I'm designing a device that will receive and digitize analog video signals. In analog video, it's recommended to capacitively couple inputs. A typical input might look like this:

Video input circuit

In this case the signal is on connector pin 2, then there is a termination resistor, and a capacitor connects the signal to the rest of the receiver. DC current returns via connector pin 1 and the video cable.

Such a configuration is recommended by this Maxim app note (page 2) as well as various other places. But none of them seem to discuss grounding. It seems as though this app note (and others I have found) are trying to isolate the two devices, while also assuming they have a common reference!

The problem is that there's no obvious reference or return path back to the source for the signal that passes through the capacitor. If the receiver (or the source, or both) were battery-powered and not connected to anything else, the whole electrical potential could float and follow the signal waveform and the ADC would see nothing.

It would be possible to connect the signal return (pin 1 of the connector) to the receiver's circuit ground, but this could create a ground loop, would defeat isolation among the connected devices, and would expose the receiver's ground to EMI/ESD coming through the connector. It seems better to keep the cable isolated from the receiver's circuit ground.

So I think it might be necessary to include capacitor C2, as shown here:

proposed alternative circuit

Otherwise the AC signal passing through C1 might have to go all the way through the power mains, or some other undesired route, to return to the source. So my question is whether capacitor C2 should be included in the circuit (and if so, what value it should have; I guess it should be large relative to C1), or if it is not necessary for some reason that I have missed, or if something else should be done instead.

  • \$\begingroup\$ Use the extra cap (so the charge thru C1 has a return path), and duplicate the cap value. Will the pulse-top droop be acceptable? (the sync pulses)? \$\endgroup\$ Commented Jun 21, 2019 at 16:11
  • \$\begingroup\$ I think you are on the right track here. The C2 capacitor is an adequate return path, but effectively C1 and C2 are capacitors in series for this path even though the ADC is between them. So the capacitance is reduced, but reactance is increased. That matches my intuition that C2 should be large, to minimize this. \$\endgroup\$ Commented Jun 22, 2019 at 10:54
  • \$\begingroup\$ I'm finding some additional information such as: epanorama.net/documents/groundloop/equipment_grounding.html which says basically that most of the time, the grounds are all just connected together and the user must deal with any resulting ground loop problems. Maybe I'm just trying too hard. \$\endgroup\$ Commented Jun 22, 2019 at 11:13

1 Answer 1


Video receivers use an active DC restore circuit during Hsync interval to restore the video white level to 0V.

A less precise method uses video with negative sync tips to restore DC with the series cap and shunt reverse biased diode to clamp the negative sync tip to -0.6V such that back porch DC level is stable.

The more precise active clamp uses and suitable analog switch or FET transistor to short circuit the video during the back porch for ~1us or less with suitable RC time constant from 75 Ohms such that DC is restored after sufficient lines of video in a repetitive pulse clamp process.

You get to choose the C value based on x<1% droop at the end of the video line from input load current and choose switch resistance << 75 Ohms to respond as fast as 75C =Tau can respond and result in x% IRE DC offset for white level.

  • \$\begingroup\$ This is all true, but not really relevant. DC restore only trims the charge on the input cap so that the signal remains within the range of the ADC and so the digitizer knows where black/white are. Put another way, if you pretend this is an audio signal which is intriniscally pure AC and doesn't have/need any DC restore, the same grounding problem arises. (Maybe I can find an answer already out there for someone working with audio!) \$\endgroup\$ Commented Jun 22, 2019 at 2:35
  • \$\begingroup\$ DC restore is standard practice as I defined and very relevant. Why do you not understand? It is a precise 0V clamp. \$\endgroup\$ Commented Jun 22, 2019 at 2:39
  • \$\begingroup\$ It solves a different problem. Think of it another way: DC restore operates only during the porch and/or sync periods. It's completely inactive during the video portion. How could it have any effect? \$\endgroup\$ Commented Jun 22, 2019 at 10:30
  • \$\begingroup\$ @fluffysheap don’t assume I’m wrong until you ask a better question. The short circuit to cap to ground stays at the potential because there is no sag current thru the cap in between to shift that 0V reference dV/dt=I(dc bias)/C=0 during each video line or <<1%. Or in other words the cap is an AC video short circuit with DC restored. This is 50 yr old technology. \$\endgroup\$ Commented Jun 22, 2019 at 14:27
  • \$\begingroup\$ Of course differences in grounds between source and load are corrected with this method and may include some line frequency ground noise as long as it is not common mode to signal. \$\endgroup\$ Commented Jun 22, 2019 at 14:37

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.