Standalone barcode verification of model labels without a PC

You stick the same model-code barcode on every finished unit in a lot, and the failures you fear are dull ones: a label from the wrong model, or a print so light it barely reads. Both are catchable without a PC, a database or a PLC program. A fixed code reader or a smart camera can store the master string once, compare every read against it, and switch a discrete output to a buzzer or stack light on NG. This guide gives the decision rules: what the compare must check, how good the print needs to be, the pixel math behind the mounting distance, the wiring, and where a dedicated verifier still beats a reader.

Can a code reader verify a label with no PC at all?

Yes. A modern fixed code reader runs its decode, string match and I/O entirely on board: teach the master model code once in its setup tool, leave it powered from its DC supply, and it switches one output for OK and another for NG with nothing else connected.

Standalone here means the reader carries three things on board: the decode engine, the stored master string and the I/O that announces the verdict. You configure it once from the vendor's setup tool over Ethernet or USB, unplug the computer, and the station then runs on its supply, one trigger and two outputs. The OK line drives your green light; the NG line drives a buzzer or red stack light. The whole category is called fixed code readers, and compare-on-board is standard across it.

Handheld terminals with a display and a stored code list do exist, and they are fine for sample checks at a bench. They cannot do the line job: every read needs someone to aim and press, so a unit that passes unseen is never checked.

A compact reader such as the MV-ID3013PM-12M-WBN pairs a 12 mm lens with 1.3 MP resolution over Fast Ethernet; after teaching, the station needs only power, a trigger and the output line.

The line-side pattern is the same one used for full barcode reader verification for label inspection; the difference is that here nothing downstream listens, and the buzzer is the whole report.

What does the reader compare each read against?

The master string you teach it: the exact characters of the correct model code, held in reader memory and compared character by character against every decode.

Teach from a golden label — a sticker straight off an approved reel — or type the string into the setup tool. Then lock the compare down: exact match on, fixed length on, and the symbology list trimmed to the one code type your labels use. Fixed length turns a half-decoded symbol into a clean no-read rather than a plausible-looking string, and a trimmed symbology list stops a misread bar from being reinterpreted as a different code type. Trim any prefix or suffix the printer adds — a carriage return is common — in the output format, not in the compare: the master string stays the pure model code.

If the label carries more than the model code, say a lot number after a separator, splitting the string into fields before the compare keeps one master per field and one clean verdict per field.

How do you catch the wrong model's sticker when it reads perfectly?

Exact string matching is what catches it: a wrong model's barcode decodes perfectly, but its characters differ from the master, so the compare fails and the NG output fires even though the read itself was clean.

This is precisely the failure a decode-only setup misses: a reader told to beep on "something decoded" will bless the wrong model all day. Identity lives in the characters, not in read success. Two cases reach the station and both must end in NG. A different model's sticker decodes cleanly and fails the compare. The right model with damaged print fails to decode at all, through the no-read path. Funneling both into one NG output is the correct behavior — any doubt stops the unit. Codes differing by a single digit are the classic trap for loose matching; exact compare with fixed length catches them.

For assemblies where several models share a chassis and only the code tells them apart, barcode to part verification for similar assemblies runs on the same compare.

How good does the printing have to be?

Set the bar at a decode grade of 1.5 (C) or better on the ISO/IEC 15416 scale, where 4.0 (A) is flawless and 0.0 (F) fails: a label that reads but grades below 1.5 is one press drift away from failing on the line.

GS1's general specifications set 1.5 (C) as the minimum grade for its logistics labels, which is why most customer specs copy that number. Decode success and print quality are different measurements: a reader reports confidence and read rate, while a verifier measures edge contrast, modulation, defects and decodability under standardized illumination and prints a grade.

Your station sits between the two. It proves the code is the right one and readable today; the incoming grade says how much margin tomorrow has. So grade each new label reel at incoming: 1.5 (C) or better means line-side matching alone is enough, re-checked at every new reel. A reel drifting toward 1.0 (D) gets rejected before it reaches the applicator. The station-level tell: when reads only succeed after raising exposure or gain well above what a known-good reel needed, the print is drifting — grade the reel rather than tuning around it. For 2D codes the equivalent scale is ISO/IEC 15415.

Barcode read confidence covers judging a decode you are not sure about.

How many pixels does your label need?

Two sensor pixels across the narrowest bar is the floor for any 1D decode; design to 3-5 pixels per module so ordinary print wobble still decodes, and check it on a saved image rather than by eye.

Worked example. The LS-CRD-0028 captures 2560 x 2048 pixels through an 8 mm lens on a 1/1.7" sensor. Stand it 100 mm from the label. A 1/1.7" format sensor's active width is about 7.4 mm, so:

• field of view ≈ 100 mm × 7.4 / 8 ≈ 93 mm
• pixel pitch ≈ 93 mm / 2560 px ≈ 0.036 mm
• a 0.25 mm module spans 0.25 / 0.036 ≈ 7 px — comfortable
• a 0.10 mm module spans ≈ 2.8 px — above the floor, below the design band

For that 0.10 mm module, shorten the stand-off to about 60 mm: FOV ≈ 60 mm × 7.4 / 8 ≈ 56 mm, pitch ≈ 56 mm / 2560 px ≈ 0.022 mm, and the module spans 4.6 px — inside the band. The geometry is the decision; the one check left is that the reader's focus range reaches the distance you pick.

Deeper oversampling buys nothing: once modules span 10 px or more, extra pixels mostly magnify print noise, and grossly oversized modules can jam some decoders — downscaling before barcode decoding covers when that bites. Code 128 module size, reading angle and reader distance runs the same math for Code 128, quiet zones included.

What wiring does a standalone station need?

Three wires do the job: mains into a DIN-rail DC supply, the reader's rated DC in, and one discrete output out to a buzzer or stack light — plus a photoelectric sensor into the trigger input if you want a read only when a unit is present.

A DIN-rail supply sized for the reader, the buzzer and the trigger sensor covers the whole station; the reader's manual prints its own consumption. Hold the NG state until someone acknowledges it — a momentary beep on a line that keeps moving verifies nothing. If the buzzer or stack light draws more than the output is rated for, switch it through an interposing relay; the output's load rating is in the manual.

Trigger from a photoelectric sensor at the station instead of free-running: free-run reads whatever passes, wastes the sensor and makes the OK/NG history meaningless. The trigger modes of a fixed barcode reader spell out sensor, serial and free-run behavior. On a smart-camera station the light wires to the camera's strobe output — enabling a smart camera's strobe output and wiring the light covers it, and SC3000-series stations pair with the square red light board and semi-polarizing lens cover from the reader accessories range.

If you would rather skip external lighting entirely, the MV-ID813M-05S-WBN-SR carries its own white light behind a 4.9 mm lens at 1.3 MP on GigE.

Which reader or smart camera fits your line?

Pick a compact fixed-focus reader when the label sits at one known spot, a reader with built-in light when you cannot add illumination, and a smart camera when the same station must also check position, presence or a second feature on the unit.

For the glossy-film case the mount distance is settled by the optics: the MV-IDB005EXI-05NRP focuses at 70 mm, so the label passes 70 mm from the window and the decision is made.

The smart cameras range covers the vision-sensor side: when the station must confirm presence or placement as well, the MV-SC3050XC-16M-ATN-H provides a 5 MP color, global-shutter platform on an M12 mount, and the same compare rules apply to a position or presence tool.

If you are unsure which way to jump, the configurator walks camera, lens and trigger choices in order:

When is a dedicated verifier the better buy?

Buy verification-grade checking when a customer or standard demands a graded report, not when you only need the right model on every unit: a dedicated verifier measures the ISO/IEC 15416 parameters and issues the grade, while an inline reader proves identity and readability.

Let the audit decide. If the only judge is your own line, the inline compare is the whole job and a verifier adds cost without adding a decision. If a customer specification, a retail logistics requirement or a regulator asks for graded proof of print quality, that report only comes from a verifier — standardized illumination, calibrated optics, a documented grade. The architecture that covers both is two tools: a verifier at incoming quality on sampled reels, standalone readers on the line for 100 percent identity.

For the bench-side spot check of an incoming reel, the LS-CRD-0047 reads 1.2 MP images over USB 2.0 at a 60-120 mm reading distance.

What do you check before trusting it on the line?

Run four tests before you trust the station: twenty clean reads on a golden label, one wrong-model label that must fire NG, one unlabeled unit that must also fire NG, and a run of 100 good units with zero false rejects.

Teach and match

• Teach from a golden label off an approved reel, not only a typed string — a typo in the master verifies the wrong code for the whole lot.
• Fire NG with a known wrong-model label, then with no label at all; a station that has never said NG is not a checker.
• Confirm exact match, fixed length and a trimmed symbology list before the first production read.

Optics and mount

• Save an image of a good read and count pixels across the narrowest bar; three or more, or move closer. The 2 px floor leaves nothing for print wobble.
• Set the stand-off from the pixel math above and confirm the focus range reaches it.

Wiring and trial

• Hold NG until acknowledged, and test the buzzer path from the setup tool rather than by waiting for a real failure.
• One unit, one frame: watch the trigger through ten units at full line speed.
• Run the 100-unit trial and record the label reel lot with it, so a later print drift traces back to a reel.

The decision in one line: teach the master, compare exactly, grade the reels at 1.5 (C), and let the NG output stop whatever it cannot vouch for. To size a station for your own label, start from the module width and stand-off in the pixel math above, then browse the camera and reader finder or the code-reading guides; if the label is an odd size or the line runs fast, describe it through the contact page and we will work the numbers with you.