Keithley Test Equipment Buying Checklist: DMM6500, True RMS, and Insulation Tester vs Megger

I review test equipment requests for a living. Not just approve-and-sign, either: I read the specs, check the uncertainty math, and send things back when they don't make sense. Over 4 years, that's been roughly 200 unique instruments a year, and I've rejected about 12% of first submissions.

When I first started doing this, I assumed a 6.5-digit meter was automatically better than a 5.5-digit meter. Simple, right? Then a detector readout failed verification, and the 'high precision' meter we'd bought was the reason. Since then, I've used the same checklist. It applies whether you're buying a Keithley DMM6500 bench digital multimeter, a true RMS multimeter, or trying to settle the insulation tester vs megger question.

The Checklist

Seven steps. Each one can be done in the time it takes to drink a coffee.

1. Define the signal before you define the box

Before you look at any Keithley test equipment, write down what you're actually measuring. DC voltage? AC current? A chromatographic detector output that changes slowly but needs to be logged for hours? Source impedance? Expected amplitude?

For example, chromatography detectors often output a low-level DC voltage from a high-impedance source. If you connect a 10 MΩ meter to that circuit, you might see an error larger than the spec you're trying to verify. In that case, input impedance is a bigger deal than the number of digits.

My rule: if the source impedance is above 1 kΩ and the signal is low level, check input impedance first.

2. Compare accuracy specs, not digit counts

Six and a half digits describes resolution, not accuracy. A meter can display 6.5 digits and still be the wrong tool for the job.

The number I look for is the 1-year basic DC voltage accuracy, usually stated as '% of reading + % of range.' On the Keithley DMM6500's published datasheet, basic DCV accuracy is 0.0020%. That's a number you can make decisions on. If a vendor only quotes '6.5 digit' and gets quiet when you ask about 1-year accuracy, that's a red flag.

Also check the temperature range. A spec quoted at 23°C ± 5°C is not the same as one quoted at 18–28°C. For quality work, the tighter the band, the better.

3. Decide if true RMS matters

A true RMS multimeter measures the heating value of a waveform, not just the average rectified value calibrated to a sine wave. If you're checking ripple on a DC supply, verifying a motor drive output, or looking at any distorted waveform, an average-responding meter can read 10–15% low.

So yes, buy a true RMS multimeter if you'll be anywhere near non-sine signals. But 'true RMS' alone isn't a complete spec. Look at the bandwidth and crest factor. A meter with true RMS up to 1 kHz is not the same as one rated to 100 kHz. For bench work, I'd rather have a slightly less 'accurate' meter with a higher crest factor than the reverse. The Keithley DMM6500 bench digital multimeter includes true RMS ACV, which makes it a solid bench unit for lab work.

4. Don't use 'megger' when you mean 'insulation tester'

This is the insulation tester vs megger trap. 'Megger' is a brand name that became generic. A megger is a megohmmeter, but an insulation tester is a broader category. If your purchase order just says 'megger,' you might receive a hand-cranked 500 V megohmmeter when you actually need a battery-powered insulation tester with 250 V, 500 V, and 1000 V test steps.

I did this in my first year. It cost us a week of waiting and a redo.

When you write the spec, state the test voltages, resistance range, pass/fail thresholds, and the safety rating. Then, and only then, call it whatever you want.

5. Check loading effect before you trust a reading

Here's the step most people skip. A bench meter with 10 MΩ input impedance can load a high-impedance circuit and give you a beautiful, precise, wrong number. The DMM6500 uses high input impedance (>10 GΩ) on its lower DCV ranges precisely because low-level measurement is where loading errors hide.

I didn't fully understand loading error until a detector readout failed verification in 2023. We replaced the meter, not the sensor. That's when the checklist started.

If you're measuring a sensor or a detector output that has any significant source impedance, calculate the loading error: if source impedance is 1 MΩ and meter impedance is 10 MΩ, you lose about 9% of the signal. That's not a meter failure. It's a spec mismatch.

6. Verify data logging and connectivity before you need them

Manual readings are a quality incident waiting to happen. If you're doing a chromatography detector stability test over eight hours, you want a timestamped log, not 800 notes on a scratchpad. The DMM6500 can log to a USB drive and has a touchscreen that makes the data easy to spot-check. But the feature only helps if it's on the approved list before you buy.

Ask three questions: Can it log at the rate you need? Can it store enough samples for your full test window? Can you get the data off without proprietary software?

7. Put calibration on the critical path

If you're using a meter to accept or reject product, calibration is not a detail. Check the calibration certificate date on every new unit. If the certificate has less than 12 months remaining, that's a negotiation point, not an assumption to ignore. The certificate should be traceable to a national standard (NIST in the U.S.) and include the actual as-found/as-left data.

Here's where the time-certainty principle kicks in. In March 2024, I paid $400 extra to rush a calibration through the lab. Why? Because the alternative was missing a customer audit window and potentially delaying a $15,000 qualification run. The rush fee wasn't buying speed; it was buying certainty. If you're under a deadline, 'probably calibrated' is not a plan.

Also do the uncertainty math: if your product spec tolerance is ±1%, and your meter accuracy is ±0.5%, you've used half your tolerance before you start. That's too tight for most QA decisions. Aim for a meter that gives you at least a 4:1 test uncertainty ratio.

Mistakes I still see on requisitions

Even after all that, some patterns keep coming back:

• Buying on brand alone. Keithley test equipment is good, but the right model depends on the measurement. A DMM6500 is not a replacement for a dedicated instrument in every case.
• Skipping warm-up. A cold meter can drift for the first few minutes. For low-level DC work, let it warm up for at least 30 minutes before a critical reading.
• Assuming true RMS is always enough. Crest factor matters on pulse-like signals. Check it.
• Relying on one calibration point. A certificate doesn't mean every range is perfect.

Bottom line: the best Keithley test equipment purchase is the one that matches the signal, the environment, and the calibration plan. Get those right, and the rest is paperwork.