-
Three lines on a datasheet cause most of the money you'll waste buying outdoor CPE routers: the throughput claim, the word "unlocked," and the security feature list.
-
Why I trust this list, and what it cost to build it
-
The "300Mbps" trap
-
"Unlocked" doesn't mean "works on your carrier"
-
"Cat6 CPE" is not "Cat6 cable"
-
Authentication and security: where cheap units quietly fail
- Two quieter failures worth adding to the checklist
-
Where this checklist doesn't apply
Three lines on a datasheet cause most of the money you'll waste buying outdoor CPE routers: the throughput claim, the word "unlocked," and the security feature list.
Get those three right and you avoid roughly 80% of the re-ships, dead-on-arrival units, and client phone calls I've had to absorb over the past three years. Get them wrong, and the vendor being "cheap and fast" won't save you.
The rest of this article is the story behind that claim, what each spec actually means once you dig into a datasheet, and where this checklist stops being useful.
Why I trust this list, and what it cost to build it
I handle sourcing and OEM coordination for outdoor networking equipment at LinkScope. Between 2022 and 2024, I documented 5 order-level problems and one near-miss — not "a unit felt off," but orders that required re-ship, partial refund, lost client time, or an apology call. Total direct cost: roughly $4,200. Client trust damage: harder to price, but real. I maintain our team's pre-order checklist now because that document didn't exist before I needed it.
The first one still stings. In September 2022, I approved a 200-unit outdoor CPE order where the datasheet said "300Mbps." I read that as LTE throughput. It wasn't. It was 2.4GHz WiFi throughput, and the LTE module inside was Cat4 — theoretical max around 150Mbps down, less in outdoor conditions. The client found out on installation day. We split the cost of a re-spec, about $600 out of our side, plus a two-week delay on their rollout.
The "300Mbps" trap
Two different standards get folded into "300Mbps" in outdoor CPE listings, and almost no factory datasheet separates them:
- LTE Cat4 — 150Mbps theoretical downlink. Cat6 is 300Mbps downlink. Both are real, current, and commonly sold products.
- 2.4GHz WiFi 802.11n — also commonly listed as 300Mbps, though it requires a 40MHz channel that's rarely usable in dense deployments.
So when a listing says "300Mbps 4G CPE," you genuinely cannot tell from that line alone whether you're getting Cat6 LTE, or Cat4 LTE paired with 300Mbps WiFi hardware. Both are sold under identical phrasing.
If the spec sheet doesn't break out "LTE category" and "WiFi standard" as separate rows, treat the 300Mbps claim as unverified. Ask for the cellular module part number — that resolves it in one email. I've stopped approving orders where the module isn't named, because the answer to "which module" has told me more about a unit than any other line on the sheet.
"Unlocked" doesn't mean "works on your carrier"
The minimum meaning of "unlocked" is: not SIM-locked to a particular operator. That's it. What it does not guarantee:
- Support for the specific LTE bands your carrier uses in your region
- Support for the carrier aggregation the operator requires to hit its advertised speeds
- Stable modem behavior on that carrier's network over a 12-month deployment
March 2023, I sourced 80 units from a factory in Shenzhen (circa 2023, things may have changed). The datasheet said "unlocked, B1/B3/B5/B7/B8/B20." The client was deploying into a rural network using B1 and B20 — looked fine. Except the factory's firmware had a VoLTE fallback issue with one regional MVNO, and roughly 30% of the units dropped to 3G during peak hours. Not the spec sheet's fault. Mine, for not testing the actual SIM in the actual deployment. Cost: about $1,100 in re-flash labor plus a very awkward week.
Every sourcing guide, every colleague, and the client themselves told me to test carriers before approving. I only believed it after skipping that step once. Now it's step one on the checklist, no exceptions.
Put another way: "unlocked" describes a modem's legal state. It says nothing about practical fit.
"Cat6 CPE" is not "Cat6 cable"
This one was only a meeting, thankfully, but it's worth flagging because it nearly became an order. A client asked for "Cat6 CPE routers." I confirmed and quoted. In the kickoff call I learned they meant routers with 1000Base-T LAN ports intended to connect over Cat6 cable runs. They did not mean LTE Cat6 CPE. Both are real products. Both are listed under "Cat6" in some marketplaces.
Note to self: always confirm the reference — cellular modem category, or Ethernet interface standard. They are not the same category of device, and some listings genuinely don't distinguish.
Authentication and security: where cheap units quietly fail
"Authentication and security features" on a datasheet can mean almost anything. Two outdoor CPEs at the same quoted price can differ by:
- WPA2-PSK only, versus WPA2/WPA3 mixed mode
- No 802.1X / RADIUS client, versus full RADIUS support
- Captive portal present or absent (this alone determines whether an ISP hotspot or hospitality deployment works)
- Firmware-updateable security patches, versus "we don't push firmware to customers"
That last one is what burns people six months after delivery. In 2024, we received a batch where the vendor's position was that firmware updates required shipping units back. For a $90 outdoor CPE, that's not an update policy — it's a return policy in disguise. I don't have hard data on how common this is across the industry, but based on our three years of outdoor CPE orders, my sense is somewhere between 30% and 50% of the low-cost factories still won't push updates to end customers. That's a sample estimate from our orders, not a market statistic.
I've never fully understood why this is hard for some factories. Remote firmware update is a technically solved problem. My best guess is staffing, not engineering — if nobody's on the other end to field update questions, they won't publish updates.
Two quieter failures worth adding to the checklist
IP rating is not installation reality
IP65 and IP66 are real ratings under IEC 60529, tested under controlled conditions with defined orientations and durations. A unit rated IP65 on a horizontal pole in a temperate climate is not the same deployment as the same unit on a vertical pole in coastal air. Salt spray testing (ISO 9227, ASTM B117) matters at least as much as the IP number if the install is within a few miles of coastline. Ask for the testing lab and the standard, not just the number.
PoE budget at distance
Outdoor CPEs are typically powered over Ethernet, 802.3af or 802.3at. Datasheets will say "supports 802.3at." They will not say what happens at 80 meters with a marginal injector — which is exactly where your installer will be when the call comes in. Early 2024, we lost a three-day production window to this on a client deployment: the "standard" PoE switch dropped below the CPE's minimum input voltage at 90m. Should mention: the fix was a $12 mid-span injector per unit. Cheap fix, expensive delay.
For any deployment with runs past 60–70 meters, buy one unit early and test power at the far end before committing the order.
Where this checklist doesn't apply
This whole approach is overwrought for certain purchases. Buying one unit to test, or ten for a prototype? Order from the lowest-cost vendor you already trust and test it yourself. The checklist only pays for itself once you're moving 50+ units and the cost of a bad order includes re-shipment plus client trust.
It also only covers what I've personally seen at LinkScope. We do sourcing and OEM coordination. We don't do custom firmware development — if you need that, a firmware house is the right call, and honestly we'd tell you so in the first meeting. We also don't manage operator certification on your behalf; that process usually runs between the factory and the operator directly, and it's not our strength.
That said, none of this is secret. If your procurement team reads spec sheets carefully, tests samples against real carriers, and writes down what goes wrong, you don't need a sourcing partner for this part. The value we add is having already made these specific mistakes on our own dime and writing them down. Which is what this article is.