N2 Passive vs Active Cooling

Post Reply
hominoid
Posts: 285
Joined: Tue Feb 28, 2017 3:55 am
languages_spoken: english
ODROIDs: C2, XU4, MC1, N1, N2
Location: Lake Superior Basin, USA
Has thanked: 3 times
Been thanked: 8 times
Contact:

N2 Passive vs Active Cooling

Unread post by hominoid » Tue Mar 12, 2019 12:25 am

There has been a lot of discussion surrounding the N2’s case and heatsink on the internet lately. Even though I believe it is a great design that for most uses is more than sufficient, there might be a few use cases out there that could tax it’s passive thermal capabilities. I want to explore the N2’s passive and active cooling capabilities to further understand were, if any, the edges of these solutions are and provide more data for informed discussion.

The first order was to design a reasonable active cooling solution. I designed a simple stand to hold a fan directly under the SOC and memory. In trying to minimize the effect on the current design, I decided to keep the height as low as possible. Reusing an existing HK OEM heatsink fan seemed like a good starting point since there aren’t to many fan choices that are thinner and I had one available.
N2_Fan_Stand.jpg
N2_Fan_Stand.jpg (43.39 KiB) Viewed 2144 times
OpenSCAD FIles and stl
20190311 Odroid-N2 Fan Stand_v1.zip
20190312 Odroid-N2_Fan_Stand_v1.0.1.zip
Odroid-N2_Fan_Stand_v1.0.1.zip
(133.49 KiB) Downloaded 29 times
The fan could still be about 3-4mm closer to the heatsink while lowering the overall stance of the design by the same amount. The extra height may allow a regular 40mm x 10mm fan to fit, which was my intent. As most people know, the actual thickness of a 40mm x 10mm fan can vary from one manufacture to another and the height will most likely need to be adjusted accordingly based on the fan chosen.
S1770012-s.jpg
S1770012-s.jpg (44.67 KiB) Viewed 2144 times
A small hole drilled next to the rj-45 connection allows the fan wires to be cleanly passed through the case to the fan connector. This is important because I want to take a look at how the enclosed case top effects heat retention. This has been one of the areas of discussion at other forums and an interest to me. It’s nothing a drill can’t solve but lets find out if it’s a problem first. Almost all of the tests I have seen so far for the N2 have been shorter tests of minutes to approximately an hour. I would also like to do some long run multi-day tests to try and push the N2 to its limit.

Fan configuration
Once everything was assembled and running, I started to examine what setup and control there was on the N2. I did find 2 thermal zones with 4 trip zones each:

Code: Select all

hominoid@odroid-n2:/sys/devices/virtual/thermal/thermal_zone0$ cat trip_point_0_temp
65000
hominoid@odroid-n2:/sys/devices/virtual/thermal/thermal_zone0$ cat trip_point_1_temp
75000
hominoid@odroid-n2:/sys/devices/virtual/thermal/thermal_zone0$ cat trip_point_2_temp
95000
hominoid@odroid-n2:/sys/devices/virtual/thermal/thermal_zone0$ cat trip_point_3_temp
110000
hominoid@odroid-n2:/sys/devices/virtual/thermal/thermal_zone1$ cat trip_point_0_temp
60000
hominoid@odroid-n2:/sys/devices/virtual/thermal/thermal_zone1$ cat trip_point_1_temp
75000
hominoid@odroid-n2:/sys/devices/virtual/thermal/thermal_zone1$ cat trip_point_2_temp
85000
hominoid@odroid-n2:/sys/devices/virtual/thermal/thermal_zone1$ cat trip_point_3_temp
110000
I did not find /sys/devices/platform/pwm-fan though and after looking around awhile it wasn’t evident to me that there was a driver in the kernel.
I then tried:

Code: Select all

$ sudo apt install lm-sensors
$ sudo sensors-detect
# sensors-detect revision 6284 (2015-05-31 14:00:33 +0200)
# Kernel: 4.9.162-18 aarch64
# Processor:  (//)

This program will help you determine which kernel modules you need
to load to use lm_sensors most effectively. It is generally safe
and recommended to accept the default answers to all questions,
unless you know what you're doing.

Some south bridges, CPUs or memory controllers contain embedded sensors.
Do you want to scan for them? This is totally safe. (YES/no): YES
modprobe: FATAL: Module cpuid not found in directory /lib/modules/4.9.162-18
Failed to load module cpuid.
Silicon Integrated Systems SIS5595...                       No
VIA VT82C686 Integrated Sensors...                          No
VIA VT8231 Integrated Sensors...                            No
AMD K8 thermal sensors...                                   No
AMD Family 10h thermal sensors...                           No
AMD Family 11h thermal sensors...                           No
AMD Family 12h and 14h thermal sensors...                   No
AMD Family 15h thermal sensors...                           No
AMD Family 16h thermal sensors...                           No
AMD Family 15h power sensors...                             No
AMD Family 16h power sensors...                             No
Intel digital thermal sensor...                             No
Intel AMB FB-DIMM thermal sensor...                         No
Intel 5500/5520/X58 thermal sensor...                       No
VIA C7 thermal sensor...                                    No
VIA Nano thermal sensor...                                  No

Lastly, we can probe the I2C/SMBus adapters for connected hardware
monitoring devices. This is the most risky part, and while it works
reasonably well on most systems, it has been reported to cause trouble
on some systems.
Do you want to probe the I2C/SMBus adapters now? (YES/no): YES
Sorry, no supported PCI bus adapters found.

Sorry, no sensors were detected.
Either your system has no sensors, or they are not supported, or
they are connected to an I2C or SMBus adapter that is not
supported. If you find out what chips are on your board, check
http://www.lm-sensors.org/wiki/Devices for driver status.
In order to continue I will need to find the answers to the following questions:

1. Is there a kernel driver loaded by default or available yet?
2. What HW is represented by thermal zone 1 and thermal zone 2?
3. Are all 4 thermal trip points used for each thermal zone?
4. Is the fourth trip zone for thermal shutdown?
5. Is or will /sys/devices/platform/pwm-fan be used and if not what then?

Once I get everything configured and running, I’ll start some testing. Almost there...
Last edited by hominoid on Wed Mar 13, 2019 11:27 am, edited 1 time in total.

joy
Posts: 841
Joined: Fri Oct 02, 2015 1:44 pm
languages_spoken: english
ODROIDs: ODROID-C1+, XU4, X
Has thanked: 7 times
Been thanked: 35 times
Contact:

Re: N2 Passive vs Active Cooling

Unread post by joy » Tue Mar 12, 2019 12:18 pm

Hi @hominoid,
Awesome work! :) Cool~

About pwm fan operation, only a basic pwm-gpio driver is done with the latest kernel,
and there are some more works to do, (1) to run pwm fan automatically (2) to adjust pwm fan control based on thermal rules and cooling maps.
Only turn on/off is available so far.
Once remaining works are done, we will share the status.

Please refer to the following for your questions.
hominoid wrote:
Tue Mar 12, 2019 12:25 am
1. Is there a kernel driver loaded by default or available yet?
hominoid wrote:
Tue Mar 12, 2019 12:25 am
5. Is or will /sys/devices/platform/pwm-fan be used and if not what then?
To run cooling fan with the pwm port, you need to run some instructions.

You can turn on the fan using this instruction.
https://wiki.odroid.com/odroid-n2/appli ... iver_usage
The pwmchip port is 16.

Code: Select all

$ echo 0 > /sys/class/pwm/pwmchip16/export
$ echo 10000 > /sys/class/pwm/pwmchip16/pwm0/period
$ echo 1000 > /sys/class/pwm/pwmchip16/pwm0/duty_cycle
$ echo 1 > /sys/class/pwm/pwmchip16/pwm0/enable
hominoid wrote:
Tue Mar 12, 2019 12:25 am
2. What HW is represented by thermal zone 1 and thermal zone 2?
thermal_zone0 is assigned to CPU temp sensor of S922S,
and thermal_zone1 is the one for DDR.
hominoid wrote:
Tue Mar 12, 2019 12:25 am
3. Are all 4 thermal trip points used for each thermal zone?
Please refer to this source code.
It describes threshold temperature to change running states,
and thermal_zone0 and thermal_zone1 have their own trip points each.
So each values are different.
(1) thermal_zone0 : CPU temp sensor
https://github.com/hardkernel/linux/blo ... dtsi#L1635
(2) thermal_zone1 : DDR temp sensor
https://github.com/hardkernel/linux/blo ... dtsi#L1696
hominoid wrote:
Tue Mar 12, 2019 12:25 am
4. Is the fourth trip zone for thermal shutdown?
Yes.
cpu temperature gets reach to the trip point, dcritical,
system will shut down.
https://github.com/hardkernel/linux/blo ... dtsi#L1651

Code: Select all

# cat /sys/class/thermal/thermal_zone0/trip_point_3_temp 
110000

hominoid
Posts: 285
Joined: Tue Feb 28, 2017 3:55 am
languages_spoken: english
ODROIDs: C2, XU4, MC1, N1, N2
Location: Lake Superior Basin, USA
Has thanked: 3 times
Been thanked: 8 times
Contact:

Re: N2 Passive vs Active Cooling

Unread post by hominoid » Tue Mar 12, 2019 11:03 pm

Hi @joy,
Right now I'm working on improving the fan opening so no hand fitting is necessary. I will be posting an updated design file and stl soon. Once that is complete I will begin testing. With such a great thermal profile, I figured the fan pwm might be a lower priority and might not be complete yet. For long run testing turning the fan on and off is really the only thing needed right now for me to move forward. I'm sure things are very busy at HK as the release date for the N2 approaches so thanks for the detail and taking the time to answer my questions! :)

hominoid
Posts: 285
Joined: Tue Feb 28, 2017 3:55 am
languages_spoken: english
ODROIDs: C2, XU4, MC1, N1, N2
Location: Lake Superior Basin, USA
Has thanked: 3 times
Been thanked: 8 times
Contact:

Re: N2 Passive vs Active Cooling

Unread post by hominoid » Wed Mar 13, 2019 11:21 am

I finished making some minor but important adjustments to the design. I moved the fan 3mm closer to the heatsink. It is 1-1.5mm from the heatsink now. It should provide better cooling and lowers the stance. I also enlarged the fan opening and screw holes so that the fan should work without any fitting while the screws are easier to sink. Aesthetically I like it better because it looks like the natural bottom of the case. Unless you hear it you might not suspect there is a fan installed.
N2 Fan Stand version 1.0.1
S1780005s.jpg
S1780005s.jpg (208.51 KiB) Viewed 2005 times

joy
Posts: 841
Joined: Fri Oct 02, 2015 1:44 pm
languages_spoken: english
ODROIDs: ODROID-C1+, XU4, X
Has thanked: 7 times
Been thanked: 35 times
Contact:

Re: N2 Passive vs Active Cooling

Unread post by joy » Wed Mar 13, 2019 1:43 pm

Hi @hominoid.
Thank you for sharing your work. :)

The pwm-fan related functions are in our TODO list
and we're planning to add it on 1st release image at least.
Sorry for the inconvenience.
Once it's done, I will share the status here,
so that N2 users can try and enjoy your work easily. ;)

hominoid
Posts: 285
Joined: Tue Feb 28, 2017 3:55 am
languages_spoken: english
ODROIDs: C2, XU4, MC1, N1, N2
Location: Lake Superior Basin, USA
Has thanked: 3 times
Been thanked: 8 times
Contact:

Re: N2 Passive vs Active Cooling

Unread post by hominoid » Mon Mar 18, 2019 7:13 am

N2 Passive vs Active Test Results
Let me start by saying that in my opinion the test results are very impressive for the Odroid-N2. I tested the N2 engineering sample in a 69-71F(20.5-21.7c) degree environment with the case top enclosure on and closed. There were no fans, air conditioning or other factors I could identify to influence the test results. All tests were with the governor set to performance with no overclocking. The first test was for the CPU only with the OEM passive heatsink. The test was comprised of a Scrypt2 algorithm to a live mining pool for approximately 24 hours using @fireworm's modified veriumMiner, compiled with gcc 7.3. The standard HK kernel was used with the exception that huge pages were enabled. They provided approximately a 15% performance increase.

System and DRAM temperature data and core frequency was collected every 2 seconds. The entries were UTC time stamped so that the corresponding hash rate data, which is also UTC time stamped, could be easily correlated. Due to the amount of data collected the results had to be broke into two parts. Below are results of the N2 passively cooled CPU 24hr Scrypt2 test.
.
Passive_Scrypt_Temperatures.jpg
Passive_Scrypt_Temperatures.jpg (228.29 KiB) Viewed 1876 times
Passive_Scrypt_Hash-Minute.jpg
Passive_Scrypt_Hash-Minute.jpg (123.93 KiB) Viewed 1876 times
.
Scrypt2 utilizes memory heavily and according to HTOP was using 3.17GB at 100% CPU utilization on all cores. The temperature and hash rate was steady thru the breadth of the test and as the test data can confirm, no throttling occurred.

Active Cooling Test
The same setup was used for the actively cooled test. The Fan Stand design at the beginning of the this thread was used with the fan on maximum speed since the kernel driver is not yet complete. The initial results after approximately 3hrs were similar to the passive test. Once the system temperature stabilized around 38C no major deviations occurred. I stopped the test early for this reason to try and stress the N2 more utilizing a passive dual CPU/GPU test. If the N2 can withstand mining on a passive cooled system, running an active cooled test provides little information other than the normalized temperature of operation.
.
Active_Scrypt_Temperature.jpg
Active_Scrypt_Temperature.jpg (131.34 KiB) Viewed 1876 times

Passive Dual CPU/GPU Mining Test
The passive dual CPU/GPU mining test results are approximately the same as the passive CPU test. The Odroid-N2 stabilized around 72-73C and then over an extended period of time actually declined slightly to just under 70C. There was no observed decline in hash rates for the CPU or GPU under 100% utilization.
.
Screenshot_2019-03-15_14-42-06.png
Screenshot_2019-03-15_14-42-06.png (204.84 KiB) Viewed 1876 times
.
I did not create a chart for the dual mining test because the thermal performance and data are nearly identical to the CPU only passive test. With all the share and stratum messages over a 24hr period, the amount of time to scrub the data for charting is significant. Also, I had to make choices due to the restriction on the number of uploads and their size on the forum. I did include all the raw test data below for further scrutiny.
N2 Passive and Active Cooled Test Data.zip
(380.52 KiB) Downloaded 18 times
Again, very impressive in my experience. The fact that the N2 can dual CPU/GPU mine with passive cooling is very significant. I have never seen any passively cooled system of any architecture type be able to accomplish CPU only mining let alone dual CPU/GPU mining. It is absolutely amazing! In my opinion, the benefit of having an active cooled N2 will mostly be delegated to enclosed or high ambient temperature environments. I cannot imagine there are many general computing applications that could put more stress on a system then dual CPU/GPU mining. One last note, I spent a modest amount of time tuning for the Scrypt2 so the actual hash rates do not necessarily represent the maximum performance of the Odroid-N2.

User avatar
odroid
Site Admin
Posts: 31363
Joined: Fri Feb 22, 2013 11:14 pm
languages_spoken: English
ODROIDs: ODROID
Has thanked: 50 times
Been thanked: 184 times
Contact:

Re: N2 Passive vs Active Cooling

Unread post by odroid » Mon Mar 18, 2019 10:14 am

Really appreciate your very detail thermal analysis with both CPU and GPU computing.
We hoped most users will not need an active cooling fan thanks to the modern 12nm fabricated silicon.

joy
Posts: 841
Joined: Fri Oct 02, 2015 1:44 pm
languages_spoken: english
ODROIDs: ODROID-C1+, XU4, X
Has thanked: 7 times
Been thanked: 35 times
Contact:

Re: N2 Passive vs Active Cooling

Unread post by joy » Mon Mar 18, 2019 12:18 pm

Hi @hominoid,
What a really valuable analysis!

I have an update to share with you.
Thermal throttling logic for cooling fan with pwn-gpio is done and it will be included for next release.
When cpu temperature (thermal_zone0) reaches to 45°C, cooling fan will start to work
and fan speed will increase up sequentially from 45°C -> 50°C -> 60°C.

For this, one preparation process is needed to turn it on by switching policy.

Code: Select all

$ sudo echo step_wise > /sys/class/thermal/thermal_zone0/policy
Then thermal control logic with pwm-fan will be activated automatically.

For normal cases, cooling fan is useless
and if pwm-gpio is turned on as default, it would waste sw resource without any advantage.
That's the reason why the additional manual instruction is set.

back2future
Posts: 229
Joined: Sun Jul 23, 2017 3:19 pm
languages_spoken: english
Has thanked: 9 times
Been thanked: 0
Contact:

Re: N2 Passive vs Active Cooling

Unread post by back2future » Tue Mar 19, 2019 4:28 am

Where are DRAM temperatures on N2 from?

hominoid
Posts: 285
Joined: Tue Feb 28, 2017 3:55 am
languages_spoken: english
ODROIDs: C2, XU4, MC1, N1, N2
Location: Lake Superior Basin, USA
Has thanked: 3 times
Been thanked: 8 times
Contact:

Re: N2 Passive vs Active Cooling

Unread post by hominoid » Tue Mar 19, 2019 5:01 am

joy wrote:
Tue Mar 12, 2019 12:18 pm
thermal_zone0 is assigned to CPU temp sensor of S922S,
and thermal_zone1 is the one for DDR....
(1) thermal_zone0 : CPU temp sensor
https://github.com/hardkernel/linux/blo ... dtsi#L1635
(2) thermal_zone1 : DDR temp sensor
https://github.com/hardkernel/linux/blo ... dtsi#L1696
It was part of @joy's comments to my questions in this thread dated March 11th...unless I misunderstood what he was saying. I originally thought thermal zone1 was the GPU until I asked.

joy
Posts: 841
Joined: Fri Oct 02, 2015 1:44 pm
languages_spoken: english
ODROIDs: ODROID-C1+, XU4, X
Has thanked: 7 times
Been thanked: 35 times
Contact:

Re: N2 Passive vs Active Cooling

Unread post by joy » Tue Mar 19, 2019 8:43 am

hominoid wrote:
Tue Mar 19, 2019 5:01 am
It was part of @joy's comments to my questions in this thread dated March 11th...unless I misunderstood what he was saying. I originally thought thermal zone1 was the GPU until I asked.
So did I. :D
When I found the node thermal_zone1 in dts from S922X BSP, also I thought "isn't it for GPU? not DDR..",
because DDR parts aren't located inside CPU, apart from CPU.
But, as I checked, S922X integrates 2 temperature sensors, one is close to DDR and another is close to PLL (between CPU and GPU).
So, it looks DDR means DDR IP (controller) inside S922X.

back2future
Posts: 229
Joined: Sun Jul 23, 2017 3:19 pm
languages_spoken: english
Has thanked: 9 times
Been thanked: 0
Contact:

Re: N2 Passive vs Active Cooling

Unread post by back2future » Wed Mar 20, 2019 7:44 am

[ Curiosity: Top current rating for example H2 SoC's sram/io-logics 4.5A is even higher than ddr4 memory controller 3 amps. ]

joy
Posts: 841
Joined: Fri Oct 02, 2015 1:44 pm
languages_spoken: english
ODROIDs: ODROID-C1+, XU4, X
Has thanked: 7 times
Been thanked: 35 times
Contact:

Re: N2 Passive vs Active Cooling

Unread post by joy » Wed Mar 20, 2019 8:10 am

back2future wrote:
Wed Mar 20, 2019 7:44 am
[ Curiosity: Top current rating for example H2 SoC's sram/io-logics 4.5A is even higher than ddr4 memory controller 3 amps. ]
If the t-sensor for DDR is to reduce temperature, so get current gain, it makes sense. :)
Is the data from your measure or from common materials?
Actually I don't have much knowledge about handling current analysis of each part of memory devices.

[ updated ]
Suddenly, I realized I have no idea which component is controlled when the d-sensor's output is passed over the trip points. :?:
ex) cpu temp sensor -> control operating cpu frequency
These users thanked the author joy for the post:
back2future (Wed Apr 10, 2019 8:15 am)

back2future
Posts: 229
Joined: Sun Jul 23, 2017 3:19 pm
languages_spoken: english
Has thanked: 9 times
Been thanked: 0
Contact:

Re: N2 Passive vs Active Cooling

Unread post by back2future » Wed Mar 20, 2019 8:40 am

[ H2 SoC power rail dc specification and Iccmax, silver-celeron-datasheet-vol-1.pdf, pages 171/172 ]
If we compare surface area of Soc and ram parts (with even lower current input to these newer (denser) ddr4 packages compared to SoC) heat dissipation (within ddr4 limits -40-95°C from Samsung) should not be a problem, although refresh cycle period for data retention is set to half above 85°C to 3.8us. Memory controller has to share heat dissipation area with cpu/gpu cores and with todays higher clock demand there might be temperature awareness necessary (and that point is supported to some extent by this thermal node being exactly at the memory controller unit?).
Did you ever make thermal imaging on an idling and cpu/gpu/memory/storage intense N2?
Another note was on eeproms, loosing reliability for data stability¹ from some 100-200yrs to even below one decade (situation and device dependent) because of higher device temperatures (above default specification related to ~55°C) and especially at temperatures below freezing point at eeprom ic (mobile applications)?

[ 1) EEPROM Bit Failure Investigation, 2003 JPLab, https://trs.jpl.nasa.gov/bitstream/hand ... sequence=1 ]

joy
Posts: 841
Joined: Fri Oct 02, 2015 1:44 pm
languages_spoken: english
ODROIDs: ODROID-C1+, XU4, X
Has thanked: 7 times
Been thanked: 35 times
Contact:

Re: N2 Passive vs Active Cooling

Unread post by joy » Mon Mar 25, 2019 10:34 am

back2future wrote:
Wed Mar 20, 2019 8:40 am
[ H2 SoC power rail dc specification and Iccmax, silver-celeron-datasheet-vol-1.pdf, pages 171/172 ]
Thank you for sharing the information, back2future. :)
back2future wrote:
Wed Mar 20, 2019 8:40 am
If we compare surface area of Soc and ram parts (with even lower current input to these newer (denser) ddr4 packages compared to SoC) heat dissipation (within ddr4 limits -40-95°C from Samsung) should not be a problem, although refresh cycle period for data retention is set to half above 85°C to 3.8us. Memory controller has to share heat dissipation area with cpu/gpu cores and with todays higher clock demand there might be temperature awareness necessary (and that point is supported to some extent by this thermal node being exactly at the memory controller unit?).
I guess it is, based on descriptions of a draft version of S922X datasheet.

Code: Select all

S922X integrates 2 Temp Sensor, one is close to DDR, one is close to PLL (between CPU and GPU).
Each Temp Sensor are the same design.

hominoid
Posts: 285
Joined: Tue Feb 28, 2017 3:55 am
languages_spoken: english
ODROIDs: C2, XU4, MC1, N1, N2
Location: Lake Superior Basin, USA
Has thanked: 3 times
Been thanked: 8 times
Contact:

Re: N2 Passive vs Active Cooling

Unread post by hominoid » Wed Mar 27, 2019 9:32 am

I have been doing some testing on the recent enabled fan driver. There appears to be an issue with the fan startup. After substantial reading of the kernel documentation and examination of the code I was not able to come to a complete and definitive identification of the problem. The main symptom is the fan does not start at the first trip point. It is similar to what had been experienced on the XU4 where the fan tries to start but does not or stutters and then stops all the while creating noise. After trying to adjust the fan speed from user land unsuccessfully(is this not possible?), I changed the original device tree source file, mesong12_odroid_common.dtsi line 225 and related temperature trip points for testing:

Code: Select all

    219         pwmfan:pwm-fan {
    220                 compatible = "pwm-fan";
    221                 pwms = <&pwmgpio 0 40000 PWM_POLARITY_INVERTED>;
    222                 cooling-min-state = <0>;
    223                 cooling-max-state = <3>;
    224                 #cooling-cells = <2>;
    225                 cooling-levels = <0 170 190 250>;
    226         };
    227 };
    228 
    229 &soc_thermal {
    230         trips {
    231                 fan_0: trip-point@4 {
    232                         temperature = <35000>;
    233                         hysteresis = <5000>;
    234                         type = "active";
    235                 };
    236                 fan_1: trip-point@5 {
    237                         temperature = <45000>;
    238                         hysteresis = <5000>;
    239                         type = "active";
    240                 };
    241                 fan_2: trip-point@6 {
    242                         temperature = <50000>;
    243                         hysteresis = <5000>;
    244                         type = "active";
    245                 };
    246         };
This significantly helped and allowed the fan to run but there is still some noise at the first trip point. From the beginning I thought the noise was coming from the fan but after implementing this change and closer examination, I realized that it was coming from one of the onboard components. Maybe power management/regulator? It would appear that the original 1st trip point fan speed value of 120 does not allow the fan to start and is also stressing an electronic component, which is responsible for the noise. With time constraints and the limits of my knowledge, this is as far as I can currently investigate the problem. Speculating, with such high fan speed values needed to start the fan and accompanying component noise, it may be related to more than just too low of a fan speed, as I originally thought. This is based on the assumption that the fan speed value range is from 0-255 like the XU4.
Last edited by hominoid on Wed Mar 27, 2019 10:06 am, edited 1 time in total.

hominoid
Posts: 285
Joined: Tue Feb 28, 2017 3:55 am
languages_spoken: english
ODROIDs: C2, XU4, MC1, N1, N2
Location: Lake Superior Basin, USA
Has thanked: 3 times
Been thanked: 8 times
Contact:

Re: N2 Passive vs Active Cooling

Unread post by hominoid » Wed Mar 27, 2019 9:43 am

One other issue, doing

Code: Select all

sudo echo step_wise > /sys/class/thermal/thermal_zone0/policy
does not work unless chmod is used and then it is reverted back to read only at the next boot.

cap00k
Posts: 84
Joined: Tue May 21, 2013 10:46 am
languages_spoken: english
ODROIDs: ODROID
Has thanked: 0
Been thanked: 6 times
Contact:

Re: N2 Passive vs Active Cooling

Unread post by cap00k » Wed Mar 27, 2019 3:35 pm

Hi, hominoid,
Thank you for testing.
Cooling Map is based on the fan to be sold for the N2 in the HK.
There may be differences depending on the electrical characteristics of the fan motor.
The reason why the cooling fan does not startup and makes noise is probably due to insufficient pwm duty.
Can you fix and check the pwmfan cooling level below?

Code: Select all

	pwmfan:pwm-fan {
		compatible = "pwm-fan";
		pwms = <&pwmgpio 0 40000 PWM_POLARITY_INVERTED>;
		cooling-min-state = <0>;
		cooling-max-state = <3>;
		#cooling-cells = <2>;
		cooling-levels = <0 200 225 250>;
	};

hominoid
Posts: 285
Joined: Tue Feb 28, 2017 3:55 am
languages_spoken: english
ODROIDs: C2, XU4, MC1, N1, N2
Location: Lake Superior Basin, USA
Has thanked: 3 times
Been thanked: 8 times
Contact:

Re: N2 Passive vs Active Cooling

Unread post by hominoid » Wed Mar 27, 2019 10:16 pm

I neglected to say that running the fan at full or near full speed, like you suggest, does work fine. I was comparing the speed settings to those that work on the XU4. I can start the fan at around 30 on a XU4. Is there or will there be away to change the fan speed settings from user space?

DarkBahamut
Posts: 332
Joined: Tue Jan 19, 2016 10:19 am
languages_spoken: english
ODROIDs: XU4, N1
Has thanked: 0
Been thanked: 0
Contact:

Re: N2 Passive vs Active Cooling

Unread post by DarkBahamut » Thu Mar 28, 2019 8:02 am

Assuming it's using the generic pwm-fan driver provided by the kernel you should be able to get user space control (to the same level as the XU4) with these two patches.

https://kessuth.asuscomm.com:8002/s/s5bjg7Z9ZeLGPQ7
https://kessuth.asuscomm.com:8002/s/geBwpEeH7jj7NjW

I don't have an N2 to test them on, but in theory they should work I believe and they compile fine at least :)

hominoid
Posts: 285
Joined: Tue Feb 28, 2017 3:55 am
languages_spoken: english
ODROIDs: C2, XU4, MC1, N1, N2
Location: Lake Superior Basin, USA
Has thanked: 3 times
Been thanked: 8 times
Contact:

Re: N2 Passive vs Active Cooling

Unread post by hominoid » Thu Mar 28, 2019 8:41 am

Thanks @DarkBahamut, I will give them a try when I get a chance. It will be nice if they work.

Post Reply

Return to “Hardware and peripherals”

Who is online

Users browsing this forum: No registered users and 2 guests