Two interconnected process transmitters, one mounted in the headspace and one in the liquid, share a signal cable so the principal can compute the density-compensated level at an accuracy equivalent of up to 0.05% of the set span [S1].
The technique removes the capillary/impulse line that conventional differential pressure transmitters rely on, so gas-phase temperature swings no longer corrupt the reading through Gay-Lussac's Law, and the measured values are continuous density, process temperature, and compensated level in one device [S1][S2].
Why a single DP transmitter fails when density drifts
A standalone gauge converts hydrostatic head into level through the relationship Pressure (inches of water) = relative density × liquid height (inches), so the level reading is only as good as the assumed specific gravity [S4].
When the process fluid changes density with temperature, concentration, or agitation, that assumed constant becomes the dominant error source, and a standard transmitter with a fixed density scalar will drift by several percent of reading over a typical 30 to 60 °C operating band [S2][S7]. On pressurized vessels the low-pressure leg must be tied to the vapor space, otherwise the head pressure adds to the hydrostatic component and the calculated level jumps by whatever the gas pressure is in inches of water [S3].
How the two-transmitter architecture measures both level and density
The subordinate process transmitter sits in the vapor space and reports P1 (gas pressure) over an internal bus to the principal, which sits in the liquid and reports P2; the principal then evaluates (P2 − P1) along with two calibration offsets to recover density in real time, then multiplies that density by the bottom-leg hydrostatic pressure to output the true level [S1].
VEGA's electronic differential pressure arrangement using the VEGABAR 80 series demonstrates the same idea: two independent sensors at two fixed heights, both covered by the fluid, give a DP that scales with ρ·g·Δh, so ρ is solved from the changing DP and the level from the lowest sensor's output, with no capillary and no external signal conditioner [S2]. The math is the same whether the sensors are strapped to the side of the vessel, mounted through flanged spuds, or suspended from the top on cables, as long as the vertical separation Δh is known and stable [S2][S3].
Selection criteria: process transmitter vs. pressure transmitter pair
Two ordinary pressure transmitters will physically do the job, but they need an external calculator, a second voltage supply for the subordinate, and two pressure inputs, so most integrators move to process transmitters that already carry the math block, the second power rail, and the bus interface internally [S1].
Three decision criteria separate the two paths. First, turndown: each process transmitter can be ranged independently to its own relevant pressure, provided the turndown stays below the recommended 5:1 limit, which keeps zero and span errors inside the published spec [S1]. Second, containment: capillary-filled remote seals are a known failure mode on temperature-cycled services, so a direct-mount or top-suspended electronic DP eliminates the fill-fluid path and the air-purge step that a traditional wet/dry leg requires at commissioning [S1][S3]. Third, serviceability: when a process transmitter fails, only that unit is swapped, while a legacy DP cell plus capillary plus manifold usually goes back as an assembly, which lengthens Mean Time To Repair on critical inventory tanks [S1].
Where it works, where it does not
The two-transmitter method is the right pick for homogeneous liquids, slurries, froth, and any service where density varies with composition, temperature, or aeration, including mining flotation cells where standard ultrasonic/ball-target froth control drifts as the agitated surface changes [S2].
It is a poor fit for stratified liquids, immiscible layers, or services with heavy rag buildup, because the assumption of a single uniform density between the two sensing points breaks down and the calculation returns a blended number that fits neither layer [S1]. Viscous or crystallizing media also pose a problem if impulse lines are retained; a flanged diaphragm mount with a chemical seal on the lower tap is the standard way to keep the process out of the sensing line [S3]. On open atmospheric vessels the head-space tap can usually be eliminated by venting the low side, but on closed or slightly pressurized vessels a top tap is still required so P1 is measured rather than guessed [S5].
Comparison of the main implementation options
Four viable architectures exist for density-aware level, and the right choice depends on containment risk, temperature band, and access for service. A traditional wet-leg DP with capillary is the lowest hardware cost but the most exposed to fill-fluid temperature errors and the hardest to bleed on startup [S1][S3]. A flanged remote-seal DP keeps the process inside the vessel and removes the wet leg, but the capillary still introduces a thermal path and usually caps differential accuracy in the 0.1 to 0.2% of span band [S3].
A two-pressure-transmitter pair with an external calculator is mechanically simple but adds panel space, a second power supply, and a software mapping step, with effective accuracy in the same 0.1 to 0.15% zone once the I/O scan time is factored in [S1]. A two-process-transmitter pair with internal bus communication, such as WIKA's CPT-2X/IPT-2X link or VEGA's VEGABAR 80 electronic DP, reaches up to 0.05% of set span with no capillary, no external conditioner, and replacement of a single unit on failure [S1][S2]. The trade is higher unit cost against lower install, commissioning, and lifetime service cost, and the answer tilts toward the process-transmitter pair on any service where the planned calibration interval exceeds 12 months or where tank access for nitrogen-purging the impulse lines is restricted.
Commissioning checks that protect the 0.05% claim
Three checks have to happen before the system is trusted. Confirm the vertical separation Δh between the two taps with a tape or engineered drawing, because an error in Δh feeds directly into the density calculation and shows up as a level error that looks like a density error [S2].
Verify that the high-side tap is fully flooded at the lowest expected level, otherwise the sensor reads a partial hydrostatic column and the calculated density is biased low [S2][S3]. Finally, walk the two transmitters' zero against a calibrated reference at the planned operating temperature; WIKA's published 0.05% of set span assumes each unit has been ranged to its own working pressure and the turndown is held under 5:1, so ignoring that step erases the headline accuracy on day one [S1].
Limits, failure modes, and standards anchors
The math is exact only when the fluid between the two taps is the same fluid the level is being measured in, which is why froth, emulsions, and two-phase interfaces are flagged as the application's edge cases rather than its sweet spot [S1][S2]. Gas entrainment above 5 to 10% by volume typically pushes the effective density error past 1%, at which point a stilling well or a separate density measurement is the honest fix [S7].
On hazardous-area installs, the bus link between the two transmitters must be specified to the same protection level as the loop, and on closed pressure vessels the upper tap and its seal must hold the same containment class as the lower tap, since a leak in the vapor leg silently corrupts every reading downstream [S3][S5]. Calibration interval should be set against the worse of the two transmitters' published long-term drift, typically 0.1% of URL per year for mid-tier industrial units, so a 12-month calibration cycle is the practical ceiling for holding the system's 0.05% claim in the field [S1].
For vessels that sit inside a control cabinet or alongside other loop instrumentation, the heat-load and signal-cable routing rules around the bus pair follow the same separation practice as any 4 to 20 mA loop, with the addition that the internal bus cable must be kept short and shielded if the installation runs next to a VFD. When the loop is part of a broader plant audit, the tuning and oscillation patterns around the level PID follow the same priorities covered in any PID loop audit findings walk-down, since a density-compensated level signal normally smooths the PV and reduces controller cycling on tanks with agitator-induced surface motion.
Trackable signals to watch: OEM published accuracy curves for the next generation of electronic-DP transmitters crossing into the 0.025% of span band, and IEC 61511 proof-test interval guidance for two-transmitter architectures, which today is still handled case-by-case rather than by a dedicated clause. Either move would let the density-compensated level method replace a separate densitometer on more services.
Spec-level background on the components involved: two hand control, and checking and testing apparatus.