Differential Pressure Transmitter

What is a Differential Pressure Transmitter

A differential pressure transmitter is a field instrument that converts the difference between two applied fluid pressures into a standardized electrical signal, most commonly a two-wire 4-20 mA loop with HART digital overlay. It has two process connections, conventionally labeled the high (H) and low (L) side, and an internal sensing capsule whose output depends only on the pressure difference H minus L. The static, or line, pressure that both sides experience together is treated as common-mode and, ideally, ignored. This is the defining property that separates a DP transmitter from a gauge or absolute transmitter, which has a single port and one fixed reference.

Structurally the capsule is symmetric. Each side has a flush isolating diaphragm, typically stainless steel 316L, a corrosion-resistant nickel alloy such as Hastelloy C-276 or Monel, tantalum, or gold-plated material, behind which a fill fluid (silicone or inert oil) transmits pressure to a single central sensing element. The sensing element, whether a capacitance cell, a piezoresistive silicon chip, or a silicon resonant beam, deflects according to the net force, which is proportional to the pressure difference. Signal-conditioning electronics then linearize, temperature-compensate, and digitize the reading and drive the output loop. When this full signal chain is integrated into one housing with a 4-20 mA / HART output, the industry calls it a transmitter, the term that appears on purchase orders.

The value of DP measurement is that very small differences can be resolved while both ports sit at very high static pressure. A reference-grade cell can resolve a 0 to 100 mbar span while both sides float at 16 MPa (160 bar) line pressure. That is roughly one part in 1,600 of the static load, which is why the static pressure zero and span effects are explicitly specified and why DP transmitters are zero-trimmed at line pressure during commissioning rather than on a bench at atmosphere.

Historically, DP measurement grew out of the force-balance pneumatic transmitters of the 1950s and 1960s, where a flapper-nozzle servo balanced process force against a feedback bellows and output a 3-15 psi pneumatic signal. The 1969 introduction of electronic capacitance DP cells, and Honeywell and Rosemount work through the 1970s, moved the industry to 4-20 mA. The 1985 HART protocol overlaid digital configuration on the analog loop. The modern era is defined by microprocessor-based smart transmitters with characterization curves, multi-point temperature compensation, and self-diagnostics, plus all-digital electronic remote sensor systems that subtract two networked pressure readings in firmware instead of through oil-filled capillaries.

Four engineering metrics determine DP transmitter quality and total cost of ownership: reference accuracy at the calibrated span, static pressure effect (the penalty for high line pressure), turndown capability (how far the span can be compressed below the upper range limit), and long-term stability expressed as percent of upper range limit per multi-year interval. A cheap unit with poor static-pressure rejection or 0.2 percent annual drift can cost more in recalibration labor and process variation within three years than a reference-grade unit bought upfront.

Applications: Flow, Level, Filter

The reason one DP transmitter dominates process plants is that flow, level, and pressure-drop monitoring are all, physically, the same differential pressure problem. The transmitter does not know whether it is reading flow or level; the meaning is supplied by where the two impulse lines are tapped and by the math the DCS applies. The table below maps the three classic DP derivations to their primary elements and configuration.

Flow by differential pressure is the oldest and still the most common flow method among industrial flow meters. A primary element such as an orifice plate constricts the pipe and creates a recoverable pressure drop. The governing relation, captured in ISO 5167, is that volumetric flow is proportional to the square root of the differential pressure: doubling flow quadruples DP. Because of this square-law, the transmitter or control system must apply square-root extraction to linearize the signal, and the practical usable flow range of a single orifice tap is limited (often about 3 to 1 to 4 to 1) by the DP turndown and by accuracy collapse at low DP. Installation must respect ISO 5167 straight-run requirements, typically on the order of 10 to 20 pipe diameters upstream and 5 downstream, reduced by a flow conditioner. On steam service, condensate (seal) pots keep hot vapor off the diaphragms.

Level by differential pressure is one of the most common operating principles behind a hydrostatic level transmitter, exploiting hydrostatic head: the pressure at the bottom of a liquid column equals column height times density times gravitational acceleration. In an open tank the low side is vented, so the transmitter reads head directly. In a closed or pressurized tank the vapor space adds its own pressure, so the low side is tapped at the top to subtract it. The reference leg between the top tap and the low port is either a dry leg (gas-filled) or a wet leg (deliberately liquid-filled for a stable reference); its static column is entered as a configuration offset, called zero elevation or suppression, rather than corrected with a large and drift-prone zero trim. Because the reading depends on density, a temperature change that alters liquid density introduces an error unless compensated.

Filter, strainer, and pump monitoring is the simplest DP duty: tap upstream to the high port and downstream to the low port, and the transmitter reports the pressure drop directly with no math. Rising DP signals a clogging element or fouled bed in an industrial filter; a sudden drop can indicate a ruptured element or bypass. Interface measurement between two immiscible liquids of different density (oil over water, for example) uses a DP span placed across the interface layer; as the interface rises or falls between the two taps, the net hydrostatic difference changes linearly, letting one DP transmitter track the boundary. These four derivations, from one symmetric capsule, are why the DP transmitter remains the workhorse of process instrumentation.

Sensing Technologies Compared

Three sensing technologies dominate modern DP transmitters: capacitance cells, piezoresistive silicon (with metallic isolating diaphragms), and silicon resonant (resonant beam) sensors. Each balances accuracy, static-pressure rejection, stability, and cost differently. The table below compares representative reference-grade and process-grade series whose figures are taken from current published manufacturer datasheets; verify the exact ordering code before purchase, because every series spans several accuracy and material options.

Capacitance cells place a sensing diaphragm between two fixed capacitor plates filled with stable fill fluid. Differential pressure deflects the center diaphragm, increasing capacitance on one side and decreasing it on the other; the differential capacitance is converted to a linear digital value. Capacitance is extremely sensitive at low DP, making it the long-standing standard for low-span flow and level service. The Rosemount 3051 family, the most widely installed DP cell in the world, uses this principle and reaches 0.025 percent reference accuracy with 200 to 1 turndown on the 3051S. The trade-offs are structural complexity, fill-fluid temperature dependence, and the need for careful overpressure protection.

Piezoresistive sensors with metallic isolating diaphragms implant strain-sensitive resistors in a silicon chip that sits behind oil-filled metal diaphragms; the chip forms a Wheatstone bridge whose imbalance tracks the net pressure. The Endress+Hauser Deltabar PMD75 uses a welded metallic measuring diaphragm and reaches 0.05 percent reference accuracy in standard form and 0.035 percent in the Platinum version, with a one-sided overload rating to 420 bar. Piezoresistive cells are robust and cost-effective and span 1 mbar to several hundred bar across a product family, but raw silicon piezoresistance is more temperature- and aging-sensitive than resonant sensing, which the better units offset with multi-point compensation.

Silicon resonant (resonant beam) sensing, marketed by Yokogawa as DPharp, etches two tiny resonating beams into a single-crystal silicon chip. Pressure puts one beam in tension and the other in compression, shifting their resonant frequencies in opposite directions; the frequency difference is measured digitally. Because frequency can be counted very precisely and the two-beam arrangement is differential, the EJX110A achieves 0.04 percent of span standard (0.025 percent optional), 0.1 percent of upper range limit per 15 years stability, and a 90 millisecond response. The published over-pressure effect is about 0.03 percent of URL and the standard maximum working pressure is 16 MPa. Resonant sensing offers very low temperature coefficient and impurity-ion sensitivity compared with piezoresistance, at a higher unit cost.

For most general flow, level, and filter loops a process-grade piezoresistive unit at 0.075 to 0.1 percent of span is more than adequate. Reserve capacitance or resonant reference-grade cells for custody-transfer flow, fiscal measurement, low-span level on tall vessels, and any loop where high turndown or strong static-pressure rejection is decisive.

Impulse Lines, Manifolds, and Seals

A DP transmitter is only as accurate as its connection to the process. More DP loops fail at the impulse lines and manifold than at the capsule. The two impulse lines that carry process pressure from the tap points to the high and low ports must be equal in length, equally sloped, and at the same temperature, because any difference in fill-fluid column height or density between the two legs adds directly to the measured DP as a false zero. Sloping rules depend on phase: liquid lines slope up toward the process (about 1 in 12) so gas bubbles purge back into the line, while gas lines slope down toward the process so condensate drains away from the transmitter.

For steam and other hot or condensing services, a wet leg is used. The reference (low) leg is deliberately kept full of condensate or a fill liquid to provide a stable, known hydrostatic reference. Its static column is not corrected with a zero trim; instead it is entered into the transmitter as a zero elevation or suppression offset, so the loop reads true level even though the low side carries a permanent liquid head. A dry leg (gas-filled reference) is used when the vapor space stays non-condensing. Getting the leg type and offset wrong is a leading cause of systematic level error.

Between the impulse lines and the transmitter sits a valve manifold. A three-valve manifold has two block (isolation) valves, one per side, plus a center equalizing valve. The five-valve manifold adds two bleed/vent valves, often fine-threaded needle valves, so each side can be vented independently, which is necessary for safe commissioning and zero checks on hazardous or high-pressure service. The equalizing valve is the heart of correct commissioning: opening it ties the high and low chambers together so the transmitter sees zero DP at full line pressure, allowing a zero trim at the actual static pressure. The standard sequence is fill the lines, open both block valves, open the equalizer, perform the zero trim, then close the equalizer and reopen to service. Performing this trim at line pressure cancels the static-pressure zero effect for that operating point; skipping or reordering these steps is the number one cause of incorrect startup readings.

When impulse lines are impractical, because the medium is hot, viscous, crystallizing, slurry-laden, or aggressively corrosive, the transmitter is fitted with remote diaphragm seals. A flush or extended seal diaphragm contacts the process, and an oil-filled capillary transmits pressure to the capsule. Seals isolate the medium and keep the electronics cool, but they add fill-fluid temperature error, slow the response, and demand matched capillary lengths on both sides to avoid head-induced zero shift. Fill-fluid selection (silicone, halocarbon, high-temperature, food-grade) is governed by process temperature and contact requirements. The table below summarizes when each isolation strategy applies.

The newest alternative to oil capillaries is the electronic remote sensor system, for example the Rosemount 3051S ERS. Two pressure transmitters are linked by an electrical cable, and the DP is computed in firmware by subtracting their two readings rather than through a mechanical oil column. This removes capillary fill-fluid temperature effects, eliminates heat tracing and insulation, and, on tall vessels, cuts response time substantially compared with long capillaries. The cost is two sensing elements and the networking cable, justified when capillary length or temperature swing would otherwise dominate the error budget.

Key Specification Parameters

Reading a DP transmitter datasheet means separating the headline reference accuracy from the terms that actually dominate at your operating point. A DP spec sheet may list 20 to 30 parameters, but a defined set governs the selection. The table below collects the parameters that drive DP decisions, with representative values from current reference-grade datasheets; the text that follows decodes each.

Reference accuracy and total performance are not the same number. Reference accuracy (for example plus or minus 0.04 percent of span on the EJX110A) is a best case measured at reference conditions and moderate turndown. The figure that matters in the field is total probable error, which combines reference accuracy with ambient temperature effect, static pressure effect, and stability over the calibration interval. Manufacturers publish a total-performance equation; always evaluate it at your exact span, turndown, line pressure, and ambient range, not the headline number.

Turndown (rangedown) is the ratio of the cell upper range limit to your calibrated span. Modern cells advertise 100 to 1 or 200 to 1, but accuracy degrades at high turndown because the zero and span uncertainty terms are referenced to the upper range limit. The common form is a fixed reference term plus a turndown-dependent term that scales with URL divided by span. A 250 mbar cell run at 10 mbar span (25 to 1) carries far more relative error than the same cell at 100 mbar span (2.5 to 1). Choose a cell whose upper range limit sits a comfortable factor above your span, rather than maximizing turndown on a single oversized cell.

Static pressure effects are the DP-specific error terms a gauge transmitter never has. The static pressure zero effect is a zero shift caused by the common line pressure deforming the capsule symmetrically (for example plus or minus 0.04 percent of URL per the rated static-pressure step), and it is correctable by a zero trim at line pressure. The static pressure span effect alters the sensitivity itself and is corrected by the manufacturer characterization or, if needed, a span adjustment at line pressure. On high-line-pressure, low-DP loops, these two effects can exceed the headline accuracy, so read them explicitly.

Overpressure and over-range ratings protect against one-sided pressure when a block valve is shut or a manifold is mis-sequenced. The Deltabar PMD75, for example, rates 420 bar one-sided overload. A DP capsule must survive full line pressure applied to one side only, which is mechanically the most demanding case, so verify the one-sided overpressure rating against your maximum credible mis-operation, not just the normal DP.

Output signal remains predominantly two-wire 4-20 mA with HART for configuration, diagnostics, and secondary-variable upload (static pressure and sensor temperature are read over HART on most smart DP units). Digital fieldbus options, FOUNDATION Fieldbus and PROFIBUS PA, suit large multi-drop DCS projects, while Ethernet-APL is the emerging two-wire digital backbone. Supply voltage for a 4-20 mA loop is typically in the 10.5 to 42 V DC window depending on load resistance and HART burden.

Selection Decision Factors

To turn the preceding five chapters into a specific model and order code, work the decision sequence below in order. Most DP selection mistakes come not from a single wrong value but from deciding a downstream parameter before an upstream one is fixed. These steps double as a complete RFQ template.

1. Process variable and DP span: First fix whether the duty is flow, level, interface, or filter monitoring, because that sets the tap arrangement and whether square-root extraction is needed. Then derive the DP span from the primary element sizing (flow) or the hydrostatic head and density (level), keeping the normal operating point within 30 to 70 percent of span.
2. Static line pressure rating: Specify the maximum common-mode line pressure both ports will see. This must be rated independently of the small DP span. Standard process cells rate 16 MPa (160 bar); high-pressure and special variants reach 420 bar one-sided. Under-rating here is a safety issue, not just an accuracy issue.
3. Accuracy class and turndown: Distinguish loop monitoring (0.1 percent of span is sufficient) from control and custody transfer (0.025 to 0.05 percent), and verify the total-performance equation at your actual turndown. Avoid forcing extreme turndown on one oversized cell; select a cell whose upper range limit sits a modest factor above your span.
4. Isolation strategy and seals: Decide direct mount with impulse lines, remote diaphragm seals with capillaries, an electronic remote sensor system, or a wet-leg reference, per Chapter 4. Hot, viscous, crystallizing, slurry, or corrosive media drive toward seals or ERS; clean ambient media stay direct-mount.
5. Wetted materials and connections: Select diaphragm and process-part material (316L for water, steam, and light hydrocarbons; Hastelloy C-276 or tantalum for chlorides and acids). Choose the manifold (three-valve or five-valve), process connection (1/2 NPT, oval flange, DN/PN flanged seal), and electrical entry.
6. Certifications and functional safety: Hazardous-area protection (ATEX, IECEx, NEPSI, FM intrinsically safe Ex ia or flameproof Ex d), functional safety SIL2 or SIL3 with the manufacturer safety manual, sanitary 3-A or EHEDG for hygienic service, and pressure-equipment compliance (PED 2014/68/EU) as applicable.
7. Output, protocol, and diagnostics: 4-20 mA with HART is the default; consider FOUNDATION Fieldbus, PROFIBUS PA, or Ethernet-APL for large digital architectures. Confirm HART secondary-variable upload (static pressure, sensor temperature), plugged-impulse-line and advanced diagnostics if condition monitoring matters.
8. Total cost of ownership: Purchase price plus installation (impulse lines, manifold, seals), commissioning, recalibration labor over the stability interval, and the process cost of drift. A reference-grade cell with 0.1 percent of URL per 15 years stability can undercut a cheaper unit needing annual recalibration within a few years.

One dimension teams routinely overlook is manufacturer serviceability: local spare-capsule and seal inventory, field-calibration service, HART DD/DTM file registration with the FieldComm Group, firmware upgradability, and documented capillary-fill data sheets for seal systems. These seem secondary at purchase but determine repair turnaround after 5 to 10 years in service. Emerson Rosemount, Yokogawa, Endress+Hauser, ABB, Honeywell, Siemens, and WIKA all maintain calibration laboratories and parts centers in China and major industrial regions, making them defensible choices for large projects; verify local support before standardizing on any one brand.