High speed conveyor barcode reading: choose the reader and setup

A code reader on a filler exit has one job and a deadline: every bottle or drum that leaves the line must yield its batch number, material number and sequence number, not just prove a label is present. At filler speeds the gap between containers shrinks to a little over a hundred milliseconds, so choosing the reader is a timing calculation before it is a catalogue decision. This guide works the numbers for a 2 m/s belt — window, pixels, exposure, light, mount, wiring — then maps the result onto fixed code readers and the accessories that close the remaining gaps. The same arithmetic scales to any belt speed.

The read window your line gives you

Start with the belt geometry, because it sets every other number. Containers arrive on a fixed pitch, and the time between them is pitch divided by belt speed.

Take a filler exit with 250 mm between container centres and the belt at 2 m/s (2000 mm/s). The window is 250 mm / 2000 mm/s = 125 ms and the demand is 8 containers per second. A reader framing at 30 fps produces a frame every 33.3 ms, so each container gets 125 / 33.3 ≈ 3.7 looks.

The decode budget has to clear the same window. At 60 codes/s a reader finishes one decode every 16.7 ms, which leaves about seven decodes in reserve per container on this line. The decision follows directly: any reader whose frame period plus decode time fits inside 125 ms will work, and the exposure, not the processor, is what usually breaks the budget. Site the trigger photo-eye one to two frame times before the field centre; at 33 ms per frame the bottle moves about 67 mm between looks.

Field of view and pixels per module

Frame the label, not the bottle. Add label width plus positioning wander on both sides: a 70 mm bottle with ±15 mm of wander needs a field about 100 mm across. On a 2560 px wide frame that field gives 100 mm / 2560 px = 39 µm per pixel.

Module size sets the decode floor. A 30 mm-wide 24 x 24 DataMatrix carries 1.25 mm modules, so the 100 mm field yields 1.25 mm / 39 µm ≈ 32 px per module, far above the 4 px per module we plan for on 2D codes. Below roughly 3 px per module, decode confidence decays fast, so judge trials on decode confidence rather than lucky hits. Cropping an ROI to the label lane buys frame rate and steadier decodes on the same sensor; the ROI tuning guide for inline code reading shows the trade.

If you would rather solve backwards from a required module size to the pixels you need, the calculator below runs the same division in reverse.

Exposure: stop the blur

Motion blur is belt speed multiplied by exposure time, and one pixel of smear is the usual budget.

With 39 µm pixels and the belt at 2000 mm/s, one pixel allows 39 µm / 2000 mm/s ≈ 19.5 µs. Set 10 µs: the smear is 2000 mm/s x 10 µs = 20 µm, about half a pixel. No continuous lamp holds that exposure, so the strobe carries the illumination. With the flash doing the optical work, open to f/4-f/5.6 for depth of field and keep gain near unity so sensor noise stays away from the module edges.

The same pixels-versus-blur arithmetic for sorting belts, with its own worked examples, is in the sorting conveyor camera selection guide.

The LS-CRD-0028 was built for exactly this duty: 30 fps, up to 60 codes/s, a 2560 x 2048 global-shutter sensor behind an 8 mm lens, on Fast Ethernet.

Imager or laser scanner

The payload decides this, not preference. Batch, material and sequence numbers usually live in one 2D symbol, and a laser scanner cannot return a 2D payload at all: it reads 1D symbologies only, and it struggles on damaged codes and reflective surfaces. An imager decodes both formats, takes several codes per frame, and archives the image for when a read is disputed.

Shutter type matters at speed. A rolling shutter exposes rows at slightly different moments, so a fast container shears its own image; choose a global shutter for moving belts. A rolling-shutter 12 MP reader such as the MV-ID5120RM-08M-RBN belongs on dense multipacks at modest speeds, where its resolution pays for itself.

Light for the label

Red is the default for paper labels: white stock bounces red strongly while printed bars stay dark, which maximises contrast. The MV-SC3000-L-RB square red board suits that duty. Blue or white suits coloured plastics and printed film; judge the choice on the image histogram, and let the LED colour contrast guide settle the physics.

Glossy film labels throw specular glare that saturates the bars, so pair a polarized light with the MV-SC3000-C-HP polarizing cover on the front. Shiny caps and foil respond to the blue MV-IDA-L-8-47-41-B board and its 8 LEDs. Size the strobe pulse to the exposure budget, not to factory defaults; the strobe-pulse sizing guide for high-speed verification has the arithmetic.

On slow logistics lines you can drop strobe sync altogether: a 24 V continuous light board made for the ID6000-series logistics readers covers that duty. At filler speeds, strobe only.

Three accessories that answer the common label problems:

Mounting distance, angle and margin

Convert the field into a working distance with the lens equation, FOV = WD x sensor width / focal length, rearranged as WD = FOV x focal length / sensor width. For a 100 mm field, a 12 mm lens and a 4.8 mm-wide (1/3-inch class) sensor, WD = 100 mm x 12 mm / 4.8 mm = 250 mm. Readers like the MV-ID3013PM-12M-WBN, a 12 mm lens on a 1.3 MP sensor, land in exactly this range. At 250 mm and f/4-f/5.6 the depth of field swallows belt bounce; focus only gets tight on close-focus models.

Tilt the reader 20-30° off perpendicular so specular return from a glossy label slides past the lens instead of into it. Keep the whole label inside the field at maximum wander: the ±15 mm you budgeted is mounting tolerance, not negotiation.

Where the filler corridor cannot give up 250 mm, go short-focus. The MV-IDB005EXI-05NRP has a 70 mm focus distance and a 4.63 mm focal length in a polarized optical path, so the mount clears the guard rail and the polarization tames film labels.

Getting the string out before the next container

A decode that never leaves the reader is a missed read. Spend the same 125 ms deliberately: 33 ms to frame, 17 ms to decode, about 75 ms left for the reader to push the string and the PLC to accept it. Fast Ethernet gives that headroom easily; a serial link at 9600 baud would burn 42 ms on one 40-character string (40 chars x 10 bits / 9600 bits/s ≈ 42 ms) and leave nothing for retries.

Plan the link, not just the read. The barcode reader to PLC communication guide covers protocol choice and retry behaviour, reading barcode reader data into PLC string tags handles the string-tag side, and routing containers from reader data on the line shows the branch logic when a read steers the route.

Reader picks by line speed

The same arithmetic, scaled to four common duties. Exposures assume a one-pixel blur budget on each class's typical pixel pitch:

Mapped to models:

• Filler exit: the LS-CRD-0028, 30 fps and 60 codes/s on a global-shutter 5 MP sensor.
• Carton duty: the MV-ID3013PM-12M-WBN, a 12 mm lens and 1.3 MP sensor over Fast Ethernet.
• Dense multipacks on slower belts: the MV-ID5120RM-08M-RBN, 12 MP over GigE.
• Short, bright mounts: the MV-ID813M-05S-WBN-SR, a 4.9 mm lens with white light over GigE.

Values as printed in each product's data sheet; open a part for its drawing and full specifications.

Checks before you order

• Measure belt speed over a marked metre with a stopwatch and pitch with a ruler; every later number divides from these two.
• Trial on the worst labels pulled from the line — glare, skew, partial print — before you commit the panel cut-out.
• Decode one label by hand and confirm every field the MES expects sits inside the payload.
• Site the photo-eye so the flash fires with the label centred in the field; prove it with the frame counter, not by eye.
• Lock the mount to the machine frame, not the guard rail, so filler vibration cannot walk the focus.
• Push a long lot number through the whole link into the PLC before sign-off.

When the usual answer is wrong: if labels arrive misplaced or unprinted, no reader upgrade recovers the data, so fix print-and-apply first and re-run the window maths. If every container carries several codes at random positions, step up resolution and decode budget before you step up strobe power.

Size your own line: divide pitch by belt speed, set the exposure from pixel size, then let the code reader finder, the accessories finder and the configurator do the filtering. Send speed, pitch and a photo of the worst label through the contact page and we will check the numbers with you. The code-reading topic page collects the rest of the decoding guides.