An analog camera and a frame grabber that pass their own tests and still refuse to show a picture are not describing a dead signal. The error "can't lock on video source" is precise: video is arriving, but the board cannot find stable horizontal and vertical sync in it. That points the diagnosis at four things, in order of probability — a half-seated capture board, a damaged 75 Ω video cable, a camera power supply that sags under load, and a video-standard setting that does not match the camera. Each check below takes minutes, and the power check comes with the arithmetic you need to judge a sagging rail honestly. The guide ends where most analog repairs end: the swap test that names the failing part, and the current hardware that replaces an analog chain you can no longer keep alive.
What "can't lock on video source" means
Analog composite video carries its own timing. Every line ends in a horizontal sync pulse and every field in a vertical sync block, and the grabber rebuilds its pixel clock from those edges. In the usual phase-locked-loop mode the board derives all of its timing from the sync information inside the video, which is why no lock means no picture at all.
The two monochrome standards fix the numbers. RS-170 is 525 lines at 60 fields per second, 2:1 interlaced; CCIR is 625 lines at 50 fields per second. Either way the signal is 1 V p-p into 75 Ω, with the sync pulses occupying the bottom 0.3 V or so of the waveform. "Can't lock" means the board's sync circuit saw no clean edges in its search window: the pulses are missing, too small, buried in noise, or the board is searching in the wrong standard.
That separates this fault from "cannot detect video source", where the grabber sees no signal at all and the fault lives in the cable path — Frame grabber cannot detect video source: signal checks covers that ladder. A lock failure is subtler: the digitizer test passes, the board works, and the analog timing chain does not.
Check 1: reseat the capture board and try another slot
A board that fails the basic digitizer test outright is often not faulty at all — it is not fully seated. Power the host down, pull the capture card, inspect the edge connector for oxidation, reseat it firmly and screw the bracket down so the connector cannot creep out of alignment. Then run the diagnostic again.
If the test now passes but lock still fails, move the card to a different slot before you condemn it; a marginal slot contact can pass a self-test and still disturb the timing reference. If a second computer is available, one more move separates the board from the host entirely: same board, same cable, same camera, different machine.
Watch for one more pattern: if the behaviour changes as devices are added around the card, or a second capture board shares the host, suspect a resource conflict before the video path — the guide on resolving machine vision frame grabber resource conflicts works through that diagnosis.
Check 2: swap the video cable and check the 75 Ω termination
Composite video is unforgiving about cabling. Use 75 Ω coax with BNC connectors; a crushed patch cord or a broken shield attenuates the small sync pulses long before it visibly harms the picture, and 50 Ω coax from a test-equipment drawer is not a substitute.
The test takes two minutes: run the camera to the grabber through a short, known-good 75 Ω cable with nothing else in the path. If lock appears, the original cable — or something hanging on it, a splitter, an extension, a damaged BNC ground tab — is the fault.
While you are at the input, check the termination. An input left in high-impedance mode without a 75 Ω terminator shows roughly twice the proper amplitude, with reflections that scatter sync edges. A correctly terminated input shows about 1 V p-p at the grabber connector, with the sync tip at the bottom of it.
Cable faults leave picture signatures of their own. An image torn in half by a black line points at the cable first, as Analog camera image split in two with a black line: fix the cable explains, while a connected camera that produces no picture at all is a different path again — see Camera connected but no image: check the frame grabber cable wiring.
Check 3: measure the camera's power supply under load
This is the check that settles most intermittent lock errors. A camera that loses its rail for even a moment restarts its sync generator, and the grabber reports a lost lock while the video cable was never the problem. The classic field signature is a lock error that appears exactly when another machine on the bench starts.
Measure at the camera connector, with the camera running:
- Read the camera label for nominal voltage and power draw. Most box-style analog cameras run 12 V DC with roughly +/-10 % tolerance.
- Measure the DC rail at the camera connector under load, not at the supply terminals; cable drop and shared loads hide sags.
- Watch the rail while other equipment starts. A visible dip at that moment is the fault, even when the average reading looks fine.
An enclosed, regulated industrial supply sized with that headroom from day one ends this class of fault; the power supplies category lists the enclosed options:
- Type
- Power cords and supplies
The LRS-75-48, for instance, is an enclosed 48 V unit rated 76.8 W — the sizing arithmetic, not the rail voltage, is what carries over.
Check 4: match the video standard and sync mode
If seating, cable and rail are clean and lock still fails, configuration is the remaining suspect. Two settings dominate.
Video standard: a board set for CCIR, at 625 lines and 50 fields per second, hunts in the wrong window for an RS-170 camera running 525 lines at 60 fields per second, and reports no lock or locks onto nothing usable. Set grabber and camera to the same standard, then record it in the machine file. The older application notes on selecting an analog frame grabber for LabVIEW vision remain a clear map of which standard and sync options belong to which camera family.
Sync source: most composite set-ups run in phase-locked-loop mode and derive timing from the video itself. Some cameras instead expect external HD/VD sync lines driven from the grabber's I/O connector, and if yours does, that harness is part of the timing chain. A frame grabber I/O lead such as the one below carries those lines from the board's connector to the camera:
- Type
- Frame grabber I/O cables
- Length
- 6-8 m
One expectation to reset: reinstalling acquisition drivers or the configuration utility cannot fix a lock error, because the failure sits in the analog front end before digitization. Three driver versions produce three identical errors. If the board does not appear in the configuration tool at all, that is the driver-level fault covered in Frame grabber driver manual installation: Device Manager fix.
Once the board does lock, picture-quality faults have a separate ladder — rolling, tearing and the settings behind analog video scanline distortion live in the grabber's acquisition settings rather than in the timing chain, and how the digitized frames reach your vision software is covered in Analog frame grabber integration with vision software.
Isolate the camera from the board, then decide on replacement
With the first four checks done, a swap matrix names the failing part:
- This camera into a second grabber input, or a second board: lock appears — the original input or board is suspect.
- A second camera into this board and input: lock appears — the camera's sync generator is failing.
- This camera and board in a different computer: lock appears — the host is the problem, not the video chain.
- Everything unchanged except the short test cable from Check 2: lock appears — the original cable was the fault.
If two boards behave differently in the same slot with the same cable and camera, the failing board is the part — and at this age, the honest question is whether to repair at all. Analog frame grabbers and RS-170 cameras are out of production. If a legacy machine must run while you plan, Identify a frame grabber model from its part numbers helps you order the exact board, and the host side is covered by PC requirements for a frame grabber: PCIe slots lanes and chipset.
The cleaner decision is migration to a digital interface. Camera Link is the closest direct successor for this generation of monochrome cameras, and the current parts are a grabber and one cable:
- Interface
- Camera Link
The MV-GC1102IOL is the Camera Link frame grabber; the LS-CBL-0191 is its SDR-to-SDR camera cable at 5 m. GigE, USB3 and CoaXPress are equally current — the frame grabbers category lists them all — and the full migration from an NTSC-era analog chain is walked through in Replacing an obsolete analog NTSC frame grabber with USB3 Vision. For bring-up of the new board, Frame grabber image acquisition for Camera Link and CoaXPress is the reference. To have the replacement specified end to end, start at the configurator:
The configurator matches a camera, lens, light, light controller and cables to each other, step by step, and lists only parts that fit what you already picked.
Prevention checklist for the next rebuild
- Reseat and screw down every capture card after any chassis work. Cards creep out of their connectors slowly and silently, and the digitizer test finds it before you do.
- Run one 75 Ω coax type, strain-relieved, with no daisy-chained extensions. Sync pulses die at connectors first.
- Give the camera a dedicated, regulated supply with at least 3x its running current in headroom. Startup dips then cannot reach the sync generator.
- Tie the camera supply's 0 V to the grabber chassis ground. A floating reference shifts the sync slicing threshold — the same principle behind trigger and power faults in Camera GPIO wiring: prevent damage and triage power faults.
- Write the video standard and sync mode into the machine file. The next engineer should not rediscover CCIR versus RS-170 the hard way.
- Keep a known-good spare video lead on the shelf from the cables category. The two-minute test in Check 2 needs it.
- Label which grabber input feeds which camera. Swap tests are only fast when the map is obvious.
When lock errors follow the board rather than the cable or the supply, stop repairing: the migration path above is the exit, not one more driver reinstall. If a legacy board must be identified from its markings before you order anything, contact us with photos of the board and its connectors and we will help read them.
Questions
Can a failing camera cause can't lock on video source with a healthy grabber?
Yes. A camera's sync generator can weaken or die while its video amplifier still produces a picture. Move the camera to a second grabber input or a second board: if the error follows the camera, replace it. Some box cameras also provide a test-pattern output; if the board locks cleanly onto that pattern, the camera's timing side is fine and suspicion moves back to the cable or the input channel.
How long can a composite video cable run before sync lock becomes marginal?
Keep thin patch cords below about 10-15 m. Proper 75 Ω video coax of the RG-59 class is rated for composite runs of a few hundred metres, because attenuation rises with length and bites hardest at the small sync pulses at the bottom of the waveform. If lock is intermittent on a long run, measure the video at the grabber connector: much under 1 V p-p, or a soft sync tip, condemns the run before any board is blamed.
What should a healthy 12 V camera rail measure at the camera?
Between roughly 11 and 13 V with the camera running, measured at the camera connector rather than at the supply terminals, and with no visible dip when other equipment on the bench starts. Ripple should be small enough that the sync tip stays clean on an oscilloscope trace. Below about 10.8 V, the -10 % point of a 12 V rail, a camera's sync section can drop out while the picture still looks present — the classic signature of an intermittent lock error.
Why does my grabber lock when cold and lose sync after an hour?
That pattern points at thermal drift in an aging board: component values in the sync-recovery circuitry shift as the chassis warms. Rule out a supply that only collapses when warm first, then swap boards while keeping the slot, cable and camera identical. If the fault follows the board, replace it — and plan the digital migration at the same time, because analog boards of this generation are out of production.
Do current digital cameras remove this failure mode entirely?
This one, yes. GigE Vision, USB3 Vision, Camera Link and CoaXPress links carry pixels with protocol-level timing, so there is no composite sync loop to fall out of lock. The equivalent faults move to link bandwidth, cable quality and trigger wiring, which are checked with different tests. Migrating an NTSC-era analog chain end to end is routine work; plan host slots and cable lengths before ordering parts.