Keithley Multimeters, Function Generators & Beyond: 7 FAQs from a Quality Manager’s Bench

You’ve got Keithley gear — and a bunch of questions. Here’s what I’ve learned from rejecting $22k batches and rushing calibrations.

I’m a quality compliance manager in a test equipment company. I review every deliverable — from calibration certificates to final assembly — before it reaches customers. Roughly 200+ unique items a year. I’ve rejected about 12% of first deliveries in 2024 due to out-of-spec components or unclear documentation. Over 4 years, I’ve learned that precision tools demand precise thinking. This FAQ covers the questions my team and I hear most often. No fluff — just what you actually need to know.

1. Is the Keithley 2002 multimeter worth the investment for everyday bench work?

Short answer: only if you need 7½-digit resolution and extremely low noise. The 2002 is a reference-grade meter — I use it alongside a DMM6500 for critical comparisons. But if your typical measurement is 4½ digits, a Keithley 2110 (5½ digits) will do the job at a fraction of the price.

Here’s the thing nobody tells you: the 2002’s 1 µΩ resolution is useless unless your test leads and thermal EMF management are equally pristine. I once saw a lab spend $12,000 on a 2002 and then use clip leads. The noise floor was — well, embarrassing. We ended up redoing their setup with low-thermal cables and a $600 scanner card. The meter itself is a beast, but only if the rest of your chain matches.

3. Can I use a metal detector near my Keithley measurement setup?

Yes — but watch the EMI. Metal detectors (especially pulse-induction types) generate strong magnetic fields that can couple into your measurement loop. In our QC lab, we had a walk-through metal detector 3 meters from a Keithley 2700 datalogger. The logged readings jumped by 12 µV during scans — enough to ruin a 6½-digit measurement.

The fix? Move the detector to a separate room (or at least 5 meters away) and use twisted-pair shields. If you absolutely must operate them together, schedule measurements during detector idle time. Or, invest in a handheld metal detector — those emit way less interference. That’s what we did after the first incident, and our 6.5-digit repeatability came back to spec.

4. How do I integrate a function generator with a Keithley DMM for automated testing?

Pretty straightforward: use the IEEE-488 (GPIB) or USB/LXI interface. Most Keithley DMMs come with dedicated trigger I/O lines. You can set the function generator to output a stimulus waveform, then trigger the DMM to measure at a precise time after the stimulus starts. A no-brainer setup: Keithley 3390 function generator -> DMM6500 -> Python script.

One pitfall I see: failing to synchronize settling time. A fast function generator (e.g., 20 MHz square wave) can jump the output before the DMM’s internal ADC settles. We had a $3,500 project where the readings were consistently 0.2% low because the DMM was triggered 50 µs too early. The fix was adding a 100 µs delay after the trigger edge. Oh, and use low-capacitance cables — that improvement alone cut our noise by 40%.

5. What’s the correct way to use a FLIR thermal camera for circuit diagnostics with Keithley equipment?

I’ve got a love-hate relationship with thermal cameras in test labs. They’re super useful for spotting hot spots on a PCB under power — but only if you interpret emissivity correctly.

Here’s what I learned the hard way: shiny metal components (like a Keithley SMU’s heatsink) have low emissivity (~0.1). The camera will show a temperature way lower than reality. Actually, way off — by 20°C or more. We once flagged a capacitor as “cool” while it was running 40°C above spec because we forgot the tape.

My rule: always apply black electrical tape (emissivity ~0.95) on the target before taking a reading. Also, make sure the camera is in the right distance range — FLIR’s manual says minimum focus distance is about 30 cm. Closer than that and the image blurs completely.

And serial number: I keep a FLIR E8-XT as a quick-check tool before powering up a new Keithley module. If a SMU’s internal temperature rises faster than 0.5°C/min after power-on, that’s a red flag. It’s saved me from shipping three units that had fan failures — a $22,000 recall avoided.

6. When should I pay extra for rush calibration or delivery?

Almost always when a deadline is tight and the cost of failure is high. In March 2024, we paid $400 extra for expedited calibration of a Keithley 2002. The alternative was missing a $15,000 customer demo. The regular turnaround was 5 days; we needed it in 2. The $400 felt painful, but the demo — we landed the contract.

But here’s the twist: sometimes “rush” just means shorter queue, not better accuracy. I’ve seen calibration labs that rush and skip the 48-hour warm-up stabilization. That’s a deal-breaker for precision work. So when you pay for rush, ask the lab if they still follow full stabilization procedures. If they say “we can do it in 24 hours flat,” that’s a red flag. Our standard protocol is 4 hours warm-up at 23°C ±1°C before any calibration. Anything shorter, and the numbers may drift by 0.005% — small but meaningful at 7½ digits.

7. What’s the biggest mistake engineers make when specifying Keithley multimeters?

Assuming higher resolution always means better measurements. I’ve reviewed projects where a 7½-digit meter was purchased for a 1% tolerance test. The extra digits added $2,000 to the cost and actually decreased throughput because the measurement took longer. The conventional wisdom says “more precision = better.” But for production go/no-go tests, a 5½-digit meter with a 1 ms reading rate is often the better choice.

I should add: low-level current measurements (nA, pA) require triaxial cables and guarding, which the DMM6500 can provide — but many engineers forget to order the guard kit. We had a $3,500 redo because someone specified a Keithley 6485 picoammeter but used standard BNC cables. The leakage current through the cable insulation was 2 nA — totally swamping the signal. The fix cost $600 for triax cables, but the project was delayed by two weeks. Should mention: always budget for accessories.

Bottom line: match the instrument to the real measurement need, not the spec sheet’s maximums. And when you’re under a deadline, pay for certainty — not just speed.