R_silicon_to_ceramic

The value R_silicon_to_ceramic is determined from a 3D finite element model using the thermal software Flotherm.  There is a standard value if the active array is fully illuminated, which for this device is 0.5 ºC/W.  The example shown in the application report for a partially illuminated array covering 960 x 475 pixels is 0.9 ºC/W.  The value increases because the area available to transfer heat to the back of the package is reduced.  

Since TI already has a thermal model, if the customer can share the illumination profile, we can easily re-run the model to determine the R_silicon_to_ceramic resistance for this scenario.  It would be helpful to know the size and location of the of the illumination spot and also if it is uniform in power across this area.  

One point I will emphasize is that these calculations apply to CW (continuous wave) laser illumination.  If a pulsed laser is used, there can be transient heating of the mirror which can be significant if the pulses are of high enough power density.   

Best Regards,

Scott Overmann

Hi Matt, Scott,

Good Day. Please see below the response of our customer to all of your replies. Thank you very much.

I can tell you that we intend to illuminate part of a DLP2000 DMD. This area will be centered in the array area, and will be 553 pixels wide and 157 pixels high. This turns out to be about 4.18mm by 1.19mm. We plan to use a continuous wave, 532nm laser, and uniformly illuminate the area. Would you mind running your software and giving me the appropriate value for R_silicon_to_ceramic?

Best Regards,

Ray Vincent

Hi Ray Vincent,

Scott will be running his model today. Please give us another 24 hours or so to provide feedback.

Regards,

Matt

Hello Ray Vincent,

Assuming full array illumination of the DLP2000 DMD, R_silicon_to_ceramic = 8.0 ºC/W.  This value can be found in the DLP2000 datasheet.  I ran the model with an illuminated area of 4.18 x 1.19 mm and the resistance only increased slightly to 8.7 ºC/W.  

Best Regards,

Scott

Hi All,

Good Day. Another query from the customer. Please see below. Thank you very much.

I don't know the value of R_mirror-to-silicon. It says that this value should be included on the DMD data sheet, but it's not on the datasheet for the DLP2000. I also need to know both R_silicon-to-ceramic and R_mirror-to-silicon for the DLP9000, with a uniformly illuminated area of 19.3mm x 5.5mm in the center of the DMD.

I have another question about thermal damage of DMDs. The silicon SRAM underneath is relatively fragile. Is it a known problem that the SRAM becomes damaged from high irradiances of light?

We are equally concerned about the impact of thermal gradients if, e.g., half of the Pond of Mirrors is illuminated. For instance, will we crack the DMD near the regions where the illumination transitions from full to zero? Are there strategies for mitigating this risk, such as slowly increasing the incident laser power from to 0 to its operating power?

Best Regards,

Ray Vincent

Ray Vincent,

What is the power density of the laser illumination? I see the questions about "cracking" the DMD, but we are not aware of any issue with normal power densities.

The SRAM and the POM will not be damaged unless there are power densities in the hundreds of Watts/cm2.

I have shared the latest questions with Scott Overmann. We will be able to reply tomorrow.

Regards,

Matt

Hi Matt,

Good Day. Customer is considering up to 35 W/cm^2 in power density. They just want to make sure they cover all our bases. Thank you very much.

Best Regards,

Ray Vincent

Ray Vincent,

The DLP9000 with uniformly illuminated area of 19.3mm x 5.5mm has R_silicon-to-ceramic = 0.95 ºC/W.  

R_mirror-to-silicon is only dependent on the pixel node.  The DLP2000 and DLP9000 both use the 7.56 um pixel, so R_mirror-to-silicon is the same for both of these parts.  R_mirror-to-silicon = 4.42e5 ºC/W.  Note this resistance is a large value because it accounts for absorbed heat load going through a single mirror.  Also note that the mirrors are highly reflective, so only a portion of the incident illumination will be absorbed in the mirror.

Regarding incident power density of 35 W/cm^2, we have customers who illuminate DMDs with this much visible light for traditional projection applications.  

Best Regards,

Scott