Buyer's Guide 📅 September 5, 2026 ⏱ 14 min read

OEM and ODM UHF RFID Readers: What Actually Gets Customised

A UHF RFID reader opened on a manufacturing bench beside its enclosure, antenna ports and a laptop showing firmware configuration settings

Most OEM enquiries that reach a UHF RFID reader factory open with a sentence borrowed from the tag business: “we want our logo on it, what’s the MOQ?” It is a fair question, and it has a completely different answer for a reader than it does for a card, a label or an inlay. A tag programme is a print-and-convert decision. A reader programme is a hardware programme, with firmware, radio approvals, an SDK and a support term attached to it.

This guide is written from the manufacturing side of that conversation. It sets out the five levels at which a UHF reader actually gets customised, what each level does to tooling and to radio conformity, who legally holds the FCC ID and the CE technical file when your brand goes on hardware someone else designed, what a reader SDK should contain before you commit to it, and how to write an RFQ a factory can quote in one pass instead of three.

The Indian and US regulatory figures below are quoted from the instruments themselves — the Gazette of India notification and the US Code of Federal Regulations — because those are the numbers that decide whether a badged reader ships in one quarter or three.

]]>

Why reader OEM is a different conversation from tag OEM

Tag and label OEM is a mature, high-volume, well-documented business, and the published terms show it. JN RFID’s OEM/ODM page puts minimum order quantity at “generally…1,000 to 5,000 pieces,” samples at “usually 5 to 7 business days,” and production at “2 to 4 weeks after sample approval.” The customisation menu is chip selection, substrate, printing, encoding and packaging. Every item on that list is a print-and-convert decision, because a passive tag backscatters the interrogator’s carrier rather than generating one of its own — the Indian rules put the tag response at −20 dBm e.r.p., in a frequency range around the interrogator channels.

A reader is the transmitter. It is the licensed, type-approved, certificate-bearing half of the system, and every customisation decision has to be tested against that fact. Five questions set the entire quote, and a factory can price a programme within a day once it has all five: which band, which enclosure, whose firmware, whose certificate, and how many.

	Tag / label OEM	Reader OEM
Unit of customisation	Artwork, inlay, substrate, encoding	Firmware, I/O, enclosure, board
Typical MOQ	1,000–5,000 pieces	Far lower; tier sets the practical minimum
Sample turnaround	5–7 business days	Depends on tier — days at Tier 1, tooling-bound at Tier 4
Tooling	Print plates, die-cut tools	Injection or sheet-metal tools, test fixtures
Radio approval impact	None — the reader is the transmitter	Central. Decides cost and calendar
Software scope	Data encoding	Firmware, SDK, host protocol, update path
Support horizon	Per shipment	Measured in years

Set that frame in the first email. Naming the conversation you are actually in on day one is worth weeks in the quotation cycle, and manufacturers who ship both tags and readers — see our manufacturing and export overview — will tell you which one you are in straight away.

The five customisation tiers, from label swap to new board

Almost every reader OEM request lands on one of five tiers. Naming the tier is the single most useful thing you can do to speed up a quote, because it fixes the tooling, the NRE and — decisively — whether the radio approvals travel with the product or have to be recreated.

Tier 1 — badge and box

Your logo on the product label, your carton and manual, your boot splash and web UI branding, your part number in the packing list. The certified electrical configuration is untouched, so the existing grant and technical file continue to apply and the unit runs on a standard production slot. One point that catches most first-time private-label buyers: putting your own FCC ID on the product is a certification event in its own right, and it sits in the certification section below.

Tier 2 — firmware and defaults

Preset read power per antenna port, session and target, antenna dwell and cycle order, Q value, host protocol and endpoint, boot behaviour, heartbeat interval, and the payload the unit ships with after a factory reset. This is where most of the perceived product differentiation actually lives, and it is firmware engineering with no tooling attached. Provided the defaults stay inside the certified power and channel envelope, the radio paperwork is unaffected.

Tier 3 — interface and I/O

Relay count, Wiegand, RS485, RS232 and GPIO fit-out, PoE against DC input, connector types, cable-gland sizes, antenna connector choice. Usually a PCBA variant or a daughter board plus a new test fixture. The RF chain is unchanged, but a materially different cable set can call for a conducted-emissions retest, so budget for lab time even though the radio itself has stayed where it was.

Tier 4 — enclosure

A new housing, a different mounting pattern, a changed IP sealing scheme, an integrated-antenna form factor. This is the first tier with real tooling cost, amortised across the first order and recovered on reorders. It is also the first tier where radiated performance genuinely changes: move the antenna, change what metal sits near it, or alter the radome material and you have changed characteristics that were reported at certification.

Tier 5 — new PCB or module carrier

A full programme: schematic and layout, RF matching, thermal design, design validation, tooling, fixtures, and a fresh conformity file for every destination market. Quoted as a development project with milestones.

Tier	Tooling / NRE	Effect on radio approvals	What sets the calendar
1 — Badge and box	Artwork setup only	None; existing grant and DoC apply	Standard production slot
2 — Firmware and defaults	Firmware engineering, no tooling	None within the certified envelope	Build and regression test
3 — Interface and I/O	PCBA variant, test fixture	Possible conducted retest	Component lead time, lab slot
4 — Enclosure	Injection or sheet-metal tool	Radiated retest; filing where reported characteristics change	Tool build and trials
5 — New board	Full NRE, tooling, fixtures	New application, new conformity file	Design validation and radio testing

Ask any prospective manufacturer to place your requirement on this ladder in writing, with the retest position stated. A supplier who can do that quickly has done it before.

Band configuration belongs on the order line

The band is fixed by the country where the reader will be switched on. It is a per-order configuration — ETSI-band 865–868 MHz and FCC-band 902–928 MHz variants are built to the destination named on the purchase order, and the choice cascades into antenna tuning, tag selection and the conformity route.

The two plans are structurally different radios. In the United States, 47 CFR 15.247 governs the 902–928 MHz band and permits a maximum peak output of “1 watt for systems employing at least 50 hopping channels; and, 0.25 watts for systems employing less than 50 hopping channels,” with hopping carriers “separated by a minimum of 25 kHz or the 20 dB bandwidth of the hopping channel, whichever is greater,” and average occupancy on any one frequency limited, for systems using at least 50 hopping frequencies, to 0.4 seconds within a 20-second period. That is a wide, frequency-hopping plan.

Europe runs a narrow, listen-before-talk, fixed-channel plan built on EN 302 208. In practice European UHF RFID sits in 865–868 MHz at no more than 2 W, as the export-certification guidance from RFID Label summarises.

Why one ETSI-tuned SKU covers two large markets

India’s licence-exempt allocation now sits inside the same band edges. Notification G.S.R. 853(E) of 10 December 2021, published in the Gazette of India, made the Use of Low Power Equipment in the Frequency Band 865–868 MHz for Short Range Devices (Exemption from Licence) Rules, 2021, expressly “in supersession of” the older 865–867 MHz RFID rules of 2005. Its Table-IV, covering Radio Frequency Identification Applications, sets 865–868 MHz at 2 W e.r.p. with channels of 200 kHz or less, and names EN 302 208 in the harmonised-standard column.

The detail that matters on the production line is in the footnote: interrogator transmissions at 2 W e.r.p. “are only permitted within the four channels centred at 865.7 MHz, 866.3 MHz, 866.9 MHz and 867.5 MHz; each with a maximum bandwidth of 200 kHz.” The same table caps continuous interrogator transmission on a channel at 4 seconds and requires at least 100 ms between consecutive transmissions on the same channel. A European-tuned reader therefore serves India with a channel-plan and dwell configuration, which is precisely why the band belongs on the order line. Rule 5 of the same notification still requires equipment type approval, and our UHF reader models carry WPC ETA and BIS registration for India’s de-licensed UHF band.

Put the destination country in your first email. It locks the variant, the tag tuning and the conformity route before anything is committed to production.

Who owns the FCC ID and the CE technical file

This is the question that separates a serious private-label buyer from a hopeful one, and the answer is written into US federal regulation in language that leaves little room for interpretation.

The FCC ID is bound to a party

Under 47 CFR 2.926, the grantee code “is assigned permanently to applicants/grantees and is valid only for the party specified as the applicant/grantee.” The second half of the FCC ID is the equipment product code, which “shall consist of a series of Arabic numerals, capital letters or a combination thereof, and may include the dash or hyphen,” with the total not exceeding 14 characters. A grantee code is obtained electronically from the Commission, with the fee due within 30 days or the code is removed from the records. And if “the grantee name is changed or ownership is transferred, the circumstances shall be reported to the Commission so that a new grantee code can be assigned, if appropriate.”

The consequence for a rebranding buyer is in 47 CFR 2.1043. A Class I permissive change covers modifications that do not degrade reported characteristics and needs no filing; a Class II change degrades them while still meeting the limits and requires a filing and FCC acknowledgment before marketing. Changes to “equipment identification” — explicitly “with or without circuit changes” — always require a new application and a new grant. Changing the ID on the label is a certification event in its own right.

The two clean routes

Operate under the manufacturer’s grant. The reader ships bearing the manufacturer’s FCC ID and grantee code, with your brand on everything else. Fastest to market, no new application, and the paperwork already exists. Ask for the grant, the test reports and the label artwork specifying ID placement, and make sure your own label design leaves that marking legible.
Hold the grant in your own name. You obtain your own grantee code, and a new application is filed against the same design with you as the applicant. This is the route to take when your brand must be the regulatory identity of the product. It costs a filing cycle and lab time, and it needs the manufacturer’s active cooperation on the technical file — agree that cooperation in the purchase contract.

CE, UKCA and the national regimes

The European route is a self-declaration backed by evidence. As RFID Label’s export-certification guidance summarises, CE marking under the Radio Equipment Directive turns on health and safety, electromagnetic compatibility, and efficient use of the radio spectrum. Behind the mark sits a technical file and a signed Declaration of Conformity, and Directive 2014/53/EU puts the accountable parties inside the Union: under Article 11 a manufacturer “may, by a written mandate, appoint an authorised representative,” who is established within the Union, while Article 12 defines the importer as a person “established within the Union who places radio equipment from a third country on the Union market.” A brand owner based outside the EU therefore works through one of those two roles, and it is worth naming which one at contract stage. Great Britain runs the parallel UKCA route: the UK government’s guidance covers “placing a product on the market in Great Britain (England, Scotland and Wales) under the UK Conformity Assessed (UKCA) or Conformité Européenne (CE) regimes.” Australia and New Zealand use the RCM under the ACMA regime, which the same RFID Label guidance covers. Several other markets operate their own type approvals, so list every destination country in the RFQ and ask the manufacturer to name the route that applies to each one.

What to write into the purchase agreement

Which party is named grantee/applicant for each destination market, listed by market.
Copies of test reports and the technical file, and the right to use them for derivative filings.
Notice period and re-test responsibility if the manufacturer makes a design change during your supply term.
Label artwork control, including who signs off ID placement and marking legibility.
What happens to the file if either party’s ownership changes — the reporting duty in 2.926 makes this a live scenario worth drafting for.

Firmware and SDK: what an OEM buyer should insist on

A reader you can integrate in a week earns its place immediately, so evaluate the software before you evaluate the enclosure. The most useful public benchmark is Zebra’s Android RFID Reader SDK, which ships as an Android .aar library described as a “class library, sample apps and source code to enable developers to easily build apps,” with documentation covering inventory, read and write access operations, block and permalock, tag storage settings, antenna configuration, singulation, pre-filters, LLRP configuration, region settings, firmware updates, trigger-mode switching and device discovery over Bluetooth, USB and serial. Hold every SDK you are offered against that list.

Five things to check before you commit

Platform coverage and a real reference app. Android, Windows and Flutter is the practical set for most reader programmes. A reference app should do more than list EPCs: it should let you change power per port, switch session and target, run a filtered inventory and push results to a host endpoint, because that is what your first integration actually needs.
How the reader speaks. LLRP, a vendor SDK, or raw TCP and serial are three different amounts of work. LLRP gives you a standard vocabulary and portability across compliant hardware. A good vendor SDK is usually faster to a working prototype and exposes features that sit above what a generic layer addresses. Raw TCP or serial works well when the protocol is documented and stable — ask for the protocol document early and you will know which case you are in.
On-reader logic. Filtering, deduplication windows, direction logic and simple rules running on the reader mean the network carries events instead of raw reads. That is the difference between a site that works over a thin link and one that needs a server in every room. Our edge management platform exists for exactly this: pushing configuration, rules and firmware to a fleet from one place.
Update mechanism and support term. Ask how firmware is delivered, whether images are signed, whether the device can roll back a bad update, and how long a released version is supported. Get the answer in the contract with a defined support window.
Long-term maintenance. Source access, escrow terms, and a named engineering contact. For a programme you will be selling for five years, the maintenance answer matters as much as the unit price.

Integration scope is also where the deployment target shows up. A reader destined for warehouse operations needs portal and dock-door logic; one going into asset tracking needs low-duty-cycle scheduling and clean last-seen semantics. Say which, early.

MOQ, tooling, NRE and lead times in a reader programme

Reader MOQs sit far below tag MOQs. Tag and label OEM starts at the 1,000 to 5,000 pieces JN RFID publishes; a reader programme starts a long way under that, and the reason is arithmetic. A reader carries a much higher unit value and a much lower per-unit setup burden, so even a modest first order can absorb firmware NRE. Tooling is the one item that needs volume behind it, which is why Tier 4 changes the shape of the conversation.

What NRE buys, tier by tier

At Tier 2 it buys engineering hours: a defaults set, a protocol adapter, a regression run against your configuration. At Tier 3 it buys a PCBA variant, a bill-of-materials change and a production test fixture. At Tier 4 it buys a tool — a physical asset with a working life long enough to serve years of reorders, quoted once, amortised across your first order and then recovered on every reorder after it. Ask three questions about tooling every time: what does the tool cost, who owns it, and where does it live. At Tier 5, NRE is the programme.

What actually sets the critical path

In practice the critical path is one of three things: radio testing, where lab slots are booked ahead and a failed run costs a full cycle; enclosure tooling, where first samples off the tool are followed by trials and adjustment; and long-lead components, where a single controller, radio module or connector with a long quoted lead time dictates the whole schedule. Ask for the bill-of-materials risk list — the parts with the longest quoted lead time and their second sources — at quotation stage.

Making the second order faster than the first

The first order pays for discovery, and the structure of the second order is what carries that discovery forward. Set it up as a schedule agreement: a total quantity with call-off dates, a rolling forecast the factory can buy long-lead parts against, and an agreed buffer of the two or three components with the worst lead times. Add a firmware baseline — a named, frozen version that ships unless you approve a change — so every reorder arrives with exactly the behaviour you signed off. Once the long-lead parts are pre-bought, the remaining schedule is the assembly and test window.

Quoting terms: EXW, FOB, CIF and DDP in a reader programme

A unit price becomes comparable the moment an Incoterm is attached to it. The term decides where cost, risk and paperwork transfer, and a USD quote against a stated Incoterm is the format that lets you put three suppliers side by side honestly.

Term	Manufacturer handles	You handle	Best when
EXW	Goods packed and ready at the works	Export clearance, freight, insurance, import duty, delivery	You have a freight forwarder and consolidate with other suppliers
FOB	Export clearance and delivery on board at the named port	Main carriage, insurance, import duty, final delivery	The most common baseline for comparing quotes across countries
CIF	Carriage and insurance to the destination port	Import clearance, duty, inland delivery	You want one line for freight but keep control of customs
DDP	Everything to your door, duties included	Unloading	Pilot batches, trade-show units, and buyers who prefer the supplier to run the import desk

Build the landed-cost model before you choose. Freight on readers is often volumetric rather than by weight, because the enclosures and their packaging are bulky relative to mass, so carton design can measurably move the freight line. Duty depends on your own tariff classification, which is your call to make and worth confirming with your customs broker in advance.

Ask for the document set with the quote, so it travels with the shipment: commercial invoice, packing list with net and gross weights and carton dimensions, certificate of origin where a preferential rate applies, and the conformity documents and test reports for the destination market. Those documents are what turn a shipment into a clearable one, and getting them right the first time is a large part of what a practised exporter is actually selling. If the readers feed a wider deployment, the same discipline pays off downstream in supply chain tracking, where every consignment repeats the pattern.

A specification questionnaire you can send to any manufacturer

Copy this, fill in what you know, mark the rest “advise,” and send it to three suppliers. It is vendor-neutral by design. A factory that can answer it in one reply is a factory that has run reader programmes before.

Radio and performance

Destination market or markets, listed by country.
Band variant: ETSI 865–868 MHz, FCC 902–928 MHz, or both as separate SKUs.
Form factor: fixed multi-port, integrated reader-antenna, desktop, handheld or embedded module.
Number of antenna ports, connector type, and cable lengths.
Output power range required at the port, and per-port control.
Air-interface standard and any protocol or memory-bank requirements.

Interfaces and software

Host interface: Ethernet, PoE, Wi-Fi, cellular, RS485, USB.
Digital I/O: input count, relay count, Wiegand or serial peripherals.
Protocol: LLRP, vendor SDK, MQTT or HTTP push, or a documented raw socket.
SDK platforms required, and whether a reference app with source is included.
On-reader filtering, deduplication or rules capability.
Firmware update mechanism, signing, rollback, and the supported-version window.

Physical, branding and commercial

Branding scope, stated as a tier from 1 to 5.
Enclosure, mounting pattern, cable entry, IP rating and operating temperature range.
Certification route and which party is the named grantee or responsible party per market.
Quantity for the first order plus a 12-month forecast.
Required delivery date, Incoterm and destination.
Supply term, spares and service-part availability, and warranty period.

Those last six lines are what turn a browse into a quotable RFQ. Send the whole thing and the three quotes you get back will be comparable line for line.

One vendor for the hardware and the software

The reason to care about who writes the firmware becomes clear the first time a deployment needs a change. Take a real pattern from a portal installation: the site wants dwell time per antenna adjusted so a forklift passing at speed is read reliably, and the host wants events pushed to a different endpoint format. When the hardware, the firmware and the integration layer come from one engineering team, that request is a single firmware build with two changes in it, tested together and shipped once — one thread, one owner, one release note.

That is the whole argument for building both under one roof, and it is the one Identium makes: hardware and software developed in-house, one vendor and one point of accountability, so a defaults change, a protocol change and a firmware release are the same conversation.

Test the claim before you commit

Measure it during evaluation. Ask for a specific firmware change on the sample unit: a different default power profile, an added field in the event payload, a changed heartbeat interval. Then time it, and note three things. How long until someone who can actually make the change is on the thread. How long until a build arrives. And whether the build comes with a version number and a note describing what changed. A supplier who turns that around in days, with a versioned image and a changelog, will do the same in year three of your programme. The answer tells you exactly where the firmware is written, and how close you sit to the people writing it.

That single test, run during evaluation, is worth more than any capability slide — and it costs you one email.

Frequently asked questions

What is the difference between OEM and ODM for RFID readers?

OEM means the manufacturer builds to your design and specification — you own the product definition and they supply production capacity. ODM means the manufacturer’s own engineering team designs the product and you commission a version of it, usually badged and configured for you. JN RFID frames the same split on the tag side as production “based entirely on your technical specifications” against “turnkey product development managed by our internal RFID engineering team.” For readers the practical difference is who owns the schematic, the firmware source and the certification file. Many reader programmes described as OEM are ODM in practice — branding and firmware customisation on top of an existing certified platform — which is usually the faster and cheaper answer.

What is the minimum order quantity for a custom UHF RFID reader?

Far lower than for tags. Tag and label OEM MOQs are published in the 1,000 to 5,000 piece range; a reader programme starts well below that, because the unit value is much higher and the per-unit setup burden much lower. Tier, more than quantity, sets the sensible minimum. Branding and firmware customisation can be economic at very small volumes because there is no tooling to amortise. A new enclosure introduces a tool, and the tool cost then has to be spread across your first order and reorders, which is what lifts the practical minimum.

Can I put my own brand and FCC ID on an RFID reader I did not design?

Your brand, yes, at any volume. Your own FCC ID is a separate matter, and US regulation is explicit about it. Under 47 CFR 2.926 the grantee code “is assigned permanently to applicants/grantees and is valid only for the party specified as the applicant/grantee,” and under 47 CFR 2.1043 a change to equipment identification — “with or without circuit changes” — requires a new application and a new grant. So there are two workable routes: ship under the manufacturer’s existing FCC ID with your branding on everything else, which is fastest, or obtain your own grantee code and file a new application against the same design with the manufacturer’s cooperation on the technical file. Agree which route you are taking in the purchase contract.

Does changing the enclosure of an RFID reader require re-certification?

It depends on whether the change alters the characteristics that were reported at certification. Under 47 CFR 2.1043, a Class I permissive change covers modifications that do not degrade those reported characteristics and needs no filing, while a Class II change degrades them — still within limits — and requires a filing and FCC acknowledgment before marketing. In engineering terms: a cosmetic change to a non-conductive housing that leaves the antenna, its position and the surrounding metal untouched is usually the benign case, while moving the antenna, adding metal near it or changing radome material affects radiated performance and calls for retesting. Get the manufacturer to state the retest position in writing alongside the tooling quote.

How long does a custom RFID reader programme take from spec to first shipment?

Tier sets the calendar. Branding and firmware work runs on the standard production schedule. An interface variant adds component lead time and possibly a conducted-emissions retest. A new enclosure adds tool build plus tool trials, which is normally the longest single item on the plan. A new board adds design validation and full radio testing on top of that. The three things that actually control the date are lab availability for radio testing, tooling, and any long-lead component in the bill of materials — so ask for the long-lead parts list and their second sources at quotation stage.

Can an OEM RFID reader be supplied in both FCC 902-928 MHz and ETSI 865-868 MHz variants?

Yes — band variants are configured per order against the destination country. The two plans are structurally different: 47 CFR 15.247 governs the US 902–928 MHz band with frequency hopping, permitting 1 watt peak output for systems using at least 50 hopping channels and 0.25 watts below that, while the European plan built on EN 302 208 uses fixed narrow channels in 865–868 MHz at up to 2 W. India’s Gazette notification G.S.R. 853(E) of 10 December 2021 places licence-exempt RFID in 865–868 MHz at 2 W e.r.p. in channels of 200 kHz or less, with 2 W permitted only in the four channels centred at 865.7, 866.3, 866.9 and 867.5 MHz — so a single ETSI-tuned platform serves both Europe and India through channel-plan configuration. Name the destination on the order and the variant, antenna tuning and conformity route are fixed from the start.

What should an OEM RFID reader SDK include?

Use Zebra’s Android RFID Reader SDK as the benchmark: it ships as a class library with sample apps and source code, and documents inventory, read and write access operations, block and permalock, tag storage settings, antenna configuration, singulation, pre-filters, LLRP configuration, region settings, firmware updates, trigger-mode switching and device discovery over Bluetooth, USB and serial. For your own programme, insist on the platforms you will actually build for — Android, Windows and Flutter covers most cases — a reference app with source that demonstrates power and session control rather than just an EPC list, a documented host protocol, and a stated firmware update and support policy. Ask about source access or escrow if the product will be in your catalogue for years.