Auditing industrial ethernet switch manufacturers is not a tour of a tidy assembly line; it is a document exercise that ends with a nameplate check against the report in your hand. A datasheet is a claim. A type test report, a derating curve and a serial-number traceability record are evidence, and the distance between the two is exactly what an audit measures.
Rayin (Shenzhen Rayin Technology) is a manufacturer of GPON OLTs and industrial Ethernet switches.
Seven documents decide the outcome: the EMC type test report, the environmental test report, the derating curve, the MTBF calculation sheet, the bill of materials with its revision history, the certification list with named holders, and the warranty plus end of life clause.
The temperature number that matters is operating temperature at full port load with the PoE load applied, not storage temperature and not an idle unit on a bench.
MTBF is a failure-rate statistic, not a service life. A 500,000 hour MTBF still implies roughly 1.7% of units failing inside the first year under the usual exponential assumption.
A datasheet answers "what does the vendor say". An audit answers "what can the vendor prove, and does the unit shipping next month match the unit that was tested". Those fail differently: a vendor can be entirely honest about a laboratory sample and still ship a production unit with a different power module or a firmware branch that was never re-tested. That gap between sample and shipment is why every line below asks for a document and a physical cross-check against the label on the box. The checklist runs in audit order: documents first, then bench verification, then commercial terms.
Print this. Every row is a request for a specific artifact, and every row has a condition that ends the conversation.
| Check item | Document to request | Pass criterion | Red flag |
|---|---|---|---|
| EMC immunity | Third-party type test report covering IEC 61000-4-2, -4-3, -4-4, -4-5 and -4-6 | All five sub-tests listed with test levels, ports exercised and performance criterion A or B | A declaration of conformity with no test data, or a report listing only two of the five |
| EMC emission | Emission report to IEC 61000-6-4 or CISPR 32, with the exact model on the cover | Model number, hardware revision and photo on the report match the sample | Report covers a "family" with no individual model identification |
| Temperature rating | Operating temperature range stated explicitly, plus a derating curve | Curve shows load versus ambient, and states whether all ports run at line rate | Only a storage range is quoted, or "-40 to 75 °C" with no curve and no load condition |
| PoE load condition | Thermal report or chamber log with PoE output active | Rating holds with the stated PoE budget drawn at maximum ambient | Wide-temperature claim measured with PoE disabled on an industrial poe switch |
| Ingress protection | IEC 60529 test report for the claimed IP code, with the test lab named | Second digit verified by immersion or spray test, gasket part number recorded | IP67 claimed with no report, or gasket shown as a "tooling sample" |
| Reliability number | MTBF calculation sheet naming the method and its inputs | Telcordia SR-332 or MIL-HDBK-217F named, with ambient, confidence level and fan treatment stated | A single MTBF figure with no method, no temperature and no confidence interval |
| Environmental | Vibration and shock reports (IEC 60068-2-6, IEC 60068-2-27) where the site needs them | Mounting orientation and duration recorded | Report performed on a bare PCB rather than the finished enclosure |
| OEM and ODM terms | Written MOQ, firmware ownership clause, board change cycle, certification holder | Firmware and source ownership stated in the contract, not implied | "We own the firmware" with no clause, or certification held by a third party |
| Warranty and repair | Warranty text defining board-level or whole-unit replacement, plus MTTR commitment | Replacement level written down, turnaround in working days, spare pool described | "Lifetime warranty" with no defined remedy or period |
| Consistency | Serial number traceability from finished unit back to component lots | Two random samples from two batches share the same hardware revision | Same model, two batches, two different PCB revisions and no change notice |
Five sub-tests carry almost all of the diagnostic value, and a report missing any one of them is incomplete rather than slightly shorter.
| Standard | What it simulates | Typical industrial level | What to verify in the report |
|---|---|---|---|
| IEC 61000-4-2 ESD | Operator and handling discharge | Contact 4 kV to 6 kV, air 8 kV | Which ports were discharged, and whether the unit recovered without a reboot |
| IEC 61000-4-3 radiated RF immunity | Handheld radios and nearby transmitters | 10 V/m across 80 MHz to 1 GHz, sometimes 20 V/m | Whether the test ran with traffic on all ports, not on an idle unit |
| IEC 61000-4-4 EFT | Relay and contactor switching on power and signal lines | 2 kV on power ports, 1 kV on signal ports | Whether the DC input was exercised, not only the AC adapter |
| IEC 61000-4-5 surge | Lightning induced transients and grid switching | 2 kV line to ground, 1 kV line to line | Whether both differential and common mode were applied to the Ethernet ports |
| IEC 61000-4-6 conducted RF immunity | RF coupled onto cable harnesses | 10 Vrms from 150 kHz to 80 MHz | Whether the cable length and grounding arrangement are documented |
The performance criterion matters as much as the level. Under the criteria used by the generic industrial immunity standard IEC 61000-6-2, criterion A means the switch performed within specification throughout; criterion B means it degraded and then recovered by itself; criterion C means someone had to intervene. For an unattended cabinet, B is the practical floor and C is a purchase stop.
Three documents look similar on a printed page and are worth very different amounts.
| Document | Who issues it | What it proves | Weight in an audit |
|---|---|---|---|
| Third-party type test report | An accredited laboratory, with accreditation number on the cover | A named sample was tested to a named standard and the data is on file | Primary evidence; ask for the lab's accreditation scope |
| In-house test report | The manufacturer's own laboratory | The design was exercised, but the result is not independently witnessed | Supporting evidence; acceptable only alongside third-party data |
| Declaration of conformity | The manufacturer | The manufacturer asserts compliance | Not evidence on its own; it is a statement about evidence that must exist elsewhere |
A useful trick: ask for the report's file number and the test date, then ask whether the hardware revision on the report matches the revision shipping today. Vendors who keep both records answer in minutes. Vendors who cannot are telling you their change control is informal.
This is the single most common datasheet ambiguity in the category, and it costs about 30 °C of honesty.
Storage temperature describes a powered-off unit in a warehouse. Operating temperature describes a powered unit forwarding traffic. A range quoted without the word "operating" in front of it should be treated as unanswered until the vendor confirms it in writing.
Three follow-up questions close the gap:
Does the rating hold at full port load? An 8 port industrial ethernet switch dissipates noticeably more heat with eight gigabit links active than with two, and the difference is exactly what a rating measured on an idle unit hides.
Does it hold at maximum ambient? The honest number is stated as "full load up to X °C, derated above", with a curve showing what is given up. A rating with no curve is a headline, not a specification.
Does it hold with PoE output drawn? On an industrial poe switch, the PoE supply is often the dominant heat source. If the wide-temperature claim was measured with PoE disabled, the number does not describe the product as deployed.
Ask for the chamber log: ambient setpoint, soak time, traffic generator rate and port count. Four numbers, one page.
IEC 60529 defines the code, and both digits mean something narrow. The first digit covers solid ingress: 4 means protection against objects of 1.0 mm and larger, 5 is dust-protected, 6 is dust-tight. The second covers water: 4 is splashing from any direction, 5 is low-pressure jets, 6 is powerful jets, 7 is immersion at 1 m for 30 minutes.
IP40 is the normal result for a DIN-rail switch living inside a cabinet, and it says nothing about water because the digit is 0, meaning not rated. IP67 appears on sealed field enclosures and requires a real immersion test, real gaskets and a documented torque for the cover screws.
What neither digit covers: ultraviolet exposure, salt mist, corrosion, or the mechanical shock of a cabinet door slamming. Where the site needs those, ask separately for salt mist testing (IEC 60068-2-11) and for the gasket material's stated service temperature, because a gasket that hardens at 70 °C turns an IP67 enclosure into an IP54 one.
MTBF is a prediction, and a prediction is only as good as its inputs. Ask for the sheet, not the number.
| Input to declare | Why it moves the result | What to accept |
|---|---|---|
| Method: Telcordia SR-332 or MIL-HDBK-217F | The two methods weight part counts, stress and quality factors differently and can differ by 2x on the same BOM | A named method and issue number |
| Reference ambient temperature | 25 °C and 40 °C are the common reference points; the same BOM can differ by roughly a factor of two between them | The temperature written next to the figure |
| Fan included or excluded | A fan is usually the shortest-life item in the box; excluding it flatters the number | A separate figure with the fan counted |
| Confidence level | Predictions are quoted at a confidence interval, commonly 60% or 90% two-sided | The interval stated on the sheet |
| Operating profile | Ground benign, ground fixed and mobile profiles carry different stress factors | The profile named on the sheet |
Then the arithmetic that matters to a buyer: MTBF is not service life. Under the usual constant failure rate assumption, a 500,000 hour MTBF gives a first-year failure probability of 1 − e^(−8760/500000), which is about 1.7%. On a 200 unit rollout that is roughly three failures in year one, and the number that determines whether that hurts is the repair turnaround, not the MTBF.
Commercial terms decide what happens in year three, and they are where audits quietly fail after passing every technical check.
| Question | Ask it as | Good answer | Weak answer |
|---|---|---|---|
| Minimum order quantity | The MOQ for a custom configuration, and the MOQ for a custom PCB | Written number for both | "Depends, we can discuss" |
| Firmware and source ownership | Who owns the image, and whether you get source, build scripts and update rights | Stated in the contract | Ownership not mentioned anywhere |
| Board change cycle | How long a schematic change takes, and who pays for tooling | Stated weeks and cost split | "Fast, no problem" |
| Certification holder | Whose name appears on the certificate you will ship under | Your name, or a documented sub-licence | No answer, or a name you cannot verify |
| Component change notice | How much notice before a BOM substitution | Written period, typically 30 to 90 days | "We will tell you when it happens" |
If you are comparing OEM programs across vendors, keep the questions identical and score them side by side; the spread between vendors on these five lines is usually wider than the spread on price.
Four terms do the work, and all four are worthless unless they are written:
Replacement level. Board-level replacement means the vendor swaps a module inside your enclosure; whole-unit replacement means the box is exchanged. On a sealed IP67 enclosure the difference is hours of field labour.
Repair turnaround (MTTR). Ask for it in working days, measured from receipt, with a stated advance-replacement option. Ten to thirty working days is the common range for industrial gear; "as soon as possible" is not a commitment.
Spare pool. Ask whether the vendor holds a consignment of units for your project, and for how long after the last delivery.
End of life notice. Ask for a written notice period before a model is discontinued, and a last-time-buy window. Six to twelve months of notice is the range worth having in writing.
Three are unambiguous, and any one of them is enough to pause:
Same model, two batches, different hardware revision, with no engineering change notice. This means you cannot know which version you will receive next quarter.
Report does not match the nameplate. Model code, input voltage range or serial prefix on the unit differ from the test report. If the tested unit is not the shipped unit, the report is decoration.
Refusal of on-site spot testing. A vendor confident in its own line will let you pick two units at random from finished goods, ideally two 8 port industrial ethernet switch units from different production weeks, and run a 30 minute traffic and PoE load test in the factory. Refusal is information.
What is the fastest way to test whether a manufacturer's claims are real?
Ask for the EMC type test report file number and the hardware revision it covers, then compare that revision with a unit pulled from finished goods. Vendors with working change control answer in minutes; vendors without it cannot.
Why is operating temperature different from storage temperature on a switch?
Storage describes a powered-off unit in a warehouse, so it tolerates a much wider range. Operating describes a powered unit forwarding traffic, where self-heating pushes internal temperatures well above ambient. The operating number is the only one that predicts field behaviour.
Does a high MTBF mean the switch will last that long?
No. MTBF is a statistical failure rate, not a service life. With a 500,000 hour MTBF the first-year failure probability is about 1.7%, and on a 200 unit rollout that is roughly three failures in year one regardless of how large the headline figure looks.
What should an industrial poe switch audit check that a non-PoE audit does not?
Whether the wide-temperature rating was measured with PoE output actually drawn. The PoE supply is usually the dominant heat source in the enclosure, so a rating measured with PoE disabled will not hold in service.
Who should hold the product certification in an OEM agreement?
The name on the certificate must be the name you ship under, or there must be a documented sub-licence naming you. If the certificate sits with an unrelated entity and no paperwork connects it to your shipment, you carry the compliance risk.
An audit is won or lost on documents, not impressions: seven artifacts, each matched against the unit in your hand, and three signals that end the conversation. Rayin publishes port configuration, power input range and environment ratings for its GPON OLT and 8+4-port gigabit industrial switch lines, and answers these same requests document by document during technical evaluation, which is the only fair way to compare industrial ethernet switch manufacturers.
OEM and ODM Programs: What a Switch Maker Will Actually Change for You
Six Dimensions That Matter More Than the Brand Name on the Box
Choosing a Switch for Harsh Sites: Heat, Dust, Vibration and Voltage
Inside Rayin: A Shenzhen Maker of GPON OLTs and Industrial Switches
Industrial vs Commercial Ethernet Switches: Where the Design Diverges
From Console Cable to Config Backup: The Deployment Checklist
The specifications, thresholds and standard references quoted above come from public industry sources — the IEC 61000-4 immunity series, IEC 60529 ingress codes, the IEC 60068 environmental test methods, and published vendor documentation — rather than from marketing copy. The method, worked examples and wording are written by Sara, who checked each figure against those sources before publication. Anything presented as an industry-typical value is labelled as such; nothing here is a measured figure from a Rayin product unless it says so.
About the author: Sara — Sara is a Customer Manager at Rayin with over 10 years of experience in the communications field. In her free time, she enjoys badminton and swimming.
About Rayin: Shenzhen Rayin Technology Co., Ltd. — Company Profile
