Beaglebone Ai Heating Issues

Hi ipv1 ,

I  looked into this and this is a known issue:

BeagleBone AI - Frequently Asked Questions (FAQ)

They recommend adding a larger heat sink and cooling via fan.   

My investigations showed much the same. My questions are:

1. How was this not caught during testing ? Some minor overheating might be missed but shutting down as described on many of the comments makes it obvious that there is a serious problem.

2. This is happening when you are only running the OS on the ARM Cortex-A15 cores.  What happens to the temperature when you start using some mixture or all of the ARM Cortex-M4s cores,

      DSPs, EVEs and PRUs too?

3. If it needs a bigger heat sink and fan just running the OS why doesn't it HAVE a bigger heat sink and fan (see 2.).

After reading the earlier comments in this discussion and checking for other comments on the web I hopped over to Ebay and bought a couple of 25mm fans. The heat sink on the AI is 25mm and

it looks like there are threaded holes to attach the fan with. I'm not sure how you mount a larger heat sink  other than a adhesive thermal pad or thermal epoxy. There is an adhesive thermal pad

between the heat sink and SoC but I'm not a big fan of thermal pads and in this case I would much rather have a heat sink that uses mount points on the PCB and a wafer thin layer of quality thermal

paste.

It also looks like it could use some serious thermal management on the OS side. I'll give them the benefit of the doubt and hope that somebody is looking into this.

"But, since they knew that people would also want to run it immediately (i.e. without needing additional heat management), there is a small heatsink already mounted, and with threads to attach to a custom solution if required."

The problem is that the small heat sink "to allow running it immediately" does not work. The custom solution is called a fan, and from the ongoing discussions I've read it is required right out of the box for even minimal operation.

Software frequency throttling should allow it to run without a fan and it will be useful later for overall power management but why it was not implemented or implemented properly (I have no idea which is correct) before shipping

product "that people could run immediately" is a very good question. At least it should be a rather quick and easy fix. For now I have a couple of 25mm 10K RPM fans on order (from Ebay) and I will not even think about powering

up my AI until one is installed. At 10K RPM it will probably scream like a banshee, luckily it will be a very small banshee.

I think ultimately that the BB-AI would operate in two (or more) modes, e.g (1) without the fan, any throttling could be more aggressive, or would run at a slower rate., and (2) with the fan cape (or if configured otherwise) then the BB-AI would operate in a higher performance, or a sustained performance mode.

That seems plausible, especially since some of the suggestions discuss upgrading the software image, to allow people to run the BB-AI without any fan or additional heatsink. The fan cape has been planned for a long time, so that too gives credence to the thought that they were aware that two modes would be used in software.

gdstew  wrote:

 

.. but why it was not implemented or implemented properly (I have no idea which is correct) before shipping

product "that people could run immediately" is a very good question. At least it should be a rather quick and easy fix.

You're right, it would have been a better experience for users if it was in the shipped image. Perhaps it just missed some integrating and testing window for their release timeframe. Presumably even without the fix, the BB-AI won't self-destruct, so the software release didn't need to match the hardware release timeframe. Again I'm just speculating (but it seems semi-plausible).

The first thing I do with a BeagleBone Black is to download the latest image, but then I'm used to that. Maybe with the BB-AI, there could be an assumption (true or not I don't know, not checked to see any release notes) that the pre-installed image is still very close to the released image, since it is a new product. I too intend to install a fan, but maybe a 40mm one, in the hope that it is quieter.

I've got a couple of 40mm fans but what I found was they have large motor hubs and very small blades which puts the blades too far away from

most of the heat sink for my liking although the SoC is offset from center by quite a bit and that may help. That and it will require some kind of

special mounting bracket or modified prototype cape. I hadn't thought about modifying a prototype cape until I wrote this and I have a couple

of Adafruit capes so I might do that if the 25mm fan doesn't work well enough. I was originally thinking about using tall nylon machine screws

for a temporary solution with the 40mm fan so I could get it up and running fast. The inside of the GPIO headers are a few mm more than

40mm apart and that could used to help wedge the fan in and prevent it from "walking" around too much.

gdstew  wrote:

 

I've got a couple of 40mm fans but what I found was they have large motor hubs and very small blades which puts the blades too far away from

most of the heat sink for my liking

That's a good point.. subconsciously that must have been bothering me too, so I've been thinking of trying to make the area larger (with an aluminium plate). The height of the heatsink that comes attached to the BB-AI clears the height of the USB connector too, so quite a large plate could be glued on top. It might be inefficient though.. I'll try it anyway, and if it doesn't cool sufficiently, I'll switch to the 25mm fan.

The actual location that the blades are pointing to is not very critical. The goal is to keep the air flow going so that the hot air gets evacuated and replaced by cold air. The air volume plays the bigger role here.

> And again, so far this is not even close to running it under full load

Apparently the GUI environment runs by default on the BeagleBone AI, and disabling this helps the temperatures to subside somewhat.

> The trick is to find that Cape Fan.  I have not been able to locate it.

The official one is a couple of months away, so I'm told.

Until then, mayermakes has been designing an add-on board

> luckily it will be a very small banshee.

Seems appropriate for Halloween - Projects and Videos

I was thinking of making one, when board arrives >>>

> without the active cooling, it would throttle more, especially if it detects no cooling is attached

I wish I could get my head around cpufreq a bit more to get a system service package to handle this, I would've perhaps hoped that one existed in debian to manage this for us already but I'm not certain/conviced, I think laptop-utils would/used to do something like this but perhaps not for ARM based controllers?

This seems to be something which other boards manage to do, but I'm not sure of the intricacies of how. If anyone knows I think it'd be pretty useful for this board to work on it at least - or point me in the right direction, dabbling with linux is fun.

> without the active cooling, it would throttle more, especially if it detects no cooling is attached

I wish I could get my head around cpufreq a bit more to get a system service package to handle this, I would've perhaps hoped that one existed in debian to manage this for us already but I'm not certain/conviced, I think laptop-utils would/used to do something like this but perhaps not for ARM based controllers?

This seems to be something which other boards manage to do, but I'm not sure of the intricacies of how. If anyone knows I think it'd be pretty useful for this board to work on it at least - or point me in the right direction, dabbling with linux is fun.

I'm convinced that this is a powerful processor with DSP cores similar to pc processor but smaller version (Didn't got hands on board yet). Surely it will heat up and demands similar active cooling like in pc's/laptops. You can try copper pipes and fan to transfer heat, setup will be bulky but you will be able to use Caps.

Yes, it is basically a tweaked BeagleBoard X15 chip mounted on a smaller board.

Hi Christopher,

It's something new to me too, I'm not really sure how it is done in Linux with ARM.

There's this web page that is for BBB discussing cpufreq, and it mentions how some parameters of the policy can be changed by writing to different devices in a cpufreq/ondemand path. But I've never seen a good explanation of how to create a custom policy, because it would be nice to have a couple of defined ones, and swap between them if heatsink/fans are attached or not.

It's an important topic, because some may prefer no throttling, and just run the device slower, whereas others may prefer to have more dynamic behaviour.

That's right, it's sparked a memory, basically a configuration of the governor needs to be setup in the configuration files, and then the governor profiles can be switched between, I think it then calls the necessary scripts depending on the parameters and automagically handles them? probably requires some time with the man pages.

My fallback tends to be to rely on Arch linux documentation because it's cutting edge / up to date, though apparently there's some kernel documentation so it's probably managed there for the BeagleBone

I believe the issue is that they are trying to get a modern OS to support 6+ year old technology.  These problems have been solved by other vendors thus they do not have the same issue.  The NVIDIA Jetson Nano greatly exceeds the capability of the BB AI and they had the mindset to place a good size heat sink on it and thus far I have not seen a need to add a fan even when running on a JetBot and using vision navigation.  The Google Coral at least comes with heatsink and fan attached since they understood the issues with heat on the chip.  Hack, the processing power in the Nintendo Switch Lite is amazing but is packaged in a nice slim package and no fan.

But, really, if this board is heating up when jut running idle, what the heck is going happen when one uses it for AI or ML as it is advertised for?  You need to either keep a fire extinguisher handy or place the board in a chryo chamber to keep it from burning down the house.

I was looking to showcase this Board at the local Maker Faire coming up, but I think not and wish not to go the route of the BB Blue last year where I shown it running with the EduMIP and folks bought one for themselves from what I had shown and later I see them and find all boards are dead just like mine.  Me thinks this board will experience a similar fate in a short time.

shabaz & cstanton I am going to tackle this problem over the weekend. If I get good results, I will publish on Monday. Fingers crossed...

Surely you can change its frequency and governer to low and power saver respectively, but beaware in low power mode it might shut down few peripherals. Also it surely needs active cooling source, read thermal considerations for AM57xx http://www.ti.com/lit/an/sprac53a/sprac53a.pdf 
Better get proper cooling method to test it at full efficiency.
Caution don't let it get too hot, there are chances of EOS damage.  >> http://www.ti.com/lit/an/sprace7/sprace7.pdf

For any upcoming show maybe it could be better to use something more mature.. most of the Pi releases had software teething issues at launch, the software took a while to be in a mature state.

NVIDIA and BB-AI's chips are playing in different markets I think.. the chip in the BB-AI (AM572x) is designed for automotive applications, so will likely have the guaranteed X amount of years availability, and comes in an industrial temp version too, and might use more traditional methods of processing (multiple CPU cores and some general purpose accelerators in H/W, and some accelerators for image recognition features). NVIDIA's chip may not be available in many years time for use in customer designs (this is a guess though) if they keep churning out newer chips and want to sell those instead. NVIDIA has fantastic technology for doing state-of-the-art computing though.

The BeagleBone Black's AM3358 had the longevity advantage too, because that chip was (is) in production or at least offered for sale, for 10 years (might be 15, I can't recall).

The AM3358 was easy-ish to prototype with, to create custom designs (I tried to create my own too, using the OSD3358 which integrates that chip, PMIC, Flash and RAM), but the project went away - lots of time was wasted with Intel's Edison, we had a working prototype, and it went end-of-sale : (

Prototyping a custom board will be much harder with the BB-AI's chip, but there should be modules with that same chip I hope.

I believe development kits are basically to test processor functionality, to learn about new things - How it performs, its features usability, How to code them, most importantly to find mistakes/errors before using it for applications.. But the main thing is, they are made to be used for specific applications. They went good with previous boards similar to other board designer, but this is one mischievous processor. They just need to design the board properly.
This is just board designing gone bad.

I don't know how useful it is, but there's some Sitara Bootcamp, with Power Management training ppt : )

https://training.ti.com/sitara-arm-processors-boot-camp-linux-power-management-overview-and-hands

It's for AX335x, but the concepts mentioned in there are supported in the BB-AI/X15 (AM572x) chip too.

There's some Linux power management training link here:

Sitara Linux Training: Power Management - Texas Instruments Wiki

But it's using TI's Linux image, I don't know if the same drivers are integrated in the normal BB-AI Linux image yet, and if not, what the plan for that is.