Detecting active video

Security/surveillance and professional broadcast video systems commonly use remote video cameras that output just black screen while there is no activity in front of the camera, and many consumer devices output a blue screen when no signal is present. The need to distinguish between inactive and active cameras or signal feeds is common. When video activity is detected in a security system, for example, it needs to automatically switch over to the camera with activity.

The challenge is to detect the presence of active video on a NTSC/PAL composite signal when the camera is sending a black or blue screen with no active video present. During non-active video, the video feed still supplies H and V syncs but the active video field is black.

There are two subtle issues to consider when working with broadcast video and designing an active video detector.

First, TV transmitters, up link satellite, and commercial video cameras typically will combine the composite sync and the active video just before transmitting and need to insure that the video channel is working. They must detect the loss of video at the transmission point in the video broadcast and be notified immediately of such a failure. Second, broadcast TV will insert a blank field just before and just after the inserted advertisement. The design must be able to allow for this without indicating a loss of active video.

The solution is to design a system to detect active video above “blank” and generate a logic level indicating active video present for the entire field.

This article can serve as a guide to assist in designing such a circuit to fit the designer's objectives; it is not offered as an all-inclusive design. We also make the assumption the designer is familiar with standard video design.

View full size

Figure 1: Block diagram

There are four basic requirements for the design.

  1. Compensate for un-terminated and double-terminated input video feeds.
  2. Monitor the video for active content and generate a logic level when detected.
  3. Operate from a single 5V supply.
  4. Account for blue versus black screen detect.

Components required for this design will include a level shifter that will shift the negative Hsync tip levels above ground; a video buffer to reduce the input impedance's influence on the active video detector; a level detector to detect the active video field and signal active video presence; and a video amp.

An optimal solution will keep the IC count to a minimum by using a single highly integrated Video Front End (VFE) incorporating all of the key signal conditioning functions for analog video signals. This device can fulfill the level shifter and video buffer functions and act as a basic interface to the incoming video feed. The delay timer is a combination of this device and a rail-to-rail video amplifier. Ideally, the video amp should be able to double as the logic signal indicator of active video.

Keeping the IC count to a minimum, we selected the EL4501 VFE as the level shifter, video buffer and basic interface to the incoming video feed. The delay timer is a combination of the EL4501 and the EL8100. The EL8100 also doubles as the logic signal indicator of active video.

Using the EL4501 also has another advantage, since it provides a LOS (Loss Of Sync) output that goes high if there is no video sync signal and has a video amp output that supports driving cable.

We have added a simple video switch (the ISL43110) to isolate any loading of the video on the next stage when no active video is present.

Next: Key design considerations

Key design considerations

Several important design considering govern the design. First, you must compensate for different terminations. Video feeds can be un-terminated or double terminated both having an impact on the Hsync tip negative level. Since we are limited to a single 5V supply, we need to take into account the impact of the termination and adjust for negative Hsync tip levels. Refer to Figure 2 , NTSC Standard Video Wave Form, below.

The key to this overall design is using a single supply VFE such as the EL4501 with internal VREF =1.3V offsets which can be used to offset the incoming video. Doing so will allow the device to support a negative incoming HSYNC tip (-40 IRE or ~ -300mV) by adding +1.3V offset on the back porch. Now the VFE will support an un-terminated video feed which will have a 2X HSYNC tip and it will also accommodate a double terminated video cable with a 0.5x HSYNC tip. We set the gain of the video input buffer to 2x to recover the losses from a back-terminated output video feed.

The un-terminated video sync tip will be ~600mV times an amp gain of 2 will output sync tip of ~1.2V. The +1.3V minus the +1.2V leaves +0.1V from ground for the sync tip. More details on the actual circuitry appear below in the section on termination and detecting reference levels.

The second consideration is to monitor for active video. Reviewing the NTSC/PAL standards, you will find that not all lines will contain active video and some of the lines are reserved for other than active video functions. The filter needs to be designed not to trigger on these non-video functions, even though, encoded into the first 20 horizontal lines, sometimes this small amount of data might be detected as active video. If we delay the detection for about the first 20 lines in the vertical interval, the detector will be monitoring the correct active video lines. Also, TV stations normally have one video field that is black before and after advertisements so the filter may need to have enough delay to not detect ~20ms of black level video (one field).

The video black level is 7.5 IRE, which is ~54mV for NTSC. Thus, the video detector will have to be set to some value above 54mV.

Figure 2: NTSC standard video wave form

We selected about 75mV for the active video detect level. This gives us about 20mV of noise margin over the typical black level. To set this level on the EL4501, we use the Data Slicer comparator reference input (DS REF) to fix the threshold by simply adding a resistor divider from VCC to VREF as shown in Figure 3 .

The 30kΩ resistor and the 680Ω resistor will set DS REF to 75mV, with respect to the BACK PORCH. The comparator will now detect video levels about 22mV above the black level and thus detect active video. You can also use this technique to monitor for detecting active video above a blue screen by setting the threshold above the blue level.

Figure 3: Black level threshold

Next: Hysteresis for slow ramp edges, delay circuitry

Hysteresis for slow ramp edges

We want to insure the detection circuitry is not responding to noise during a slow edge. We need to design in some level of hysteresis to help prevent false triggering.

The video detector senses the output level of a filter with slow edges. The series 10kΩ output resistor and 0.33uF capacitor to ground control how fast the DS output transits to ground. If some of the blanking lines contain digital data, such as TTY, this series 10kΩ will slow the response and reject this digital data and noise.

Figure 4: Hysteresis

The EL8100 is a fast recovery amp so can be used as a comparator to detect the level of the DS output filter (10K series to the 0.33uf cap to ground). The filter will have slow rise and fall edges so hysteresis will be needed. The EL8100 has a 750kΩ resistor from the output to the '+' input to generate hysteresis. The EL8100 may be set up to give the reversed logic output; its inputs will be reversed to the 1.3V and filter output with the 1kΩ resistor in series with the '+' input to the 1.3Vs to allow for hysteresis.

We add about 6.8mV offset to the DS comparator output by adding a 300kΩ?'resistor feedback to the DS REF pin. This will help prevent the DS Output from jitter due to slow changing video through the comparator's transition point.

Termination and detecting reference level
The incoming video may come from a cable source with normal or double or no termination. This will cause the HSYNC tip to change from 286mV to double or reduce to 2/3. The back porch is always at zero volts so is the most stable reference point. The back porch pulse is from the sync separator to the DC restore to set the back porch to 1.3V. The 1.3V allows the sync double level of 572mV times an amplifier gain of 2 to have a 1.14 volt HSYNC tip with 160 mV above ground so as not to induce clipping. The black level can be 54mV times 2 for no termination and times 2 for amp gain to give 216mV. The video from the amp output is internally connected to the Data Slicer (DS) — input. The data slice detector can have the DS Ref set to ~250 mV. If the EL4501 amp gain is one the DS REF can be ~145mV. Figure 5 is for an amplifier gain of one.

There is also the video detect function. The Data Slicer (DS) inverts the video. The output will be high (open drain) with no video. If data is present in the vertical blanking intervals, then the DS output will go low and the filter input resistor value (10K ohm) can be set to be fast enough to detect the data or slow enough to reject the data. When the filter is set to be slow then it will take longer to detect video. The large pull up resistor (300kΩ) on the filter allows dark video pictures to not give false outputs.

Non-video delay time
We need to design the input filter into the video amp to be slow enough to delay any detection for a number of non-active video lines. A total of 1.3mS is the time necessary to wait for 20 video lines added to the 1 field time of one field of 16.6us (the advertisement black field), for a total of 18ms or more. When video is not present (no active video luminance above 75mV) the DS output is open drain. Since the DS OUT becomes an open source, you can conclude the charge path is the sum of two paths: 300kΩ from 5V and the parallel 310kΩ from the ~2V node. We can make this simplification because the current through the 30kΩ and 680Ω resistors is large as compared with the 300k from 5V supply rail. Thus, the node at the junction of 30kΩ and 680Ω can be viewed as ~2V supply regardless of the charge on the capacitor. Also, the amp's trip point is at 1.3V or about 1 RC time constant for the 2V charge path. So, the total current to charge the 0.33uF capacitor to 1.3V is essentially 22uA.

For standard and double cable termination we advise using 680Ω.

Figure 5: Non-video delay circuitry

A first order approximation of the delay time can be determined from I = C(δv/δt). So, the time it takes to charge to 1.3V is about 22+ms or the delay time. If video is present, then DS Out would be at ground, not allowing the RC network to charge. Thus the output of the amp would be low for active video. This 22+ms delay is one field (16.6us) plus 1.3ms into the second field to insure we are looking at an active video field and also ignoring the black field before and after an advertisement.

Concerning VCR/DVD blue screen detect, the 680Ω or 1.5kΩ DS REF set resistors would be set to a value to sense blue screen as the same as non-video. NTSC blue luminance is 18.5 IRE, by setting the threshold to 150mV (Blue luminance is 132mV), we added 18mV of noise margin. Thus, using a 1.5kΩ in place of the 680Ω resistor will set the threshold at 150mV. The circuit will now detect blue screen as non-video. The blue detect level may not detect some dark video pictures, as the average voltage level for the active video may be too low.

Next: Complete system diagram and test performance

Test performance
The results of testing the circuit: Loss of video to output high for no video was ~25ms; and new video input to output low for video active was ~3ms. These agree with the design goals.

This basic design using a video amp with a video front end provides a system to detect active video above “black” and generates a logic level indicating active video present for the entire field.

View full size

Figure 6: Complete system diagram using the EL4501 and ISL8100

About the authors
Rudy Berneike is a Senior Applications Engineer at Intersil, specializing in video and wideband amplifiers. He has more than 40 years of experience in the analog industry including more than five years at Intersil and 11 years at Elantec. Rudy has been designing analog circuits since the “mostly analog” days–he began his career building amplifiers using discrete transistors and resistors. He has authored numerous technical papers and articles. He can be reached at .

David Laing has been employed by Intersil Corp for the past 15 years in Analog, Multimedia and WLAN as a Senior Product Marketing Manager and for the last three years as a Senior Principal Engineer in Central Applications. Prior to Intersil, he was employed with Analog Devices and GenRad in a variety of design engineering and product marketing posts. 5 years as a Design/Application Engineer and finally Product Marketing Engineer. He
served in the US Navy for seven years as a Submarine Electronic Technician working on Poseidon navigation and computer systems and holds a BSEE from
Northeastern University. He can be reached at .

0 comments on “Detecting active video

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.