Request Calibration Where to Buy Region: Global ▾
Application Note

Starrett Micrometer and Dial Indicator Checklist: 9 Steps Before the First Measurement

Posted on 2026-11-20 by Sofia Lindberg

Who this checklist is for

If you're the one who signs off on incoming measuring tools—micrometers, dial indicators, blade mics, anything with a thimble—this is for you. Nine steps. About 30 minutes per batch. It catches the stuff that shows up three weeks later as a rejected production run.

I'm a quality compliance manager at a mid-size machining supplier. I review roughly 200 incoming measuring tools a year. Since I started tracking rejections in 2022, I've sent back about 14% of first deliveries. Not because the tools were bad. Because the specs didn't match, the certificate was off, or someone along the supply chain forgot how to measure.

The checklist below is what I actually use. Not a theoretical framework. A working document.

Step 1: Match the certificate to the tool

First thing: serial number on the calibration certificate against the serial on the tool body. They should match. If they don't, stop. Everything after this step is wasted effort.

Then traceability. The certificate needs an unbroken chain back to a national metrology institute—NIST in the U.S., NPL in the U.K., or the equivalent. A certificate that says "calibrated" without a traceable path is a piece of paper, not a quality record.

One note on claims. If a vendor tells you a tool is "certified accurate to 0.0001 inches," that claim needs support. Per FTC guidelines (ftc.gov), accuracy claims must be substantiated with evidence. Ask for the calibration report. If they can't produce it, believe the report, not the pitch.

Step 2: Look at the contact surfaces (most people skip this)

This is the step people blow past. They zero the tool and assume it's fine. But zero is not the same as clean, square, and undamaged.

Pull out a loupe. Check the anvil and spindle faces for nicks, burrs, or embedded chips. On a micrometer, a burr on the anvil face will skew every measurement you take, and it'll do it consistently enough that you might not notice. You'll just think the part is out.

We ate about $2,400 in rework on this exact problem. We saved a few minutes by skipping the visual check on a batch of imported micrometers. Two weeks later, a gauge dimension was off by three tenths across an entire production run. The tools got sent to calibration after the fact. The faces cleaned up fine. The damage was already done.

Step 3: Zero, then zero again

Zero the tool, lock it, read it. Unlock it, read it again. The reading should be the same.

If the lock changes the reading, the spindle might be loose, or the thimble might be dragging. Either way, the tool isn't ready for the floor. Repeatability matters more than a one-time zero. Anybody can make a tool zero once. Making it zero every time is the job.

Speaking of repeatability—here's where a lot of people get sloppy.

Step 4: How to use a Starrett micrometer (the parts the manual glosses over)

If you're new to Starrett micrometers, or you're handing one to someone who is, this is the practical version. The company's been making these since the 1880s. The mechanics haven't changed much, but the training still gets skipped.

Use the ratchet stop. Not the thimble. You turn the thimble until the spindle is close, then use the ratchet at the end to finish the measurement. The ratchet provides consistent measuring force—about eight to ten clicks is enough for most Starrett models. If you're tightening past that, you're not getting a measurement. You're getting a deformation.

The lock is for holding a reading, not for clamping during measurement. Lock, read, unlock. Don't measure with the lock engaged.

And here's the one that gets missed: check the ratchet before you trust it. Spin it lightly three times against a known zero. If it doesn't slip consistently, the ratchet is sticky. That's a calibration issue, not a user issue. Don't try to fix it on the floor.

One more thing. Store the Starrett micrometer with the measuring faces slightly open—about 0.010" to 0.020". Fully closed storage traps moisture between the faces. Over a year, you get rust spots that no amount of calibration can fix.

Step 5: Starrett blade micrometer—what's different

The starrett blade micrometer uses a thin blade anvil instead of a flat face. That's the whole point—it fits into narrow grooves, keyways, and slots that a standard micrometer can't reach.

Check the blade edges under magnification. Sharp, parallel, no chips. Blades chip easily in shipping if the tool gets dropped. A chipped blade will still zero, but it'll measure the wrong thing.

Set the zero with a cylindrical standard, not a flat gauge block. The blade contacts a cylinder along a line, not a flat plane. Using the wrong standard gives you a false zero every time.

Also match the blade thickness to the feature you're measuring. A 0.030" blade in a 0.030" groove with a 0.001" tolerance? You're going to skew every reading. Blade mics need clearance. Check the drawing before you measure.

Step 6: Dial indicator checks

Dial indicators are mechanically simpler than micrometers, but they fail differently. Check the stem for straightness. A bent stem shifts the entire measurement axis, and the reading goes off in a way that doesn't repeat.

Check backlash. Push the contact point in, let it spring back. The reading should return to the same spot. If it doesn't, the jewel bearings or the rack might be worn. Backlash is the most common dial indicator failure, and it's the one people miss because the indicator still moves.

Mounting matters too. A dial indicator on a magnetic base needs to be square to the surface. If it's tilted, the cosine error quietly adds to your reading. On a 1" travel indicator, a 5-degree tilt adds about 0.004" of error at full extension. That's enough to pass a bad part.

Step 7: Check the chain, not just the tool (signal converter)

If your measurement goes into a data logger, a PLC, or a quality system, the signal path matters as much as the tool. A signal converter that turns a 4-20mA output into a 0-10V input is part of the measurement. If it's not verified, it doesn't matter how good the micrometer is.

Verify the converter against a known input. Compare the observed output to the expected output at two or three points across the range. If the converter drifts, the whole measurement chain drifts with it, and the micrometer gets blamed for it later.

I've got a rule: any tool in the chain gets a check, or the chain doesn't get signed off. No exceptions.

Step 8: Bring everything to thermal equilibrium

This is the one that requires patience. And the one your team will want to skip.

A micrometer that's been in cold transit reads differently than one that's been sitting in your shop for a day. So does the part. If you're measuring a part that's still warm from a machining operation, and the micrometer is cold from the delivery van, the measurement is wrong before you even touch the thimble.

Give it time. Four hours minimum in the same environment for most shop conditions. If you're not sure whether the part is still hot, put a thermal camera on it. Hand-held thermal cameras are usually what industrial teams look for—when someone asks where to buy FLIR thermal cameras, the safe answer is authorized distributors or the FLIR site directly. Gray-market units often come without a valid calibration record. Fine for a spot check, useless for a quality audit.

And no, a hand on the part is not a thermometer. I've seen that argument. It doesn't win.

Step 9: Log it and set the rejection line

Every tool that passes this checklist gets a line in the log: date, serial, certificate ID, operator, pass/fail. If it fails, the log should say why. Not "out of spec." Which spec. Which reading. What the reading was.

Why bother? Because when a tool gets recalibrated six months later, you want to know whether it drifted slowly or was out of spec from day one. That's the difference between a calibration issue and a supplier issue.

There's something satisfying about a clean acceptance log. Checking the last box, closing the file, knowing the next batch doesn't need to be reopened—that's the payoff. Five minutes of verification beats five days of correction. Every time.

Common mistakes worth avoiding

  • Skipping the visual check. Zero without contact surface inspection is a false pass.
  • Measuring with the lock engaged. You'll get a reading. It won't be a measurement.
  • Using the wrong standard for blade mics. Flat blocks against blade anvils give you a false zero.
  • Forgetting the signal path. A verified micrometer into an unverified signal converter isn't a verified measurement.
  • Measuring hot parts. The tool's expansion and the part's expansion don't cancel out. They compound.

None of these are hard to catch. They're just easy to skip. That's the whole reason a checklist exists.

Leave a technical note