Purethermal2 Lepton 3.5 RPi 3 - Loss of connection after 30 seconds to a few minutes

[Erik Archibald]

Hi Everyone,

I have a problem where I keep losing contact with my Lepton3.5/PureThermal2 on my Raspberry Pi 3 B+.  It cannot stay connected for more than a minute or so when I run uvc-radiometry.py.  It ends up giving me a queue empty error.

I followed the instructions from https://github.com/groupgets/purethermal1-uvc-capture and getthermal for setup onto a clean Raspbian setup from NOOBS.   I had problems reliably connecting and maintaining a connection on my Raspberry 2 as well.  I am running Python 2.7.13 and OpenCV 2.4.9.1

Does anyone know how to fix this? 

Thanks,

Erik Archibald

[Andrew V. Jones]

Hi Erik,

Did you have any luck with this?

Just so you are aware, I also experienced this issue with a 3B+ and a
3.5.

Using (a modified version of) uvc-radiometry.py, I was able to
consistently capture images from the Lepton 3.5. However, if I connected
another device to the 3B+ (in this instance, a Logitech webcam), then
the Lepton would become unreliable with respect to the queue being
empty.

Just so you are aware, the uvc-radiometry.py is “asynchronously”
filling up a buffer of a certain length (2 is the default). If this gets
filled, then the later frames are ignored. Equally, if you read from it
quicker than it can get filled, then you get the empty queue.

In my usage (and when I had the other webcam connected), once it “got
upset” and stopped receiving frames, it would never work again until
the script was restarted. On occasion, I think this even required a full
restart of the Pi.

I actually “gave up” using a Raspberry Pi for the “dual camera”
scenario because it felt very much like it as a Raspbian (or potentially
USB 2.0 or ARM-related) issue. Using _exactly_ the same Python code on a
commodity PC does not have the issue where the Lepton “gives up”.

However, I have not experienced the issue you mention when _only_ using
a Lepton and nothing else on the USB bus (except ethernet, which goes
via USB) on the Raspberry Pi — I have been consistently capturing
images every 20 seconds for > 2 weeks without issue.

Are you experiencing this issue when _only_ using the Lepton?

Cheers,

Andrew

> --
> You received this message because you are subscribed to the Google
> Groups "Flir Lepton" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to flir-lepton...@googlegroups.com.
> Visit this group at https://groups.google.com/group/flir-lepton.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/flir-lepton/2c2823b1-9646-4086-8515-599e3e3e335d%40googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.

[Kurt Kiefer]

This is interesting guys. The other component I would cast doubt on here is libuvc. I've been meaning to take a look back upstream to see if we still need to be maintaining the fork, which at this point is pretty old.

Have either of you tried to use GetThermal on the RPI3 B+ in a long-term test? It still uses libuvc, but the data acquisition is structured in a different way, so we can at least cut the problem there. You'll have to build it from source, but that should be pretty straightforward now that Raspbian has a modern QT: https://github.com/groupgets/GetThermal/wiki/Building-for-Raspberry-Pi

Kurt

[Andrew V. Jones]

Hi Kurt,

As I say, I’ve found uvc-radiometry.py + libuvc.so to be extremely
stable (ignoring any Lepton “hiccups”). I have multiple weeks’
worth of data (capturing every 30 seconds) that has worked fine.

However, if I try to capture from the Lepton + another webcam (using
OpenCV), the Lepton just dies (within an hour or so, sometimes a lot
quicker). Using this setup (i.e., exactly the same code!) with Ubuntu on
x86 is working fine. I blame either ARM, the USB on the Pi or Raspbian.
I found a number of issues on Github related to image acquisition issues
with Raspbian (I can try and find these again if this is interesting).

I did actually hack GetThermal to try to dump frames to jpg/png (rather
than creating video), but I just got frustrated with Qt (potentially due
to the conversion to/from a Qt video frame to an image frame) and stuck
with using Python.

Cheers,

Andrew

>>> an email to flir-lepton...@googlegroups.com <javascript:>.

>>> Visit this group at https://groups.google.com/group/flir-lepton.
>>> To view this discussion on the web visit
>>>
>> https://groups.google.com/d/msgid/flir-lepton/2c2823b1-9646-4086-8515-599e3e3e335d%40googlegroups.com.
>>
>>> For more options, visit https://groups.google.com/d/optout.
>>
>
> --
> You received this message because you are subscribed to the Google
> Groups "Flir Lepton" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to flir-lepton...@googlegroups.com.
> Visit this group at https://groups.google.com/group/flir-lepton.
> To view this discussion on the web visit

> https://groups.google.com/d/msgid/flir-lepton/3199f45f-b75e-40ca-a79c-1cb8adb2e896%40googlegroups.com.

[Andrew V. Jones]

I should add: I found it stable, _on a Raspberry Pi 3 B+_, with
weeks’s worth of data.

It only became unstable when I was capturing from two devices at the
same time. This led me to suspect (with no hard facts!) the USB on the
Raspberry Pi was “choking” and unable to do the throughput of both
cameras.

With only PureThermal 2 connected to the Pi (and serving a simple
webpage for the images), it is working really well!

Cheers,

Andrew

[AB Peter]

Hi.

I am running PT2, Raspberry Pi 3B and Lepton 3.5. This is my first run with Raspberry Pi so i am quite the noob at this point. I got the opencv-capture.py example working after a while. However, when I tried the the uvc-radiometry.py i got stuck. I tried to get the "modified version of libuvc, as described, but obviously I', not doing it right. I cloned the https://github.com/groupgets/libuvc (but to the python folder, not shure if that is right), then:

cd libuvc

mkdir build

cd build

cmake ..

 

and get this in the terminal window:

pi@raspberrypi:~/Downloads/purethermal1-uvc-capture/python/libuvc/build $ cmake ..

-- The C compiler identification is GNU 6.3.0

-- The CXX compiler identification is GNU 6.3.0

-- Check for working C compiler: /usr/bin/cc

-- Check for working C compiler: /usr/bin/cc -- works

-- Detecting C compiler ABI info

-- Detecting C compiler ABI info - done

-- Detecting C compile features

-- Detecting C compile features - done

-- Check for working CXX compiler: /usr/bin/c++

-- Check for working CXX compiler: /usr/bin/c++ -- works

-- Detecting CXX compiler ABI info

-- Detecting CXX compiler ABI info - done

-- Detecting CXX compile features

-- Detecting CXX compile features - done

-- No build type selected, default to Release

-- No target type selected, default to shared library

CMake Warning at CMakeLists.txt:58 (message):

  JPEG not found.  libuvc will not support JPEG decoding.

CMake Error: The following variables are used in this project, but they are set to NOTFOUND.

Please set them or make sure they are set and tested correctly in the CMake files:

LIBUSB_INCLUDE_DIR

   used as include directory in directory /home/pi/Downloads/purethermal1-uvc-capture/python/libuvc

   used as include directory in directory /home/pi/Downloads/purethermal1-uvc-capture/python/libuvc

   used as include directory in directory /home/pi/Downloads/purethermal1-uvc-capture/python/libuvc

   used as include directory in directory /home/pi/Downloads/purethermal1-uvc-capture/python/libuvc

   used as include directory in directory /home/pi/Downloads/purethermal1-uvc-capture/python/libuvc

   used as include directory in directory /home/pi/Downloads/purethermal1-uvc-capture/python/libuvc

   used as include directory in directory /home/pi/Downloads/purethermal1-uvc-capture/python/libuvc

   used as include directory in directory /home/pi/Downloads/purethermal1-uvc-capture/python/libuvc

LIBUSB_LIBRARY_NAMES

    linked by target "uvc" in directory /home/pi/Downloads/purethermal1-uvc-capture/python/libuvc

-- Configuring incomplete, errors occurred!

See also "/home/pi/Downloads/purethermal1-uvc-capture/python/libuvc/build/CMakeFiles/CMakeOutput.log".

Where did i go wrong? :P 
Sorry if its a bit dumb this, but I'm new to Pi so... :P

Thanks

Arne

[Andrew Jones]

Hi Arne,

Try installing (via `apt-get install`):

libusb-1.0-0-dev

And:

libjpeg-dev

And then repeat the steps below.

Cheers,

Andrew

--

You received this message because you are subscribed to the Google Groups "Flir Lepton" group.
To unsubscribe from this group and stop receiving emails from it, send an email to flir-lepton...@googlegroups.com.
Visit this group at https://groups.google.com/group/flir-lepton.

To view this discussion on the web visit https://groups.google.com/d/msgid/flir-lepton/b98acc08-32c7-4a48-b0f9-b6129d91a5aa%40googlegroups.com.

[Pure Engineering]

I would try a couple things out. 

1. on a desktop verify you can run longer to rule out a bad board or camera

2. on the pi try using a better power source 1.5A+, although the lepon power is lowish, a usb power glitch can still mess things up. 

To unsubscribe from this group and stop receiving emails from it, send an email to flir-lepton+unsubscribe@googlegroups.com.

Visit this group at https://groups.google.com/group/flir-lepton.
To view this discussion on the web visit https://groups.google.com/d/msgid/flir-lepton/b98acc08-32c7-4a48-b0f9-b6129d91a5aa%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

You received this message because you are subscribed to the Google Groups "Flir Lepton" group.

To unsubscribe from this group and stop receiving emails from it, send an email to flir-lepton+unsubscribe@googlegroups.com.

Visit this group at https://groups.google.com/group/flir-lepton.

To view this discussion on the web visit https://groups.google.com/d/msgid/flir-lepton/A71BE8BF-711E-42AE-BB80-2AFFFA2E01E9%40gmail.com.

For more options, visit https://groups.google.com/d/optout.

--

Sashi Ono

Principal Embedded Systems Engineer

(805) 699-6690

www.pureengineering.com 

[AB Peter]

Thanks for the quick reply guys!! :)


As Andrew suggested I tried:

sudo apt-get install libusb-1.0-0-dev

sudo apt-get instal libjpeg-dev

The "libusb-1.0-0-dev" seemed to fix the "libusb error" so the "cmake .." command ran through :D Still the JPEG part was not found though. 

Then i ran the "make && sudo make install" and went to the python directory and tried to run both the "./uvc-radiometry.py" and "./uvc-deviceinfo.py". Result:

pi@raspberrypi:~/Downloads/purethermal1-uvc-capture/python $ ./uvc-radiometry.py

Error: could not find libuvc!

Tried to reinstall the whole Pi and start over, just to for the fun (It was Saturday night you see).

All commands i ran since reinstalling noobs was (in order):

~$ sudo apt-get purge wolfram-engine

~$ sudo apt-get autoremove

~$ sudo apt-get update && sudo apt-get upgrade

~$ sudo apt-get install cmake

~$ sudo apt-get install xrdp

~$ sudo apt-get install libusb-1.0-0-dev

~$ sudo apt-get install libjpeg-dev

~/Downloads $ git clone https://github.com/groupgets/libuvc

~/Downloads $ cd libuvc

~/Downloads/libuvc $ mkdir build

~/Downloads/libuvc $ cd build

~/Downloads/libuvc/build $ cmake ..

~/Downloads/libuvc/build $ make && sudo make install

~/Downloads $ git clone https://github.com/groupgets/purethermal1-uvc-capture

~/Downloads/purethermal1-uvc-capture/python $ ./uvc-radiometry.py

Error: could not find libuvc!

Pure engenineering, thanks for you suggestions! I have tried a few different power supplys now. But I don't know if any of them are ideal. I tired a 5v 1a power supply from an earlier Pi model. I have also tried tow different samsung mobile chargers, both with 5v 2a. No change.

As for your suggestion with trying the camera with desktop i tried running it in vlc just as a webcam. That worked just fine.

I'll add a copy of the code (after the "sudo apt-get update && sudo apt-get upgrade"-command), if it is of any use.

Sorry my reply was a bit long :P Thanks again! Arne

[Kurt Kiefer]

The python script is reporting that it can't find libuvc, so it's failing before it even gets to anything hardware related.

Can you try copying the libuvc.so binary that the build created (make sure it's named this, if you have to rename it do that, and copy the actual binary and not the symlink) into the working directory for uvc_radiometry.py?

Kurt

[Andrew Jones]

Hi Arne,

Okay, before you do `make install`, do you have a file called `libuvc.so` in the current build directory?

If so, copy this to the same directory as `uvc-radiometry.py` and then run:

export LD_LIBRARY_PATH=$(pwd)

and then re-try running the script.

Running `make install` isn’t actually necessary, given you just need a local copy of the shared object.

I’ve never done `make install` for my usage, as setting `LD_LIBRARY_PATH` works (and, in my opinion, is preferable).

Let me know how you get on — cheers,

Andrew

--

You received this message because you are subscribed to the Google Groups "Flir Lepton" group.

To unsubscribe from this group and stop receiving emails from it, send an email to flir-lepton...@googlegroups.com.

Visit this group at https://groups.google.com/group/flir-lepton.

To view this discussion on the web visit https://groups.google.com/d/msgid/flir-lepton/294f2861-1434-492a-8b8d-4d283b276463%40googlegroups.com.

[Paul S]

Hi Andrew,

I'm in a similar position to Arne (- (Lepton 2.5 on PT2 into a Rasp Pi 3) I have installed GetThermal and it runs without a problem as a video streamer with radiometric side bar and a central cross hair. I can change palettes, Range and gain state, and seen-based NUC and nothing else. 

When I run uvc-radiometry.py I had the same error as Arne -  Error: could not find libuvc!. 

When I copied the libuvc.so file into the same directory as uvc-radiometry.py (/home/pi/purethermal1-uvc-capture/python) this still produced the same error, but then running export LD_LIBRARY_PATH=$(pwd) gave a different error:

Traceback (most recent call last):

    File "./uvc-radiometry.py", line 6 in <module>

       inport cv2

ImportError: no module named cv2

All I want to do is capture 14bit image so I can process it

interesting when I connect the PT2 to my Macbook pro and run Getthermal I get a different capability  - the full feature set of the GetThermal is available but no radiometric info (side bar or central cross hair), 8bit image streaming is excellent  (1st image) and the same as I get from GetThermal on the Pi (but the Pi gives me radiometric info as well) but the display of the 14bit image is garbage 2nd image

 - I presume (?) because the system is only capable of displaying 8bit  RGB?. Do you know where the radiometric data has gone? 

rgds

Paul

[Andrew V. Jones]

Hi Paul,

You need to install opencv:

apt-get install python-opencv

should do the trick!

Not sure about the radiometric data when using it via a Mac (I think I
also have the same issue); it appears for me via Linux, so I moved on
with my life ...

Cheers,

Andrew

On 18 Jun 2018, at 16:58, Paul S wrote:

> Hi Andrew,
>
> I'm in a similar position to Arne (- (Lepton 2.5 on PT2 into a Rasp Pi
> 3) I
> have installed GetThermal and it runs without a problem as a video
> streamer
> with radiometric side bar and a central cross hair. I can change
> palettes,
> Range and gain state, and seen-based NUC and nothing else.
>
> When I run uvc-radiometry.py I had the same error as Arne - Error:
> could
> not find libuvc!.

> When I copied the *libuvc.so* file into the same directory as *uvc*
> *-radiometry.py* (/home/pi/purethermal1-uvc-capture/python) this still

> produced the same error, but then running export
> LD_LIBRARY_PATH=$(pwd)
> gave a different error:
>

> *Traceback (most recent call last):*
> * File "./uvc-radiometry.py", line 6 in <module>*
> * inport cv2*
> *ImportError: no module named cv2*

>
> All I want to do is capture 14bit image so I can process it
>
> interesting when I connect the PT2 to my Macbook pro and run
> Getthermal I
> get a different capability - the full feature set of the GetThermal
> is
> available but no radiometric info (side bar or central cross hair),
> 8bit
> image streaming is excellent (1st image) and the same as I get from
> GetThermal on the Pi (but the Pi gives me radiometric info as well)
> but the
> display of the 14bit image is garbage 2nd image
>
>
>

>> email to flir-lepton...@googlegroups.com <javascript:>.

>> Visit this group at https://groups.google.com/group/flir-lepton.
>> To view this discussion on the web visit
>> https://groups.google.com/d/msgid/flir-lepton/294f2861-1434-492a-8b8d-4d283b276463%40googlegroups.com

>> <https://groups.google.com/d/msgid/flir-lepton/294f2861-1434-492a-8b8d-4d283b276463%40googlegroups.com?utm_medium=email&utm_source=footer>
>> .

>> For more options, visit https://groups.google.com/d/optout.
>>
>>
>

> --
> You received this message because you are subscribed to the Google
> Groups "Flir Lepton" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to flir-lepton...@googlegroups.com.
> Visit this group at https://groups.google.com/group/flir-lepton.
> To view this discussion on the web visit

> https://groups.google.com/d/msgid/flir-lepton/2f24cbc7-29ee-4a2b-8d03-a006c30fbe9d%40googlegroups.com.

[Paul Sacker]

Thanks Andrew - its now running. Now for 14bit capture…..

rgds

Paul

[Andrew V. Jones]

Sorry! Can’t help there! `uvc-radiometry.py` (with some minor tweaks)
is 100% sufficient for my needs :)

Cheers,

Andrew

>>>> email to flir-lepton...@googlegroups.com <http://googlegroups.com/>

>>>> <javascript:>.
>>>> Visit this group at https://groups.google.com/group/flir-lepton

>>>> <https://groups.google.com/group/flir-lepton>.

>>>> To view this discussion on the web visit
>>>> https://groups.google.com/d/msgid/flir-lepton/294f2861-1434-492a-8b8d-4d283b276463%40googlegroups.com
>>>> <https://groups.google.com/d/msgid/flir-lepton/294f2861-1434-492a-8b8d-4d283b276463%40googlegroups.com>

>>>> <https://groups.google.com/d/msgid/flir-lepton/294f2861-1434-492a-8b8d-4d283b276463%40googlegroups.com?utm_medium=email&utm_source=footer

[AB Peter]

I could not find the libuvc.so in the build directory. Is it suppose to be located somewhere in the build directory i created with the 'mkdir build'?

[Andrew V. Jones]

Is this a screenshot _before_ or _after_ you’ve run `make`?

I would expect there to be at least two files in that directory after
running `make`: libuvc.so and libuvcstatic.a.

Cheers,

Andrew

> <https://lh3.googleusercontent.com/-5eTPBrHNyD0/WyfjnkUIliI/AAAAAAAAAEI/wUJKe6A2I-QRNNUMS5fYHPsGK-vypVTVACLcBGAs/s1600/build.png>

>
> --
> You received this message because you are subscribed to the Google
> Groups "Flir Lepton" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to flir-lepton...@googlegroups.com.
> Visit this group at https://groups.google.com/group/flir-lepton.
> To view this discussion on the web visit

> https://groups.google.com/d/msgid/flir-lepton/0b54eaf6-b7db-4911-a64a-5b22b31eb259%40googlegroups.com.

[AB Peter]

Sorry, forgot the make, oops :P 

Now that i did the 'make' command the file 'libuvc.so' appeared. I copied (right clicked on it in file manager) to the 'python' directory where the './uvc-radiometry.py' is located. Then i did the 'export LD_LIBRARY_PATH=$(pwd)' (when I was in the 'python' directory). Then i ran './uvc-radiometry.py'. The terminal then said 'uvc_open error'.

[Andrew Jones]

Okay, this could be a few things …

What does `sudo lsusb` given you? How about `sudo LD_LIBRARY_PATH=$(pwd) ./uvc-radiometry.py`?

Cheers,

Andrew

On 18 Jun 2018, at 18:34, 'AB Peter' via Flir Lepton wrote:

Sorry, forgot the make, oops :P 

Now that i did the 'make' command the file 'libuvc.so' appeared. I copied (right clicked on it in file manager) to the 'python' directory where the './uvc-radiometry.py' is located. Then i did the 'export LD_LIBRARY_PATH=$(pwd)' (when I was in the 'python' directory). Then i ran './uvc-radiometry.py'. The terminal then said 'uvc_open error'.

--

You received this message because you are subscribed to the Google Groups "Flir Lepton" group.
To unsubscribe from this group and stop receiving emails from it, send an email to flir-lepton...@googlegroups.com.
Visit this group at https://groups.google.com/group/flir-lepton.

To view this discussion on the web visit https://groups.google.com/d/msgid/flir-lepton/b8e5e45a-1af7-4399-b6f1-838f7533c9fe%40googlegroups.com.

[AB Peter]

Then i get:

pi@raspberrypi:~/Downloads/purethermal1-uvc-capture/python $ sudo lsusbBus 001 Device 005: ID 1e4e:0100 Cubeternet WebCam

Bus 001 Device 003: ID 0424:ec00 Standard Microsystems Corp. SMSC9512/9514 Fast Ethernet Adapter

Bus 001 Device 002: ID 0424:9514 Standard Microsystems Corp. SMC9514 Hub

Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub

pi@raspberrypi:~/Downloads/purethermal1-uvc-capture/python $ sudo LD_LIBRARY_PATH=$(pwd) ./uvc-radiometry.py

device opened!

Version gpp: 3.3.26 dsp: 3.3.26

FLIR part #: 500-0771-01

FLIR serial #: '\x9e\xb0$\x00\x00\x00\x00\x00'

format: UYVY

  frame 160x120 @ 9fps

format: Y16 

  frame 160x120 @ 9fps

  frame 160x122 @ 9fps

format: Y8  

  frame 160x120 @ 9fps

format: RGBP

  frame 160x120 @ 9fps

format: }�6

  frame 160x120 @ 9fps

Estimated / selected altsetting bandwith : 18 / 962. 

No protocol specified

(Lepton Radiometry:3354): Gtk-WARNING **: cannot open display: :10.0

Regards
Arne

[Andrew Jones]

Okay, so, do you have X11 forwarding? How are you connecting to the Pi? From Windows? What “ssh client” are you using?

To resolve the needing `sudo` on `uvc-radiometry.py`, you can set the udev rules as listed here:

https://github.com/groupgets/GetThermal/wiki/Building-for-Raspberry-Pi

(Search the page for “1e4e”).

--

You received this message because you are subscribed to the Google Groups "Flir Lepton" group.
To unsubscribe from this group and stop receiving emails from it, send an email to flir-lepton...@googlegroups.com.
Visit this group at https://groups.google.com/group/flir-lepton.

To view this discussion on the web visit https://groups.google.com/d/msgid/flir-lepton/a43f9907-461a-43f7-8e46-b5ac0e013e27%40googlegroups.com.

[AB Peter]

I am using remote desktop in Windows. I have used Putty earlier, but didn't think id have to use it since I was just doing the commands over remote desktop. Since i don't know what X11 forwarding is, i suppose I don't have it...?  

[Andrew Jones]

Sorry, how are you connecting the Pi? Via Putty?

You might need to follow a guide like this:

http://www.geo.mtu.edu/geoschem/docs/putty_install.html

I’m not sure your issues are PureThermal related now though (which is good!).

On 18 Jun 2018, at 21:19, 'AB Peter' via Flir Lepton wrote:

I am using remote desktop in Windows. I have used Putty earlier, but didn't think id have to use it since I was just doing the commands over remote desktop. Since i don't know what X11 forwarding is, i suppose I don't have it...?  

--

You received this message because you are subscribed to the Google Groups "Flir Lepton" group.
To unsubscribe from this group and stop receiving emails from it, send an email to flir-lepton...@googlegroups.com.
Visit this group at https://groups.google.com/group/flir-lepton.

To view this discussion on the web visit https://groups.google.com/d/msgid/flir-lepton/a9eacf17-838b-460e-8317-0a6960c2d9b2%40googlegroups.com.

[Erik Archibald]

Thanks for the reply Andrew. My experience with the Raspberry Pi 3 B + seems to confirm the hypothesis that is the usb controller on the Pi.  I found that that the connection was unreliable when I was running the pure thermal along with having the wired network connection going.  The wired connection apparently is also on the usb controller chip.  The other thing that maybe sometimes was an issue for me was not having the lepton fully seated.

-Erik

[Andrew V. Jones]

How much traffic were you pushing through ethernet?

I hadn’t noticed it “crap out” when just doing ethernet, but my
ethernet stuff is _extremely_ low bandwidth (mainly console-based SSH).

>>> an email to flir-lepton...@googlegroups.com <javascript:>.

>>> Visit this group at https://groups.google.com/group/flir-lepton.
>>> To view this discussion on the web visit
>>>
>> https://groups.google.com/d/msgid/flir-lepton/2c2823b1-9646-4086-8515-599e3e3e335d%40googlegroups.com.
>>
>>> For more options, visit https://groups.google.com/d/optout.
>>
>
> --
> You received this message because you are subscribed to the Google
> Groups "Flir Lepton" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to flir-lepton...@googlegroups.com.
> Visit this group at https://groups.google.com/group/flir-lepton.
> To view this discussion on the web visit

> https://groups.google.com/d/msgid/flir-lepton/4ce0b321-75f4-413f-aea5-2c5ce67d4711%40googlegroups.com.

[David nunya]

With the newest RPi, I highly recommend using a 3 amp power supply.  With the last gen RPi3, at least a 2.5 amp power supply.

<<<snip>>>

[Angus Fraser]

There is a pad on the back of the breakout board that carries a sync signal. I have found using that to drive the spi sync is just bulletproof. I cant claim the idea as mine it is actually from here . It was amazing when from loosing sync often to never loosing it.

Gus

[Tomas K]

Hello

we have exactly the same issue. Is there any solution for it?

We tried two different Pi's . Raspberry Pi 3 B and Pi 3 B+, 2.5A power supply used. There are also connected keyboard and mouse via USB, but the problem appears even when both are disconnected.

The problem appears regardless the used software and library - libusb+libuvc, GetThermal SW, v4l2 lib - all of them have the same problem. The streaming starts correctly, image is displayed, but it gets stuck after some short time (but it seems random). libuvc starts reporting error -9 (Pipe error) on any following communication with the camera. Using v4l2 just stops receiving the frames (querying the buffer returns EAGAIN error in non-blocking mode) and /var/log/kern.log shows the following errors:

kernel: [21418.056754] Transfer to device 15 endpoint 0x1 frame 1099 failed - FIQ timed out. Data may have been lost.
kernel: [21421.932856] Transfer to device 15 endpoint 0x1 frame 879 failed - FIQ timed out. Data may have been lost.
kernel: [21443.020767] Transfer to device 15 endpoint 0x1 frame 1487 failed - FIQ timed out. Data may have been lost.
kernel: [21447.019344] uvcvideo: Non-zero status (-32) in status completion handler.

where "-32" corresponds to pipe error. Since this time, it is even impossible to make the camera disconnect. When stream is stopped and device descriptor closed, the LED continues blinking fastly and camera does not work even after application restart.

However, it is seems to be somehow related to the device load. When streaming with a simple v4l2 application which only prints "frame-center" temperature into the console, it works correctly for an hour. When frame image is being displayed (@9fps) or I start browsing in Chromium (or load the system somehow), the problem appears after a while.

Does anybody knows what to do in such situation? Thank you.

[Pavelas Volginas]

Any progress? I have same issue with rpi3b+ and PT2 wuth lepton 3.5 works random amount of seconds and stops.

[Andrew V. Jones]

What else do you have the USB bus?

I have noticed that the PT2/L3.5 “gets grumpy” if I have too much on
the USB bus on my RPi (e.g., PT2 + a webcam is enough to make things
become very unreliable).

Remember, Ethernet and USB share the same bus on an RPi.

Cheers,

Andrew

> --
> You received this message because you are subscribed to the Google
> Groups "Flir Lepton" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to flir-lepton...@googlegroups.com.

> To view this discussion on the web visit

> https://groups.google.com/d/msgid/flir-lepton/47da15c0-a3ed-4cbc-9180-4d014438086d%40googlegroups.com.

[Pavelas Volginas]

I have bluetooth adapter for my mouse/keyboard and ] I am using LAN port, will try without ethernet, thanks!

> an email to flir-...@googlegroups.com.

[Pavelas Volginas]

Eliminating ethernet and other USB devices does nothing, this is frustrating because Getthermal works perfectly, but not V4l2.