Difficulties Connecting

[tcottle]

Half of the time when I try to connect, the blue light stops flashing and the IDE hangs. The other time is get the picture below



In between lockups, I exit the IDE and unplug the module

[kwagyeman]

Hi Sorry about these issues.

I can't really tell what the problem would be from the error info you've given me. That said, something to try would be to reflash the device. See this: https://github.com/openmv/openmv/wiki/U ... r-firmware

Let me know if that helps.

[deshipu]

From the random nature of the problem, I would infer that it may be something wrong with the USB connection. I would try to use a different USB port, different USB cable, etc. Hope that helps.

[kwagyeman]

I'll be posting a Kick Starter update with a link to the preview version of the new IDE for windows. I don't have all the bells and whistles complete yet. But, the serial connection is good and it's more or less usable. If the new IDE gives you problems then its either the firmware wasn't flashed just right at the factory or the USB cable is bad. A lot folks have had bad USB cables.

[tcottle]

Thanks. I'm running Win7 64 bit

Here is what I tried last night (without success)
1. Reflash firmware
2. Erase and reflash booloader (from IDE). Reflash firmware
3. Change USB cables
4. Change USB ports
5. Reload IDE

When connecting the USB cable, the led flashes green a couple of times, white once and then drops into a repeating blue flash pattern. It mounts as USB drive and I can see the files. The blue flash stops when I press the connect button. Out of many many attempts I have only loaded a python script (one of the examples) successfully once or twice

When I first connected the module identified as Teensy Serial Port. I uninstalled and deleted the Teensy driver before I installed your serial driver.

Will try on a different computer this weekend

T

[kwagyeman]

The Cam behavior seems fine. Mmm, okay, well, when the IDE preview comes out give it a try.

[tcottle]

So I've downloaded and installed the 1.0.0 IDE. I've verified that the OpenMV USB serial port is present (COM17)

I'd didn't see any way to set the serial port from the IDE. When I hit the connect icon i get the following error: Timeout while getting firmware version! The blue LED does stop blinking though

Does the IDE have a diagnostic mode that shows/captures the communication between the PC and the python board?

[kwagyeman]

So, if you got a timeout error then that means the IDE was able to open a serial port on the board. Then it sent the firmware version command and got no response. So, the firmware on the cam is likely the problem.

Um, I'll post in a bit how to update the firmware if I haven't already.

[kwagyeman]

Oh, the Ide automatically finds the correct serial port for the cam. It will only ask if there are two cams connected. You don't have to worry about serial ports anymore with the new Ide. It filters serial ports by PID and VID so that only cam ports are considered.

[kwagyeman]

Okay, just try updating the firmware using this:

https://github.com/openmv/openmv/wiki/U ... r-firmware

Make sure to click the correct box when updating the firmware.

>> To load the firmware image onto your OpenMV Cam open the DfuSe Demo application and click the Choose... button in the Upgrade or Verify Action menu box (do not click the Choose... button in the Upload Action menu box). Then choose the above firmware image dfu file to select it to be loaded onto your OpenMV Cam. Finally, click the Upgrade button once you are ready to start the firmware update process. After you've finished updating the firmware you can optionally verify it if you like by clicking the Verify button.

As for the firmware to use, please see here: https://github.com/openmv/openmv/blob/m ... openmv.dfu.

Please don't use the new IDE to connect afterward. Just use the DFU tool to re-flash. Then use the new IDE to connect. If it doesn't work after this then I'll send you a new one.

[tcottle]

Upgraded and verified the firmware.
Removed the jumper and power cycled the module
I get the same error with the new IDE
It looks like I have a bum module ...

[kwagyeman]

Alright, email us with your info and I'll send out another as a replacement next week.

[Kermit666]

Hi,
I just connect my OPEN VM cam on my Comp., I have the following observations :
When I connect the card on few USB ports I obtain few green blinks, then one red/purple blink and then blue blinks ...it seems normal
I can see a new disk (31,0 Ko) with 3 files inside ... it seems normal
when I launch OPEN MW IDE (1.1.0) : no OPEN MV Cams found !
I try almost everything to connect OPEN VM with IDE (using the Connect button) but I didn't find the solution
Thanks

[kwagyeman]

Hmm, the IDE filters out USB ports by the VID/PID of the Cam. Okay, try this:

Click the connect button while the Cam is not connected to the PC. The IDE will then ask you if you have a bricked OpenMV Cam. Say, yes, then select if you want to erase the flash drive or not. Doesn't matter. Then the IDE will wait for you to connect the Cam to your computer. Do that. The boot loader will not overwrite the cam.

After that everything should work fine.

Note that the bootloader on the IDE won't work on a VM. But, the IDE should normally run on a VM.

[tcottle]

Sorry I should have sent an update when the replacement module arrived. It arrived (thanks much) but has the same problem as the first one.

I downloaded and installed the latest IDE and followed your unbricking instructions above. Selected the M4 board. That process completed fine on both modules. I can read the directory and there is a selftest.py files that wasn't there before

On both modules the original fault remains - after I press connect, I get "Timeout while getting firmware version"

I believe the problem is caused by a conflict in your USB serial driver and pjrc.com Teensy USB Serial Driver. With your module attached, windows grabs the Teensy driver ....

[kwagyeman]

Hmm, okay. Might be able to fix this by installing our driver now. Please look into the IDE install folder:

see /share/drivers/openmv/openmv.inf. Right click on that and click install. It's digitally signed now so windows should install it without any issues.

Also, the pycdc driver is there to if you want to install that also.

[kwagyeman]

Note that you need to install the latest GUI from our website.

[tcottle]

Thanks for quick reply. I could not install the .inf file by right clicking. Got the following error "The INF file you selected does not support the method of installation". Same issue with the pycdc file

I did update the serial driver through Device Manager but still have the same problem - TImeout error while getting firmware revision. Also restarted the IDE and disconnected/reconnected the module ...



Notice the title on driver version window (Teensy USB (COM17) Properties) Maybe this is leftover from how I updated the driver ...

[kwagyeman]

Can you try the firmware update process again since the driver is installed? Also, please delete the flash file system.

If the boot-loader works then the camera units are fine. Getting a timeout error on connect is very weird. This only happens if the cam takes over 500 ms to respond. Not sure why that would be the case however.

[tcottle]

Reflashed bootloader.dfu and openmv.dfu using the DfuSe tool. The files came from here https://github.com/openmv/openmv/tree/m ... re/OPENMV2
Could not connect
Reflashed using the bricked module process. I can occasionally connect but cannot run any examples.

On the times that I can connect here is the error message from Serial Terminal

[kwagyeman]

That's not an error message. It's supposed to print that out. The cam prints that when the IDE connects to it.

I don't know what's going on with the serial driver for you. However, I'm testing with windows 7 right now to get to the bottom of this. On windows 10 we're not seeing any problems because windows install the driver itself.

I can increase the serial timeout, but, that shouldn't make a difference... 500 ms is more than enough time for the cam to connect. Is it possible the USB cable/port is bad? I've been testing with 10 ft USB extension cables over here and I haven't had any issues.

[kwagyeman]

Actively working on this problem. I installed a windows 7 VM and I encountered the same behavior. Can you try out this version of the IDE? http://openmv.io/upload/openmv-ide-wind ... -1.2.0.exe. I made the get firmware timeout much longer.

However, even after making the timeout longer I still saw failures with the get firmware command. I'm thinking of making the IDE try more than once before giving up, this might solve the problem. Also, possibly moving the first get firmware command away from the same time the port was opened.

[tcottle]

Confirming that my modules work OK on Win10. Win10 maps in a generic USB Serial driver and does not use your driver though

[kwagyeman]

Yeah, windows 10 works a lot better. Our driver is actually the same as the Arduino.inf driver. I copied the code for what they did more or less. The problem is likely the interaction our camera firmware and the old operating system driver.

[kwagyeman]

The pyboard driver is the same too. Sigh... I believe there are some real problems in the serial firmware in the OpenMV Cam. But, we need a reproducible issue.

[tcottle]

Downloaded and installed IDE 1.2

If I have the module connected before the IDE starts I always get timeout errors
If I start the IDE, attach the module, wait for the flash drive to mount, then press connect I get a connecting text box and nothing else. If I disconnect the module and reconnect I get a warning message that no cams were found and then it connects
If I start the IDE, attache the module, wait for the device drivers to start (windows make a sound), press connect right before the flash drive mounts (I get an OS prompt when removable storage is attached), it usually connects

In all cases I am unable to run example scripts after it connects

I am out on travel and will not be able to do additional testing before Wednesday

[kwagyeman]

This is on windows 10? Or windows 7?

[tcottle]

Sorry - Win7. Windows 10 seems to be fine (it connects and runs example scripts)

[kwagyeman]

I'm getting some the connect issues your having with a windows 7 VM. I'm adding in code to retry commands right now. I don't think it will help to much. The firmware needs to be able to handle split USB serial packets which I think cause the problem.

[kwagyeman]

Okay, spent all day on this. A new version of the IDE has been uploaded for the above link: http://openmv.io/upload/openmv-ide-wind ... -1.2.0.exe. Let me know if version works better. I'm aiming for another release of the IDE by the end of the week with all these fixes.

I updated the driver and fixed the driver installer. It has perfect behavior except for windows 8 where it displays an info box when done - all else are silent. Also, I tested on windows vista, windows 7, and window 8. All of them are working fine. I added some hard 2 ms delays between serial flush calls for the serial thread in the IDE so that USB packets have to be sent out before another gets built up. Additionally, I also put retries into the IDE so that it will re-execute a command up to 5 times before giving up. Each time it will wait 200 ms. So, a total failure will take the IDE 1 second to see.

Windows XP is completely broken for me. Not even the pyboard works on windows XP. I thinks its because of the version of the OS I have. Its an old service pack 2 disk. It won't let you install drivers by clicking on the inf. You have to go through the add/remove hardware menu. Anyway, the IDE installs fine on windows XP and can even open the serial port. But, no data is transmitted on the USB link.

[Kermit666]

Thank you for your quick answer.

"Hmm, the IDE filters out USB ports by the VID/PID of the Cam. Okay, try this:

Click the connect button while the Cam is not connected to the PC. The IDE will then ask you if you have a bricked OpenMV Cam. Say, yes, then select if you want to erase the flash drive or not. Doesn't matter. Then the IDE will wait for you to connect the Cam to your computer. Do that. The boot loader will not overwrite the cam.

After that everything should work fine.

Note that the bootloader on the IDE won't work on a VM. But, the IDE should normally run on a VM.
Nyamekye,"

So I tried, but it didn't work ... no results.

So, I installed the OpenMV IDE 1.2.0
Now when I connect the Cam I have green blinks, one red/purple blink then 2 blue blinks and the blinks stop.
When I launch OpenMV IDE, it does not detect the Cam, if I click on the connect button : connecting ... and nothing else
If I disconnect the Cam, no result, if I connect the cam while IDE is connecting, if I connect the cam after relaunching IDE .. no results

Before I install the IDE 1.2.0 the Cam was always blinking (blue led) but now, it blinks only when I connect the Cam on the USB port and after 2 blue blinks the blinks stop.

Thanks

[kwagyeman]

I'm working on something else that can fix this all. So, the OpenMV Cam debugging protocol was written for libsub originally. It worked for that. But, when libusb was not usable on windows the debugging protocol was then embedded in the USB CDC calls. Anyway, this means that commands must be sent to the Cam in separate USB packets to work. USB CDC packets can be at maximum 64-bytes.

Ibrahim threw this code together and it sorta works as long as the USB packets come in slow. I've been trying to work around this on the PC by sending USB packets slowly so that the host doesn't combine two commands into one packet. But, I think I have a more stable fix which is to just send 64 byte packets always. The USB debugging code for the Cam doesn't care if the packets are longer than necessary, only that one command per USB packet is sent. So, by zero padding the command everything works much better. I already tested this on my PC and it let me remove all the delays I added in between sending packets.

However, this fix is not alone enough to guarantee the system works on all PCs. Ibrahim will... fix this... in the next version of the firmware and boot loader by adding in a small 64-byte buffer which will hold commands sent that are shorter than 64-bytes. This will remove the assumption we depend on that the computer will send 64-byte packets in separate frames and not generate short packet lengths.

In reality, the firmware should have never been such that it required debugging packets to be in separate USB transfers. Terrible code right there. But, since that's what's there out in the wild right now I have to put fixes in the IDE to get around this.

[kwagyeman]

@Kermit666 - Inability to connect is a separate issue from the USB packets. When you plug in your OpenMV Cam... does it appear as a device in your device manager? If so, what device does it appear as? There's no IDE fix I can do to solve any problem related to the Cam not connecting to the PC initially.

Two things can be happening:

1. The wrong driver is getting loaded for your OpenMV Cam.
2. The Cam's firmware is bad and you need the bootloader to fix it. If 2 is the case the bootloader should be able to connect to the Cam, but, might have trouble re flashing given the packet issue. You'd get a failed to connect to bootloader error here.

Please let me know - more info the better.

[tcottle]

Additional Win7 Testing

I uninstalled the IDE and reinstalled using the link you provided a couple of post up. This new version still IDs as 1.2.0 (?)
OpenMV driver version 5.1.2600.3

No improvements from my end. The only way I can connect is to attach the module and then quickly press the connect button before the flash drive mounts

When connected, the helloworld script will not run

[Kermit666]

Hi,

"@Kermit666 - Inability to connect is a separate issue from the USB packets. When you plug in your OpenMV Cam... does it appear as a device in your device manager? If so, what device does it appear as? There's no IDE fix I can do to solve any problem related to the Cam not connecting to the PC initially."

When I plug in the Cam it appears as a removable disk, with 31 KB and 3 files inside.
This works perfectly from the beginning

"Two things can be happening:

1. The wrong driver is getting loaded for your OpenMV Cam.
2. The Cam's firmware is bad and you need the bootloader to fix it. If 2 is the case the bootloader should be able to connect to the Cam, but, might have trouble re flashing given the packet issue. You'd get a failed to connect to bootloader error here."

I have the lcd module, what happens on the lcd module when I connect the cam ? on my cam + lcd I have only a white lcd, nothing else !

Thanks

[wminarik]

...just to report that the new IDE (v. 1.1.0) installs fine on a MacBook Pro (mid-2012) running OS 10.8, and installs the updated firmware. The key is to connect the OpenMV to the Mac via a powered USB 2 hub. Directly connecting OpenMV to the built-in USB ports gives only grief -this is a known feature of this generation USB 3.0 controller.

[kwagyeman]

I can report that it works fine on my Mac Mini. The new model with USB 3.0 ports. (2014?)

Anyway, updated IDE coming soon this weekend. If you installed v1.1.0 it will let you know when the new version is available.

Thanks wminarik

[deshipu]

I just tried the new IDE, downloaded from http://openmv.io/upload/openmv-ide-linu ... -1.1.0.run -- is that the most recent working one?

I couldn't get it to work on my Ubuntu. Trying to connect tells me "no openmv cams found" and then asks me if I'd like to unbrick the camera and delete the filesystem, when I click yes, it displays a full progress bar and tells me to disconnet and reconnect the camera, but nothing happens when I do. Any hints? It worked fine with the old IDE.

[deshipu]

Oh, I have the udev rules installed for it, should I remove them?

[kwagyeman]

Yeah, I'm looking for a good and vid pair that makes a serial port. Use rules might mess this up.

[kwagyeman]

The uploader that worked for you doesn't filter by the PID and vid so it didn't care.

[deshipu]

I removed the udev rules, but still no joy. The log in dmesg when I connect my cam is:

Code: Select all

[875424.766530] usb 2-1.2.2: new full-speed USB device number 36 using ehci-pci
[875424.877244] usb 2-1.2.2: New USB device found, idVendor=1209, idProduct=abd1
[875424.877254] usb 2-1.2.2: New USB device strings: Mfr=1, Product=2, SerialNumber=3
[875424.877260] usb 2-1.2.2: Product: OpenMV Virtual Comm Port in FS Mode
[875424.877265] usb 2-1.2.2: Manufacturer: OpenMV
[875424.877269] usb 2-1.2.2: SerialNumber: 000000000011
[875424.877737] cdc_acm 2-1.2.2:1.0: ttyACM0: USB ACM device
[875424.878104] usb-storage 2-1.2.2:1.2: USB Mass Storage device detected
[875424.878335] scsi host35: usb-storage 2-1.2.2:1.2
[875425.876142] scsi 35:0:0:0: Direct-Access     uPy      microSD Flash    1.00 PQ: 0 ANSI: 2
[875425.876602] sd 35:0:0:0: Attached scsi generic sg2 type 0
[875425.877351] sd 35:0:0:0: [sdc] 97 512-byte logical blocks: (49.6 kB/48.5 KiB)
[875425.877968] sd 35:0:0:0: [sdc] Write Protect is off
[875425.877975] sd 35:0:0:0: [sdc] Mode Sense: 00 00 00 00
[875425.878592] sd 35:0:0:0: [sdc] Asking for cache data failed
[875425.878598] sd 35:0:0:0: [sdc] Assuming drive cache: write through
[875425.888105]  sdc: sdc1
[875425.891214] sd 35:0:0:0: [sdc] Attached SCSI removable disk
[875431.746159] FAT-fs (sdc1): Volume was not properly unmounted. Some data may be corrupt. Please run fsck.

[kwagyeman]

I'll make a version of the Ide tonight that will print out the info it sees. This will let me know what device type I need to look for.

Does the OpenMV cam appear as a try serial port? In dev?

[deshipu]

Yes, it appears as a disk and as a serial on /dev/ttyACM0. I can use screen to have normal python console there.

[kwagyeman]

Please try this version of the IDE out (assumed you had 64-bit linux) http://openmv.io/upload/openmv-ide-linu ... -1.2.0.run.

I don't have all the fixes I've been working on in this version of the IDE. I just disabled the port filtering. Please let me know what you see in the port list. It should print out this:

name=ttyACM0,hasVID=1,hasPID=1,VID=1209,PID=abd1

[deshipu]

Thanks, I will try it. By the way, you could add a mime type to your web server configuration (or a content-disposition header), so that it doesn't serve those files as text, but downloads them instead. Just a small thing.

[deshipu]

After clicking "connect" I get a message box with "name=ttyACM0,hasVID=1,hasPID=1,VID=8086,POD=1e26" with this version. Then it prompted me to upgrade the firmware, the to erase the filesystem, then it began programming. Once finished, it asked me to wait for reconnect before continuing. Then it displayed an empty message box with a comic balloon with blue letter "i" in it. Finally, an error message popped up with text "No openmv cams found" and it asked if I have an openmv attached and it's bricked -- just like the previous time. I clicked "cancel", and clicked on "connect" again. Then it asked me to associate a drive with the camera, showing my EFI partition as the default (the camera was the second option). Then it finished, and I was able to run the hello world example.

[kwagyeman]

Okay,

Um, about the EFI partition. Is that a FAT partition? I'd like to filter that out possibly for you. Right now the IDE looks for FAT partitions that are writable.

As for the PID and VID. That's odd... Pretty sure it was tje Foss one before.

[deshipu]

I think this is one of the first cameras with the new sensor. You sent it to me for testing, after I tested the fire-sale camera a bit. Perhaps it's somehow different from the kickstarter ones (I haven't yet unpacked those).

[deshipu]

The EFI partition has "EFI System" as the parition type:

/dev/sdb1 2048 999423 997376 487M EFI System

But I think that it actually has a FAT filesystem on it.