A procurement manager doesn't usually get asked how to calibrate an blood pressure machine. But the phrase lands in our support inbox more than you'd expect. One email from a service vendor was direct: three of twelve annual calibrations failed. My first thought was that the calibration procedure needed fixing. That was my first mistake.
Calibration does not fix bad components. It reveals them.
After tracing those failed units, I realized the problem wasn't the calibration protocol. It was the quiet parts on the PCB—relays, capacitors, connectors—that gave every accuracy check a hidden starting point.
What people actually mean by calibration
When someone searches 'how to calibrate an blood pressure machine,' they usually want a checklist. You connect a reference manometer, pump the cuff to a known pressure, hold it, and compare the reading.
That's the surface problem. The checklist matters, but it's not where accuracy is won.
In my experience, calibration failures are symptoms. The underlying cause is something on the board that stopped behaving like the datasheet said it would.
The part of the board nobody budgets for
I used to think a failed calibration meant the sensor was drifting. Not always. Actually, the sensor was usually fine. The failure was in the support electronics around it.
Take relays. A blood pressure machine uses relays to switch its pump and valves. Every inflation cycle stresses the contacts. A relay can pass a bench check and still fail under the load of a real cycle. That's why it matters where you buy Kemet relays from. The datasheet gives a rated operating life only when the part is real.
Tantalum capacitors are another one. Kemet T491 series parts show up all over medical power circuits. They have published ESR, leakage, and capacitance limits. But if someone replaces a T491 with a part that only 'looks the same'—same capacitance, same voltage, undefined ESR—it can pass a bench test and fail under ripple or temperature.
Look at a BOM for a small non-invasive blood pressure board and you'll see a mix: Kemet relays, T491 tantalums, and odd-numbered parts like the C300 and the G310 5G. They look random until you realize each one is a contract with a datasheet. A suffix that looks unimportant is usually a voltage, tolerance, or contact material code. I'm not an engineer, but I've learned to ask exactly which variant a part number is before ordering twenty of them.
Why does this matter? Because a blood pressure monitor is measuring a small pressure signal on top of electrical noise. If a capacitor is out of tolerance or a relay contact has drifted up in resistance, the reading drifts too.
Calibration shows the drift. It doesn't remove it.
What a failed calibration really costs
From my chair, calibration cost is not the calibration fee. It's what happens after a unit fails.
Say the calibration vendor charges $80 per device. A fleet of twelve devices is $960. Fine. Two fail. Now you have troubleshooting time, replacement components, a backup unit, a second calibration round, and a rescheduled clinic schedule. That failed pair can end up costing more than four successful calibrations.
I only took this seriously after ignoring it once. We approved a low bid for generic capacitors to save about $400. The units worked for a month. One failed calibration later, the redo cost $1,200 and a chunk of trust.
The upside was $400 in savings. The risk was a failed re-certification. I kept asking myself whether $400 was worth disrupting a clinic's schedule. It wasn't.
Honestly, I'm not sure why some boards drift more than others. My best guess is it comes down to design margin and lot traceability. But the financial pattern is not hard to read.
A practical order of operations
So what do you do? I don't recommend calibration first. I recommend components first, calibration second.
- Check the cuff and hoses. Leakage will make any calibration look bad. No electronic component replaces a sealed bladder.
- Check the power supply. Noise or ripple at the rails can look exactly like sensor error on the output.
- Inspect relays and connectors. Contact corrosion, cracked solder joints, and damaged harnesses cause intermittent failures. Kemet relays are usually not the first suspect, but the socket might be.
- Verify capacitors with an ESR meter. Capacitance alone is not enough. This is especially true with Kemet T491 tantalum parts.
- Then calibrate. Use a NIST-traceable reference manometer, generate pressure, hold at 250 mmHg, and compare at 50, 150, and 250 mmHg.
Oh, and I should add: don't skip the cuff leak test twice. It's the first thing I look for when a unit fails at high pressure.
Will that pass every unit? No. If the pressure sensor itself is damaged, or the cuff is leaking, no passive component will save the calibration.
Know when a component is not the answer
I'll say this even though it sounds like an odd sales pitch: Kemet relays and T491 capacitors are not a universal fix. If the real issue is a cracked sensor mount or a worn cuff, replacing parts is an expensive detour.
I recommend board-level checks when failures are inconsistent. A device passes one calibration, fails the next, passes again—the variable is often in the support electronics. A device that fails the same way every time? Suspect the sensor.
There's no single best component kit for every blood pressure machine. There's only what the circuit was designed to tolerate.
At least, that's been my experience on the procurement side. And when you're paying for calibration, a quick board audit is a small price to keep failed units out of your budget.
