Weekly Watch: EEV Blog



The entire EEV Blog channel is brilliant; it's some rando Strayan in a big electronics engineering lab, with the occasional intern, who fishes things out of junk piles (or out of his mailbox, where viewers send them) and gleefully takes them to bits in front of a camera. Once in a great while he repairs something, or tries to.

This particular entry of Teardown Tuesday, I highlight because he is taking apart the front bits of a 1985 Sony Video8 camcorder, which unexpectedly turns out to be a work of origami wonder. Despite being packed almost solid with boards and parts it is entirely, beautifully, technician-serviceable. I think I'm even a bit too young to remember the days when you could take consumer electronics to a shop and have someone fix them when they got sad, but we're now past the days when even my father, the pig-headed professional circuit-board designer, could do it. Everything's irrevocably glued together now, because you're just supposed to throw it away when it spits up.

He does not seem to know what all of the adjustment pots on the boards are for. Iris is for the lens and AGC (automatic gain control) is for massaging the signal submitted by the front lens assembly. The rest of them are things he'd recognize immediately if he had to dig around in the guts of more old television cameras. The thing labeled MPX is a mixer or multiplexer of some kind, but BM, RM, B-Y, R-Y, and the like all have to do with adjusting various aspects of the video signal it's feeding into the back end. Given the specific adjustments, it looks like it handles everything as YPbPr (component) video internally, which separates out the luma (Y, the black and white part of the picture) and chroma (B-Y, the difference between blue and luma; R-Y, the difference between red and luma; and G, deduced from what's left over once you've figured out red and blue). Something somewhere also has to be inserting horizontal and vertical sync pulses into the video, or nothing would line up correctly. The signal that gets laid down on the Video8 tape is some kind of ugly composite, I'm sure -- the recorder takes down the exact signal a TV would see, which has everything smashed together -- but the viewfinder would be fed component video so they could get the sharpest possible picture onto the tiny tiny screen.

[When he gets the CCD sensor pried out later, he sticks it under the microscope and notes that it's a Bayer sensor layout, otherwise known as RGGB. This attempts to mimic the response of the human eye, which has two kinds of cones sensitive to green (the medium and long wavelength cones) and only one kind that isn't (the short wavelength ones), and therefore is quicker to detect weirdness in the intensity of greens. But if you look at Bayer's description of his own filter, he refers to the G sensors as "luminance-sensitive elements" -- he's using the doubled green sensor pixels to pick up the luma channel of the picture. It gives the equivalent of 4:2:2 chroma sampling, which is at least no worse than anything else at the time.

Getting the video straight off a sensor that has the same effective resolution as the Video8 tape could carry also conveniently gives you separate R G B signals, with a stop at the end of each horizontal line to insert HSYNC, and a refresh at the end of each full read to insert VSYNC.]

What the G1 and G2 pots are for, I'm not sure. The context in which I'm familiar with G1 and G2 adjustments are on the yoke of a cathode ray tube. I see the deflection controls on the board jammed into the viewfinder stem, but no gain controls, so G1 and G2 all the way over in the sensor assembly might be for the wee little CRT. The signal sent to the viewfinder is flipped left to right, incidentally; the picture you see is bounced off that mirror jammed into the right angle. Otherwise it would be like looking down the barrel of a rifle. Long vacuum tubes are long.

Comments