Why Your Keithley 2700 Multimeter Drifts (And Why It's a Business Risk)

Thursday morning. You connect a thermocouple to your Keithley 2700 multimeter and read 2.31 µV. You short the input, zero it, reconnect, and now it reads 1.08 µV. Same wires. Same settings. The meter doesn't look broken. But you trust it a little less.

That doubt is the real problem.

In my role coordinating rush instrument support for test labs and manufacturers, I've spent seven years helping people who called at the last minute because a measurement suddenly looked wrong. More often than not, the meter they were using wasn't obviously dead. It was just quietly drifting. And by the time they called, the drift had already become a schedule problem, a quality problem, or both.

The surface problem: a Keithley 2700 multimeter that doesn't look broken

A meter that still powers up, displays digits, and passes a front-panel self-test can still be giving you bad data. The Keithley 2700 multimeter is an excellent instrument—but excellent is a state of the world at a specific time, not a permanent condition.

Here's what I see most often:

• Calibration drift. Every voltage reference ages. Keithley specifies accuracy for 90-day and 1-year intervals after calibration, not forever. According to the Keithley user manuals now hosted on tek.com (accessed April 2025), those intervals are part of the official accuracy definition, not just a recommendation.
• Thermal EMFs. This one surprises people. Different metals in connectors, relays, and even the meter's input terminals generate small voltages when there's a temperature gradient. A 1°C difference at a junction can cause microvolt-level offsets. That's why a microvolt measurement can change if someone opens the lab door.
• Input loading. A DMM's input impedance can change the circuit you're trying to measure. If your source impedance is high, the meter becomes part of the circuit.
• Worn scan-card relays. In a 2700 system, the problem is often in the multiplexer card, not the main DMM. A relay with high contact resistance gives you stable-looking readings that are still wrong.

All of these have one thing in common: they don't trigger an error message.

The problem behind the problem: nobody knows what the meter knows

When a Keithley 2010 digital multimeter drifts slightly, you usually can't see it without a known reference. That's the scary part. A meter without a current calibration record is not a measuring instrument. It's a guess with a backlit display.

I remember one case where a client was sure their 2700 multimeter was fine because the readings seemed consistent. The surprise wasn't that a batch had failed. The surprise was how long the drift had been hidden. Their internal data showed acceptable readings for two months after the calibration interval had lapsed. The meter was wrong by just enough to matter, but not enough to look crazy.

If you ship a product based on an unverified measurement, you're not shipping a measurement. You're shipping an opinion.

What a drift problem actually costs

In March 2024, a client called at 4:47 p.m. on a Wednesday. They needed a Keithley 2010 digital multimeter calibrated, certified, and delivered to a facility 600 miles away by Friday morning. Normal turnaround was ten days.

The upside was simple: meet the deadline, keep the audit on schedule. The risk was also simple: spend $420 in rush fees and still fail if the auditor rejected the paperwork. I kept asking myself: is $420 worth potentially losing a six-figure contract? The answer was yes—because the alternative was a lot full of components whose readings no one could defend.

That's when the deeper truth hit me. The calibration fee was never the real cost. The real cost was the week of uncertainty, the unhappy calls, and the sleepless night before the auditor arrived. In my experience, a meter you can't defend is more expensive than any calibration service.

The Keithley 2010 digital multimeter isn't exempt

I like the Keithley 2010 digital multimeter for low-level measurements. It's one of the better tools on the bench. But better only helps if the accuracy spec is valid. Once the calibration interval lapses, the spec is no longer guaranteed. To some extent, you can still use the meter for relative comparisons. But if you're publishing data, certifying a product, or setting a tolerance, you need traceable proof.

And don't forget the older units. The 187 multimeter that's been on your shelf for years can still be reliable—if you know its history. That 16 multimeter in the drawer might also be fine. But it still turns on is not a calibration record. If you ask me, a working meter with an unknown calibration state is a red flag, not a spare instrument.

How to calibrate an Extech pH meter (the same logic in a different package)

This isn't just a DMM issue. A pH meter is also a measurement instrument, and its electrode drifts with age, temperature, and chemical exposure. An Extech pH meter will happily show 7.01 in a buffer that's actually 7.00—until you check it. So if you're running a process that depends on pH, calibrate it before you need it.

The short version of how to calibrate an Extech pH meter (verify your model's menu at extech.com):

1. Rinse the electrode with distilled or deionized water. Blot it gently with a lint-free tissue; don't wipe the glass bulb.
2. Immerse it in fresh pH 7.00 buffer. Stir once, wait for a stable reading, then press CAL. Many Extech meters recognize the buffer automatically.
3. Rinse again.
4. Immerse the electrode in pH 4.00 buffer (or pH 10.01 if you're measuring high-pH samples). Wait for stability and confirm the second calibration point.
5. Rinse the electrode one more time and store it in storage solution, not in plain water.

According to Extech's pH meter manuals (extech.com, accessed April 2025), a two-point buffer calibration is the standard routine. If your model has automatic temperature compensation (ATC), it will adjust for buffer temperature, but you should still check the temperature range on the buffer bottle.

Make calibration a production step, not a chore

So what's the solution? It's not more expensive gear. It's a simple rule: before any critical measurement, verify the instrument. Keep a known reference on the bench. Check the calibration sticker. If it's expired, don't justify it. Swap in a calibrated Keithley 2700 multimeter or a Keithley 2010 digital multimeter from your rotation and send the other one out.

Have a spare. The extra meter is annoying until the day a deadline depends on it. When I set up our first spares program, I thought we were being paranoid. Last quarter alone, we processed 47 rush calibration jobs with a 95% on-time delivery rate. The people who planned ahead were the ones who stayed calm.

Bottom line: if you use a Keithley multimeter to qualify a product, or a pH meter to release a batch, you are making a quality promise. A measurement you can't defend is a product defect you can't see. In my opinion, the $200 for calibration and the ten minutes for buffer checks is a no-brainer. The brand that has to explain unverified data in an audit? That's the expensive part.