A pressure sensor datasheet carries a maximum working pressure, a proof pressure and a burst pressure, and purchase orders frequently quote the wrong one. The three are not points on a scale of quality. They answer three different questions, and only one of them tells you what happens after the event.
Proof pressure asks whether the sensor still meets its specification after being overpressured. Burst pressure asks whether it stays in one piece.
The three levels
| Level | Typical relationship | Acceptance after the test |
| Rated pressure | The calibrated range | Full specification |
| Proof pressure | Often 1.5 to 2 times rated | Specification must still be met, including zero offset |
| Burst pressure | Often 3 to 5 times rated | No leakage and no fragment ejection, function not required |
| Pulse cycling | Hundreds of thousands of cycles between rated and zero | No shift in span or zero |
| Leak test | At rated pressure, held for a minute | Bubble-free or a defined leak rate |
| Zero shift after proof | The number that matters most | A sensor that survives but shifts has failed |
The zero shift is the part that gets ignored. A sensor taken to proof pressure and back may still read correctly at mid-range while its zero has moved by 1 percent of full scale. On a gauge that is a visible error; on a 4 to 20 mA transmitter it is a 0.16 mA offset that quietly biases every reading.
Why a higher burst number is not automatically better
Burst pressure is set by the diaphragm thickness and the body wall. Making both heavier raises the burst number and lowers the sensitivity, which means more signal conditioning and more noise. A sensor with a burst of 5 times rated and a sensitivity of 1 mV/V is not better than one with a burst of 3 times rated and 3 mV/V; it is a different compromise.
The useful ratio is proof to rated, because that is the number that predicts survival in service. The burst number predicts the consequence of a catastrophic event, which is a safety question rather than a performance one.
What to specify for a pulsing application
- Pulse rating, not burst rating. A hydraulic system at 200 bar with a 250 bar spike needs a fatigue number at 250 bar, not a burst number at 800 bar.
- Cycles and waveform. A square pulse is harder than a sine, and a pressure spike with a 1 ms rise time is harder than either.
- Zero shift limit after the pulse test. Write it as a percentage of full scale, and test it, because it is not on most datasheets.
Running the tests
Proof pressure is applied slowly, held for a defined period, released, and followed by a full calibration check. Burst is applied until failure, with the sensor behind a guard, and the result is the pressure at which it leaked or ruptured, not the pressure at which it stopped responding. Record which of the two happened: a diaphragm that ruptures at 3 times rated and a weld that leaks at 3 times rated are different failure modes with different design implications.
Temperature changes both numbers. A burst test at 20 °C says little about the same sensor at 80 °C, where the diaphragm is softer and the seal is closer to its limit. A pressure sensor with a proof and burst rating stated at the operating temperature is the only kind that can be specified from a datasheet alone, and the site index lists the ranges and output types.
Standards referenced: IEC 61298-2.
