This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

DLPC3479: After RunOnce for many times, it will not respond

Part Number: DLPC3479
Other Parts Discussed in Thread: DLP4710LC, , DLPA3005, DLP4710, DLP3010, DLPC3478, DLPA3000

Hello TI support team.

We use the dlpc3479 + dlpa3005 + dlp4710lc + MSP430 product combination to switch to the internal mode and free mode. MSP430 realizes that the external input signal sends a RunOnce projection, and then continues to trigger after all the projections are completed. In this way, after the cyclic test is about 1 million times(about 1-3 hours), the system may not work. Cannot continue projection.

After that, we found that I2C communication can be partially normal, and some data queried by GUI can be returned normally. But RunOnce cannot continue to light up the projection. It cannot be lit even if it is switched to test mode. DLPA3005 INT_ Z is always high.

In addition, we always send the I2C command of RunOnce through cypress USB on the computer side. This problem will also occur after a period of time.

If we use “Run Continuously” to project all the time, we won't have this problem (at least run for more than 24 hours).We would like to know someting of clue.

Regards,

andy

  • Hi Andy,

    How quickly are you sending these commands? With some commands which cause the DLPC3479 the fetch data from memory (such as on the external SPI flash device) there is a period during which the DLPC device is busy and cannot accept additional I2C commands. In these cases, sending too many commands too quickly may cause the bus to hang up.

    Since the 'Run continuously" command works. I suspect that you may be sending commands too quickly to the DLPC3479.

    Regards,

    Arthur

  • Hi ,

    I have doubts that when the projection cannot be lit, the I2C command can still work for a period of time. "06" and other commands to read data can still work normally.
    In addition, I send I2C at a fixed frequency or wait for "RunOnce" projection before continuing to send. There should be no competition.
    Is there any way for me to query the specific status information when the normal projection cannot be performed? Is dlpa3005 inoperable, or is dlpc3479 internal chaos?

    Regards

    Andy

  • Hi Andy,

    Sorry for the delay. Please allow me a couple days to discuss with my team.

    Regards.

    Arthur

  • Andy,

    Can you elaborate a little bit on the conditions which cause failure case?

    From your description above, it sounds like you are saying that the MSP430's I2C commands are a concern to you. If necessary, note that you can actually modify the MSP430 code by making changes to the source code and uploading your new version to the MSP430. You can grab the sample code from the EVM tool folder: https://www.ti.com/tool/DLP4710EVM-LC

    I hope this helps.

    Regards,

    Philippe

  • Hi  

    thanks .My problem is not that I2C doesn't work.
    Hundreds of thousands of I2C commands "9E 00 00" (RunOnce) have been called to DLPC before the error occurs. It's a problem after repeated cycles for many times. DLPC can't light up the DMD projection. At this time, I2C can continue to work. In addition, we can successfully send the RunOnce command with the evm-gui program on the computer side, but it is not bright at all.
    I suspect that dlpa3005 stopped working for some reason. I would like to ask you what caused it?.

    Work steps:

    (0) Start

    (1) MSP430 sends I2C command (RunOnce) to DLPC

    (2) Wait for the DLPC internal pattern RunOnce projection to complete.

    (3) Continue step (0)

    After many cycles, it may not work, but i can also send the RunOnce command

    Regards

    Andy

  • Andy,

    Thanks for clarifying. If I2C is still functioning, it would imply that the controller is still OK so perhaps there is a sequence problem. What kinds of error status bits are triggered during this failure case? There are a couple status registers documented in the DLPC3479 programmer's guide that you can check: https://www.ti.com/lit/ug/dlpu081a/dlpu081a.pdf 

    Regards,

    Philippe

  • Hi 

    We used the board of Dlp3010 to verify that this problem did not occur. I ran the same way for a week and continued to project. Dlp4710 is easy to be unable to project after running for 2 hours. I would like to ask you, at this moment, what state or IO pin state can I analyze to locate the fault of DLPA or DLPC?

    When I operate on Evm-Gui, sometimes the following errors appear, but after continuing "get", the errors disappear again.

  • Hi Andy,

    these errors would indicate that the controller did not understand the op code sent from the msp430. is it possible for you to look at the I2C bus when this issue occurs? if you can do this and then compare to working conditions it may shed some light.

    Also you say that " we can successfully send the RunOnce command with the evm-gui program on the computer side, but it is not bright at all." Does the pattern project but it is dim? or not at all?

    Regards,

    Arthur

  • Hi  

    1)“Invalid Command error ” is caused by the "get" operation on the information page in evm-gui. It has little relevance to this problem. I2C bus can still be used.


    2)When there is a problem, the internal mode RunOnce operation, the projection light is not bright at all, and other "get" operations are normal, which will not cause an error prompt. Finally, it doesn't light up after using test mode or splash mode, but after these two operations, the GUI starts to report I2C error.

    Additional information:

    Through the logic analyzer and counting board, we record the number of "RunOnce" sent and the output of "DLPC out2" and "pattern ready" signals. After 15884 cycles, the DLPC stops projecting. It is normal to continue other I2C operations at this time.

  • Andy,

    It sounds like you tried this same operation on the DLP3010-LC chipset (DLPC3478) with no problems. Is that correct?

    Are you using a custom design or are you performing both tests using the TI.com EVMs (DLP3010-LC EVM and DLP4710-LC EVM)? If you are using a custom design, what optical engines are you using?

    Regards,

    Philippe

  •  Hi 

    We refer to Ti design and self-developed optical engine. Some tests were done as follows:
    1) Dlpc3478 + dlpa3000 + dlp3010, in the same way, run “RunOnce” for nearly a week without problem
    2) Dlpc3479 + dlpa3005 + dlp4710lc. Run more than 20000 times “RunOnce” in the same way, it may hang up, and then continue to send. There is no error in I2C and the LED light is not on.
    3) Dlpc3479 + dlpa3005 + dlp4710lc, download the latest official firmware of Ti, and then burn the same pattern.bin. Start the test, run more than 200000 times, and RunOnce stops responding.


    The timing configuration in pattern.bin is shown in the figure below.

  • Andy,

    Thanks for the summary. We may need a couple days to attempt to reproduce the problem. Will post back here as more information arises.

    Best Regards,

    Philippe

  • Hi Philippe Dollo,

    Can you reproduce this problem,Or is there something wrong with our control?

    In addition, if we reduce the trigger frequency, the projection survival time is relatively long. I hope this information is useful for your analysis. Thank you very much for your help.


    Best Regards,
    Andy

  • Hi Andy,

    We apologize for the delay, we are still looking into this.

    Thank you,

    Chris

  • Hello Andy, 

    Can you please share your contact information so that we will reach out to you in order to help you in addressing this issue.

    Regards,

    Mayank 

  • Hi Mayank,

    My personal email address is junglezg@outlook.com. You can also send it to my forum bound email.

    Thank you,

    Andy

  • Andy,

    Thanks for your feedback. We'll reach out to you as soon as we are able to.

    Regards,

    Philippe

  • Andy, 

    We have root caused this issue and the fix for it will be available in the next firmware release. In the meantime, a potential workaround that you can use is to power cycle the system when you encounter this issue. 

    Regards,

    Mayank