The Biggest Concern: How Much Will Hardware Retrofit Cost?
After visiting hundreds of factories, "Do we need to replace all sensors?" is one of the most common questions from equipment managers. The concerns are very practical:
The good news is: **most old factories do not need to replace sensors**. The core strategy is "reuse what you have, fill the gaps."
Which Sensors Can Be Reused Directly?
1. Temperature Sensors (Highest Reuse Rate)
PT100, K-type thermocouples, NTC thermistors — these sensors do not really "expire." As long as wiring is intact and resistance is normal, accuracy usually does not drift much.
**Only two things need confirmation:**
As long as the signal type is clear, the DTU acquisition terminal can handle it. Different signals just need different acquisition modules — no need to replace the sensor itself.
2. Current Transformers (Nearly 100% Reusable)
Split-core CTs clamp onto cables without power shutdown. Old factory electrical panels usually already have CTs — they just are not connected to any acquisition system.
**Note:** If existing CTs are used for power metering, their accuracy class is usually sufficient for monitoring. But if they are old electromagnetic types, verify the ratio and accuracy class.
3. Vibration Sensors (Depends on Installation)
If magnetic or bolt-mounted vibration sensors are already installed, they can likely be reused. Confirm:
**A common misconception:** Some factories mistake "vibration switches" for "vibration sensors." Vibration switches only output an on/off alarm — no continuous waveform data. These do need to be replaced with continuous-output vibration sensors.
4. Pressure / Flow Sensors (Depends on Signal Type)
Air compressor discharge pressure, pump inlet/outlet pressure — if existing sensors output 4-20mA or 0-10V, they can be reused by connecting to DTU analog channels.
When Must You Replace or Add?
| Scenario | Solution | Estimated Cost |
|---|---|---|
| Only local indicators, no signal output | Add sensors + DTU | ¥800-2,500/point |
| Vibration switch (only on/off alarm) | Replace with continuous vibration sensor | ¥1,500-4,000/point |
| Sensor damaged or severely drifted | Replace with same model | ¥500-2,000/point |
| Need additional monitoring points | Add sensors + wiring | ¥1,000-3,000/point |
| Existing PLC has data but no comm port | Add DTU to read from PLC | ¥1,500-3,000/unit |
A Real Case: A Hardware Processing Plant
Site conditions:
Solution:
**Result: 65% sensor reuse rate. New sensor cost about ¥12,000, DTU and installation about ¥5,000, first-year platform fee about ¥8,000. Total investment about ¥25,000 — roughly the cost of one unplanned downtime event.**
Three-Step Quick Assessment
**Step 1: Inventory existing sensors**
List sensor type, brand, output signal, and installation location for each machine. If you do not know the signal type, take a photo and ask the manufacturer.
**Step 2: Confirm DTU compatibility**
Send the sensor list to the platform vendor for a free compatibility assessment — which can connect directly, which need adapter modules, which must be replaced.
**Step 3: Do the "reuse vs. replace" math**
Do not just look at sensor unit price — look at total investment. Reusing 10 sensors may save enough to cover a year of platform fees.
Common Misconceptions
**Myth 1: "Sensors have been used for 5 years, they must be inaccurate"**
Industrial sensors typically last 10+ years. With periodic calibration (every 1-2 years), accuracy is perfectly adequate. Predictive maintenance looks at trend changes, not absolute precision.
**Myth 2: "Different brand sensors cannot connect to the platform"**
Platforms connect to signals (4-20mA, 0-10V, RS-485), not brands. As long as signal standards are consistent, brand differences do not affect integration.
**Myth 3: "Go all-in or do not bother"**
Predictive maintenance can be rolled out in phases. Start with 1-3 machines most prone to failure, validate results, then expand. Do not pursue one-step perfection.
**For old factories, the biggest cost of predictive maintenance is not sensors — it is overthinking the problem. Inventory first, assess second, deploy in phases. 95% of sensors can be reused.**