---
Participants:
---
Marcus: Welcome, Elena. It's great to have you here. I know our audience — calibration technicians, lab managers, and quality engineers — will benefit enormously from your experience. You've spent over 18 years directing accredited calibration laboratories and have trained hundreds of specialists. Today, I want to focus on the mistakes you see repeatedly, even in well-funded labs, and how to avoid them. Let's dig in.
Elena: Thank you, Marcus. I appreciate the opportunity. You know, the calibration world has evolved enormously — tighter tolerances, more complex instruments, greater regulatory scrutiny — but I still see the same foundational mistakes surfacing year after year in lab audits. Some of them are costly. Some of them are dangerous. Most of them are entirely preventable. I'm happy to walk through the big ones.
Marcus: Perfect. Let's start with what I consider the backbone of any credible calibration operation.
---
Marcus: Elena, in your experience, where do labs typically fail first when it comes to environmental conditions?
Elena: Temperature. Without question, temperature is the number one environmental variable labs underestimate. I've walked into facilities that calibrate precision resistance standards — instruments with temperature coefficients as sensitive as ±1.5 ppm/°C — and found their lab fluctuating by ±2°C throughout the day. That's a potential drift of 3 ppm, which for a 100-ohm standard translates to an error of 0.3 milliohms. If your target measurement uncertainty is 0.1 milliohms, you've already blown your budget before you've even connected a wire.
Marcus: That's a concrete example. What's the practical fix?
Elena: The fix is layered. First, you need a properly sized HVAC system that accounts for equipment heat load — not just room volume. I worked with a lab in Stuttgart that had a beautifully designed climate control system but didn't account for the 8 kW of heat generated by their aging power supplies. The room would heat up by 3°C every afternoon. Second, you need continuous environmental monitoring, and I mean continuous, not a thermometer on the wall someone reads twice a day. Use data loggers that record at least every five minutes, with calibrated sensors themselves. I recommend placing sensors at the actual workbench height, not at ceiling level where most labs mount them.
Third — and this is where many labs stumble — you need to give instruments time to thermally equilibrate. If an instrument has been stored in a 20°C warehouse and you bring it into a 23°C lab, you cannot begin calibration immediately. The rule of thumb for precision instruments is a minimum of two hours per degree of temperature difference, but I've seen high-accuracy DC voltage references require 24 hours or more to fully stabilize. Rushing this step introduces errors that get attributed to the instrument rather than to poor practice.
Marcus: Humidity is the other one, right?
Elena: Absolutely. Most ISO/IEC 17025 accredited labs target 40–60% relative humidity, but what matters is stability more than absolute value. Rapid humidity swings cause condensation on high-impedance circuits and alter surface leakage resistance on insulators. I investigated a case where a lab was calibrating high-value standard resistors — 10 GΩ range — and kept getting inconsistent results. Turned out their humidity was swinging 15% in an hour because of an open loading dock door nearby. Surface leakage across the insulator mounting was varying by orders of magnitude.
---
Marcus: Let's talk traceability. This is foundational to ISO/IEC 17025, and yet labs still get cited for traceability deficiencies. What are the common mistakes?
Elena: The biggest mistake is treating traceability as a paperwork exercise rather than a metrological one. I've audited labs where every certificate was present, properly formatted, and totally meaningless because the calibration chain was either broken or irrelevant to the measurement being performed.
Here's what I mean. Let's say a lab calibrates pressure gauges using a dead weight tester. Their dead weight tester was calibrated by a primary lab that certified the mass of the weights. Sounds fine, right? But nobody verified the local gravity, the air density correction, or the surface tension effects on the piston-cylinder assembly. The primary lab's certificate assumed standard gravity of 9.80665 m/s², but the user's lab in Denver, Colorado, is at an effective gravity of 9.79613 m/s². That's a 0.1% difference in pressure — enormous for a Class 0.1 gauge. The chain looked complete on paper, but the realized value at point of use was wrong.
Marcus: That's a powerful example. So traceability isn't just about having a certificate — it's about understanding every link.
Elena: Exactly. And another critical point: calibration intervals on your reference standards must be justified. Too many labs simply adopt whatever interval the previous lab used, or they use arbitrary annual cycles without any data to support them. ISO/IEC 17025 doesn't mandate a specific interval — it mandates that the interval be appropriate. I recommend labs track their reference standard performance over time. If a reference voltmeter drifts less than 10% of its specification over two years, there's no metrological reason to calibrate it every 12 months. Conversely, if it's drifting 60% of spec in six months, annual calibration is too infrequent. Use historical data. Apply statistical methods. The ILAC-G24 guidance document is an excellent resource on interval determination.
---
Marcus: Uncertainty budgets are perhaps the most misunderstood requirement in calibration. What do you see labs getting wrong?
Elena: Oh, where do I start? The most pervasive mistake is cherry-picking contributions to make the final uncertainty look impressively small. Labs will carefully evaluate Type A contributions — the statistical repeatability — and completely ignore or underestimate Type B contributions like resolution, reference standard uncertainty, environmental effects, or operator influence.
I reviewed a budget from a lab calibrating torque wrenches. They reported a measurement uncertainty of ±0.3% with a coverage factor of k=2. When I drilled into the budget, they had included repeatability (0.15%) and reference standard uncertainty (0.2%). But they had completely omitted the contribution from the alignment error between the torque transducer axis and the wrench axis — which, for a mechanical wrench with a 300 mm handle, can easily contribute 0.15–0.5% depending on the angle. Their actual expanded uncertainty was probably closer to ±0.7%. They were issuing certificates with falsely tight uncertainty statements.
Marcus: That has real consequences for the end user.
Elena: Massive consequences. The user was making pass/fail decisions based on ±1% tolerance. With a properly evaluated uncertainty of ±0.7%, their guard band should have been significant, meaning a substantial percentage of wrenches near the tolerance limits could have been misclassified. And I'll tell you — this isn't rare. In my experience, roughly 30–40% of first-time assessments I conduct reveal underreported uncertainty budgets.
The practical advice I always give: sit down with your measurement process and honestly ask, "What could influence this result?" Walk through it physically. Is the instrument at exactly the same height as the reference? Could cable resistance matter? Is there vibration from a nearby machine? Is your operator reading the display from an angle that introduces parallax? Then evaluate each contribution quantitatively. Use sensitivity coefficients. Don't just assign arbitrary percentages — derive them from specifications, experiments, or published data.
Marcus: And document the entire reasoning.
Elena: Precisely. An auditor should be able to trace every number in your budget back to a source — a specification sheet, a published paper, an experiment, or a calculation. If your budget says "environmental contribution: 0.05%" with no explanation of how you arrived at that number, it might as well be fiction.
---
Marcus: Let's shift to equipment. What mistakes do you see when labs select their reference standards and calibration equipment?
Elena: The classic rule is the 4:1 test uncertainty ratio — your reference should be at least four times better than the device under test. In practice, I see labs violating this regularly, sometimes because of budget constraints, sometimes because of ignorance.
A specific example: a lab was calibrating digital multimeters with a 6½-digit specification using a calibrator that was itself only characterized to 6½-digit accuracy. The test uncertainty ratio was essentially 1:1, which means you cannot meaningfully distinguish the calibrator's contribution from the instrument under test's performance. Any measurement result is essentially ambiguous. The lab was issuing calibration certificates with values, but those values were dominated by the calibrator's own uncertainty.
Beyond selection, handling is a major issue. I've seen technicians carry precision voltage references across a lab floor that had anti-static carpet — generating triboelectric effects on high-impedance inputs. I've watched people set reference standards on top of running oscilloscopes, introducing electromagnetic interference. I've seen high-value capacitors calibrated on a bench next to an operating ultrasonic cleaner, the vibration from which was mechanically disturbing the capacitor's dielectric geometry.
The fix is threefold. First, select equipment with adequate capability — don't just buy what's cheapest. Second, establish handling procedures specific to each reference standard. Third, train your people. The best equipment in the world is useless in untrained hands.
---
Marcus: Documentation is one of those areas where labs often do the minimum. What's your perspective?
Elena: Documentation serves two masters: the auditor and the future technician. Labs that only write procedures to satisfy ISO/IEC 17025 clauses miss the point entirely. The procedure should be detailed enough that a competent technician, unfamiliar with the specific instrument model, could perform the calibration correctly on the first attempt.
I'll give you an example of what good looks like. One of the best labs I've ever audited — a German aerospace calibration facility — had their procedures written to include not just the method, but the rationale. So instead of saying "set the source to 10.00000 V, wait 30 seconds, record the reading," their procedure said:
"Set the source to 10.00000 V. Allow the source to settle for a minimum of 30 seconds to permit thermal recovery of the output amplifier after the range transition. If the output display shows instability (last digit varying by more than ±2 counts), extend the settling time to 60 seconds and note the delay in the comments field. Record the reading only after the source display has shown stability for 10 consecutive seconds."
That level of detail is what prevents mistakes. It tells the technician why they're waiting, what to look for, and what to do if something goes wrong. It's the difference between following instructions and understanding a process.
Marcus: And on the calibration certificate side?
Elena: The certificate must contain enough information for the user to make informed decisions. That means the measurement results, the associated measurement uncertainty, the traceability statement with specific reference to the standards used, the environmental conditions during calibration, and — critically — the "as found" and "as left" data when adjustments are made. I see labs omitting "as found" data when they perform adjustments, which destroys the historical trending information the client needs.
---
Marcus: Let's talk about the people. You can have the best procedures and equipment, but if the technicians aren't competent, it all falls apart.
Elena: That's absolutely right, and this is an area where many labs are dangerously complacent. I've seen labs that qualify a technician based solely on years of experience. Ten years of doing something wrong is not competency — it's a decade of practiced error.
True competency assessment, as required by ISO/IEC 17025, involves three components: education, training, and demonstrated experience. You need to verify that the technician understands the theory behind the measurement — not just the procedure. Can they explain why we use a four-wire connection for resistance measurement? Can they articulate what happens if they don't? Can they recognize an anomalous result and know when to stop and investigate rather than just recording whatever the instrument displays?
I once watched a junior technician calibrate a precision thermometer using a stirred liquid bath. He recorded the readings perfectly, followed every step of the procedure. But the bath had developed a temperature gradient — about 0.05°C from top to bottom — because the stirrer motor was failing. The thermometer under test had a 2-inch sensing element, while the reference PRT had a 1-inch element. They were immersed at slightly different depths and therefore seeing different temperatures. The technician didn't catch it because he didn't understand the significance of immersion depth and thermal gradients.
Marcus: So how should labs approach competency?
Elena: Formal training programs supplemented by supervised practice, followed by independent evaluation. I recommend a mentoring system where senior technicians oversee new staff for a minimum of six months, even if the new hire has years of experience from another lab. And competency must be reassessed periodically — not just at hire. Technologies change. Methods evolve. A technician qualified on analog instruments in 2005 is not automatically competent on today's digital signal processing-based devices.
---
Marcus: Before we wrap up, can you share the top nonconformities you find during laboratory audits?
Elena: Gladly. Based on roughly 150 audits I've conducted or supervised:
Notice these overlap — many labs have multiple deficiencies. The encouraging thing is that every single one of these is correctable with proper attention and investment in process.
---
Marcus: Elena, this has been incredibly valuable. What's your parting advice for our readers?
Elena: I'd say three things. First, respect the measurement. Every calibration result that leaves your lab may be used to make a decision — a safety decision, a quality decision, a financial decision. Treat that responsibility seriously.
Second, invest in understanding, not just compliance. The labs that consistently produce reliable results are the ones where the technicians understand the physics, not just the procedure. Compliance is the floor. Understanding is the ceiling.
Third, build a culture of continuous improvement. Use your data — your interlaboratory comparisons, your proficiency testing results, your trend analyses — to identify weaknesses and address them proactively. The best labs I've worked with are never satisfied with "good enough." They're always asking, "How can we be more accurate? More reliable? More efficient?"
Calibration is both a science and a discipline. Get the science right, maintain the discipline, and you'll build a lab that genuinely serves its purpose — providing confidence in measurements.
Marcus: Couldn't have said it better. Thank you, Elena, for sharing your expertise and your time. This conversation will benefit our entire engineering community.
Elena: My pleasure, Marcus. Accurate measurements make the world work — quite literally. Keep asking the hard questions.
---
This interview has been edited for clarity and length. For more technical discussions and industry insights, follow BRIDZA's Engineering Series.