It is always another little problem. But, on occasion, a problem is an excuse to have a little fun while getting the job done.
Enter an old WYKO interferometer with a composite video output. The gear is a critical piece of kit used to monitor the deformable mirror used in the adaptive optics system. The images from the interferometer are analyzed by a windows PC with special software. Windows, as in Windows 98!
Updating this computer has been on our to-do list for way too long. Time to get it done… When installing the new computer and video board to update this system I found that the video quality was quite bad. It would tear about halfway down the image, something wrong with the horizontal sync? Looking back it was not great on the old system, but usable. On the new system it was just not going to work.
Something wrong with the video input card? No. Breaking out a ‘scope it is clear that the video amplitude from the WYKO is much too low, about half of what is needed for good 1Vp-p video. The sync pulses are only about 150mV below the reference level, this should be closer to 300mV. Additionally there was a notable DC drift in the bias level, neatly explaining why the image would break up halfway down the screen.
I DO NOT want to work on the interferometer. Not only is it a major chore to get out from under the optical bench where it lives, a major chore to get it back and optically align it, it is also a piece of twenty year old tech with no spare parts available. I had to replace the laser in the unit last year, doing that again is to be avoided. Visualize worming your way out from under an optical bench with a 75 pound piece of gear on your chest! The video signal looks fine other than the gain and bias shift. Maybe just an amplifier to clean it up?
A quick check finds many video amplifiers available online, but not with the needed features, or limited documentation as to exact performance. We need one channel-in and three out. One for the computer, one for a local monitor used in optical alignments, one for an Axis web-based monitor used for remote operation. We had been using a cheap distribution amp, but it did not provide any gain, bad video in-bad video out.
A little more research quickly locates a chip that looks like just the thing! The THS7374 is a nice little chip. A four channel video amplifier with 6dB gain, DC restore and filtering built in. Exactly what I needed! It just needs a handful of capacitors and resistors to set it up. I did not even need to design anything, just build the evaluation circuit shown in the data sheet. How can I get so lucky?
There is only one problem with the THS7374, it is only available as a TSSOP-14 package. This means tiny! The entire part is 5x6mm in size, with leads 0.65mm apart (0.025 inch). This can be soldered by hand without too much trouble, if I had a circuit board, which I did not.
My answer was to cut a small piece of single sided copper-clad circuit board just a bit bigger than the part to create a breakout board. Into the cladding I would cut a pattern that I could solder the chip to. This included a decent ground plane under the part, and a nice power lead in with a place to solder filter capacitors right at the power pin. The individual pins I broke out with 30AWG lead wire to pads at the edge of the breakout that have about 0.1″ spacing.
A bit of fine work under the microscope with a very sharp knife was needed to create the breakout board. A careful inspection to insure there was no little copper shavings left over creating a short circuit. Then another round of very tedious work under the microscope to mount the chip
Fortunately the THS7374 has a very convenient pin layout, a number of the pins that could be soldered to ground were grouped together and could be soldered to the same pad. This included a pair of enable lines. Likewise, since I was setting up as a distribution amplifier, with one input feeding four outputs, I could solder all of the input pins together to the same pad. All four inputs pins are grouped together. I like the team that designed this chip!
The breakout was then soldered to the project perfboard with copper braid (solder wick) to create a low impedance ground plane connection. The perfboard was then installed in the case of the cheap distribution amplifier that also donated connectors and a power supply.
Plug it all together, hook up an oscilloscope to check the output… A very nice looking video waveform. Unfortunately the video on the computer still looks terrible! Something is still wrong. It is better than it was, the horizontal sync is good, but there is still some sort of bias shift between lines. Back to work, and keep looking…
Update: I just took a little more gain. I cascaded two of the amplifier stages in the THS7374 to get x4 gain, all is good!