IWR1443: gui uniflash

Hello Al,

Sorry to hear you are having issues connecting to the IWR1443 from Linux.

You mentioned the mmWave Demo Visualizer, the High Accuracy lab GUI, and Uniflash. All three are different applications which require the mmWave device to be in a different state (running a specific firmware or in flash mode in the case of Uniflash).

1. Are you having issues with all three or just one of the applications? How often do you see the issue with each application?

2. Can you describe the exact sequence of events that happens before the failure to connect with the application that is experiencing the issue most frequently?

3. Are you using Linux in Virtual Machines or native Linux PCs? What OS and version?

4. What mmWave SDK version are you using?

5. When the EVM is powered up and connected to the Linux machine we have seen that the Linux serial port seems to take up to 20 seconds to completely initialize. Can you try waiting that long before sending a config to the device to see if that helps?

6. What power supply are you using for the EVM? (it must be able to supply at least 2.5Amps of current) Can you try a different power supply to see if it affects the issue?

Regards,
John

Hi John,

Thanks for the reply.

1.  Yes, I am having issues with all three applications.  Yes, I am aware that different firmware is required for the gui application.  If the wrong firmware is on the part, the gui will error out, unless the uniflash gui is being used which will erase the firmware and load new firmware.  Very offten, sometime the connection is made but I am getting tried of wasting my time.

2. With the two visualizer gui's, the ports are defined (/dev/ttyACM0 and /dev/ttyACM1; note that the cfg and data ports are switched in their screen location between the two gui's), the uniflash requires only /dev/ttyACM0. Listing /dev/ttyACM* shows that the ports are present.  The lower left hand corner indicates the ports have been specified, however the oval is split with no internal bar.  Clicking in the are indicates that the the sensor is disconnected.  For uniflash, a message pops up to indicate that there is a permission or some other problem.  I have confirmed that these are not the problem.  Besides, sometime it works but not at other times.

3. I am on a native fedora 27 platform.

4. The sdk is mmwave_sdk_01_02_00_05.

5. I am using D3's starter kit.  I have waited along time.  I have tried power cycling and pressing the reset button.

6. I will check the power supply and try another one.

Thanks.

Al

Hi Al,

Ubuntu 16.04 is the only Linux OS we have tested with Uniflash and the mmWave Demo Visualizer web GUI. The high-accuracy web GUI is not tested for Linux. I have seen issues when using the mmWave Demo Visualizer web GUI from Chrome in Ubuntu and found that Firefox worked better.

Please also try using only the USB cable that comes with the EVM as opposed to using a longer one or adding an extension cable as they may cause connection issues.

Please tell me if replacing the power supply resolves the issue.

Regards,
John

Hi John,

I use only firefox and I am offline.   The power supply I am using came with the d3 unit and it is rated at 3 amps.  I am using the usb cable that came with the d3 unit as well.

So I guess I am not up and running yet.

Al

Al,

Let's start by focusing on the Uniflash issue since it is not dependent on anything pre-programmed in the flash.

What is the exact sequence you follow when attempting to flash the EVM?

Can you send a snapshot/copy of the error message and a copy of the contents of the console from Uniflash when the issue occurs?

Regards,
John

John,

Yes, uniflash is the only app that provides a little feedback.  I have attached and image of the gui. It appears it is waiting for something from the sensor.   The order is

1.  I have the flashing toggle in the flashing state

2. I connect the usb and the the 5 v power

3. Bring up uniflash 4.4.0

4. Specify /dev/ttyACM0

5. Specify two binaries (I used the prebuilt ones in this case)

6. I did do a power cycle but I assume that is not neccessary since  the flashing switch was already set before power up

7. The gui times out trying to connect.

I would use wireshark to monitor the usb traffic if I know what to expect.

Hope this helps

Al

John,

I forgot to attach the image referred in my last posting.

John,

My last functional activity with the sensor was when I was using the CCS debugger.  I had made a change, did a clean and rebuilt.  I got the error shown in the attached image.    Hopefully this gives a little more information.

Thanks.

Al

Hi Al,

Thanks for sending the detailed sequence you are using for Uniflash. How much time do you wait after step 6 (power cycle EVM) before clicking the Uniflash "Load Image" button?

I plan to install the same version of Uniflash on a native Ubuntu 16.04 PC and try the same sequence to see if I can reproduce the issue. I will reply with the results once I try it out (most likely tomorrow).

Regards,
John

Hi John,

I hooked up the sensor again and I waited 30 seconds between power cycling the sensor and clicking the load image button.  Same result.

The uniflash console indicates that a connection to /dev/ttyACM0 is being made, a reset connection to the device is being made and a break signal was set.  I assume no connection was ever made.  Correct?

Did you examine the console message I set when I used ccs in the debugger mode?

This weekend, I put on  windows 7 using vmware and installed the software.  No ability to connect either.  However I did get this error message which I have attached.  Did some other firmware get corrupted?  I tried another usb cable too but I got the same result.

Thanks.

Al

Hi Al,

I have not tried Windows 7 on vmware so not sure if it might have additional requirements/challenges over using a native Windows machine.

In order to maintain an organized/focused approach, I will try to reproduce the Uniflash issue first.

Regards,
John

John,

I found a windows 7 platform and loaded the mmwave software on it. I get the same issues with it as with the linux platform. I then went back to the linux platform. Strangely, I found out that I can connect to the high accuracy gui but not to the oob demo visualizer gui or uniflash. Since I had the ccs debug load on the part, the high frequency would naturally not work.

Al

Al,

Just to clarify, are you saying that on a native Windows 7 PC with Uniflash installed you are unable to connect to the mmWave sensor via Uniflash in order to flash an image onto the EVM?

If so, are the XDS110 emulation drivers installed so that the COM ports are seen in the Windows Device Manager?

XDS110 Drivers
• For EVM XDS device support
• Included with CCS Installation, or standalone through TI
XDS Emulation Software
processors.wiki.ti.com/.../XDS_Emulation_Software_Package

Regards,
John

John,

The device manager sees the com ports as com11 for control and com12 as data.

Al

Al,

If you are seeing issues on Windows then it may be more useful to try to debug any issues there (rather than Linux) since Windows is our standard platform.

What version of Uniflash are you using on Windows? What does the error message and console output look like? How often do you see the issue?

Thanks,
John

John,

Using windows is fine for this.  I am using uniflash 4.2.2.1692.  Attached are a couple of images illustrating the setup.  Does uniflash use the control port (com11) or the data port (com12) and does uniflash set the baud rate?  Same error as with windows.

Al

Al,

Uniflash requires the "User UART" COM port for programming (looks like COM11 in the pic you sent). Do you still see the error when selecting that COM port on the "Settings & Utilities" tab before clicking the Load Images button?

Also, are the jumpers in the following position when flashing? (and EVM power cycled after placing jumpers in that position?)
SOP2: ON
SOP1: OFF
SOP0: ON

Regards,
John

John,

I specify the port first, then specify the binaries , then power cycle and then click the load button.  The error appears after the load button is clicked.  I have the switches in the proper positions.  I have down a number of loads on the linux platform before this issue came up.

Question, in the upper right hand corner of the uniflash gui, I get a message of indicating that the cortex is suspended.  If I click on this area, I get an option if I which to be connected after the operation or disconnect.  If I disconnect, the message becomes tha the cortex is disconnected and halted.  I assume the cortex should be just suspended during flashing.   Should the box be checked or not or does it make a difference?

Al

Al,

When I use Uniflash the status in the upper right corner in Uniflash says "Cortex_R4_0" after selecting the IWR1443BOOST board from the first screen. When I click the Load Images button the small dot to the left of the status changes from grey to a more solid green color and then the status changes to "Cortex_R4_0 Suspended" during the flashing operation and then changes to "Cortex_R4_0 Disconnected: Halted" after the flashing is complete. You should not have to check the box.

I'm going to have one of my colleagues look at the thread to see if he has some suggestions.

Regards,
John

Hi John,

Is there any advice you can provide to examine if there has been a hardware failure?  Perhaps you have only considered a software issue.  Can you provide me what traffic there should be on the usb bus?  I can put an analyzer on it to inspect.  Is there another flash tool that I could use?

Al

John,

Additional actions I would like to pursue:

1.  Examine the usb line with an analyzer.  Can you provide me with any detail on the traffic I should see?

2. On the ccs side, I can start the debugger session but I get and error as shown in the attachment.  What does the mean?  Can it shed any guidance.  I googled the error and I got something pertaining to a low power mode.

3. I came across a reference that ccs could flash the part.  Might be a possibility since I can get the debugger going.  I tried to follow the thread but I could not get any head way.  Perhaps you could provide a reference of some detailed guidance.

Thanks.

Al