Hello all,
i have a relatively simple question, or at least I hope it is one of those. Currently in our lab, we are utilizing a DLP 3000EVM and we are dissatisfied with the speed in which we can display images to act as a channel scanning spectrometer. If we interface by USB and control it with Lab view the quickest we can go between images is ~ 150ms when using static image format. Now we have tried to use a work around using a Raspberry Pi 3 through the HDMI port, but we have run into image corruption/ inconsistent transfer of images over two different HDMI cables. As such we are looking into an alternative DMD that will suit our needs and based on recommendations, the DLP 4500 EVM may be the next logical step.
As such, when reading the manual, it states " Flash: Images stored in flash memory. The flash memory can store up to sixty-four 24-bit compressed images. "
Then goes further to state that a single 24-bit image can be composed of 24 1-bit images. Does this mean that we can take a sequence of 24 1-bit binary images and have them displayed in such a way that we can correctly and reproducibly regenerate our spectrometer filters?
What devices does the external streaming mode support? Can it be streamed directly from a Win8 laptop more reliably than a Ras-Pi 3?
Does the DLP 4500 support direct Lab View control over the mirrors in a manner consistent with the DLP 3000EVM, such that we can correlate a channel scan with a PMT?
To further state why we are looking into this newer system, say that our original spectrometer utilizes a Hadamard transfrom spectral set to collect a spectrum, of lets say Benzene. Our original process is constrained by the time limitations for our Labview code to create the path to the image, load the image, and wait for the image to fully loaded despite it being a 1-bit bitmap image. This results in a a time delay of about 150 ms, but we pad it to 200ms to ensure we don't accidentally add mirror flipping/jittering to the acquired channel. We repeat this process for a total of 680 wavelength channels to collect, as we have binned the mirrors to pixels 2 columns to increase the apparent speed. We do not gain much more information when our spectra are no binned. This sadly results in a minimum time of ... ~13.6 minutes for a 1Hz (1s/channel) to acquire a relatively high signal to noise spectrum, that is further used as a part of a training set for other applications. This dead time is something we are trying to mitigate and hopefully, someone with more knowledge into the workings of a DMD will be able to provide an insight to the query proposed.