Why I Stopped Treating Sick Thermometers and DMMs as Commodity Purchases

I used to think a 'sick thermometer' was just a thermometer. I was embarrassingly wrong.

When I first started reviewing inbound instrumentation for our assembly lines, I treated every device with a 'temperature' or 'voltage' label as essentially the same. A DMM is a DMM, right? A 'sick thermometer' (and yes, I know the brand is SICK, but the keyword search volume on that typo is too big to ignore) measures fever temperature or process temperature. What's to mess up?

I learned the hard way that assuming two things that measure the same variable are interchangeable is the most expensive shortcut in industrial procurement. Put another way: you can buy a Fluke 15B+ digital display multimeter for $80 or a professional-grade industrial DMM for $300. Both measure voltage. They are not the same thing.

The 'Sick Thermometer' Confusion That Cost Us a Shift

Let me give you an example from our Q1 2024 quality audit. A requisition came through for a 'sick thermometer fever temperature' sensor. The team assumed they needed a medical-grade temperature probe. What they actually needed was a SICK thermoMETER—a non-contact industrial infrared sensor for monitoring conveyor belt component temps. The difference? About $600 per unit and a completely different interface protocol.

We ordered six of the wrong devices. That quality issue cost us a $22,000 redo and delayed our launch by two weeks. (I don't have the exact breakdown in front of me, but trust me—it stung.)

Now I force every spec to include the exact model number and protocol. 'Temperature sensor' isn't a specification. ‘SICK PROFINET encoder’ is a specification. '15B+ digital display multimeter' is a specification. 'Accurate within 0.01°C over -20 to 100°C range' is a specification. Everything else is just a wish.

Why I Now Pre-Test Every 'Commodity' Instrument

The surprise wasn't that cheaper options existed. It was how much hidden technical risk came with the 'good enough' option. Take the Fluke 15B+ digital display multimeter. It's a fine meter for general electronics work. But if you're measuring 480V three-phase in a factory? The CAT rating matters. The input protection matters. The response time on the display matters.

Here's what I do now before any order for a 'sick' anything or a random DMM goes through:

1. Force a protocol check. Is it PROFINET? IO-Link? Analog 4-20mA? (I learned this one after ordering a SICK encoder that spoke PROFINET for a system that only spoke SSI. Ugh.)
2. Verify the environment rating. IP65 vs IP67 vs IP69K—they are not the same. An IP65 device in a washdown zone will die in three months. I've seen it on a bottling line.
3. Test one unit before ordering a batch. This seems obvious. It is not common practice. We received a batch of 50 pressure gauges (think a 1008 pressure gauge or similar) where the zero offset was visibly off. The vendor claimed it was 'within industry standard.' We rejected the batch, and they redid it at their cost. Now every contract includes a zero-point verification requirement.

The 5-Minute Rule: An Anecdotal Cost-Benefit

I wish I had tracked this more carefully over the years, but here's my rough sense: spending five minutes verifying a spec before ordering saves, on average, about four hours of troubleshooting or return processing later. That's a 48x return on time investment (if you want to be math-y about it).

I don't have hard data on industry-wide defect rates for first deliveries of industrial sensors, but based on our five years of orders, my sense is that about 10-15% of first shipments have a spec mismatch. That might sound high. It might just be our supply chain. But I'd argue most procurement teams are suffering from the same optimism bias.

You might be thinking: 'This sounds like a lot of overhead for simple stuff like a thermometer or a multimeter.'

To which I'd say: the 12-point checklist I created after my third mistake has saved us an estimated $8,000 in potential rework in 2024 alone. Five minutes of verification beats five days of correction. Every time.

So yes—I now treat a 'sick thermometer fever temperature' query as a serious spec challenge, not a simple purchase. And if you're buying a 15B+ digital display multimeter for anything beyond hobby work, please check the CAT rating first. Your line won't go down over a $12 thermocouple. It can absolutely go down over the wrong SICK PROFINET encoder (even though I might be misremembering the exact protocol specs on that one).