LCD 7" and LM Guitar case BananaHome

cyryllo · Post time Oct 07, 2015 12:12

Today make new case for Guitar and module LCD 7"
http://www.thingiverse.com/thing:1058921

cyryllo · Post time Oct 08, 2015 07:21

And second case http://www.thingiverse.com/thing:1059695 ;)

Vexelius · Post time Oct 10, 2015 08:12

Amazing work as always, man! Can't wait to have a 3D printer to test them!

cyryllo · Post time Oct 11, 2015 03:15

Vexelius replied at Sat Oct 10, 2015 15:12
Amazing work as always, man! Can't wait to have a 3D printer to test them!

Thx

Also remake case for Banana PRO

tkaiser · Post time Oct 12, 2015 03:08

cyryllo replied at Thu Oct 8, 2015 07:21
And second case http://www.thingiverse.com/thing:1059695 ;)

BTW: You should be aware that unlike other ARM SoCs, the Guitar's main components are really 'hot stuff' and some mechanisms that should prevent overheating randomly fail to jump in.

I managed to get the SoC's temperature above 110°C even with factory settings clock speeds and a simple benchmark running for half an hour. And thermal throttling didn't work then for unknown reasons to me: http://forum.armbian.com/index.p ... kers-guitar/?p=2044

In my eyes it's important to keep that in mind when designing enclosures for this board. Both SoC and PMU can be used as water boilers since their temperatures might easily exceed 100°C

When testing an enclosure I would recommend to monitor core temperatures and cpufreq settings by using

while true ; do echo "SoC: $(awk '{printf ("%0.0f",$1/1000); }' </sys/devices/virtual/thermal/thermal_zone1/temp)°C, PMU: $(awk '{printf ("%0.1f",$1/1000); }' </sys/devices/b0170000.i2c/i2c-0/0-0065/atc2603c-hwmon.0/ic_temperature)°C, clock $(awk '{printf ("%0.0f",$1/1000); }' </sys/devices/system/cpu/cpu0/cpufreq/scaling_cur_freq) MHz" ; sleep 1; done

Copy the Code

While torturing the board using either

while true ; do sysbench --test=cpu --cpu-max-prime=20000 run --num-threads=4 ; done

Copy the Code

or

cd / && stress -c 4 -m 2 -i 2 -d 2

Copy the Code

tkaiser · Post time Oct 12, 2015 03:51

Scary results. This is Lemuntu's default kernel and Lemuntu's default cpufreq settings (rather conservative with the CPU's upper clock limit at 1.1 GHz) and the emulation of an enclosure not providing any airflow around SoC and PMU while running both 'stress' and 'sysbench' in a loop (maximum CPU load -- remember that when you put also significant load on the GPU then temperatures might easily increase a lot more):

SoC: 113°C, PMU: 94.6°C, clock 1104 MHz
SoC: 112°C, PMU: 95.0°C, clock 1104 MHz
SoC: 112°C, PMU: 94.6°C, clock 1104 MHz
SoC: 113°C, PMU: 94.6°C, clock 1104 MHz
SoC: 113°C, PMU: 94.8°C, clock 1104 MHz
SoC: 113°C, PMU: 94.8°C, clock 1104 MHz
SoC: 112°C, PMU: 95.0°C, clock 1104 MHz
SoC: 112°C, PMU: 94.8°C, clock 1104 MHz
SoC: 112°C, PMU: 94.8°C, clock 1104 MHz
SoC: 112°C, PMU: 95.0°C, clock 1104 MHz
SoC: 113°C, PMU: 95.0°C, clock 1104 MHz
SoC: 112°C, PMU: 95.0°C, clock 1104 MHz
SoC: 114°C, PMU: 95.0°C, clock 1104 MHz
SoC: 114°C, PMU: 94.8°C, clock 1104 MHz
SoC: 113°C, PMU: 95.2°C, clock 1104 MHz
SoC: 113°C, PMU: 95.2°C, clock 1104 MHz
SoC: 114°C, PMU: 95.2°C, clock 1104 MHz
SoC: 112°C, PMU: 95.4°C, clock 1104 MHz
SoC: 112°C, PMU: 95.2°C, clock 1104 MHz
SoC: 112°C, PMU: 95.4°C, clock 1104 MHz
SoC: 113°C, PMU: 95.4°C, clock 1104 MHz
SoC: 112°C, PMU: 95.4°C, clock 1104 MHz
SoC: 113°C, PMU: 95.2°C, clock 1104 MHz
SoC: 113°C, PMU: 95.4°C, clock 1104 MHz
SoC: 114°C, PMU: 95.0°C, clock 1104 MHz
SoC: 113°C, PMU: 95.2°C, clock 1104 MHz
SoC: 113°C, PMU: 95.6°C, clock 1104 MHz
SoC: 113°C, PMU: 95.4°C, clock 1104 MHz
SoC: 112°C, PMU: 95.4°C, clock 1104 MHz
SoC: 112°C, PMU: 95.2°C, clock 1104 MHz
SoC: 114°C, PMU: 95.0°C, clock 1104 MHz

Copy the Code

Unfortunately no thermal throttling happens. This is an area where LeMaker has to investigate. And what has to be considered when designing enclosures. Unlike nearly all other ARM SoCs Actions Semi's S500 is freaking hot!

The very same board still running stress and sysbench but with enough airflow and standing upright looks now like this a few minutes later:

SoC: 97°C, PMU: 82.4°C, clock 1104 MHz
SoC: 98°C, PMU: 82.0°C, clock 1104 MHz
SoC: 99°C, PMU: 82.2°C, clock 1104 MHz
SoC: 99°C, PMU: 82.4°C, clock 1104 MHz
SoC: 98°C, PMU: 82.2°C, clock 1104 MHz
SoC: 97°C, PMU: 82.2°C, clock 1104 MHz
SoC: 97°C, PMU: 82.4°C, clock 1104 MHz
SoC: 97°C, PMU: 81.8°C, clock 1104 MHz
SoC: 98°C, PMU: 81.8°C, clock 1104 MHz
SoC: 98°C, PMU: 82.2°C, clock 1104 MHz
SoC: 99°C, PMU: 82.2°C, clock 1104 MHz
SoC: 97°C, PMU: 82.2°C, clock 1104 MHz
SoC: 98°C, PMU: 82.2°C, clock 1104 MHz
SoC: 98°C, PMU: 82.2°C, clock 1104 MHz
SoC: 98°C, PMU: 81.8°C, clock 1104 MHz
SoC: 98°C, PMU: 82.2°C, clock 1104 MHz
SoC: 97°C, PMU: 82.0°C, clock 1104 MHz
SoC: 98°C, PMU: 81.6°C, clock 1104 MHz
SoC: 98°C, PMU: 82.0°C, clock 1104 MHz
SoC: 97°C, PMU: 82.0°C, clock 1104 MHz
SoC: 97°C, PMU: 82.0°C, clock 1104 MHz
SoC: 97°C, PMU: 81.8°C, clock 1104 MHz
SoC: 97°C, PMU: 81.8°C, clock 1104 MHz
SoC: 98°C, PMU: 82.0°C, clock 1104 MHz

Copy the Code

Airflow and optionally the use of a fan should be the most important design criteria when building enclosures around Actions Semi's little stoves...

tkaiser · Post time Oct 12, 2015 04:42

To test 'worst case' conditions appropriately one has to set the system under full load for at least 5 minutes, better are 10-15 minutes. I adjusted my monitoring setup and tested again starting with an idle system (60°C SoC, 52.5°C PMU) using again both stress and sysbench. It took more than 12 minutes until the maximum (94°C SoC, 76.3°C PMU) was reached. Lower values than before since I managed to get even more airflow (stack-effect):

10:14:46: SoC: 60°C, PMU: 52.4°C, clock 408 MHz
10:14:56: SoC: 68°C, PMU: 55.5°C, clock 1104 MHz
10:15:06: SoC: 72°C, PMU: 58.4°C, clock 1104 MHz
10:15:17: SoC: 73°C, PMU: 60.3°C, clock 1104 MHz
10:15:28: SoC: 75°C, PMU: 62.3°C, clock 1104 MHz
10:15:39: SoC: 78°C, PMU: 63.3°C, clock 1104 MHz
10:15:49: SoC: 78°C, PMU: 64.4°C, clock 1104 MHz
10:15:59: SoC: 80°C, PMU: 65.6°C, clock 1104 MHz
10:16:09: SoC: 81°C, PMU: 66.6°C, clock 1104 MHz
10:16:20: SoC: 82°C, PMU: 67.0°C, clock 1104 MHz
10:16:30: SoC: 82°C, PMU: 67.4°C, clock 1104 MHz
10:16:41: SoC: 84°C, PMU: 68.7°C, clock 1104 MHz
10:16:53: SoC: 85°C, PMU: 68.7°C, clock 1104 MHz
10:17:03: SoC: 84°C, PMU: 69.3°C, clock 1104 MHz
10:17:13: SoC: 85°C, PMU: 69.7°C, clock 1104 MHz
10:17:23: SoC: 86°C, PMU: 70.1°C, clock 1104 MHz
10:17:33: SoC: 86°C, PMU: 70.7°C, clock 1104 MHz
10:17:44: SoC: 87°C, PMU: 71.1°C, clock 1104 MHz
10:17:55: SoC: 87°C, PMU: 71.1°C, clock 1104 MHz
10:18:05: SoC: 88°C, PMU: 71.7°C, clock 1104 MHz
10:18:17: SoC: 87°C, PMU: 71.8°C, clock 1104 MHz
10:18:29: SoC: 88°C, PMU: 72.0°C, clock 1104 MHz
10:18:41: SoC: 88°C, PMU: 72.4°C, clock 1104 MHz
10:18:52: SoC: 88°C, PMU: 72.4°C, clock 1104 MHz
10:19:02: SoC: 89°C, PMU: 73.0°C, clock 1104 MHz
10:19:12: SoC: 88°C, PMU: 72.8°C, clock 1104 MHz
10:19:22: SoC: 89°C, PMU: 73.0°C, clock 1104 MHz
10:19:33: SoC: 90°C, PMU: 73.4°C, clock 1104 MHz
10:19:43: SoC: 89°C, PMU: 73.6°C, clock 1104 MHz
10:19:54: SoC: 90°C, PMU: 74.0°C, clock 1104 MHz
10:20:07: SoC: 90°C, PMU: 74.0°C, clock 1104 MHz
10:20:17: SoC: 89°C, PMU: 73.8°C, clock 1104 MHz
10:20:27: SoC: 90°C, PMU: 73.8°C, clock 1104 MHz
10:20:38: SoC: 90°C, PMU: 74.2°C, clock 1104 MHz
10:20:48: SoC: 90°C, PMU: 74.2°C, clock 1104 MHz
10:20:59: SoC: 90°C, PMU: 74.2°C, clock 1104 MHz
10:21:11: SoC: 91°C, PMU: 74.6°C, clock 1104 MHz
10:21:22: SoC: 90°C, PMU: 74.6°C, clock 1104 MHz
10:21:33: SoC: 90°C, PMU: 74.8°C, clock 1104 MHz
10:21:44: SoC: 91°C, PMU: 74.6°C, clock 1104 MHz
10:21:54: SoC: 92°C, PMU: 75.0°C, clock 1104 MHz
10:22:04: SoC: 91°C, PMU: 75.2°C, clock 1104 MHz
10:22:14: SoC: 91°C, PMU: 74.8°C, clock 1104 MHz
10:22:25: SoC: 92°C, PMU: 75.2°C, clock 1104 MHz
10:22:35: SoC: 92°C, PMU: 75.0°C, clock 1104 MHz
10:22:45: SoC: 92°C, PMU: 75.2°C, clock 1104 MHz
10:22:55: SoC: 92°C, PMU: 75.4°C, clock 1104 MHz
10:23:06: SoC: 91°C, PMU: 75.4°C, clock 1104 MHz
10:23:17: SoC: 92°C, PMU: 75.4°C, clock 1104 MHz
10:23:28: SoC: 92°C, PMU: 75.5°C, clock 1104 MHz
10:23:39: SoC: 92°C, PMU: 75.5°C, clock 1104 MHz
10:23:49: SoC: 91°C, PMU: 75.5°C, clock 1104 MHz
10:23:59: SoC: 92°C, PMU: 75.9°C, clock 1104 MHz
10:24:10: SoC: 92°C, PMU: 75.7°C, clock 1104 MHz
10:24:20: SoC: 92°C, PMU: 75.9°C, clock 1104 MHz
10:24:31: SoC: 92°C, PMU: 75.9°C, clock 1104 MHz
10:24:43: SoC: 91°C, PMU: 75.7°C, clock 1104 MHz
10:24:53: SoC: 92°C, PMU: 75.7°C, clock 1104 MHz
10:25:03: SoC: 92°C, PMU: 75.5°C, clock 1104 MHz
10:25:13: SoC: 91°C, PMU: 75.9°C, clock 1104 MHz
10:25:23: SoC: 92°C, PMU: 75.9°C, clock 1104 MHz
10:25:34: SoC: 92°C, PMU: 76.1°C, clock 1104 MHz
10:25:45: SoC: 92°C, PMU: 75.9°C, clock 1104 MHz
10:25:55: SoC: 92°C, PMU: 76.1°C, clock 1104 MHz
10:26:05: SoC: 91°C, PMU: 76.3°C, clock 1104 MHz
10:26:15: SoC: 92°C, PMU: 75.9°C, clock 1104 MHz
10:26:25: SoC: 92°C, PMU: 76.3°C, clock 1104 MHz
10:26:36: SoC: 94°C, PMU: 76.3°C, clock 1104 MHz
10:26:46: SoC: 92°C, PMU: 76.3°C, clock 1104 MHz
10:26:57: SoC: 92°C, PMU: 76.3°C, clock 1104 MHz
10:27:07: SoC: 93°C, PMU: 76.3°C, clock 1104 MHz
10:27:18: SoC: 93°C, PMU: 76.3°C, clock 1104 MHz
10:27:28: SoC: 92°C, PMU: 76.1°C, clock 1104 MHz
10:27:38: SoC: 92°C, PMU: 76.3°C, clock 1104 MHz

Copy the Code

destroyedlolo · Post time Oct 12, 2015 07:35

HHHooooo, such temperature is DANGEROUS for the CPU.
I'm not sure the board can survive for long time without heatsink ... potentially voiding the installation of any hat on this board

(It's a quite common situation on 4core ARM unfortunately).

tkaiser · Post time Oct 12, 2015 07:55

destroyedlolo replied at Mon Oct 12, 2015 07:35
HHHooooo, such temperature is DANGEROUS for the CPU.
I'm not sure the board can survive for long tim ...

I don't know whether that's dangerous or not. Such temperatures might be quite normal. Let's look what's in S500's manual... surprise: Nothing. We might ask LeMaker but they won't answer as usual.
And no, that's not quite normal for quad core ARM SoCs. When I do tests I always monitor things like temperatures and clock speeds since if you don't do this you get results you wouldn't expect when throttling jumps in:

These are the temperatues the 4 thread sysbench run produced on 4 quad core SoCs:

LeMaker Guitar 1.3 GHz: 75.5°C (heatsink applied)LeMaker Guitar 1.1 GHz: 79°C (no heatsink back then)Raspberry Pi 2 1.0 GHz: 56°C (no heatsink)ODROID-C1+ 1.7 GHz: 62.5°C (standard heatsink applied)Wandboard Quad 1.0 GHz: 42°C (standard heatsink applied)

The problem is that the S500 reaches its maximum temp after minimum 10 minutes full load. So using benchmarks that do not run long enough you don't get the real picture whether a device can be clocked at its advertised top speed longer than just a few minutes without dropping performance due to throttling (and unfortunately this is the case with Guitar @ 1.3A -- without a fan performance gets either worse than with 1.1 GHz or when throttling decides to fail you might burn your SoC)

At the moment I would forget about the marketing phrase 'up to 1.3 GHz' since it might only work with load peaks but will get worse if you need full CPU load over longer periods of time. Unless LeMaker or Actions Semi provide a proof that the S500 withstands weeks under full load with 1.3 GHz I won't use this setting. But since using 1.1 GHz seems also too hot if the device is under heavy load longer than 10 minutes... I don't know what to say regarding the S500. I don't like the idea that a device is advertised as being pretty fast since it's able to be clocked at 1.x GHz and I myself have then take care to lower cpufreq settings since otherwise I would've to use a fan to cool it down if software starts to amok for whatever reasons.

tkaiser · Post time Oct 12, 2015 07:57

WTF?! The forum software also sucks! Destroyed list formatting and no way to edit the post. Why the hell LeMaker is constantly decreasing the quality of this forum?

Maybe it's the best idea to stop writing here now that LeMaker gave up Banana Pi/Pro and start completely over at armbian.com!

Login

LCD 7" and LM Guitar case BananaHome

[Other projects] LCD 7" and LM Guitar case BananaHome