In response to tvropro's interest in having duplicate a/v outputs:
Originally Posted by tvropro
You could add a second set of composite video outputs to any CECB. Just take the video before the 75 ohm matching resistor to ground and install a buffer amp there. Then feed it to 2 outputs. It isn't that hard to do.
In some cases, depending on the drive strength of the driver, you can get away with just connecting another 75 ohm resistor there. Often there is a buffer amp IC and the simple way to construct a second buffer is to take an identical IC and bend the pins so you can piggyback it on top of the other IC. You bend the output lead up so it isn't connected to its counterpart and solder a wire or a resistor there to go to the jack. You may need to double up on the decoupling caps. The piggybacking won't work if the amplifier needs feedback resistors. If you can find a board with the connectors and buffer amplifiers on it, you are golden.
The audio is pretty easy. Just connect another 1K ohm output resistor where the first one connects. On some boxes, such as the shenzhen MTC AT2016 family (tivax, rjtech, etc.) the mute is implemented by shorting across the output of that 1K ohm resistor so your second audio output won't be muted (but that may be a good thing).
For both audio and video, you may find a series cap in series with the output resistor and that you may want to duplicate as well. For audio, it is simpler to use an external Y adapter. You may be able to use one on the video as well. If your cables are short (under 6 feet) they are lumped sum capacitance rather than a transmission line at 6Mhz. For both audio and video, your signal will be somewhat weaker but the boxes on the other end may boost the gain to compensate (or you adjust brightness and volume) so you can get away with stuff like this.
These changes would be pretty simple on the shenzen ATC family. It uses a FMS6143 video amp ($0.96 digikey). The schematic is labeled "change to eliminate the IC" (and let the zoran drive directly), but since it wasn't changed before certificaitn they are probably stuck putting the IC on the board especially since it is both filter and driver. If they did eliminate it, that would probably have involved shorting C34 and U6 pin 1 to pin 6 which affects us very little. If there is an empty pad we either cut the trace and piggyback two ICs or leave it there and install one with pin 6 bent up into the air. The IC is surface mount but it is a SOIC (0.05" lead spacing) so it isn't hard to work with.
But wait, there is more. Though it doesn't show it on the schematic, that IC is actually a 3 channel amplifier and only one channel is used. Short Pins 1,2, and 3 (the inputs) together and you can tap off two extra video signals on pins 6 and 7 with a series resistor and capacitor. This chip doesn't actually appear to need the input capacitor shown on the schematic. The zoran chip appears to have some extra video inputs as well, though there are two problems: 1) software probably doesn't support, and 2) they are on BGA pins you can't get too unless the board designer included vias for every pin whether it was used or not and it does not look like that was the case.
The schematic is included in the FCC filings. Can't link but lookup UVD20071228001 under eas reports.
That board has a rather high parts count, unnecessarily. Each extra component costs you 1) the price of the component, 2) board area costs, 3) part placement cost, 4) yield reduction, 5) solder, and 6) reel feeder setup cost for each unique component. If I had designed the board, I might have been able to shave the cost enough to be able to throw in all the features allowed by law. There are lots of 2N3906 transistors (each accompanied by several resistors) that could have been replaced with 1/6 of a hex inverter and maybe 1 resistor. The RS-232 is one example, it produces a substandard 0-12V (instead of +/-12V output) which doesn't work any better than a 0-5V output. 9 components, could have used 1/3 of a hex inverter plus 3 resistors for 3-1/3 components. The smart antenna circuit could be done similarly. The mute circuitry (4 components, plus 10 on another page) should be totally unnecessary - you can do that in software, though it looks like they may have had a problem with spikes on power up that can probably be eliminated by being smarter about your DAC initialization sequence. The audio amplifier/filter uses lots of components. A mini-din connector for s-video costs $0.40 qty 1000 US from an expensive distributor, plus soldering. And the tuner/modulator with built in pass thru can't cost much more than the version without since it is just a few extra components stuffed on a module that already has sites for them and CECB's are about the only devices on the planet that don't do passthru. Asian companies do this a lot. They are too cheap to spend $0.50 on an IC so they try to roll their own, but if you need 21 extra components at $0.36 to avoid using a $0.11 IC, you are $0.25 in the hole even before you count stuffing cost. 12 components to mix left and right audio for the RF modulator when two 100K resistors and a cap would have done the job better. Why buffer something (with an input impedence of 10K) when you are going to put it through a 100K resistor? And why use two transistors and two resistors to make two copies of the video signal when you have two unused amplifiers in an IC on the same signal?