I was looking at adding a Y/C video input to an old TV. But it's got a hot chassis so it's not a trivial task. I thought i could just capacitively couple the Chroma line, but then i looked at the "input isolation board" and the circuit looks like this:

enter image description here

The tuner is capacitively coupled as usual hot chassis TVs. But the Audio and Video lines have active circuitry for coupling to the TV. Why is this? Why not just a couple of capacitors like the antenna input uses?

And if I were to replicate such a circuit, the only references I found to "video isolation" are from this part by Broadcom. The posted circuit only calls the part "AVM-1 IVM-2 (TLP651)". TLP651 refers to an old (no longer in production) Toshiba optocoupler. I'm guessing the module has a TLP651 inside?


3 Answers 3


Why not capacitors? Because huge values are required: notice the 0.47 and 47µF parts feeding audio and video to the isolators. Even if we don't know what impedance these isolators use, we can observe that signals are fed to them through coupling capacitors. Similarly for the output side, C133 and C123. Such values would terribly violate the ground leakage required for safe operation, and the injected ground current would utterly stomp any input signal, leaving garbled or absent video and total audio buzz.

Remember that the chassis has about 120VAC on it, plus a DC offset. Or 240V as the case may be.

The IVMs likely contain some manner of isolation amplifier circuitry; given the pin connections, they're certainly not just a bare TLP651 (a photodiode receiver with integral BJT amplifier, comparable to 6N136, SFH6345, etc.), but there is at least some bias circuitry and likely a transistor or few, or op-amp, driving the LED and receiving the photodiode. (The integral transistor, with base wired to pin, could even be wired as a transresistance amplifier, which would be helpful to attain the required video bandwidth -- about 6MHz.)

Optoisolators can be problematic for analog isolation, as they aren't well controlled in terms of gain (CTR). In this case, it would manifest as unreliable audio and video levels. Video is typically AC coupled into a DC-restore or active sync-clamp circuit, restoring its baseline value, so that's no problem; and audio is AC-coupled in general, so the mean (DC) value is irrelevant. The video amplitude can either be adjusted a similar way (based on "front porch" voltage for example), or left alone for the user to solve (within reason, a weak video signal can be compensated by increasing brightness/contrast and tint). Likewise the user won't notice a difference in audio level: users don't typically set audio by number on the volume control, they adjust it until it's loud enough to hear. So despite the limitations, optoisolators are an effective solution here.

From a design perspective: nowadays, digital isolators might be used for the same functionality; a sigma-delta type analog isolator, with digital isolator core, might be used for the audio. But the video is a bit more problematic: HDMI requires quite high bitrates, challenging even for modern digital isolators. It's much easier to ground the chassis and use standard interface chips. An LCD TV doesn't require anything at mains voltage, either (rectified i.e. 160/320V DC, in contrast to analog TVs that were quite economical to build that way), they only use low voltages -- so a power supply is needed anyway, and making that supply isolated is only a small cost adder.

A grounded chassis also makes it practical to have oodles of inputs; a typical TV/monitor might have several input channels to select from, in multiple formats (composite, component, VGA, DVI, HDMI..), and having to isolate each one would be very expensive. Back in the day, with only two signal inputs, isolation was still an effective solution though.

  • \$\begingroup\$ do you think the example circuit on page 10 of the PDF linked in my question is similar to what's going on in these modules? it's basically an input buffer, a video-specific optocoupler, and an output stage built with 3 2N3904s including Biasing for the optocouplers transistor \$\endgroup\$
    – hjf
    Apr 25 at 10:43
  • \$\begingroup\$ Yes, that is possible. They're also using the internal transistor as part of a transresistance amplifier, which checks out. \$\endgroup\$ Apr 25 at 11:44
  • \$\begingroup\$ Broadcom does a 17 MHz BW optical isolator (HCPL-4562) which is similar to the part in the schematic. The datasheet has example circuits (typical of HP) and Mouser has one off's for a few £ \$\endgroup\$
    – D Duck
    Apr 26 at 8:05

The antenna signal and ground can be capacitively isolated as RF only requires high frequencies to pass and low frequncies like mains freqency is not needed, so a small capacitor value passes high frequecies and does not pass the mains frequency which must be isolated.

For baseband video and audio, the passband of the signals extend down to mains frequencies, so the capacitance needed would be larger to pass these frequencies, and thus larger cap would not isolate the audio/video grounds.

Therefore, there seems to be a local transformer isolated supply for audio and video isolator modules, used to pass the signals from isolated input side to non-isolated chassis side.


if you put the audio through a capacitor the signal will be dominated by the half-wave mains power voltage that the hot chassis operatas at.


simulate this circuit – Schematic created using CircuitLab enter image description here


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.