I had to put the 16 bit driver circuit on hold for right now, there is an over heating problem with the ADC that I cannot solve right now, I have 2 boards made up and both have the same issue.
So I have decided to go ahead and build the fast 8 bit drive instead because I already know that it works flawlessly, and can still read all 3648 pixels from the CCD, It reads 3694 pixels, but the first 30 and the last 16 are dark reference or dummy pixels and are discarded.
I have already drawn the schematic using Design Spark and will post the gerber files so you can order your own PCB and build it. I am including the BOM also.
This is the schematic I drew up on Design Spark, this is an original design by David Allmon, whom I have been working with the last few months now. He developed this on a breadboard as a prototype and never took it further, so I went ahead and saw the potential of this design and decided it would work perfectly as an alternative to the 16 bit driver I have now.
This next set of panels are from the 16 bit driver circuit, when it works they beautiful and sharp as you will be able to see…When the AD7667(ADC,) heats up, they are not!
This is a spectral capture of my CFL (compact fluorescent lamp) using the 16 bit driver circuit. the peaks are inverted (a problem somewhere in the circuit,) but they are sharp and the resolution is apparently very good, thanks in part also to the optical mirrors and lenses in my set up.
Now this is a capture with no zoom, you can see further to the left the inverted peaks of the CFL.
Here you an see what is called the “dark counts,” on the 2nd plot line is just the CCD with NO light source, which allows you too save it and subtract it from your sample scan. The 1st plot line is our sample (CFL,) both plots shown are un-zoomed here, this principle will work in the same manner with the 8 bit CCD circuit driver.
As I stated earlier, I’ll still be working on this problem, but I need my project to start moving forward and it will with the 8 bit driver and still have very good resolution because of the optics and 1800 ln/mm holographic grating I am using the Raman edge longpass filter, the only difference will be digitizing a frame will take 32ms (8 bit) instead of 16ms (16 bit).
Other than that, there isn’t too much of a real difference.
Complete details on both of these projects can be found here;
Also, I am including this projects gerber files which will enable anyone to build and manufacture the PCB ciruit boards 🙂
This is the firmware: