Pressure · 2.4

Diaphragm seals: when you need them and the temperature errors they bring

26 October 2026 · 9 min read · IEC 60770-1, EN 837-1

In short

A diaphragm seal keeps the process away from the transmitter. A fill fluid carries the pressure, either close-coupled or through a capillary.
Seals add errors. The fill's own head (ρ_fill g h) and the change of that head with temperature are the big ones.
The head is fixed and can be calibrated out. The temperature change cannot, and it can be several percent of a level span.
Keep capillaries short, equal and at the same temperature, and use the largest seal diameter that fits.

Some fluids should never reach a transmitter's sensor: slurries that block, corrosive acids, polymers that set, hot fluids or anything that must stay sterile. A diaphragm seal solves that, but it turns a simple pressure measurement into a small hydraulic system, with errors of its own.

Where this matters

Corrosive, viscous, crystallising or solids-laden fluids that would block or attack an impulse line.
High temperatures beyond the transmitter's limit.
Hygienic service needing flush, cleanable connections.
DP level on tall vessels with remote seals, where the temperature error is largest.

How it works

A thin flexible diaphragm sits flush with the process. Behind it a sealed fill fluid (silicone oil, a high-temperature oil, an inert fluorocarbon, or a food-grade oil) carries the pressure to the transmitter. The connection is either direct mount (close-coupled) or through a capillary, a thin armoured tube several metres long.

Head of the fill

ΔP_head = ρ_fill × g × h

ρ_fill is the fill density (kg/m³), g = 9.80665 m/s² and h is the vertical distance (m) between the seal and the transmitter. A seal above the transmitter adds pressure, and one below subtracts it. In a two-seal DP level system, the difference in height between the two seals gives a fixed offset that is set in the calibration.

Temperature effects

Density change: the fill's density falls as it warms (about 0.1 %/°C for silicone oil), so the head changes. This is the largest error for tall remote-seal systems.
Volume change: the fill expands and pushes against the diaphragm. A stiffer (smaller) diaphragm turns that into a larger false pressure, so large diaphragms are better.
Response time: long, thin capillaries and viscous fills slow the response, especially in cold weather.

What the standards and practice say

Vendors give seal systems their own total error and response time, calculated for the specific seal, fill, capillary length and temperatures. Ask for that calculation for each critical application.
EN 837-1 covers gauges, and seals for gauges follow vendor and application standards. IEC 60770-1 covers the transmitter.
Choose fill for the process temperature range and vacuum. Some fills boil off under vacuum at high temperature.
For DP level, tuned systems (matched capillaries) and electronic remote sensors (two transmitters linked digitally) reduce or remove capillary errors.

Worked example: DP level on a 5 m tank with two seals

Input	Value
Seal spacing (vertical)	5 m
Process	Water-like, ρ = 1,000 kg/m³
Fill	Silicone, ρ_fill = 930 kg/m³, density change ≈ 0.096 %/°C
Temperature swing of the low-side capillary run	30 °C (night to midday sun)

Step 1, span. The level range is 0–5 m, so the span is 1,000 × 9.80665 × 5 = 49,033 Pa = 490 mbar.

Step 2, fill head. 930 × 9.80665 × 5 = 45,601 Pa = 456 mbar on the low side. This is constant, so it is taken out by the calibration (elevated zero).

Step 3, temperature error. The density change over 30 °C is 0.096 % × 30 = 2.9 %. The head therefore changes by 0.029 × 456 = 13 mbar.

Step 4, as a share of span. 13 / 490 = 2.7 % of the level span, or about 13 cm of level from the sun alone.

Error source	Size	Can it be calibrated out?
Fill head (456 mbar)	93 % of span	Yes, fixed offset
Fill density with temperature	≈ 2.7 % of span	No, it changes all day
Transmitter reference accuracy	≈ 0.1 % of span	Small by comparison

What this example shows

The seal system, not the transmitter, sets the accuracy.
Tall vessels in the sun are the worst case. Trace-heat or insulate both capillaries together, or use an electronic remote sensor.
Ask for the vendor's seal-system error calculation before you accept a seal for a trip or an inventory measurement.

When it fails: the fixes in order

Shorten or remove the capillary. Direct-mount where you can.
Keep both capillaries together, the same length and at the same temperature (shade, insulation, tracing).
Use a larger diaphragm to reduce the volume-change error.
Use an electronic remote sensor system on tall vessels.
Choose a lower-expansion or more suitable fill for the temperature range.

Common mistakes

Specifying seals by habit for clean services that do not need them.
Long capillaries coiled in the sun on one side only.
Small seal diameters on low ranges.
Forgetting the fill head when ranging, which gives a large zero error at commissioning.
A fill fluid unsuitable for vacuum at temperature.
Kinked or damaged capillaries. They cannot be repaired in the field.

Checklist

A seal is really needed: corrosive, solids, viscous, hot or hygienic.
The diaphragm material and seal type suit the process.
The fill suits the minimum and maximum temperature and vacuum.
The capillary length is as short as possible, and both sides are equal.
The vertical heights between seals and transmitter are recorded for calibration.
The seal-system total error and response time come from the vendor.
The ambient temperature swing is stated, and the capillaries are protected.
Critical tall-vessel levels consider an electronic remote sensor.

Quick reference

Head: ΔP = ρ_fill g h (≈ 91 mbar per metre for silicone at 930 kg/m³).
Temperature error ≈ (density change %/°C × ΔT) × fill head.
Silicone: density change about 0.1 %/°C.
Rules: short, equal, same temperature, big diaphragm.