DLP3010: Defect of DMD

In addition, I have a question.

I saw a Write Mirrors Lock Command (39h) in DLPC 3478 Software Programmer's Guide, which stated that this command is only used in factory to prevent DMD damage.

Excuse me, is this command used by DMD factory or Optical Engine factory? How can I use it correctly?

Hello Junfa,

Thank you for reaching out to us. I have assigned this thread to a colleague, but please expect a delay in response because of the holidays.

Thank you for your patience.

Regards,
Aishwarya

Hello Aishwarya，
That' s all right, thanks.

Junfa,

Thank you for your understanding. We will get back to you next week.

Regards,
Aishwarya

Hi, in my experience (and confirmed by TI in some other thread) this is caused by improper power down sequence.

See here:

https://e2e.ti.com/support/dlp-products-group/dlp/f/dlp-products-forum/811235/dlpc3433-when-the-temperature-is-higher-than-60-c-the-noisy-screen-happens/3002248#3002248

"Also, I noticed your image has vertical lines down the screen. This is a fail mode experienced by the DMD when the power down sequence doesn't have enough time to complete (wall power removed unexpectedly) or when the DMD pins have a loose connection. To prevent this from happening on other devices, please ensure the power down is executed properly with enough time and that the DMD has a good connection to the PCB. "

Kind regards,
Kajetan

Hi Kajetan,

Thanks for jumping in and helping us out!

Junfa,

I am not an expert on this device, but maybe my colleague will be able to comment on the issue. Please give us a little more time.

Regards,
Michael Ly

Hello Junfa,

As your test states, you don't turn off the DMD, so I don't see this as a power down issue at all.

junfa ye said:

The aging process is to switch to the dark field observation picture with low LED current. After no defects are found, switch to the internal chessboard test picture and continue to work for more than 8 hours. After the end of the time, switch to the dark field observation picture without interruption, and observe whether there is any defect in the picture.

Could you please advise what temperature the DMD was seeing at the end of this test? The Tarray and Twindow during operation is a max of 90C, I'd imagine after 8 hours of exposure the DMD could be getting hot.

Are these bright spots at edge of the checkboard boxes or are they random? It appears to be in that pattern now.

Also, do these failures correct over time?

Best,

Aaron 

Hello  Aaron,

Sorry, I have been busy with other things recently, and my reply is late.
Yes, as I described earlier, I didn't cut off the power, so it showed a bright spot, so I don't think it has anything to do with the timing of the power outage.

Recently, we are rearranging the production workshop and experimental platform to ensure that testers can see a clear picture without moving or touching the machine. After finishing, we will continue to test according to this condition. Later, the aging test will also be carried out with the dark field.
In the previous aging test, the machine was put in the cabinet, and when it was finished, it was taken out to adjust the distance and then observed.
Because my colleagues and I are busy with other things, and it is also close to our Spring Festival holiday, we may not be able to have the test results soon. Please wait for a while and give feedback.

Regarding the temperature of DMD at the end of the test, I can't confirm and tell you the exact value because we didn't monitor it. But I can confirm that the temperature is about 50 degrees Celsius, because the radiator surface of DMD is about 40 degrees Celsius.

Are these bright spots on the edge of the chessboard or random? I didn't notice this. I think it should be random. I will confirm the experiment afterwards.

The failures may disappear with the passage of time, because the bright spots of some machines will always exist, and the bright spots of some machines will disappear at the next power-on, or the bright spots will disappear when the DMD is replaced with other machines (excluding the dirty problem).

In addition, there is a known problem on our machine, that is, there is liquid at the metal point on the back of DMD.

As shown below:

I guess it may be that the cooling pad on the back of DMD reacts with air when it is heated, thus producing water.

Because there has always been such a phenomenon from the design section, the probability of bright spots before is very low.

I can't confirm whether this water has pure water or other ingredients, and I can't confirm whether it has anything to do with the bright spot of DMD.

Hello Junfa,

There are quite a few probabilities here but I'll reach out to my team and I'll give you a response if we have an idea. When you finish testing please share your results and I'll be happy to respond.

We've seen some pixels be sticky in situations like this but they do correct themselves over time, I cannot give a rate of this happening as we don't test in this fashion, but it has been seen in the lab.

I'll be looking forward to hearing back from you!

Best,

Aaron

Hello  Aaron,

Ok, recently, we also learned about BBF failure, because there are many directions to be investigated, so we need to spend some time testing and verifying.
If you have good suggestions, we can test them.

If our test has certain results, we are also willing to share and study together. Thank you.

Hey Junfa,

For investigating BBF pixels you will need to know confirm the location of the vertical lines.

When you share information with us it is most necessary for us to have the exact sequence of events for us to try and replicate the issue ourselves. This especially includes the modes of operation.

I'd ask you to report a problem through ti.com and that will pull my team and myself into the situation. That can be much more efficient and tracked well between multiple groups.

Are you in touch with any of our FAEs? If not, I can put you in touch with them.

Best,

Aaron

Hello  Aaron,

Yes, I have received a picture to verify the pixel position of BBF, so I already know how to confirm the position.

I contacted FAE engineers in China, and they provided some reference documents, such as BBF and design guidelines and other possible reasons.

Because we know the design guidelines and precautions, we also pay attention to the design and use, and the bad point of DMD here is not entirely BBF, so I hope to communicate this problem with the engineers of TI R&D team.

I am reporting the problem through TI E2E now. Where else can I report this problem through ti.com?

I will share the aging process of our test with the group.

We will test the power supply, signal and timing, and also do some tests on connection stability. In addition, we will re-classify the defective products as BBF.
As we are approaching the spring break in China, the verification of the experiment and the reply of the results will be delayed. Please understand and thank you.

Let me add and share some more information:

The abnormality of DMD's bad points, some appear after assembly, some appear after aging, some are found after re-testing after storage for a period of time, and some are found in the use of end customers.

Our production environment and personnel have done ESD grounding or ESD protection.

The aging process we tested is as follows:
1. A batch of optical engines have been assembled, and the test picture has no bad points;
2. A batch of optical engines are tested for aging at the same time, and connected with a 12V and 10A power supply (from a good power adapter). The optical engines enter SPLASH mode, and the LED current is configured at 100 by default.
3. Use DLP EVM GUI or use the buttons on the driver board (MCU sends IIC command) to switch Test Patterns, switch Checkerboard to dark field, and observe whether there are any bad points on the DMD display; If there is no bad point in DMD, switch the Foreground to white and the Background to black.
4. The optical engine keeps this projected picture all the time.
5。 After 8~10 hours, use GUI or key to switch Checkerboard to dark field, and observe whether there are any bad spots on DMD display.
6. When there are many machines aging at the same time, a small proportion of machines will see that DMD shows one bad point or multiple bad points.
7. In addition, there is a smaller proportion of machines. When switching to dark field, observe that there is a bright spot on the DMD display, and then the bright spot disappears instantly.
8. Pull down PORJ ON and soft shut down the machine.

Because our optical engine is used in industrial inspection, DMD is a big defect. We hope to solve the problem, thank you.

Hello Junfa,

Thank you for giving a descriptive test sequence for us to try and replicate ourselves. I'll align with my team to see how we can try this ourselves.

I notice that you also mention:

junfa ye said:

6. When there are many machines aging at the same time, a small proportion of machines will see that DMD shows one bad point or multiple bad points.

Does this not occur in lower batch numbers?

Best,

Aaron

Hello  Aaron,

Yes, in the old batch, the probability of occurrence seems very low. Let me share our usage:
In the first four years, we used 3010 more, and last year (2024), we started to use 3010LC more.
In the first four years, we used 4710 more, in 2022, we used 4710A, in 2023, we used 4710LC more, and in 2024, we almost used 4710LC.

We didn't find the packaging information for the DMD batches purchased in the first half of 2024 or before. In the second half of 2024, we have recorded the batch information of DMD, they are 2347 and 2338.

Then, before 2024, the proportion of bad spots is relatively small, and it is more obvious that BBF or abnormal pictures appear, which mostly occur in product defects or terminals.

Hello Junfa,

Sorry, maybe there is some misunderstanding. When I asked, 

Aaron Black said:

Does this not occur in lower batch numbers?

This had to do with your testing batch, not the production batch. As you you've stated before: 

junfa ye said:

6. When there are many machines aging at the same time, a small proportion of machines will see that DMD shows one bad point or multiple bad points.

When you test with a lower amount of machines are there no issues?

I'd like to hold a quick meeting with you and your FAE. Could you have them reach out to me and we resolve this off the E2E forum?

Best,

Aaron

Hello  Aaron,

Sorry, I misunderstood. You mean, if there are fewer machines in my test batch, will there be problems?

My answer is that he will also appear, because sometimes only a few sets are tested.

As I mentioned the production batch, this information should also be the key information.

I'd also like to take part in a meeting to discuss, but I'm sorry, we're going to have a vacation.

After the middle of next month, I will invite my FAE to come to see this problem. Then, if it is convenient for him, we can also have a video conference together.

Please wait patiently, thank you!