Fixed-Mount Code Reader Selection — Resolution, Lighting, DPM and RFP for Thai Factories

Introduction — Fixed-Mount Code Reader Selection Is Not About Brand and Pixel Count

At the end of a process, an operator holds a handheld scanner and reads labels one part at a time. The old laser reader mounted on the machine cannot read the codes on marked parts. The customer is asking for serial-level traceability. A no-read stops the line, and there is no record of what was read and what was not. Discussions about fixed-mount code reader selection usually start from problems like these.

Here is the main point of this article up front. Fixed-mount code reader selection is not decided by brand name or pixel count. It is a matter of writing down the following four conditions first, then lining up the models that meet them and testing them.

• (1) What to read: 1D, 2D or DPM (direct part marking on the part itself), and the minimum cell or bar width
• (2) Surface and lighting: paper labels, glossy surfaces, marks on metal, curved surfaces
• (3) Motion: conveying speed, exposure time, trigger, variation in distance
• (4) Connection: PLC, network, MES, behavior on a no-read, image storage

The acceptance criterion is not “it read during the demo”. It is decided by the read rate and read margin on your own actual parts, at your actual conveying speed and under your actual lighting, together with how the system rejects, stops and records on a no-read.

This article is not a recommendation or ranking of specific products. Manufacturer specifications are based on each company’s published materials and are presented in the form “the company states that…”. The cost estimate in the second half uses hypothetical numbers for this article. How to choose between barcode, QR and RFID is covered in our article on traceability barcode selection. This article focuses on what comes after the code has been decided, namely how to read it automatically on the line.

Why Thai Factories Are Revisiting Automatic Code Reading Now

Business sentiment in Thai manufacturing is recovering. The S&P Global Thailand Manufacturing PMI was 54.3 in September 2026, up from 53.8 in August and the highest level since December 2025 (according to the aggregator site Trading Economics). Employment, on the other hand, declined slightly, and companies cited natural attrition as the reason. Work is increasing, but headcount is not.

Labor-related costs and rules are also changing. Contributions to the Employee Welfare Fund began on October 1, 2026, and until September 30, 2030, employers and employees each contribute 0.25% of monthly wages. On July 14, 2026, the Thai Cabinet approved extending the work permits of about 770,000 migrant workers from Laos, Myanmar and Vietnam. The reported aim is to secure labor for the manufacturing and service sectors.

On the investment side, according to the Thailand BOI, there were 1,299 investment promotion applications in January to June 2026, worth USD 43.6 billion. Of these, applications under the “Smart and Sustainable Industry” measure (machinery upgrades, adoption of digital technology, integration of automation and robotics) numbered 132, worth USD 507.6 million. Both figures are based on applications and differ from approved amounts.

Reading parts one by one with a handheld scanner tends to be a weak point in both labor and record keeping. Against this background, more factories are considering moving the “reading” task to the equipment side first.

Food and consumer goods factories face a separate trend. A GS1 position paper (July 2026) sets an industry goal that, by the end of 2027, retail POS systems will be able to read GS1-conformant 2D codes (such as GS1 DataMatrix and QR codes with GS1 Digital Link) in addition to existing 1D barcodes. This is a goal for the reading capability of retail POS. During the transition, 1D and 2D codes are to be printed side by side, and it is not a legal obligation. Nor is it a requirement for part codes in factories. Still, for manufacturers that print 2D codes on the packaging of products they ship, it is reasonable to assume that the readers checking the print on the line will also need to support 2D codes. For how to approach the transition, see our article on GS1 Digital Link implementation.

What Is a Fixed-Mount Code Reader — Differences from Handheld Scanners and Verifiers

A fixed-mount code reader is a device mounted on a conveyor or machine that automatically reads the codes on parts as they pass. It sends the results to a PLC or a higher-level system.

The difference from a handheld scanner is not only “held by a person or fixed to equipment”. With a handheld, the operator points it at the code, so a person adjusts the distance and angle. With a fixed-mount reader, the equipment side has to absorb variation in part position, distance and speed. In return, it can record the read result for every cycle and drive the equipment on a no-read. Handheld terminal operation and costs are covered in our article on handheld terminal implementation.

Another device that is easily confused with a reader is the verifier. KEYENCE’s barcode course describes a verifier as “a device that inspects barcodes”. The company states that “even barcodes judged to have low print quality can often be read by a barcode reader”, and recommends a verifier when the purpose is quality checking.

In summary:

Print quality standards include ISO/IEC 15415 for 2D (2024 edition), ISO/IEC 15416 for 1D (succeeded by the 2025 edition) and ISO/IEC 29158 for DPM (2025 edition, published March 25, 2025). These are standards for verification. The IEC Webstore abstract describes ISO/IEC 29158:2025 as intended for verifier manufacturers and for those who develop application specifications. The AIM DPM quality guideline that was used in the past is shown as “OBSOLETE” in AIM’s official store, which directs users to follow the ISO document.

Note that some readers also have a print verification function. KEYENCE states that the SR-X has print verification functions for 2D codes and barcodes and supports ISO/IEC 29158, 15415, 15416 and others. This is useful for tracking trends on the line. On the other hand, if you need a formal quality evaluation to submit to a customer, consider a separate verifier. How to choose a verifier is explained in our article on barcode verifier implementation.

Laser Readers vs Imagers (Camera-Based)

Fixed-mount readers fall broadly into two types, laser readers and imagers (camera-based readers).

According to KEYENCE’s barcode course, a fixed laser reader consists of a laser diode, a polygon mirror, a motor and a light-receiving element. The rotating polygon mirror sweeps the laser beam across the barcode, and the reflected light is received and decoded. The company explains that the motor rotates continuously while power is on so that parts in motion can be read at high speed. It then lists as points of caution that, over long-term operation, motor deterioration can prevent correct operation, and that shocks can damage the bearings. Having moving parts is something to keep in mind from a maintenance perspective.

A camera-based reader locates the code in a captured image and decodes it. The company describes the 2D code reader process as image capture, grayscale image, binarization and decoding. For fixed CCD readers, the company explains that fast LED pulse emission “makes objects moving at high speed appear stationary”, which is an advantage for reading moving objects.

As a rough guide, if you only read 1D barcodes, laser readers remain an option. For 2D codes and DPM, camera-based readers are generally used. The problem mentioned at the start, “the old laser reader cannot read marked parts”, may require rethinking the reading method itself.

The company also explains that even when a print looks fine to the human eye, a 2D code may not be recognized depending on how the light hits it, the surface condition of the part and the camera settings. Choosing a camera-based reader does not solve everything. The following sections look at the four conditions one by one. The overall flow is shown in the figure below.

Difficulty Depends on What You Read — 1D, 2D and DPM

Even though it is all “reading a code”, the difficulty varies greatly by target.

A 1D barcode printed on a paper label is a target for which conditions are relatively easy to control. A 2D code can hold a lot of information in a small area, but its cells tend to be small. KEYENCE’s FAQ lists insufficient quiet zone, dirt on the finder pattern, color combinations that make contrast hard to obtain, and mounting angle as causes of 2D codes not being read. The company explains that “a 2D code is read by first finding the finder pattern”, so if the corner markers are dirty, the code cannot be read.

DPM is a method in which the code is marked directly onto the part itself by laser or dot peen. With no label, there is no peeling or mislabeling, but reading conditions become more demanding. This is because strong contrast, such as white on black, is hard to obtain. The appearance also changes depending on whether the surface is as-cast die-cast skin, a machined surface, a hairline finish or resin.

The AIM DPM quality guideline defined DPM quality evaluation on the premise that the reading device is a 2D imager. If you read DPM, a camera-based model with selectable lighting is the starting point.

In automotive parts, there are industry guidelines for DPM such as AIAG B-17. Check first which requirement specification your customer expects.

The marking method and the reading conditions are two sides of the same coin. There is a limit to how far the reader alone can rescue a mark that is hard to read. How to set marking conditions is covered in our article on industrial laser marking machine selection. It is efficient to proceed with reader selection and marking condition development in parallel, using the same samples.

How to Decide Resolution, Field of View and Reading Distance

Resolution means “the smallest cell or bar width that can be read”. It is the first item to check in a reader’s specifications, but it is not determined by the model’s performance alone. It changes greatly with distance and lens.

The specification table for KEYENCE’s SR-X shows the minimum resolution for 2D codes at each distance.

For the SR-X300, the minimum resolution is 0.024 mm at a distance of 70 mm and 0.335 mm at a distance of 1000 mm. Even for the same model, it changes by a factor of about 14 depending on the installation distance (0.335 ÷ 0.024 ≈ 14.0). Comparing at a distance of 300 mm, the wide field of view type has a field of view about 2.6 times wider horizontally (267 ÷ 104 ≈ 2.6), but its minimum resolution becomes coarser, from 0.099 mm to 0.251 mm. Field of view and resolution are a trade-off.

What this shows is that the suitability of a model is not decided by its resolution figure alone. Even a model with a high pixel count assigns fewer pixels to each cell if it captures a wide field of view from far away.

The steps for deciding are as follows.

1. Check the minimum cell (or bar width) of the code to be read from drawings or the print specification
2. Determine the required field of view from part position offsets, height variation and differences between product types. Make it large enough for the code to fit in the field of view with a margin
3. Determine the range of mounting distances possible from the machine structure
4. At that distance and field of view, shortlist models and lenses whose resolution can read the minimum cell

When there are multiple codes in one image (multi-code), also check that they all fit in the field of view. For example, Cognex explains in the DataMan 380 manual that when multi-code is enabled, the reader reads multiple symbols within a single image.

The thinking on pixel count, lens and field of view is the same as for cameras used in image inspection. See also our article on industrial camera selection.

Lighting and Surfaces — Paper Labels, Gloss, Metal Marks and Curved Surfaces

Code reader lighting is something you choose to suit the surface. A camera-based reader captures the light and dark contrast of the code as an image. That contrast changes greatly with how the light is applied and how the surface reflects it.

KEYENCE states that the SR-X300 series has “3-way multi-lighting”, which automatically selects direct, polarized or diffuse light to suit the part. As representative examples, the company’s product page includes a table showing that direct lighting suited dot peen marks on cast skin, polarized lighting suited black resin, and diffuse lighting suited hairline surfaces. These are the company’s representative examples, not general rules. Confirm by testing what happens with your own parts.

Other companies also offer lighting options. The Cognex DataMan 380 manual lists high-output red, white, clear and polarized lighting, as well as a polarized front cover, as accessories. Datalogic states that for the Matrix 320, the color and intensity of the lighting can be selected to match the code type, the material the code is on and the color of the material.

Points to check for each surface are as follows.

• Paper labels: contrast is easy to obtain. With laminated or glossy paper, watch out for reflections of the lighting
• Glossy resin and metal: specular reflection easily causes washed-out highlights. Try polarized light or changing the lighting angle
• Marks on metal (DPM): contrast is low. The choice of lighting method strongly affects reading
• Curved surfaces (shafts, pipes, etc.): reflection and distortion tend to appear at the edges of the code. Test the mounting angle and field of view together

Factories also have ambient light, such as sunlight through windows, ceiling lights and light from neighboring equipment. As a basic rule, tests are carried out at the actual installation location. Lighting selection, including the handling of reflective parts, is explained in detail in our guide to AI visual inspection lighting selection.

Focus — Fixed Focus, Liquid Lens and Autofocus

On lines where part distance varies, or where code height changes by product type, the focus method becomes important. There are three main options.

• Fixed focus: used at the set distance. Simple structure
• Liquid lens: changes focus electrically. Some models can change focus remotely during product changeover
• Autofocus: the device adjusts the focus position automatically. However, when it adjusts differs by model

Each company’s explanation is as follows.

• OMRON lists autofocus for reading codes at different distances in the MicroHAWK V430-F datasheet and explains that fixed-focus and autofocus lenses can be selected. The autofocus range is stated as 50 to 300 mm for the 0.3MP and 5MP models and 75 to 1160 mm for the 1.2MP long-range model.
• Datalogic states that the liquid lens models of the Matrix 320 (6 mm, 9 mm, 16 mm) allow focus to be changed remotely when switching production batches, without opening or touching the reader.
• Cognex lists either a C-mount lens or the company’s high-speed liquid lens as lenses for the DataMan 380.
• KEYENCE states that SR-X autofocus “automatically adjusts the focus position during installation and tuning”. It is not described as a mechanism that refocuses each time during operation.

Do not judge by the word “autofocus” alone. Check whether the reader focuses on each part during operation, or focuses at installation and then stays fixed. On lines where the distance changes greatly from one part to the next, this difference directly affects reading. The reliable approach is to write the measured distance variation in the quotation request and have the manufacturer answer how they will handle it.

Reading on a Moving Line — Conveying Speed, Exposure Time and Trigger

When reading parts moving on a conveyor, motion blur becomes an issue. The background art section of a KEYENCE US patent (US10262177B2) states that blur depends heavily on the conveying speed of the part and, in addition, also depends heavily on the size of the cells that make up the code. It also notes that, conventionally, a table matching conveying speed to an appropriate exposure time was prepared in advance. This is a statement in a patent document, not a product specification.

The size of the blur can be roughly estimated as “conveying speed × exposure time”. The following is an editorial example calculation. If the conveying speed is 500 mm per second and the exposure time is 100 μs (0.0001 seconds), then 500 × 0.0001 = 0.05 mm. For a code with 0.25 mm cells, the blur is one fifth of a cell. At the same speed, the smaller the cell, the relatively larger the effect of blur. How much blur is acceptable depends on the model and code conditions, so it is not given here. Confirm it by testing.

The exposure time setting range differs by model. Cognex gives the DataMan 380 exposure time as a minimum of 15 μs and a maximum of 200 ms. The OMRON V430-F is divided by model into global shutter (0.3MP, 1.2MP) and rolling shutter (5MP). When reading moving parts, include the shutter type in your checklist.

The trigger decides when to capture. The DataMan 380 manual describes the following modes.

According to the same manual, an external trigger can be given by an I/O pulse, a serial command or an industrial protocol from a PLC. The trigger input is said to be drivable from either NPN or PNP photoelectric sensors or PLC outputs.

In practice, you choose between “part detection sensor + external trigger” and “continuous reading”. The former makes the timing of the read clear and makes it easier to link results to parts. For choosing the part detection sensor, see our article on photoelectric sensor selection.

Conveying speed and part pitch are determined by the conveyor design. Check that the reader’s capture and decode time fits within the time it takes one part to pass. The conveyor side is summarized in our article on conveyor system design.

Read Rate and Read Margin — Looking Beyond “It Read”

Read rate is “how many times it read successfully out of how many attempts”. It is the central metric in acceptance testing, but it is not sufficient on its own. Even with a high read rate, the reader may be reading just barely.

KEYENCE explains that the SR-X matching level is an indicator that lets you check a code’s read margin as a value from 1 to 100, rather than only whether it was read or not. In an example the company gives, “both have a read rate of 100%, but the matching levels are 75 and 43”, showing the difference in margin as a number.

There are two points to note. The matching level is the company’s own 1 to 100 indicator and is not a read rate. A value of 75 does not mean “75% readable”. Also, readers from other companies have indicators equivalent to read margin, but the names and scales differ by manufacturer, and the values cannot be compared side by side.

The value of read margin is that it lets you catch changes early. Even before the read rate drops, if the margin is falling, changes such as the following may be occurring.

• Faded print, deterioration of ink or ribbon
• Changes in mark depth or marking conditions
• Dirt on the lens or protective window, adhesion of oil or dust
• Mounting misalignment, deterioration of lighting

In our research for this article, we could not confirm any generally established rule for the number of trials in read rate testing. Decide and record the number of trials and pass line yourself. In operation, set a threshold on the margin and create a flow that sends the issue back to maintenance or to the printing or marking process when the margin falls below it.

To run this flow, a selection point is whether the model can output the margin to a PLC or higher-level system and keep it as a time series.

How to Evaluate AI-Powered Readers

In recent years, several companies have released fixed-mount readers that feature AI. The claims we could confirm in published materials are as follows.

• KEYENCE states that the SR-X300 series is equipped with an “AI filter” pre-trained on more than 100,000 images and an inference AI chip dedicated to reading codes. As examples of its effect, the company lists dirt, scratches, rough backgrounds and uneven color within cells.
• In January 2025, Cognex announced the DataMan 290 and 390 and said they deliver stable decoding performance through AI technology. The 290 is positioned as handling many applications out of the box, and the 390 as having higher resolution for more complex applications.
• The DataMan 380 was introduced by the trade publication Vision Systems Design (October 2023) as having “AI accelerated decoding”.
• Zebra states that the FS42 is equipped with a neural processing unit (NPU) and can run AI deep-learning-based tools quickly.

On the other hand, no company’s published materials give a quantitative figure for how much AI improves the read rate.

The evaluation method is the same whether or not AI is present. Collect your own hard-to-read parts (dirty, faded, with variation in marking) and compare the read rate and margin with and without the AI function. If the AI function comes only with higher-end models or a license, judge from the test results whether the improvement justifies the price difference.

Fixed-Mount Readers from Major Manufacturers — Specifications Confirmed in Published Materials

The table below lists the items we could confirm in each company’s published materials for fixed-mount readers from major manufacturers. All entries are each company’s own descriptions and do not indicate recommendations or rankings. Fields that could not be confirmed are marked “Not confirmed in published materials”.

Points to note when reading the table:

• The way items are described differs by company. The SR-X shows minimum resolution by distance rather than pixel count, so it cannot be compared directly with other companies’ pixel counts
• The DataMan 380 is a high-resolution model whose applications, according to the trade publication, also include reading codes on palletized goods and tires. Its character differs from models that read small parts at close range
• Datalogic promotes DPM reading performance for the Matrix 320 and states that it targets traceability applications
• In Thailand, KEYENCE’s Thai subsidiary has a Thai-language page for the SR-X series on its website. Check stock, lead times and local support arrangements with each company

For fixed-mount readers from SICK and Honeywell, we could not confirm official specifications this time.

This table is only an entry point for narrowing down candidates. The final decision is made on the answers to requirements describing the four conditions and on test results with actual parts.

PLC and Network Integration and Behavior on a No-Read

In code reader PLC integration, first decide “which PLC, which protocol and what to return”. For example, KEYENCE states that the SR-X supports EtherNet/IP, PROFINET, MC protocol, OMRON PLC Link, KV STUDIO, OPC UA and others. Cognex lists EtherNet/IP, PROFINET and SLMP for the DataMan 380 and also shows a wiring example connecting high-speed outputs directly to PLC inputs. Check for each model whether it fits your factory’s PLC manufacturer and network.

The wiring configuration also needs attention. According to KEYENCE, the PoE models of the SR-X (such as the SR-X300P) do not have control I/O or serial communication. While a single LAN cable can provide both power and communication, a configuration that inputs a trigger directly via I/O or outputs OK/NG via I/O is not possible. In that case, exchange signals over the network or choose a standard model.

If read results and images are exchanged over the network, include switch selection and network diagnostics in the design. Our article on industrial Ethernet switch selection and our article on PROFINET network diagnostics are useful references.

The most important thing is what happens when a code cannot be read. There are three main options.

• Reject: remove the no-read part from the line for a person to check later
• Stop: stop the line and deal with it on the spot
• Re-read: capture again, or read with another reader downstream

Which to choose is not for the manufacturer to decide but for your company to decide for each process. Evaluate in the process FMEA the impact of a no-read part flowing to downstream processes, and choose accordingly. For a process that reports serial numbers to the customer, an interlock that prevents unreadable parts from flowing to the next process will likely be needed. The thinking behind interlocks is organized in our article on poka-yoke.

Ways of communicating a no-read to operators on the floor also help. Cognex states that the DataMan 380 lights up green on a good read and red on a read failure.

Image Storage, Print Degradation Monitoring and MES/Traceability Integration

If read results end at just “OK/NG”, you cannot trace the cause when trouble occurs. Record at least the following items.

• Read result (the decoded string, or NG)
• Time
• Station (which reader)
• Read margin
• Image (at least NG images)

As for manufacturer features, KEYENCE states that the SR-X300 series can store up to 68,000 images in 4GB of memory. The company’s Web Traceability Tool is said to allow you to trace when, which reader and why a code could not be read, from statistics and error images. The company’s Link System is described as comparing the read images, mounting angle, read tact, focal distance, lighting conditions and matching level of readers in upstream and downstream processes, and displaying warnings for deviations in focal distance or mounting angle. For communication with higher-level systems, the SR-X lists support for OPC UA, FTP / SFTP and HTTP / HTTPS.

The basic form of MES integration is to link the read result, time, station and image storage location to the MES lot or serial number. Design that links equipment operating history to parts is covered in our article on production equipment traceability.

In processes that use labels, one possible configuration is to confirm the read with a fixed-mount reader immediately after the label is applied. For combining it with an applicator, see our article on automatic label applicator selection.

If you want to check the characters themselves, such as expiry dates or lot numbers, that is a separate consideration from code reading. Zebra states that AI-OCR tools can be added to the FS42 by license. How to approach print checking of characters is explained in our article on date code inspection AI.

Reading before shipment is also directly linked to preventing mis-shipments. For the design of checkpoints across the whole shipping process, see our article on mis-shipment prevention.

Implementation Steps — Part Collection, Bench Testing and Line Testing

Implementing a fixed-mount code reader in the following four stages reduces rework.

1. Collect samples of actual parts

Collect not only good parts but also hard-to-read ones. These include dirty parts, faded parts, parts with variation in mark depth or position, and representative parts for each product type. If you have actual parts that caused no-reads on the line, those are the most useful samples.

2. Measure variation in distance and angle

Measure part height, position offset and tilt. Include conveyor meandering and fixture play. The ranges measured here become the requirements for resolution, field of view and focus.

3. Compare 2 to 3 candidate models under the same conditions

Narrow down candidates from published specifications and manufacturer answers, and run bench tests with the same samples, the same distance and the same lighting conditions. Record not only the read rate but also the margin, the time taken for setup and how easy it is to review NG images. Note that manufacturer demos are often run under conditions that the company is good at.

4. Test at line speed

Temporarily install the models that did well on the bench on the actual line and read at the actual conveying speed, with actual lighting and actual ambient light. Also confirm trigger timing, communication with the PLC and behavior on a no-read at this stage.

If you skip this sequence and place an order “because it read during the demo”, conditions under which the mass production line cannot read will be found later.

Requirements for the RFP and FAT/SAT Acceptance Criteria

In the request for proposal (RFP), describe the four conditions specifically. This is so that each company’s answers can be compared under the same conditions.

Asking for the quotation to be broken down by item makes it visible that even if the reader itself is cheap, differences appear in lighting, licenses and integration work.

Acceptance testing is split into two stages.

• FAT (factory acceptance test): confirm the read rate and margin using the collected samples of actual parts. Include hard-to-read samples
• SAT (site acceptance test): run in parallel for a set period on the actual line at the actual conveying speed. During that period, continue handheld or visual checks and cross-check the results. Also confirm reject and stop behavior and recording on a no-read

Decide the numerical pass criteria (read rate, lower limit of the margin, parallel-run period, etc.) based on your own processes and customer requirements. In the RFP, put your own target values “as examples” and finalize them in discussion with the manufacturer. There are no standard values set by the industry.

Cost and Benefit Model Estimate (Hypothetical Numbers)

All numbers from here on are hypothetical numbers for this article. Because published prices could not be confirmed for most readers, the costs are also hypothetical. Actual amounts vary greatly with factory conditions and quotations.

The assumed factory is a Japanese-owned automotive parts plant in Thailand. At the end of processes on 6 machining and assembly lines and before shipment, operators read labels and marks with handheld scanners. It runs 2 shifts, 250 days a year.

The current annual costs (baseline) are assumed as follows.

Configuration A is 6 fixed-mount readers. It includes camera-based readers, lighting and brackets, PLC integration, and storage of results and images in the MES. The initial cost is set at 1,200,000 THB and the annual operating cost (maintenance, setting changes, licenses) at 120,000 THB.

Configuration B adds, on top of Configuration A, higher-end models with AI functions and lighting at 2 processes for DPM-marked parts. The additional initial cost is set at 900,000 THB and the additional annual operating cost at 60,000 THB.

(Unit is THB. All reduction rates are assumptions)

The calculation breakdown is as follows.

• Configuration A: 336,000 + 144,000 + 150,000 = 630,000. 630,000 − 120,000 = 510,000. 1,200,000 ÷ 510,000 ≈ 2.4 years
• Configuration B (additional): 240,000 × 60% = 144,000. 144,000 − 60,000 = 84,000. 900,000 ÷ 84,000 ≈ 10.7 years
• Sensitivity: if the line stop and rework baseline is 720,000 THB, 40% of it is 288,000. 336,000 + 288,000 + 150,000 = 774,000. 774,000 − 120,000 = 654,000. 1,200,000 ÷ 654,000 ≈ 1.8 years

Here are three points on how to read the estimate.

The first point is which benefit supports the payback of Configuration A. Of Configuration A’s annual benefit of 630,000 THB, a little over half, 336,000 THB, is the reduction in manual entry labor. Moving human work to equipment is the foundation of the benefit.

The second point is the Configuration B addition. Under this assumption, the addition takes about 10.7 years to pay back. This means that in factories with few DPM no-reads, adding higher-end models is hard to recover through labor savings alone. The basis for the decision lies elsewhere, namely customer traceability requirements (if the code cannot be read, the serial number cannot be recorded), marking yield and the cost of re-marking.

The third point is sensitivity. If the line stop and rework baseline is doubled, the payback shortens from about 2.4 years to about 1.8 years. How accurately you understand the actual state of line stops today changes the investment decision.

There is a caveat. “Line stops and rework −40%” and “customer response −50%” assume that the reject and recording procedures on a no-read and the monitoring of read margin are working. Do not count these as benefits until the read rate has been confirmed in the SAT parallel run.

Frequently Asked Questions

What is the difference between a fixed-mount barcode reader and a handheld scanner?

With a handheld, the operator points it at the code, so a person adjusts the distance and angle. A fixed-mount reader is mounted on equipment and reads passing parts automatically. Although the equipment side needs to absorb variation in position and speed, it can record the read result for every cycle and link a no-read to equipment actions such as rejecting or stopping.

Where in the factory should an industrial 2D code reader be installed?

Candidate locations are the end of a process, immediately after marking or label application, points where parts are fed into assembly, and before shipment. A practical approach is to prioritize processes where you want to record serial numbers, processes where you want to stop mix-ups, and processes where you want to confirm readability immediately after printing. Taking stock of where handheld reading is done today is the starting point.

What conditions are needed for DPM code reading?

Lighting that suits the marking method (laser, dot peen) and the surface (cast skin, machined surface, resin, etc.), and a resolution that can read the minimum cell. A camera-based reader with selectable lighting is the starting point. The shortcut is to tune not only the reader but also the marking conditions with the same samples. If a quality evaluation is needed, consider a separate verifier that follows ISO/IEC 29158, the verification standard for DPM.

What barcode read rate should we target?

There is no standard value set by the industry. Decide it yourself based on the role of the process (recording serial numbers, preventing mix-ups, shipping confirmation, etc.) and customer requirements. Including a lower limit for the margin in the target, in addition to the read rate, lets you catch signs of degradation early. Decide and record the number of trials and test conditions yourself as well.

How should we choose for code reader PLC integration?

First confirm your factory’s PLC manufacturer and the network in use (EtherNet/IP, PROFINET, etc.), and check in each company’s specifications whether that model supports it. Next, decide whether to exchange triggers and results via I/O or over the network. Some models, such as PoE models, have no I/O, so choose together with the wiring configuration.

Do we need an AI-powered code reader?

In processes with many hard-to-read parts (dirt, scratches, variation in marking), it is worth trying. However, each company’s published materials do not show a quantitative figure for how much AI improves the read rate. Check the difference in read rate and margin with and without the AI function using your own hard-to-read samples, and judge whether it justifies the price difference.

How much does it cost?

Published prices could be confirmed for only a small number of models. The Zebra FS42 was listed on US online retail sites at about USD 2,800 to 5,000 excluding tax (as of October 2026, varying by model number and configuration). A simple conversion at USD 1 ≈ 33.80 THB gives the equivalent of about 95,000 to 170,000 THB. This excludes taxes and import costs and is not a price in Thailand. For the other major models, we could not confirm published prices. In practice, compare quotations that include lighting, brackets, licenses and integration work.

Summary

• Fixed-mount code reader selection starts by writing down four conditions first: (1) what to read, (2) surface and lighting, (3) motion, and (4) connection
• Resolution changes greatly with distance and lens. For the SR-X300, the minimum resolution differs by a factor of about 14 between distances of 70 mm and 1000 mm. Do not judge suitability by pixel count alone
• With moving parts, blur is determined by both conveying speed and cell size. Test on the actual line, including exposure time and trigger
• In addition to read rate, use read margin as a monitoring indicator. The name and scale of the indicator differ by manufacturer
• A reader is a device that judges “whether it can read”, and pass/fail of print quality is the role of a verifier. Do not confuse the two
• No quantitative figures for the effect of AI functions have been published. Compare using your own hard-to-read parts
• Decide reject, stop and recording on a no-read for each process yourself, and write them into the RFP and the FAT/SAT acceptance criteria
• Count the benefits in the estimate only after the read rate has been confirmed in the SAT parallel run

Which code in which process should you automate first, and where are no-reads and manual entry happening today? You are welcome to consult TOMAS TECH even at that stocktaking stage. We work with Japanese-owned factories in Thailand from organizing requirements through the RFP and FAT/SAT. Please contact us here.

References

• AI-powered Code Reader SR-X Series (KEYENCE Japan, Japanese)
• SR-X Series AI-powered Code Reader (KEYENCE)
• SR-X Series (KEYENCE THAILAND, Thai)
• DataMan 380 Reference Manual (Cognex)
• Cognex Launches New AI-Assisted Barcode Reader (Vision Systems Design, 2023-10-26)
• Cognex DataMan 290/390 launch release (PRNewswire / AAP, 2025-01-14)
• FS42 Fixed Industrial Scanner Specification Sheet (Zebra Technologies)
• Zebra Technologies Launches Fixed Industrial Scanner for AI Applications (Zebra, 2024-05-28)
• Matrix 320 Series Stationary Industrial Scanners (Datalogic)
• MicroHAWK V430-F Autofocus Multicode Reader Datasheet (OMRON)
• ISO/IEC 29158:2025 (IEC Webstore)
• ISO/IEC 15416:2016 (IEC Webstore)
• ISO/IEC 15415:2024 (Singapore Standards eShop)
• AIM Direct Part Mark (DPM) Quality Guidelines (AIM)
• Barcode Verifiers (KEYENCE Barcode Course, Japanese)
• Fixed Laser Barcode Readers (KEYENCE Barcode Course, Japanese)
• Fixed CCD Barcode Readers (KEYENCE Barcode Course, Japanese)
• How 2D Code Readers Read Codes (KEYENCE Barcode Course, Japanese)
• Causes of Codes Not Being Read (KEYENCE FAQ, Japanese)
• US10262177B2 Optical information reading device (Google Patents)
• Migration to a single GS1 conformant 2D barcode (GS1 Position Paper, 2026-07)
• Sunrise 2027 (GS1 US)
• PR No.117/2569 Thailand Secures $43.6bn 1H 2026 Investment Surge (Thailand BOI)
• Thailand’s Employee Welfare Fund Contributions to Start in October 2026 (DFDL)
• Thailand extends work permits for migrant workers from Laos, Myanmar, Vietnam (The Star)
• Thailand Manufacturing Sector Hits 9-Month High (Trading Economics)
• Zebra FS42 FS42-SR20G4-2C00W (ZPS Store)
• Zebra FS42 FS42-SR20Z4-2C00W (Advanced Automation Inc.)