- Ask a related questionWhat is a related question?A related question is a question created from another question. When the related question is created, it will be automatically linked to the original question.
Hello,
I am looking at the datasheet of the DLP7000 DLP DMD ( http://www.ti.com/lit/ds/symlink/dlp7000.pdf ). In the first page description it says :
high speed pattern rates of >32000 Hz (1-bit binary) and >1900 Hz (8-bit gray).
I am obviously missing something as if I can switch the mirrors at 32khz then for 8 bits of resolution per mirror then 32,0000/256 = 125hz. How is it possible to achieve 8 bits of resolution at 1900Hz with a maximum 1 bit pattern rate of 32,000. I cannot find any documentation explaining this process or anything to shed some light on it.
In my mind if I want 8 bits of resolution at 1900Hz then for each 1/1900 time period I would need to be able to switch the pattern at 1900*256 = 486,000Hz
In additon to this there is a quoted "Micromirror switching time at 400Mhz" as 43 µs, which equates to 1,000,000/43 = 23khz, not 32khz ?
Thanks for any explaination.
Rob
Hi,
Thanks Trevor. That clears up some of my confusion. However I am still confused as to how modulating the light source will give me the extra resolution. I have attached an example of how I think this is working, and perhaps you could use this to explain to me how the extra resolution is achieved.
For the example assume 50us is the absolute minimum switching time of a mirror, and there is an array of mirrors 4x1. Assuming that 200us is the frame time, then with a minimum switching time of 50 us, it appears to me that I can only have 5 different values of intensity, 100%,75%,50%,25%,0% for this example. This is obviously with a constant light source.
What I do not understand is that if I wanted to achieve more bits of resolution, and therefore more intermediate levels of gray you suggest that this can be done by modulating the light source with pwm. So if I had a 50% duty cycle PWM on the light source, this will halve the intensity per mirror. But the light source is not per mirror but for the whole array. So this will just have the effect of giving me a dimmer image, with a smaller maximum white value. But the relative intensities will still only have 5 discreet values.
Based on the rates quoted in the datasheet, the 8 bit rates are faster than the 1 bit rate/256, suggesting that you can achieve more gray levels in the same frame period. The only way I could see to do this is if you were able to modulate the light source on a per-mirror basis.
I am obviously missing something fundamental to the mechanics of how this works :)
Thanks for any explainations.
Rob.
Rob,
Sorry for the confusion, that is not what I meant about the PWM signal.
Let me explain a little more fully. The 8-bit sequence can be represented by the image below. This gives you an 8 bit gray scale image when integrated over the frame. So for full white, the mirror would remain "on" for b0, b1, b2, b3, b4, b5, b6, and b7. The limit for how fast this 8-bit sequence can be displayed is clearly the speed with which the fastest bits can be displayed (b0).
In our 8-bit sequence, we are able to achieve faster speeds by also modulating the light through a PWM signal. This is able to increase the speed with which the smallest bits are exposed. So, without getting into the specifics of the pattern sequence, for example, b0 can be exposed by switching the mirror to the "on" state and then the exposure can be quickly stopped by turning off the light source, the light source is then turned on to expose the next bit which is already switched into place, and so on. This does introduce some dark time into the 8-bit pattern sequence. The bits are also shown in a more optimized order (i.e. not the b0 - b7 order shown above). Thus we are able to achieve 1,900 Hz 8-bit patterns on the DLP7000.
Let me know what questions you have!
Best regards,
Trevor
