Direct Answer
The site needs to connect instruments and DI/AI signals at the same time — should you buy the ESX, the CW or the CX? The three have different positions: the ESX has an OLED display, DC5V power and 4G uplink, built for monitoring-class access; the CW is a DC24V display-less industrial gateway; the CX comes with 8DI+8DO+8AI built in and can do local control directly — choose by three questions: "do you need control, do you need a display, and how does the uplink go". The models: ESX-0223-GR, CW-C1/C2/C3, CX-08R06AI08-C1/C2/C3.
A scene: a park is buying gateways and the procurement person asks "what's the difference between ESX and CW, and what is CX" — one sentence cannot explain it, and the wrong model means rework on site. The three gateways are not an "upgrade ladder"; they are a "positioning set": monitoring picks ESX/CW, control picks CX — before choosing, decide whether the site "watches it" or "manages it".
A second scene: a project first bought the ESX as a monitoring gateway, and at acceptance the customer said "when the cabinet temperature exceeds 40°C, the exhaust fan must start automatically" — the ESX is a monitoring gateway with no DO output — so a PLC had to be added, logic rewritten, and two more weeks of integration. Choosing the CX first would have covered everything in one unit: 8DI for status, 8DO for the fan, 8AI for temperature, and the local logic runs directly — reverse the selection order and the whole site reworks.
1. The Three-Model Comparison: One Table Settles It
ESX — OLED display, DC5V power, 4G uplink; fits monitoring scenarios that need on-site viewing and have no 24V supply; CW — DC24V power, no display, a pure industrial gateway; fits in-cabinet installation and platform-only viewing; CX — built-in 8DI+8DO+8AI, a "gateway + control unit" two-in-one; fits scenarios requiring local logic control.
In one sentence: ESX is the "monitoring gateway with a screen", CW is the "monitoring gateway without a screen", and CX is the "gateway that can control" — position first, then choose.
A positioning example: a switch room needed to connect meters (monitoring) plus exhaust-fan interlock (control) — buying only the ESX could only read data, not control, and a controller would have to be added; choosing the CX directly: 8DI on device status, 8DO on the exhaust fan, 8AI on temperature, and the local logic runs on board. A wrong positioning is not "one more device"; it is one more layer of logic on the site.
A one-line table: ESX = screen + DC5V + 4G uplink, "monitoring plus on-site viewing"; CW = no screen + DC24V + industrial rail, "monitoring plus platform viewing"; CX = 8DI+8DO+8AI + local logic, "monitoring plus control". Three rows and three columns to memorize: screen, power, control — three features and the position is instantly clear.
2. Pure Monitoring: How to Pick Between ESX and CW
Look at power — DC24V available on site picks CW; only DC5V, or needing 4G uplink, picks ESX; Look at display — on-site data viewing picks ESX (the OLED shows values directly); platform-only viewing picks CW; Look at uplink — wireless 4G backhaul picks ESX; wired network picks CW; General judgment — both connect Modbus instruments for monitoring; "screen plus 4G" is the ESX's bonus, not the CW's flaw.
In one sentence: ESX and CW are "two configurations of the same gateway class" — the ESX adds a screen and 4G, the CW is more industrial; for pure monitoring, pick by "power + display + uplink".
Why ESX uses DC5V: field sensors and local displays commonly use 5V; the ESX positioned as "on-site monitoring" shares the 5V source with sensors. The CW's DC24V is the industrial cabinet standard — the two voltages fit two installation scenes, not a performance difference. Do not agonize over which voltage is "better"; look at what the site has — if both exist either works, and if only one exists follow it.
A power example: a project switch room had only DC24V and platform viewing — the CW, DIN-rail mounted in the cabinet, no screen, no 4G, simple and reliable; another project had no 24V on site and needed 4G backhaul — the ESX with DC5V + OLED + 4G works where there is no cabinet power supply. The same monitoring need, different power conditions, different models — selection follows site conditions.
3. DI/AI Acquisition and Local Control: Pick the CX
8DI — 8 dry-contact inputs: device start/stop status, breaker position, door magnet signals; 8DO — 8 dry-contact outputs: exhaust-fan start/stop, audible-visual alarms, breaker open/close; 8AI — 8 analog inputs: temperature, pressure, level, as 4-20mA/0-10V signals; Local logic — control rules written inside the gateway (start the fan above 40°C), independent of the cloud, still executing while disconnected.
In one sentence: the CX is a "gateway + PLC" two-in-one — monitoring and control in one body, local logic usable offline; the CX's stage is any site with "something that must act".
How DI/DO/AI divide the work: DI handles "status" (fan running or not, door closed or not), DO handles "action" (start the fan, light the red lamp), AI handles "process values" (how many degrees, how many pascals) — the three signal classes exactly cover the three kinds of site information: status, action and value. 8+8+8 is not arbitrary: one typical switch-room control scenario is basically covered by 8 of each.
A control example: a switch room needed "start the exhaust fan automatically above 40°C, audible-visual alarm above 60°C" — the CX's 8AI reads the temperature probe, 8DO drives the fan and the alarm, 8DI reads the fan status, and the rules live in the gateway — running even when powered or network disconnected. Without the CX, this logic needs a PLC plus a gateway, two devices and two programming jobs — the CX does it in one, saving devices and integration.
4. Capacity: 30 Devices / 2000 Data Points
30 devices — single-gateway normal access is about 30 Modbus devices (within a reasonable polling cycle); 2000 data points — single-gateway data-point capacity is about 2000 (registers/points combined); A note on capacity — capacity is not "the more the better": more devices means slower polling and more points means bigger caches — beyond capacity, split gateways, do not cram; Planning advice — estimate points as "device count × points per device", add 20% margin, then decide the number of gateways.
In one sentence: 30 devices and 2000 data points are the "normal capacity" — plan by need with margin, split gateways when exceeded, never cram.
How to count points: one register/point is one data point (voltage, current and power each count one); a multi-function meter runs about 30–50 points and a switch-room instrument about 20–30. Count "device count × points per device" first and add 20% margin — do not wait until installation is done to discover the points exceeded.
A capacity example: a park had 80 meters × 12 points = 960 data points — one gateway is enough (far below 2000), but polling 80 devices takes about 4 seconds per round — for faster refresh, split into two gateways of 40 each. Capacity within limit does not equal a good experience — even if points are fine, the polling cycle must be calculated too.
5. Suffix Mapping: C1/C2/C3 and -R/-E/-G
C1/C2/C3 — communication suffixes: C1 with 4G, C2 with Ethernet + RS485, C3 full-channel (4G + Ethernet + RS485 all on); -R/-E/-G — uplink suffixes: R for 4G wireless, E for Ethernet wired, G for all-network; Mapping — C1 ≈ 4G-first, C2 ≈ wired-first, C3 ≈ full-channel — pick the suffix by the site's network conditions; CX model breakdown — CX-08R06AI08: 08 is 8DI+8DO+8AI, R is 4G uplink, 06 is six RS485 channels.
In one sentence: the suffix is the "communication configuration sheet" — C1/C2/C3 sets the communication mix and -R/-E/-G sets the uplink; read the suffix and the selection table is clear at a glance.
Why "all-network": the -G suffix means support for all three carriers' 4G (mobile, unicom, telecom), not locked to one carrier — swap the SIM card on site and it works; -R means 4G-wireless-first and -E means Ethernet-wired-first. Reading the suffix is reading the "network preference": G is "any card works", R is "wireless first", E is "wired first" — when the network condition changes, the suffix follows.
A suffix example: a site had fiber wired networking — choose CW-C2 (Ethernet + RS485, wired first); another site had no cable, only 4G signal — choose ESX-0223-GR (4G all-network uplink); a third project wanted full-channel redundancy — choose C3. The network condition decides the suffix, and the suffix decides the model — see what network the site has first, then fix the suffix.
6. Three Questions for Selection: Control? Display? Uplink?
Q1: does the site need control actions — yes (fan/alarm/breaker) picks CX; pure monitoring goes to Q2; Q2: does it need on-site display — yes picks ESX (OLED); no picks CW (no screen); Q3: which network for the uplink — 4G picks -R/G, wired picks -E/C2, full redundancy picks C3.
In one sentence: control → CX, display → ESX, uplink → suffix — answer the three questions and the model comes out.
Why the order matters: the control need is a "function-level" difference (decides the whole machine), the display is a "configuration-level" difference (two configurations of one platform), and the network is a "suffix-level" difference (a combination item) — function first, configuration second, combination last; the three-level funnel leaves a unique model. Looking at the network first, nine of ten projects pick C3 — but with CX vs ESX undecided, the model is still empty.
A complete selection: a park switch room: cabinet-temperature control needed (Q1: CX), on-site temperature viewing (CX screen optional), fiber on site (suffix C2) — the model CX-08R06AI08 adjusted accordingly. The order of the three questions must not be shuffled: decide "control or not" first (CX or not), then "display" (ESX or not), finally "network" (suffix) — reversed order lets the suffix pull you off course.
Back to FEXLINK: the ESX-0223-GR, CW-C1/C2/C3 and CX-08R06AI08-C1/C2/C3 cover the full "monitoring → control" spectrum — pure monitoring picks ESX/CW, control picks CX, and the communication suffix follows the site network. Walk into selection with the three-question checklist — need control, need a display, what network — answer the three and the model jumps out by itself; no parameter table to memorize.
Scope and Limitations
- The conclusions in this article are intended for LV distribution and electrical-safety sites. Criteria are compiled from public codes and FEXLINK product documentation, and product parameters follow the latest selection handbook and product knowledge base.
- Quantitative criteria (protection zone, discharge and residual voltage, earth resistance, thresholds and grading) hold under the preconditions stated in this article. Where the site voltage level, earthing arrangement, environmental category or load profile differs, the figures must be recalculated. Specific configurations involving ESX-0223-GR、CW-C1/C2/C3、CX-08R06AI08-C1/C2/C3 should be verified against actual circuit parameters.
- Specific product models and parameters are subject to the latest selection handbook and product knowledge base; this article does not constitute selection or procurement advice.
- Standard numbers cited here refer to the currently effective editions; when a standard is revised, the related quantitative conclusions must be re-checked.
- Deployment and implementation for specific projects must be carried out by qualified engineers based on site survey and design drawings.
Frequently Asked Questions
ESX vs. CW vs. CX Edge Gateways
The site needs to connect instruments and DI/AI signals at the same time — should you buy the ESX, the CW or the CX