Short to Power Protection for ADV7180 and ADV7393

We probably just need to protect against the short to 28V, and not the load dump on top of that.  Does this make AC coupling a better solution?

For AC coupling the important factor is the rise/afll time of the voltage. A momentary contact of 28V will have a faster rise time then just a plain load dump.  The key to this is the video chip ESD protection circuit is spec'd to handle 2000KV HBM or about 200uJ energy.  The key is to prevent 200uJ of energy getting dumped into the chip.

I'd suggest 2x buffering with a 75 ohm inline resistor, a schottkey bridge diode to ground and power then the cap.  The bridge diode will redirect the rising/falling edge energy to the power rails, the 75 Ohms limits the energy getting dumped into the buffer.  the ADV7393 would see no voltage spike at all.  Remember removing contact to the 28V will cause a negative spike on the left side of the cap which is about the same energy content as the rising edge. 

ADV7393 -> 2xbuffer -> 75 Ohm -> bridge diodes -> Cap

The basic approach can be taken with the input side.  Absorb or redirect the 200+ uJ of energy before it gets to the chip pin.  A ax buffer with a 75 Ohm input load and bridge resistor would work fine.

Cap -> bridge -> 75 Ohm to ground -> 1x buffer -> ADV7180

If my one line diagrams aren't clear I can sketch something up.

It would be helpful to have a sketch.  I was still a little unsure of requirements, but it looks like we will want to protect against 40V shorts, and forget about any transients and load dumps.  I look forward to seeing your sketches. Thanks

What value would the caps need to be?

I thought about your problem a little bit last night.  First of all the 40 Volts is a continuous connection to the video line, i.e. it's not a spike.

Attached is a sketch of what I had in mine.

The first 2 circuits block the 40V from getting to the chip inputs or outputs.  Q2 is a switch that opens the video when Q1 is turned on blocking te 40V from even getting to the input or outputs.  The trip point is based on Vthr of Q1.

The second 2 circuits are the capacitive blocking you were talking about earlier with a bridge diode to clamp the rising/falling edges off the signal.  The diodes should be plain silicon type, not schottkey. 

<fineprint>

Full disclaimer:  These are just sketches of what might work and have not been tested or tried.  I've used similar approaches with other signals but have never done this with video.  Full analysis, tweaking and testing would need to be done.

</fineprint>

Let me know if you build and test these, I'd like to know how well they work.

From the literature, I've seen cap values up to 220uF with sag compensation circuity.  One problem with capacitively coupled video is the video will have a DC bias.  The receiver has to either restore the DC bias level or be able to handle negatively biased video. 

Search the web for AC coupled video output, video output sag, video DC restore..  A lot of articles will pop up.

Okay, that is about the same size as I was thinking and it is also what I've seen used typically.  Although most of the time I've a polar electrolytic used for those caps. Won't that cap be damaged if the line was shorted to 28V?  There would be a negative 28V across the capacitor during the short.  I suppose a bipolar 50V cap could be used, but I have to imagine those are large, expensive parts...

The cap voltage rating has to meet your blocking voltage requirements.  Bipolar caps are not overly large.  You need to watch out for ESR at video frequencies like 3.57MHz.  They won't list the specs there so it has to be done experimentally.  This is why I'm still leaning to the active blocking and/or double insulation so we don't have to protect against double faults.

If you do go the cap route, you still need to check the sag circuit and handle DC restore at the receiver.