Buying Machine Vision or Lab Equipment? This 7-Point Checklist Keeps Me From Wasting Money

I'm a controls engineer who has ordered machine vision and lab equipment for about six years. I've personally made—and documented—enough procurement mistakes to fill a small binder. Total waste: roughly $4,500, give or take. Now I keep a checklist that starts with questions, not product names.

If you're comparing Cognex products, looking for a Cognex sensor, shopping for used oscilloscopes, wondering where to buy FLIR thermal cameras, or just stocking pipette tips for the lab, this list is for you. It's the practical version of 'measure twice, spend once.'

Here's what I check before I hand over a purchase order. Seven steps, plus the mistakes I've made along the way.

1. Define what you're actually trying to inspect or measure

It's tempting to start with a product category: 'We need a Cognex sensor.' But a sensor is an answer to a question. If you haven't written down the question, you're guessing.

For a machine vision system, the basics are field of view, working distance, object speed, tolerance, lighting, and surface finish. When I first started, I thought resolution was the whole game. Let me rephrase that: I spent all my time comparing pixel counts and almost no time on depth of field. The result was a camera that could see the part but couldn't focus on the surface because my working distance needed to be half a meter shorter than I planned.

So before you search for 'cognex sensor reviews' or 'cognex sensor price,' write down the part geometry and the defect you need to catch. That one paragraph will save you weeks.

2. Let the application drive the sensor choice

Different jobs call for different product lines. If you need to read a Data Matrix code on a moving carton, a fixed-mount barcode reader might be enough. If you need to detect scratches on a reflective surface, you probably need a vision system with multiple lighting directions. These are different Cognex products with different cost structures.

I've made the budget vendor mistake too—not with Cognex, but with a cheap camera module. The module looked smart until the image quality fell apart when line speed changed. Net loss: $1,600 in rework. That's the classic penny-wise, pound-foolish pattern. To be fair, budget tools have their place. But only after you've tested them under your real conditions.

3. Treat the sensor as part of a system, not a standalone part

Among Cognex products, the sensor is often the easiest part of the system to choose. It's the lens, lighting, mounting, cabling, and software that make or break the application. If you order a camera with a C-mount and the installed lens is F-mount, you're not one adapter away from success. You're one adapter away from a vibration problem.

In 2017, my first major mistake was ordering a vision sensor that looked perfect on the datasheet but didn't match the lens on our line. That error cost $890 in redo plus a one-week delay. I now list every component in the optical path before I place the PO. That includes filters, polarizers, lighting controllers, and Ethernet cable length. It sounds like overkill, but the one thing you don't want to discover during installation is that the parts don't behave like a system.

4. Verify the software and license story

This is where people get tripped up. The hardware price might look competitive, but software can change the total by a lot. Cognex In-Sight and VisionPro have different tools and licensing tiers. If your application needs deep learning for defect detection, make sure that feature is active in the trial—not just in the brochure. I once ran a free evaluation that worked fine until we tried to export results. (Ugh.)

The same logic applies to thermal cameras. Search 'where to buy flir thermal cameras' and you'll find different bundles with different software. Ask whether the software can export radiometric data or only images. That distinction matters more to your analysis than the camera body alone.

5. If you buy a used oscilloscope, ask for calibration records

Buying a used oscilloscope can be a smart move. The price is often half of a new unit. But used gear has a hidden failure mode: the screen works, the knobs feel fine, and the measurement is wrong.

In September 2022, I bought a scope that powered up cleanly. It looked like a great deal—about $300 less than the next option. Then I noticed a timebase error on a specific range during a critical test. I want to say the error was around 8%, but don't quote me on that exact figure. Either way, it was enough to ruin data. The repair cost more than the savings, and I lost two weeks of lab time.

Ask for calibration records and the original date of manufacture. If the seller says 'it was calibrated recently,' ask what was measured and what the readings were. A used oscilloscope without paperwork is a potential source of mysterious measurement errors.

6. Buy thermal cameras from a channel you can trace

If you search 'where to buy flir thermal cameras,' you'll see official distributors and marketplace listings mixed together. I'm not saying you have to buy direct from the factory. But you should know exactly who you're buying from, what support comes with it, and how the performance claim is backed up. Per FTC guidelines (ftc.gov), marketing claims have to be truthful and substantiated. If a vendor says a camera detects temperature differences of ±2°C, ask for the test evidence behind that number.

I'd rather spend ten minutes asking about the sales channel than deal with a mismatched camera later. An informed customer asks better questions and makes faster decisions.

7. Read the fine print on consumables like pipette tips

This step feels small until it's not. I once ordered 50 boxes of pipette tips without checking the pipettor brand. The tips fit a generic pipettor in the storage closet, but the lab used multi-channel pipettors, and the tips were oriented wrong. The lab manager was not amused. (Surprise, surprise.)

If you're buying pipette tips, check these four things:

• Volume range: Does it match the pipettor's stated minimum and maximum?
• Filtered vs. non-filtered: Filtered tips prevent aerosol but cost more.
• Sterile vs. non-sterile: For cell culture, the wrong choice can ruin a batch.
• Tip length and rack format: Multi-channel pipettors are picky about spacing.

That last one cost us a $3,200 assay. Let me correct myself—$3,200 was the reagent value; the wasted labor was extra. But the number stays in my head because it still hurts.

If your lab also prints labels or plates, check the print quality standard too. Commercial print is usually expected at 300 DPI at final size, and colors are often defined by Pantone numbers. A camera can't reliably read a low-contrast label that looked acceptable on a screen but failed on the press.

Final notes and the mistake that keeps showing up

The biggest mistake I keep seeing in my own projects is the assumption that one specification fits every context. It's tempting to say, 'This sensor will work for all packaging materials.' But vision inspection is almost never that simple. A matte box, a glossy pouch, and a textured shrink sleeve will each behave differently under the same lighting.

Granted, this checklist adds work upfront. But it saves time later. Since I started using it, I've caught at least 47 potential errors before they became purchase orders—though I might be misremembering the exact count. When you add it all up, the real value isn't just money. It's production time and trust with the team.

If a spec sheet makes a claim, ask for the test record. The vendor who can't produce one is giving you a wish, not a specification.

So next time you're about to order a Cognex sensor, check the optical path. When you're considering used oscilloscopes, ask for calibration records. And if you're deciding where to buy FLIR thermal cameras, trace the seller and the software. The product is only as good as the plan around it.