Black Level Clamp
“I’m looking for a simple GOOD black level clamp circuit for clamping a video signal.”
(From: Joseph H Allen (jhallen@world.std.com).)
This is actually a very interesting subject. Assuming you have a constant video level (or AGC), a simple diode clamp on the horizontal sync tips (diode clamp the sync so that the black level is where you want it) is actually a high quality black level clamp. The sync level is constant after all. I use this method in accurate video digitizers for DC restoring the video before putting it into an A/D converter. The cool thing about simple diode DC restorers is that the capacitance is usually much less than that of the analog switches needed in other types. For example one of the best analog switches is the readily available 74HC4316, but even this has 40pF (if memory serves) on its pins.
The really hard part is finding a high quality large-signal linear video amplifier. If the video is capacitor coupled to the video amp, then the average brightness level will change the voltage of the black and sync levels seen by the amplifier. This is ok if the video amp is linear, but most aren’t so the sync amplitude ends up changing depending on brightness. I.e., the black level will change depending on the brightness of the image (this is completely unacceptable for medical image capturing devices, for example).
The way linearity is measured for video amplifiers is the term “differential gain”. It gives the largest difference in percent between a constant small amplitude signal (traditionally the color carrier is used for this) measured at different voltage levels (hopefully which sweep the entire output swing of the video amplifier).
Now you have to be very careful about manufactures differential gain measurements. Many of them play games to get even a mediocre 3% – 1% differential gain. Typically they specify this parameter with a reduced output range (when you really want the parameter to apply to rail-to-rail output swings so you can get a 2V signal needed for most A/D converters) or limit the input range, the gain or flat out lie (I have no idea where comLinear got the specs for their clc520/522 variable gain amplifier, for example. It says .5% in the datasheet, but I measure it to be more like 5%).
All older video amplifier ICs (like the uA722 and NE592) are really lousy. Discrete transistor amplifiers also suck (many monitors use a cascode amplifier which is very bad). Most new ICs suck too- especially those which are labeled as video amplifiers. The only ones which I have been satisfied with are the newest current feedback OP-amps (the + side is high-impedance, but the – side is zero ohms. Usual op-amps are linear, but have a limited bandwidth because the – input is high impedance). These have enough open-loop gain so that they really are linear. One that I really like is the AD9617: .01% differential gain (!), 160MHz bandwidth, immense slew rate (settles to less than 1% of final value within 10ns or so) and only costs about $10.
If you need variable gain (for AGC perhaps), the best chip to use the AD834 500MHz multiplier. This little 8-pin chip is expensive (like $40) but it is the only thing that even approaches being linear (and even it is quite a bit worse than the AD9617).
(From: Brian Campanotti (bcampano@toronto.cbc.ca).)
Look at the Clamping ICs from Gennum Corp (part numbers GB4550 and GB4551). They do input buffering and clamping. They are a good front end to any video project.
(From: Mika Iisakkila (iisakkil@alpha.hut.fi).)
Since your application probably needs some kind of an input/output buffer anyway, you might consider some integrated DC restored video amplifier. Elantec (now part of Intersil makes excellent such chips.
EL2090 is really good as far as video quality goes, but a bit expensive for applications that don’t need 100 MHz bandwidth and near-zero droop. EL4089 is simpler and cheaper, but not quite “broadcast” quality. There’s also some new chip, but I haven’t looked into it yet.
For both chips, you’ll need to get the sample pulse from somewhere, so you can’t lose the LM1881. I’ve used the burst gate output from it to control 2090’s sample input, and quality of the result far surpasses my measuring instruments.
