2018 7 minutes ago, guidol said: I will let run the R1 idle over night and will see if it wont hang (like on kernel 4.14.x or before).... Thanks, is there a way for me to upgrade to the nightly without flashing the SD card?
guidol Posted June 11, 2018 Posted June 11, 2018 3 minutes ago, none1337 said: Thanks, is there a way for me to upgrade to the nightly without flashing the SD card? you could try and take a look inside armbian-config -> system Here in the nightly is a option "Stable -> Switch to stable builds" Maybe you could there also switch from stable to nightly But only (I think when you are on dev?) because next <--> dev
none1337 Posted June 11, 2018 Author Posted June 11, 2018 Thanks, if I have to scrub the card it wont be a total loss. I also did some reading on using the microUSB for power. The barrel style plug you use on the R1 can you take photos / link me to the product name? I did not see that for sale when I purchased the unit.
guidol Posted June 11, 2018 Posted June 11, 2018 15 minutes ago, none1337 said: I also did some reading on using the microUSB for power. The barrel style plug you use on the R1 can you take photos / link me to the product name? I did not see that for sale when I purchased the unit. :-) I did build (the worldwide first?) OPi R1 with the NAS-Expansionboard for the normal OPi Zero I did this in November 2017 (see there the pictures in my post) : (click on "guidol replied on a topic" below) Infos about the NAS Expansion-board could be found in the following thread:
guidol Posted June 12, 2018 Posted June 12, 2018 12 hours ago, guidol said: I will let run the R1 idle over night and will see if it wont hang (like on kernel 4.14.x or before).... 12 Hours later its still running (at a room temperatur of 28 degree): Time CPU load %cpu %sys %usr %nice %io %irq CPU C.St. 10:29:57: 1200MHz 0.11 0% 0% 0% 0% 0% 0% 40.3°C 0/8 10:30:02: 1200MHz 0.10 3% 2% 0% 0% 0% 0% 42.1°C 0/8 10:30:07: 1200MHz 0.09 1% 0% 0% 0% 0% 0% 39.7°C 0/8 10:30:12: 1200MHz 0.08 0% 0% 0% 0% 0% 0% 40.7°C 0/8 10:30:17: 1200MHz 0.08 0% 0% 0% 0% 0% 0% 42.1°C 0/8 10:30:22: 1200MHz 0.07 0% 0% 0% 0% 0% 0% 37.8°C 0/8 10:30:27: 1200MHz 0.06 0% 0% 0% 0% 0% 0% 40.4°C 0/8 10:30:32: 1200MHz 0.06 0% 0% 0% 0% 0% 0% 39.9°C 0/8 10:30:38: 1200MHz 0.05 0% 0% 0% 0% 0% 0% 40.7°C 0/8 10:30:43: 1200MHz 0.05 0% 0% 0% 0% 0% 0% 39.8°C 0/8 10:30:48: 1200MHz 0.05 1% 0% 0% 0% 0% 0% 38.1°C 0/8 10:30:53: 1200MHz 0.04 0% 0% 0% 0% 0% 0% 40.4°C 0/8 10:30:58: 1200MHz 0.04 0% 0% 0% 0% 0% 0% 41.5°C 0/8 10:31:03: 1200MHz 0.04 2% 1% 0% 0% 0% 0% 41.1°C 0/8 10:31:08: 1200MHz 0.03 2% 2% 0% 0% 0% 0% 39.4°C 0/8 1
none1337 Posted June 12, 2018 Author Posted June 12, 2018 guidol, thank you for your help and the temps. I think I have resolved the issue by using a different USB port. The OPi R1 started last night out at 32, climbed to 48 likely the boot process and startup, then came back down to 35-39 for the remainder of the night without the need for any fan or moving air with an ambient room temp of 26c. So far I am seeing a uptime of 11 hours this is a new high compared to the max 6 hours I would get before. I can only assume the USB port on my keyboard I was using to power the device was giving it more power than it could handel. Thank you for the information on the barrel plug adapter I will look into that if I continue to have issues.
chwe Posted June 12, 2018 Posted June 12, 2018 1 hour ago, none1337 said: port on my keyboard I was using to power the device was giving it more power than it could handel. It's more likely that your USB port can't deliver as much as your SBC need. On 6/11/2018 at 1:26 PM, none1337 said: 4.95-5.00 V (stable at 4.68V) not sure if this is a typo, otherwise 4.68V is a clear sign that your board doesn't get the power it should. This can either be related to the cable and/or PSU (in your case the USB-port of your keyboard).
none1337 Posted June 12, 2018 Author Posted June 12, 2018 12 minutes ago, chwe said: not sure if this is a typo, otherwise 4.68V is a clear sign that your board doesn't get the power it should. This can either be related to the cable and/or PSU (in your case the USB-port of your keyboard). That is the ammount that is shown on my USB voltage meter. 4.68v is where it holds at once the OPi is booted. The following layout cause the board to continually heat up. PC > Keyboard (USB Port) > USB Voltage Tester > (microusb) OPi Currently I have it plugged into a USB 2.0 port on another box I can test the A/V when I return from work. But the OPi is no where near as hot or crashing.
chwe Posted June 12, 2018 Posted June 12, 2018 57 minutes ago, none1337 said: That is the ammount that is shown on my USB voltage meter. 4.68v is where it holds at once the OPi is booted. and 4.68V is not appropriate.. See here for some examples: https://forum.armbian.com/forum/31-sd-card-and-power-supply/ and after cable and microUSB connector, there will be even a higher voltagedrop, see this one: If you've a multimeter, just set up the device again on the USB from the keyboard and check on the pin-header. You'll see this is most likely below the 4.68V. Let's assume your pc delivers according to specs.. You get a maximum voltage of 5.25V out of the pc, the first voltagdrop happens on the cable from your pc to the keyboard, then it's most likely that the powerline is just wired through the keyboard, then it's an other USB-A socket and a second cable and the microUSB, all of them are responsible for a second voltagedrop. A next thing that should be considered. According to specs: https://en.wikipedia.org/wiki/USB#Low-Power_and_High-Power_devices They're often rated at 500mA (sometimes the ports on a computer deliver more). So if you 'suck' near to the max. what normally happens is that the voltage drops even more, see 'Powering through micro USB' where I forced the PSUs to go to their maximum by charging a powerbank. Even the psu I would consider as a 'known good one' dropped somewhere to 4.8V (I think that would still be 'ok' but for sure nothing I would recommend). 'Undervoltage' of your board can end with all sorts of 'unexpected behavior' from crash to file-corruption etc. I never tested it (due to lack of proper equipment to test it), but it might be worth a test. According to this one: SoC temperature for those H2+/H3 needs a reference voltage. So if this reference voltage is to low, the whole temp. readout is bogus. I never checked, what happens to this reverence voltage for different boards when the powering is not sufficient. I think an EE with a benchtop psu (with adjustable voltage) could easily compare SoC surface temp. vs. reported internal temp. at different sufficient/insufficient voltage delivered to the pin-header. @TonyMac32 do you need a new 'pet-project'? My guess, and that's just a guess is that with a board that suffers from undervoltage, the gap between surface temp. and internal reported temp. would be higher than with an appropriate voltage. IMO a board should be powered with a dedicated (known good) PSU and not with a port from your PC. There are for sure USB ports on computers which are able to deliver sufficient power but I prefer to spend 10$ for a sufficient PSU (and probably 5$ for a known good microUSB-USB cable, from what I read friendlyArm is known to deliver 'known good' cables) than spend hours in debug a unexpected behavior of my SBC due to saved a few bucks.
guidol Posted June 12, 2018 Posted June 12, 2018 3 hours ago, chwe said: If you've a multimeter, just set up the device again on the USB from the keyboard and check on the pin-header. You'll see this is most likely below the 4.68V. Let's assume your pc delivers according to specs.. You get a maximum voltage of 5.25V out of the pc, the first voltagdrop happens on the cable from your pc to the keyboard, then it's most likely that the powerline is just wired throug 'Undervoltage' of your board can end with all sorts of 'unexpected behavior' from crash to file-corruption etc. for voltage-drops and resistance inside connectors and how to measure I could recommend the following Youtube-Video 1
none1337 Posted June 12, 2018 Author Posted June 12, 2018 6 hours ago, chwe said: and 4.68V is not appropriate.. See here for some examples: Thanks for the information. I now have a better understanding on what is going on with it now. I just checked and looks like my Razer keyboard was the issue. These is the power draw once booted. This is the draw from a USB 2.0 port on another PC in the house. Thanks everyone for the help!
WarHawk_AVG Posted July 10, 2018 Posted July 10, 2018 I notice on my Orange Pi Zero I had to go into /etc/default/cpufrequtils and set max CPU freq down to 912000 and lowered the lowest CPU to 240000 It helps quite a bit...if you can't get active cooling even with a small fan running at 5vdc just moving a tiny bit of air...then slowing down the CPU helps quite a bit Yeah...kinda sux having a governor that limits 30-40% of processor speed...but for stability it greatly enhances it Best thing to do is get a good heatsink and active cooling
guidol Posted February 28, 2019 Posted February 28, 2019 My OPi R1 is - for me - really cool with Welcome to ARMBIAN 5.76 user-built Debian GNU/Linux 9 (stretch) 4.20.12-sunxi Linux opi-r1 4.20.12-sunxi #5.76 SMP Thu Feb 28 15:04:35 +03 2019 armv7l GNU/Linux root@opi-r1(192.168.6.101):~# more /etc/default/cpufrequtils # WARNING: this file will be replaced on board support package (linux-root-...) upgrade ENABLE=true MIN_SPEED=240000 MAX_SPEED=1200000 GOVERNOR=ondemand Time CPU load %cpu %sys %usr %nice %io %irq CPU C.St. 21:53:24: 1008MHz 0.04 1% 0% 0% 0% 0% 0% 39.3°C 0/6 21:53:29: 240MHz 0.03 0% 0% 0% 0% 0% 0% 37.4°C 0/6 21:53:34: 240MHz 0.03 1% 1% 0% 0% 0% 0% 38.8°C 0/6 21:53:40: 240MHz 0.03 1% 0% 0% 0% 0% 0% 38.0°C 0/6 21:53:45: 240MHz 0.03 1% 0% 0% 0% 0% 0% 39.2°C 0/6 21:53:50: 240MHz 0.02 1% 1% 0% 0% 0% 0% 37.8°C 0/6 21:53:55: 240MHz 0.02 1% 0% 0% 0% 0% 0% 38.0°C 0/6 21:54:00: 1008MHz 0.02 1% 1% 0% 0% 0% 0% 38.7°C 0/6 21:54:06: 1008MHz 0.02 2% 1% 0% 0% 0% 0% 39.3°C 0/6 21:54:11: 240MHz 0.02 1% 1% 0% 0% 0% 0% 38.1°C 0/6 21:54:16: 240MHz 0.01 1% 0% 0% 0% 0% 0% 37.3°C 0/6 21:54:21: 240MHz 0.01 1% 0% 0% 0% 0% 0% 38.2°C 0/6 21:54:26: 240MHz 0.01 1% 0% 0% 0% 0% 0% 36.9°C 0/6 21:54:32: 240MHz 0.01 1% 1% 0% 0% 0% 0% 37.4°C 0/6
Recommended Posts