"Day 18/30: Why is a 1979 protocol still the default in 2026?"

a factory icon connected via a cable labeled RS485 + Modbus

I'll be honest about where I started today: after three days of learning RS485 and Modbus RTU, my first reaction was mild surprise that this is still the default. Modbus was originally published by Modicon in 1979 — that's older than my parents, let alone me. RS485 as an EIA standard dates to around the same era. There's Ethernet-based industrial protocols now, there's wireless everything, and yet Wei tells me most of Siliqs' customer sites are still running RS485 buses with Modbus RTU as the actual protocol talking to sensors and meters. I wanted to understand why that's not just industrial inertia.

It's simple enough to implement on almost nothing. Modbus RTU's entire frame format is a slave address, a function code, some data, and a CRC — I covered this on day 16 and it genuinely fits in a page of notes. That means it doesn't need much silicon or much firmware to implement. A cheap microcontroller with a UART and an RS485 transceiver chip can speak it. Compare that to something that needs a TCP/IP stack, a network interface, and enough compute for the associated overhead — that's a real cost difference at the scale of "every temperature sensor in a factory."

It's extremely well understood, which is its own kind of value. Because it's been around since 1979 and is genuinely simple, there are decades of engineers, integrators, and technicians who already know it cold. When something on a Modbus bus breaks, there's a huge base of people who can diagnose it with a cheap USB-RS485 adapter and free software, without needing vendor-specific tools. That matters a lot on a factory floor where the person troubleshooting at 2am might not be the person who installed it.

RS485's electrical robustness fits the environment it's actually used in. I wrote about this on day 15 — differential signaling that rejects common-mode noise and tolerates long cable runs — and industrial sites are exactly the noisy, sprawling environment that plays to those strengths. Motors, VFDs, switching supplies, long conduit runs between buildings: this is the environment RS485 was built to survive.

And maybe the real answer: universal adoption becomes self-reinforcing. Practically every PLC, sensor, power meter, and VFD sold today still has an RS485/Modbus RTU port, because everyone else's does. A vendor dropping it loses compatibility with an enormous installed base for very little gain, so nobody drops it, so it stays universal. It's the industrial-automation version of a network effect — being the lowest common denominator is a moat, not a weakness, once enough of the world has standardized on you.

None of this means it's the best protocol on paper — I get that newer alternatives exist with more bandwidth, better security, native IP support. But "best on paper" and "what's actually bolted to ten thousand existing installations" are different questions, and today I finally get why the answer to "why hasn't this been replaced" is basically: replacing it is expensive, Modbus RTU already works, and nobody's forced to. That reframes how I think about the SQC485Iv2's job too — it's not competing with RS485/Modbus, it's built to sit alongside it and extend it wirelessly, which only makes sense if the wired side isn't going anywhere.

工廠圖示透過標示 RS485 + Modbus 的纜線連接

老實說今天一開始的心態:學了三天 RS485 和 Modbus RTU 之後,我第一反應是有點驚訝這居然還是預設選項。Modbus 最早是 Modicon 在 1979 年發表的——比我爸媽還老,更不用說比我老。RS485 作為 EIA 標準大概也是同個年代的東西。現在明明有以 Ethernet 為基礎的工業協定,什麼都能無線化,但 Wei 說 Siliqs 大部分客戶現場實際上跟感測器、電表講話的協定,還是跑在 RS485 bus 上的 Modbus RTU。我想搞懂這是不是真的只是工業界的慣性而已。

它簡單到幾乎什麼都能實作。 Modbus RTU 整個訊框格式就是 slave address、function code、一些資料、加上 CRC——這是我 day 16 寫過的東西,真的一頁筆記就寫得完。這代表實作它不需要多少硬體或韌體資源。一顆便宜的 microcontroller,配一個 UART 和一顆 RS485 收發器晶片,就能講這個協定。相較之下,如果要跑一整套 TCP/IP stack、網路介面,還要有足夠算力處理相關 overhead——在「工廠裡每一顆溫度感測器」這種規模下,成本差距是實實在在的。

它被理解得極其透徹,而這本身就是一種價值。 因為它從 1979 年就存在,而且真的很簡單,幾十年下來累積了大量已經對它瞭若指掌的工程師、整合商、技術人員。當 Modbus bus 上出問題時,有一大群人可以用一個便宜的 USB-RS485 轉接器加免費軟體就診斷出問題,不需要廠商專用工具。這在工廠現場很重要——半夜兩點在排除故障的人,不一定是當初裝這套系統的人。

RS485 的電氣強韌性剛好適合它實際被使用的環境。 這是我 day 15 寫過的——差動訊號能抑制共模雜訊,也能承受長距離的纜線——而工業現場正是那種吵雜、範圍又大的環境,剛好吃到這些優點。馬達、變頻器(VFD)、切換式電源、建築物之間長長的線槽:這正是 RS485 當初被設計來承受的環境。

也許真正的答案是:普及本身會自我強化。 現在市面上幾乎每一台 PLC、感測器、電表、VFD 都還有 RS485/Modbus RTU 接口,因為別人都有。如果某個廠商把它拿掉,就會跟龐大的既有安裝基礎失去相容性,得到的好處卻很少,所以沒人願意拿掉,於是它就一直保持通用。這其實是工業自動化版本的網路效應——一旦全世界夠多人都用同一套標準,「成為最大公約數」就是一道護城河,而不是弱點。

這不代表它在紙面上就是最好的協定——我知道新的替代方案有更大頻寬、更好的安全性、原生 IP 支援。但「紙面上最好」跟「已經接在一萬套既有裝置上的東西」是兩個不同的問題,今天我終於理解「為什麼這東西還沒被取代」的答案基本上就是:取代它很貴,Modbus RTU 已經能用了,而且沒有人被逼著非換不可。這也讓我重新想了一下 SQC485Iv2 的定位——它不是在跟 RS485/Modbus 競爭,而是被設計成跟它並存、用無線把它延伸出去,而這件事之所以成立,前提就是有線那一側哪裡都不會消失。