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.

DLPDLCR2000EVM: with Raspberry Pi

Part Number: DLPDLCR2000EVM
Other Parts Discussed in Thread: DLP230NP, DLP2000, DLPC2607

Hi,

Good day.

Our customer is using the DLPDLCR2000EVM and they have a question about it. Kindly refer below.

""I am following this tutorial frederickvandenbosch.be/ and have my DPLDLCR2000EVM connected to my raspberry pi 3. I can see the raspberry pi screen but the projection is a bit quivery and some videos have a split line going through the projected image.
I have rechecked and followed the wiring in this tutorial a number of times
I also have added the code provided to config.txt and to /etc/rc.local
There are some suggestions here in this forum that I might need to change the hdmi_timings?
But I do not understand how to do this?
e2e.ti.com/.../dlpdlcr2000evm-resolution-problem-settings-with-i2c-and-raspberry-pi
Philippe Dollo TI__Mastermind says "have you tried just relaxing the timings and reducing the pixel clock rate? As I mentioned in the other thread this is generally a timing issue when us"

I have ordered a second DLP LightCrafter Display 2000 EVM and shorter 5cm connecting jumper wires to see if this might improve things""
Looking forward to your inputs here. Thank you.

DLPDLCR2000EVM_Attachement.zip

Regards,
Cedrick

  • Cedrick,

    Given you are still having video issues with shorter wires and a relaxed pixel clock, it may be necessary to use a PCB to resolve the problem. For instance, the Pi Projector PCB by MickMake would very likely resolve your signal integrity issues:

    https://www.tindie.com/products/mickmake/pi-projector/

    In this case you can eliminate the wires which are causing your signal integrity issues (though you will need to fabricate the PCB somehow). I hope this helps.

    Regards,

    Philippe

  • Hi Philippe,

    I've got a feedback from our customer.

    ""thank you for your response, I have ordered shorter 5CM wires and a second DLP LightCrafter Display 2000 EVM and power supply. As soon as they arrive I can test out the shorter wires to see if they improve the projection. I did not understand how to relax the pixel clock can you give me an outline of some instruction of how to do this? Unfortunately there are no MickMake PCBs available , I did try to order one a few months ago, I am on a waiting list but have had no response. I would also like to enquire if you think it is ok to run the LightCrafter Display 2000 EVM continuously it will be in an exhibition for 3months.
    I understand in time the bulb may die but aside from that might it overheat or have any other issues. I have ordered a second one.

    PS I am using this powersupply recommended by a colleague of yoursTE20A0503F01‎
    AC/DC DESKTOP ADAPTER 5V 15W FROM DIGIKEY


    Regards,
    Cedrick

  • Cedrick,

    If you are running the system for a long period of time I recommend you review the recommended operating conditions information in the DMD datasheet (https://www.ti.com/lit/ds/symlink/dlp2000.pdf). If your operating conditions are not within the specification, you run the risk of having the DMD fail at some point.

    How hot do you expect the system to run?

    Regards,

    Philippe

  • Hi Philippe,

    Good day.

    I've just got a feedback from our customer. Please see below.

    I now have a 2nd raspberry pi and DLP LightCrafter Display 2000 EVM set up with shorter wires and I am seeing much the exact same quivering image on the raspi screen projection. I will attach photos and video. I use the same code but had to change to i2c-11 rather than using the proposed I2c -3 to display the raspberry pi screen.
    I ran the i2cdetect -l and changed the code to

    sudo i2cset -y 11 0x1b 0x0c 0x00 0x00 0x00 0x13 i
    sudo i2cset -y 11 0x1b 0x0b 0x00 0x00 0x00 0x00 i

    instead of the proposed
    sudo i2cset -y 3 0x1b 0x0c 0x00 0x00 0x00 0x13 i
    sudo i2cset -y 3 0x1b 0x0b 0x00 0x00 0x00 0x00 i

    Do I need to go back and edit anything in the config.txt to match this change to 11

    The image is still quivering and has a split line going through it. Philipe has suggested in other answers > Reduce the pixel clock output from the Raspberry Pi (lower is easier for the RPi to drive) . Can you ask him for instructions on how to do this reduce the pixel clock ?

    Philipe also suggests adjusting the GPIO drive strenght in the command line to a value betweem 0 to 7 Adjust the GPIO drive strength. Do "gpio drive 0 <X>" in command line where <X> is a value from 0 to 7.
    Does he mean to change it to 6 first "gpio drive 0 <6>"

    then "gpio drive 0 <5>"

    Can he recommend a no. between 0 and 7 ?

    I also now have a warning sign coming up on the screen that says low voltage. I contacted one of you colleagues to suggest the correct powersupply and this is the one I am using.

    So Yes thanks I still need some help, the projector in the gallery is also now displaying the low voltage warning it is also using the AC-DC switching power supply which accepts 50-60 Hz 100-240 V AC inputs, and outputs a nominal 5-V DC at maximum 3-A output current. Which your colleague recommended. For this purpose we recommend the TE20A0503F01 Desktop Wall Adapter (or equivalent). Which I ordered from Digi Key www.digikey.com/.../5400184

    Many thanks in advance for your help. I will recheck all my wiring but its is strange that it shows almost exactly the same split through the screen and quivering. I am hoping that if Philipe can give me instructions on how to reduce the pixel clock and adjusting the GPIO drive strenght in the command line it will help . Many thanks Fiona


    I hope you could help us further with this. Looking forward to your advice.


    Regards,

    Cedrick

    DLPDLCR2000EVM.rarClick here to play this video
  • Cedrick,

    A few comments from me.

    > First, the video timings (namely, pixel clock) can be adjusted by modifying the config.txt file for the Raspberry Pi (typically, the video timing configuration is stored there).

    > Second, for the gpio drive strength I cannot give a recommendation for a specific value as it would depend on your hardware. In general, I've found that anything above drive strength 5 should be enough for both this EVM and the DLP230NP EVM (the 1080p EVM for the Raspberry Pi). You can get information about what the drive strength actually means from this page: https://www.raspberrypi.org/documentation/hardware/raspberrypi/gpio/gpio_pads_control.md 

    > The command for GPIO drive strength in command line is "gpio drive 0 <VALUE>" where the value is 0 to 7 depending on the desired output. This affects the entire bank of GPIO pins, so you cannot adjust it pin-by-pin.

    > Ultimately, the problem you are seeing (video quivering) is a noise issue. If you want to resolve the problem decisively, you may want to use a custom interfacing PCB for the DLP2000 (as the pinouts for that EVM are made for the BeagleBone Black). Even if the MickMake design is out of stock, you can still borrow the design to print your own (which may be less trouble than trying to put together a noise-free setup with jumper wires).

    If you want to determine experimentally what value to use, you could always measure the parallel interface signals to see whether the output voltage is high/low enough for a proper signal.

    I hope this helps.

    Regards,

    Philippe

  • Hello Philippe,

    Please see the response of our customer below:

    I made progress in resolving the jitter issue and also the low voltage signal displaying on the pi, however there is now another issue that I would also like some help with. As I mentioned before I bought a second DLP LightCrafter Display 2000 EVM to resolve some issues as the first one was already installed at the gallery. I connected the second DLP LightCrafter Display 2000 EVM to the pi using shorter jumper wires. To completely stop the jitter in the projection I reduced the GPIO drive strength to 2mA using the command gpio drive 0 0 & & set it to auto configure after every boot up in the sudo nano /etc/rc.local . The low power signal was still displaying on the pi projection. I measures the voltage across 5v & GND on the rasp pi , it was dropping 4.85. I then measured the voltage across the 5V on the barrel pin on the DLP LightCrafter Display 2000 EVM & GND on the pi to be 5.3volts. We concluded the issue was the power jumpers wires.

    I had been following the Frederick VandenBosch tutorial, perhaps his jumpers were shorter or thicker. He suggested "I’ve added P2_6 (5V) via GPIO pin 2 to power the EVM from the Pi or vice versa. A bridge is then required on the EVM from P2_6 to J3 using the M-F jumper wire." I made a direct connection with one jumper from GPIO pin 2 to the EVM J3. The low power issue is now resolved and the signal is no longer appearing on the pi screen.
    I would like to have one of the PCB boards printed for future projects but for my needs now I see a problem the pi zero does not have a Line out/headphone audio. (If you want to use audio currently, you'll need to make use of Bluetooth audio.) I do not think Bluetooth audio will not work for me on this occasion as it will fall out of sync.

    So to get back to the new issue, When I went to swap the original DLP LightCrafter Display 2000 EVM which had been running continuously for 3 weeks, I found that the projection image had deteriorated a lot. There were a lot of specs in the projection. Perhaps Philippe can give some idea as to why this is happening? Is it perhaps because the projector is overheating because it is running continuously? Phillipe previously asked "How hot do you expect the system to run?" Is there a way to measure how hot the system is running? Will the fixes/changes we have made to the GPIO pin strength help and the low power voltage help with this issue. I am looking into adding a power off button the the raspberry pi to shut it down safely every evening this will reduce the time it needs to run www.makeuseof.com/.../ I will need to use a different GPIO pin which is not being used by the EVM . I will attach photos of the issue in the projection.

     

     

    As you can see in the photos it is pretty bad. It would be great if Phillipe can give some idea as to why this is happening and suggest how to prevent it happening to the second EVM . Along with using a power button to safely turn off the pi every evening so we can then unplug the EMV , might it be possible to add a heatsink somehow to the EMV ?

    I have also since found that the main issue with the Jitter in the projection was being caused by the long thin 5volt power jumper wire. When I made a direct connection with one jumper from GPIO pin 2 to the EVM J3 the jitter is resolved and there is no need to reduce the GPIO drive strength. As this is the case would Phillipe suggest leaving the gpio strenght higher than gpio drive 0 0 maybe at gpio drive 0 3.

    Thanks again for your help it is very much appreciated. 


    Regards,
    Carlo

  • Carlo,

    Thank you for the thorough summary of your issues. A couple comments from my end:

    > To be clear, can you confirm that you are powering the EVM and Raspberry Pi via separate power supplies? I recommend you do so for the time being. If it provides any context, I've been able to drive video on this combination (DLP2000 EVM + Raspberry Pi) with a setup that matches the one in this particular tutorial (VandenBosch's). I recall seeing the power status but no visible issues related. I believe that so long as you are using separate power supplies for both devices (Pi and EVM) you should be OK.

    > For the "speckle" type issue you are seeing with the attached images, this almost certainly looks like a DMD problem (you can confirm this if the white dots are in the same place regardless of input content or video mode. Try with internal TPG mode if possible. These white dots are most likely dead pixels on the DMD (mirrors that have for some reason or other been damaged or broken and are no longer working). Perform the TPG mode test to confirm whether these are actually dead pixels or not.

    > If you can confirm that the issue above is dead pixels, then the most likely causes are usually either a bad DMD connection or a power-related problem. In terms of the DMD connection, this kind of issue arises if the DMD data flex cable is unplugged or tampered with in the middle of normal operation. It sounds like something could have happened during the time period that the DMD was running in this case.

    > Alternatively, it is possible that the DMD was not safely powered down. There is an array of power down capacitors on the EVM which prevent such a problem from occurring when the power jack is unplugged, but if you are trying to power the EVM via Raspberry Pi then you could have an issue there. You may want to check that the DMD power rails (VDMD, VBIAS, VOFFSET, VRST) are all coming up and are stable on the EVM side.

    I hope this helps.


    Regards,

    Philippe

  • Hi Philippe,

    Thanks for your reply , I have been following the tutorial (VandenBosch's) and as he recommended powering both the EVM and the Pi from the EVM "The EVM’s 5V/3A power supply is plenty to power both devices"I am using the recommended EVM power supply 5V/3A connected to the EMV barrel power jack. As I mentioned the low power signal is not longer coming on since I am using just "one jumper wire directly from GPIO pin 2 (5VOLTS) on the pi to J3 on the EVM.
    However I can certainly swap to powering both the pi and the EVM separately using two power separate supplies. Before I do this can you please confirm if I need to remove the jumper wire that goes from the J3 on the EMV to GPO pin 2 (5volts) on the raspberry pi?

    Yes the white dots are in the same place whether I am in video or EMV splash screen so I assume a DMD problem.Can you explain what you mean by the internal TPG mode and how to try it.How do I perform the TPG test?

    As you can see in the attached photo I have not been powering the EVM from the raspberry pi. I have been powering the pi from the EVM recommended power supply . How do I check that "the DMD power rails (VDMD, VBIAS, VOFFSET, VRST) are all coming up and are stable on the EVM side"

    It is possible that the EVM power supply was unplugged while the video looper was running on the raspberry pi. Is this a problem?

    Do you think the problem might be that the EVM is running continuously for a long period if time , 3 weeks?

    Thank you so much for checking this for us.


    Regards,
    Cedrick

  • Cedrick,

    If you run the system with two separate supplies you should probably not connect the supplies on each system together. Connecting the grounds should be OK though. For the purposes of debugging this issue, I do think you should try this approach for now as it does seem you are having a power related issue. Once you confirm you are providing enough power you could consider using a single supply afterwards.

    The white dots are likely broken DMD mirrors (which makes them impossible to fix on that particular DMD at this point. The I2C commands which should be used to switch to TPG mode (along with all other DLPC2607 commands) are covered in the DLPC2607 Programmer's Guide: https://www.ti.com/lit/ug/dlpu013a/dlpu013a.pdf 

    You can find the layout/and schematic information for this EVM on the associated reference design page: https://www.ti.com/lit/ug/dlpu013a/dlpu013a.pdf 

    You should use the above to find the DMD power rails and track their behavior.

    This kind of DMD failure most typically occurs if something happens to the DMD flex cable during operation (such as being tugged or removed while DMD commands are being sent over the cable). It is possible for this behavior to happen due to sudden power down, but the DLPDLCR2000EVM is specifically designed to power down gracefully when power is removed by unplugging the jack mid operation. 

    One possible theory I have based on your circumstances is that the weak power rail provided by the Raspberry Pi was not providing enough power to charge the supply capacitors on the DLPDLCR2000EVM. In this state, if the system turned off suddenly then there may not have been enough time for the system to shut off the DMD gracefully. For the sake of the DMD you should try to ensure you have a stable 5V-3A supply on the EVM side.

    I hope this helps.

    Regards,

    Philippe

  • Hi Philippe,

    Good day. Here's what our customer told us.

    I have been using a stable 5V-3A supply on the EVM side all along, the model was recommended by a colleague of yours before I got started on the project it is the TE20A0503F01‎AC/DC DESKTOP ADAPTER 5V 15W FROM DIGIKEY

    I no longer have any low power issue. I found the reason for the low power issue was that I had been following in Frederick Vanden Bosch tutorial, in which he suggested adding P2_6 (5V) via GPIO pin 2 to power the Pi from the EMV. and A bridge then required on the EVM from P2_6 to J3 using the M-F jumper wire." Rather than using 2 connection jumper wires I made a direct connection with one jumper from EVM J3 to GPIO PIN 2 on the pi. This resolved the low power issue, it also resolved the quiver issue.

    Perhaps it is better to unplug the plug on the 5V-3A supply on the EVM, rather than unplugging the jack from the barrel pin from the DLPDLCR2000EVM?

    As I mentioned I had been running the system continuously 24 hours a day for 3 weeks , I wondered if this might add to the problem with the DMD? In order to shut down the system correctly in the evening I have now added a power button to the raspberry pi. The raspberry pi needs to be shut down correctly ( its memory card can be corrupted by unplugging it suddenly) before unplugging 5V-3A supply on the EVM. I am hoping that shutting down the pi and the EMV every evening will help.

    I also came across another article which mentioned that "The DLPDLCR2000EVM optical engine can be driven to 30 lumens at maximum current settings, but will also run at high temperatures. It is recommended that the user add a heatsink to the system when running beyond typical operating conditions to maintain the same unit temperature.there was a heatsink "

    This area circles in red does seem to get quite hot. I am wondering if it might be an idea to add another heatsink here?

    For example a heat sink like this
    https://coolcomponents.co.uk/products/heat-sink-kit-for-raspberry-pi-b?_pos=1&_sid=b550872ad&_ss=r

    This was the review which mentions the heatsink
    www.element14.com/.../dlp-pico-display-projector-evm-beaglebone-black-review


    Additionally, The I2C commands which should be used to switch to TPG mode (along with all other DLPC2607 commands) are covered in the DLPC2607 Programmer's Guide: www.ti.com/.../dlpu013a.pdf You can find the layout/and schematic information for this EVM on the associated reference design page: www.ti.com/.../dlpu013a.pdf
    You should use the above to find the DMD power rails and track their behavior.

    I looked through these documents but cannot find The I2C commands which should be used to switch to TPG mode , or the DMD power rails and track their behavior. The document is very extensive and detailed. It would be great if he could point me to which page they are on on the document and how to use the commands.

    Also Phillipe said "the DLPDLCR2000EVM is specifically designed to power down gracefully when power is removed by unplugging the jack mid operation. " Is he suggesting it is safe to unplug the EMV at the jack end of the power supply or is it safer to unplug the power supply at the plug, or does it matter .

    Thanks again for all your help.


    Regards,

    Cedrick



  • Cedrick,

    For how to use the I2C commands, see section 2.2.1 on the I2C protocol. As for which commands to use, see section 2.4.1.3 (Input Source Selection Command).

    As for the unplug operation, it shouldn't matter where you unplug power so long as you've given enough time for the power-down capacitors to charge.

    As for the heatsink, you have the right idea. If you want to run the system at the maximum allowable brightness you should add an additional heatsink on top of what the system already has. Otherwise the optical engine/LEDs will run too hot.

    Regards,

    Philippe