Servo Press and Press-Fit Equipment Selection: Thailand RFP and FAT/SAT Guide

Servo press and press-fit equipment selection for a new or retrofitted cell in a Thai factory cannot be settled by comparing maximum force and stroke alone. Part tolerances, fixture deflection, seating criteria, and the destination of rejected parts all belong in one process specification. This guide helps production engineering, quality, and procurement teams carry the same evidence from the RFP through factory acceptance testing (FAT) and site acceptance testing (SAT) in Thailand.

Decide the process before selecting a servo press

A servo press can program ram position and speed while monitoring force and displacement. That capability does not automatically create a valid pass/fail test. When a bearing or bush is pressed into a bore, roundness, chamfer, lubrication, temperature, and material lot affect the trace. A ram may reach its target position even when the part is tilted or cracked. Conversely, force may enter the nominal range early because the part has met an obstruction before seating.

Begin with the question: which part numbers and tolerance combinations must reach which depth from which datum, and what record proves acceptance? Collect customer drawings, process FMEA, and current defects. Define the part combinations and quality criteria before comparing candidate machines for force, stroke, speed, duty, fixture envelope, and interfaces. Selecting from a catalog first leaves component variation and inspection responsibility outside the purchase scope.

Choose one representative workpiece carefully

For a cell pressing a bearing into a motor housing, show minimum and maximum bearing outer diameters and bore diameters, chamfer, insertion direction, seating datum, allowed tilt, and downstream functional inspection in one controlled drawing set. A nominal sample is insufficient. Include minimum interference, maximum interference, and in-tolerance parts close to the limits in FAT samples. Agree on who supplies the samples, how they will be paired, how many are available, and whether destructive confirmation is allowed. This prevents acceptance from turning into a debate about unavailable representative parts.

New cells and retrofits have different boundaries

A new cell can be designed around the press, frame, guarding, fixtures, PLC, infeed and outfeed, and record storage as one system. A retrofit inherits line height, electrical supply, network protocol, safety circuits, maintenance space, and permitted downtime. If the old frame is reused, check stiffness, eccentric load, fastening, and datum repeatability in addition to strength at maximum force. Replacing only the servo axis will not make the force–displacement trace represent movement of the parts if the supporting frame flexes.

Turn the force–displacement curve into an acceptance decision

The basic record comprises unit ID, part number, recipe revision, timestamp, equipment ID, force–displacement samples, result, and alarm code. Where relevant, link temperature, fixture ID, and calibration status. A curve offers clues about contact, sliding, and seating, but it does not identify root cause on its own. Approve and version control limits using actual good parts, defective samples, and measurement variation.

Start force: specify the coordinate origin

“Press start position” can mean machine home, ram tip, workpiece contact, or the point at which a defined force is reached. If displacement is zeroed at a contact-force threshold, specify the threshold, filtering, search speed, and any retract motion. A very low threshold is sensitive to lubrication and surface variation. A high threshold may place the zero after meaningful deformation has begun. Demonstrate the choice on samples and define who can change the recipe.

Middle travel: use windows, not just a peak value

Tilt, debris, and excessive interference can raise force earlier than expected. Too little interference or a missing component may leave a zone where force should rise almost flat. Define upper and lower force bounds against displacement windows for initial contact, sliding, and the approach to seating. A single maximum-force limit over the entire stroke can call different defects “force OK.” Set window location and width by part number, then validate the association with destructive or functional inspection.

Seating and end position: separate motion completion from product quality

The ram reaching its end position proves that the machine finished a movement. It does not prove that the product seated correctly. Choose quantities connected to product function: seating force, end position, dwell, or height after return. A process that intentionally bottoms out needs different criteria from one stopped at a specified depth. Do not count force from a machine hard stop as part seating force. Keep product acceptance limits separate from protective stop limits; a unit involved in a protective stop should be classified as rejected.

Specify the saved curve format

A curve visible on an HMI is weak evidence if it cannot later be retrieved and exported by unit ID. Put sampling method, number of retained points, filtering, units, time synchronization, retention, missing-data behavior, and numeric export format in the RFP. Do not copy catalog figures directly into a quality limit. Confirm at FAT that sample traces resolve the features required for acceptance. Keeping numerical samples rather than screenshots alone permits later analysis.

Treat fixture stiffness, datums, and sensors as one measuring system

Catalog position resolution is not finished-part insertion-depth accuracy. The drive, ram, load cell, upper tool, workpiece, lower fixture, and frame all deform under load. Machine travel may differ from relative movement between the joined parts. Draw whether the datum is on the workpiece or machine and where displacement is measured; evaluate deflection under process load. SCHMIDT’s ServoPress lists measurement directly at the ram and compensation for system elasticity. Verify any such function with the actual parts and fixtures rather than assuming it delivers the required assembled-part accuracy.

The fixture both holds the part and establishes the datum

Distant lower supports can bend a housing, changing bore shape and axis. Chips or oil on a seating face, clamp force, and wear of locating pins on exchangeable tools move the reference point. Fixture drawings should specify support points, load path, datums, reliefs, cleaning, and confirmation after a fixture change. Define how tool and part axes are aligned; check tilt with and without load. Interlock fixture identity and part recipe so an operator cannot select a mismatched pair.

Calibrate the load cell and verify the installed measurement chain

A calibration certificate does not prove that displayed load equals the force at the tool tip in every setup. Check load-cell orientation, preload, cable, temperature, filter, and off-axis loading. Specify range, required uncertainty, interval, traceability, and the stop condition when calibration expires. At FAT, check against a reference force instrument; repeat after shipping and installation at SAT. Record the reference device, points, measured difference, and values before and after adjustment. Reading the press’s own display is not independent verification.

Calibration and process capability are separate approvals

Calibration checks a sensor against a reference. Product yield and trace spread concern the process. Even with a calibrated sensor, fixture compliance or part variation may make good and defective traces overlap. Simply widening windows around prototype good parts risks defect escape. Once the measurement chain is checked, challenge the windows with good parts at tolerance limits and plausible defects. Quality must decide how false acceptance and false rejection are handled. State sample size and lot makeup with any statistical conclusion; do not invent a universal capability target.

Put the entire press-fit cell into the RFP

Attaching product drawings alone is not enough. Place supply scope, performance, judgment logic, data, abnormal handling, maintenance, and acceptance method in the same requirements table. Clarify what the quotation includes and what the buyer provides before comparing prices. If adjacent processes or guarding are separate contracts, name interfaces, owners, and dates.

1. Parts and production conditions

Provide part numbers and drawing revisions, material, tolerances, insertion direction, seating accuracy, pretreatment, and lubrication. Attach the definition and inspection method for good and bad assemblies. Specify target cycle time from feeding and ID reading through pressing, judgment, discharge, and recovery; distinguish this from unloaded press motion. Give the force–time combination for continuous production as well as peak force and compare it with supplier duty ratings. SCHMIDT’s range separately lists peak and continuous forces, showing why one “maximum kN” number is inadequate.

2. Press, fixture, and protection

Calculate travel from maximum part height, feed clearance, tool length, and retract position. Separate approach, contact search, pressing, and return speeds; define safe contact speed. State the sensor and action for overload, missing or reversed parts, tilt, absent fixture, and wrong recipe. Include hard stops, tool replacement, load-cell protection, and broken-tool recovery. Determine guards and light curtains through a risk assessment of the complete cell, rather than treating a catalog safety option as sufficient.

3. Recipe and quality logic

For each part number, define start force, contact origin, press speed, position and force windows, seating condition, protective stop, and whether rework is allowed. Specify edit rights, approval, change log, and restoration of old revisions. A vendor’s statement that “windows are configurable” is insufficient: ask which sample defect each window detects, demonstrated with design notes and traces. Sanyo Machine Works describes load-cell checks, multiple press programs, and load-curve display. A comparison should ask how those functions will actually be configured and exported in this cell.

4. PLC, MES, and quality records

The PLC signal list should cover cycle permit, part and recipe match, part presence, fixture match, ready, running, OK, NG, equipment fault, reset, and discharge permit. Define direction, timing, timeout, and state after power restoration. Send unit ID, part number, recipe revision, result, trace, fault code, equipment ID, and timestamp to the MES. Quality must decide whether a press can pass a part when the network is down and the record is missing; specify isolation, retry, and duplicate prevention. Delta’s electric servo press describes PLC/HMI/MES integration and force and position histories, but field mapping and failure behavior still require project-specific proof.

5. Maintenance and local support

List wear parts, spare load cells, screw or gearbox checks, lubrication, backup and restore, fault diagnostics, and software licenses. Distinguish items replaceable in Thailand from imports and parts with temporary substitutes. Specify English and Thai operating and maintenance instructions, training, inspection intervals, and access approval for remote support. An advertisement for remote monitoring does not replace factory IT approval. Name the first responder when production stops.

Isolate abnormal parts and prevent duplicate processing

If an operator can place an NG part back on the good conveyor, waveform monitoring will not stop an escape. Treat stopping the ram, retracting the tool, holding the part, moving it to a quarantine destination, locking its ID, and updating MES status as one sequence. Define by defect type whether extraction and retry are permitted, destructive inspection is needed, or a quality manager must release the unit. Avoid a reset that lets the same ID acquire a later OK result that overwrites its earlier NG history.

For “high start force,” “mid-window overload,” “low seating force,” “end-position deviation,” “sensor disconnected,” “MES timeout,” and “emergency stop,” make a table of ram position, part retention, destination, operator message, and record code. Test tools and procedures for removal of a jammed part at FAT. A machine that cannot be recovered safely cannot be operated reliably, even when it meets nominal takt. Match physical quarantine with digital status so it remains traceable across shifts.

Compare servo press offers using conditions that can be proved

Supplier news indicates possible functions, but a product announcement is not a measured result in your factory. On 23 September 2026, Masmec announced a new Armax version with simultaneous control of up to three presses from one module, Industrial Edge remote monitoring, and IoT readiness. These are manufacturer claims, not demonstrated takt, communications load, or judgment performance for three concurrent presses at a Thai site. If shortlisted, FAT should test how a fault on one axis affects the others, unit ID and trace association, and backup boundaries. Treat the Masmec announcement as candidate specification to be verified contractually.

Promess presented the M-Series and UltraPRO 2 at IMTS 2026, including a design intended to avoid manual recalibration after certain component replacements. However, Promess expects first customer shipments in Q1 2027. As of October 2026 this should not be presented as immediately available for a production cell. If it remains a candidate, confirm shipment, production references, and support start date; compare it with products that can meet the current schedule. The Promess release is the primary source for that timing.

Kistler describes real-time force–displacement monitoring and documentation for electromechanical joining systems. SCHMIDT offers ServoPress modules across several force ranges; Sanyo describes load-cell verification and judgment functions; Delta describes PLC/HMI/MES integration. These are examples of vendor capabilities, not an independent comparison under identical conditions. Use separate columns for “catalog claim,” “demo observed,” “proved on our parts at FAT,” and “proved in Thailand at SAT.” Do not fill gaps in accuracy or uptime with guesses.

Use the same cost boundary for every bid

Include the press, production and spare fixtures, feeding and discharge, guarding, PLC changes, MES connection, data retention, calibration, FAT samples, freight, import, installation, Thai-site training, and maintenance agreement. A retrofit also needs a shutdown and restart plan. Mark each item as included, excluded, or buyer supplied in one common format. Compare the same delivered outcome, then separate capital cost from future fixture replacement, calibration, and defect analysis effort.

Add an acceptance matrix to the press-fit equipment RFP

Give each requirement a verification method, pass criterion, location, evidence file, and owner. For end-position accuracy, state which parts and how many, which gauge, and which product datum will be used. Sample counts and tolerances must come from product risk and customer specifications, not generic numbers in an article.

An empty cell in the matrix is an unresolved purchasing decision. Separate what suppliers may propose from criteria the buyer must set. In particular, the buyer owns the good/bad product definition and retention policy. A machine vendor can supply judgment functions but cannot decide the acceptable quality envelope for the product.

FAT: prove the equipment before shipment

FAT establishes that equipment and controls meet the contract before shipping. Reconcile drawing, PLC/HMI software and recipe revisions, calibration certificates, I/O list, and parts list. Then test dry cycles and safety, followed by normal production with parts. Finally inject defect conditions to see whether trace judgment, stop, isolation, and records act together. Save curves and videos with sample ID and timestamp so they can be compared at SAT.

Test edge cases and faults, not only good parts

Use in-tolerance center and edge samples plus simulated insufficient seating, missing or reversed component, debris, and tilt. Agree on safe simulations for hazardous faults. For each sample, retain input condition, expected stop and judgment, observed result, and trace as a set. Do not quietly widen a window at the witness test to accept an ambiguous sample. Log cause and action as an open item. Before FAT closeout, identify items that block shipment and those explicitly assigned to SAT.

Challenge software and communications

Simulate missing unit ID, wrong part number, slow MES response, network loss, absent PLC OK handshake, and power loss. If power fails mid-cycle, verify ram and part positions and who decides disposition after restart. Check sequence numbers and IDs that prevent duplicate upload, local buffer capacity, and retry behavior. Where the customer’s MES is unavailable at FAT, specify the mock server and its limitations and require real integration at SAT.

SAT: validate the installed cell in Thailand

SAT is not a simple repetition of FAT. It proves production with the actual floor, frame, power, air, climate, network, local parts, and operators. Recheck alignment, tool position, load cell, emergency stop, guards, and sensor wiring after transport. If traces from locally sourced parts differ from the samples used abroad, examine drawing conformity, lubrication, fixture datum, temperature, and calibration in that order. Do not merely widen the pass window to make production appear stable.

Carry measurement repeatability from FAT to SAT

To transfer a judgment window approved at FAT to Thailand, document what remained the same and what changed between the two locations. With the same reference workpieces and fixture, use an independently calibrated instrument to measure finished height or seating dimension and pair each result with the press’s recorded end position. If the application wears or deforms a reference piece, do not reuse it indefinitely; prepare several masters whose dimensions have been checked. During witnessing, do not pass one good part once and declare success. Test repositioning the same finished part for measurement and processing separate good samples to distinguish fixture location effects from part variation. Record the measurer, gauge, temperature, fixture ID, recipe revision, and lot with each result so any SAT shift can be investigated later.

Because press fitting may be irreversible, distinguish repeated measurement of the same finished part from pressing several nominally equivalent unprocessed parts. The first evaluates repeatability of the finished-dimension gauge and operator; the second includes variation across the entire process and its inputs. Combining both into one variation figure conceals whether the gauge or components caused a difference. Use the same characteristic definition, units, measurement location, and record form at FAT and SAT. Define a rule that a changed result triggers investigation before anyone changes the pass window.

Verify the datum after product and fixture changeovers

A production changeover may change the tool, lower fixture, stop, and feeding position as well as the recipe. The standard change procedure should include tool and fixture ID matching, cleaning mating faces, checking fasteners and sensor connectors, slow manual interference checks, and checking contact origin and seating position with a reference workpiece. Before releasing the first part as good, have quality examine its trial trace and finished dimension. Compensating only for tool length while reusing the previous product’s quality windows obscures why force rises at a different position. Record position offsets and approvals before and after the exchange, and inhibit automatic cycling if they fall outside the approved range.

Do not attribute an immediate post-changeover fault only to operator carelessness. Consider whether tool mismatches, worn location pins, a loose clamp, and recipe revision conflicts can be detected by the cell. If barcodes or fixture ID sensors are used, FAT should prove that a read failure cannot leave the machine running on an old setup. At SAT, ask local operators to perform the real changeover and check screen guidance, required tools, recovery time, and retained records. Incorporating a datum check into routine changeover makes the process maintainable beyond its initial commissioning team.

Link traces and abnormal parts to controlled revisions

Quality windows should not be set once and forgotten. New material, fixtures, tooling, or drawing revisions can change the distribution of good traces. Attach the recipe and window revisions, calibration status, and fixture ID to every unit trace so results before and after a change can be compared. When a revision is issued, retain the reason, author, approver, affected part numbers, effective time, and sample results before and after. Keep old records readable. If a quality problem appears, unit IDs and production times must identify the products made under each version.

Quarantine also requires ongoing work after the NG signal. Make the list of unresolved units at shift handover, reconciliation between physical quarantine shelves and MES counts, retest authority, and approval for scrap or rework part of daily operation. When records buffered locally during a network outage are uploaded later, confirm that they cannot overwrite the status of a unit already quarantined. Where rework is allowed, retain original press result and post-rework result as separate events visible in traceability. Acceptance should include creating an NG part, isolating it, and having the next shift find its reason and pending disposition by ID.

Require operators and maintenance staff to run normal production, change part numbers, quarantine NG units, exchange fixtures, recover after a power failure, and retrieve records without the builder taking over. Review English or Thai screens and instructions on the floor, correcting unclear alarms. Define recipe approval on night shift and a recovery route when a manager is absent. Measure takt with real feeding, inspection, and discharge, not press movement alone.

At SAT closeout, hand over a controlled list of software, recipes, drawings, backups, calibration, spares, and training evidence. Record each open item with owner, deadline, interim measure, and production release decision. Do not leave unresolved safety or quality criteria under an indefinite “operator control.” Establish an initial period for defect and trace review, a review meeting, and parameter change control before transferring process ownership.

Run a small proof before a full purchase decision

Rather than contracting a universal cell for every part number immediately, prove the trace and fixture on one family with clear quality risk and volume. Measure existing defect types, detection point, inspection, cycle, and downtime. Take good, tolerance-edge, and defective traces on a candidate press; distinguish faults the trace can detect from those needing separate inspection. Then run a single cell with NG quarantine and unit-ID integration, and update the retrofit scope and investment case. A demonstration should end with quality evidence and operating effort, not just “the part went in.”

For a conversion from a pneumatic cell, see our pneumatic-cylinder selection guide for drive limitations. Our servo-control procurement and FAT/SAT guide covers whole-axis and system acceptance. This article addresses the next layer: purchasing specifications that make force–displacement judgment and quarantine work on actual press-fit parts.

Conclusion

Servo press selection cannot be decided by maximum force alone. Use part tolerances and fixture datums to define quality windows; distinguish sensor calibration from process variation; and isolate defective units both physically and in data. Link each RFP demand to evidence, prove equipment function at FAT, and verify local production conditions at SAT in Thailand. Use announcements and catalogs to expand the shortlist, then decide against results demonstrated on your parts.

If a new press-fit cell or retrofit is still at concept stage, drawings, current defects, and quotation boundaries are enough to start an RFP and acceptance outline. Contact TOMAS TECH to discuss fixture, controls, and quality-record boundaries for a Thai factory.

FAQ: servo press selection and acceptance

Can I select a servo press for press fitting by maximum force alone?

No. Maximum force narrows the shortlist but must be evaluated with actual workpiece load, continuous duty, stroke, tool length, fixture stiffness, force and position measurement, and abnormal-part isolation. Read each maker’s peak and continuous force conditions separately.

What defects can force–displacement monitoring detect?

Appropriate windows may detect tilt, debris, insufficient interference, and incomplete seating. Good and defective traces may overlap, depending on the defect and part variation; a curve is not a universal guarantee. Validate with good, tolerance-edge, and defective samples and retain other necessary inspection.

What belongs in a press-fit equipment RFP?

Include part numbers and drawing revisions, tolerances, good/bad definition, defect cases, target cycle, installation constraints, PLC/MES signals, retention, NG handling, and FAT/SAT pass criteria. Specify responsibility for buyer-supplied parts and fixtures.

How should press-fit FAT differ from SAT?

FAT proves controls, judgment, safety, fault recovery, and data export before shipment at the builder. SAT rechecks the installed machine, local lots, real network, neighboring processes, and Thai-site operators. Common sample IDs and trace formats make differences visible.

Are newly announced servo press systems immediately orderable?

Availability varies. Promess presented its M-Series at IMTS 2026 but expects first customer shipments in Q1 2027. For an October 2026 purchase, confirm availability and local support with the maker and compare options with committed delivery dates.

Primary sources

• Masmec, New Armax servo press (23 September 2026)
• Promess, M-Series Press System and UltraPRO 2 at IMTS 2026
• SCHMIDT Technology, ServoPress
• Sanyo Machine Works, Servo Press
• Kistler, Electromechanical joining system
• Delta, Electric Servo Press