Minimizing offset voltages

Voltage offset can be the source of significant error. For example, an offset of 3 μV on a 2500 mV signal causes an error of only 0.00012%, but the same offset on a 0.25 mV signal causes an error of 1.2%. Measurement offset voltages are unavoidable, but can be minimized. Offset voltages originate with:

• Ground currents. See Minimizing ground potential differences.
• Seebeck effect Induces microvolt level thermal electromotive forces (EMF) across junctions of dissimilar metals in the presence of temperature gradients. This is the principle behind thermocouple temperature measurement. It also causes small, correctable voltage offsets in data logger measurement circuitry.
• Residual voltage from a previous measurement

Remedies include:

• Connecting power grounds to power ground terminals (G).

• Using input reversal (RevDiff = True) with differential measurements.

• Automatic offset compensation for differential measurements when RevDiff = False.

• Automatic offset compensation for single-ended measurements when MeasOff = False.

• Using MeasOff = True for better offset compensation.

• Using excitation reversal (RevEx = True) with bridge measurements.

• Programming longer settling times.

Single-ended measurements are susceptible to voltage drop at the ground terminal caused by return currents from another device that is powered from the data logger wiring panel, such as another manufacturer's communications modem, or a sensor that requires a lot of power. Currents greater than 5 mA are usually undesirable. The error can be avoided by routing power grounds from these other devices to a power ground G terminal, rather than using a signal ground ( ) terminal. Ground currents can be caused by the excitation of resistive-bridge sensors, but these do not usually cause offset error. These currents typically only flow when a voltage excitation is applied. Return currents associated with voltage excitation cannot influence other single-ended measurements because the excitation is usually turned off before the data logger moves to the next measurement. However, if the CRBasic program is written in such a way that an excitation terminal is enabled during an unrelated measurement of a small voltage, an offset error may occur.

The Seebeck effect results in small thermally induced voltages across junctions of dissimilar metals as are common in electronic devices. Differential measurements are more immune to these than are single-ended measurements because of passive voltage cancellation occurring between matched high and low pairs such as U1/U2. So, use differential measurements when measuring critical low-level voltages, especially those below 200 mV, such as are output from pyranometers and thermocouples.

When analog voltage signals are measured in series by a single measurement instruction, such as occurs when VoltSE() is programmed with Reps = 2 or more, measurements on subsequent terminals may be affected by an offset, the magnitude of which is a function of the voltage from the previous measurement. While this offset is usually small and negligible when measuring large signals, significant error, or NAN Not a number. A data word indicating a measurement or processing error. Voltage overrange, SDI-12 sensor error, and undefined mathematical results can produce NAN., can occur when measuring very small signals. This effect is caused by dielectric absorption of the integrator capacitor and cannot be overcome by circuit design. Remedies include the following:

• Programing longer settling times.

• Using an individual instruction for each input terminal, the effect of which is to reset the integrator circuit prior to filtering.

• Avoiding preceding a very small voltage input with a very large voltage input in a measurement sequence if a single measurement instruction must be used.

The following table lists some of the tools available to minimize the effects of offset voltages:

Offset voltage compensation options
CRBasic measurement instruction	Input reversal (RevDiff=True)	Excitation reversal (RevEx=True)	Measure offset during measurement (MeasOff=True)	Measure offset during background calibration (RevDiff=False) (RevEx=False) (MeasOff=False)
BrHalf()		ü		ü
BrHalf3W()		ü		ü
BrHalf4W()	ü	ü		ü
BrFull()	ü	ü		ü
BrFull6W()	ü	ü		ü
Resistance()	ü	ü		ü
Resistance3W()	ü	ü
TCDiff()	ü			ü
TCSe()			ü	ü
VoltDiff()	ü			ü
VoltSe()			ü	ü

Compensating for offset voltage

Differential measurements also have the advantage of an input reversal option, RevDiff. When RevDiff is True, two differential measurements are made, the first with a positive polarity and the second reversed. Subtraction of opposite polarity measurements cancels some offset voltages associated with the measurement.

Ratiometric measurements use an excitation voltage or current to excite the sensor during the measurement process. Reversing excitation polarity also reduces offset voltage error. Setting the RevEx parameter to True programs the measurement for excitation reversal. Excitation reversal results in a polarity change of the measured voltage so that two measurements with opposite polarity can be subtracted and divided by 2 for offset reduction similar to input reversal for differential measurements.

For example, if 3 µV offset exists in the measurement circuitry, a 5 mV signal is measured as 5.003 mV. When the input or excitation is reversed, the second sub-measurement is –4.997 mV. Subtracting the second sub-measurement from the first and then dividing by 2 cancels the offset:

5.003 mV – (–4.997 mV) = 10.000 mV

10.000 mV / 2 = 5.000 mV

Ratiometric differential measurement instructions allow both RevDiff and RevEx to be set True. This results in four measurement sequences, which the data logger processes into the reported measurement:

• positive excitation polarity with positive differential input polarity
• negative excitation polarity with positive differential input polarity
• positive excitation polarity with negative differential input polarity
• negative excitation polarity with negative differential input polarity

For ratiometric single-ended measurements, such as a BrHalf(), setting RevEx = True results in two measurements of opposite excitation polarity that are subtracted and divided by 2 for offset voltage reduction. For RevEx = False for ratiometric single-ended measurements, an offset-voltage measurement is determined from self-calibration.

When the data logger reverses differential inputs or excitation polarity, it delays the same settling time after the reversal as it does before the first sub-measurement. So, there are two delays per measurement when either RevDiff or RevEx is used. If both RevDiff and RevEx are True, four sub-measurements are performed; positive and negative excitations with the inputs one way and positive and negative excitations with the inputs reversed. The automatic procedure then is as follows:

1. Switch to the measurement terminals.
2. Set the excitation, settle, and then measure.
3. Reverse the excitation, settle, and then measure.
4. Reverse the excitation, reverse the input terminals, settle, measure.
5. Reverse the excitation, settle, measure.

There are four delays per measurement. In cases of excitation reversal, excitation time for each polarity is exactly the same to ensure that ionic sensors do not polarize with repetitive measurements.

Read More: The Benefits of Input Reversal and Excitation Reversal for Voltage Measurements .

Measuring ground reference offset voltage

Single-ended and differential measurements without input reversal use an offset voltage measurement with the PGIA inputs grounded. This offset voltage is subtracted from the subsequent measurement. For differential measurements without input reversal, this offset voltage measurement is performed as part of the routine background calibration of the data logger. See About background calibration. Single-ended measurement instructions VoltSE() and TCSe() include the MeasOff parameter determines whether the offset voltage measured is done at the beginning of the measurement instruction, or as part of self-calibration. This option provides you with the opportunity to weigh measurement speed against measurement accuracy. When MeasOff = True, a measurement of the single-ended offset voltage is made at the beginning of the VoltSE() or TCSe() instruction. When MeasOff = False, measurements will be corrected for the offset voltage determined during self-calibration. For installations experiencing fluctuating offset voltages, choosing MeasOff = True for the VoltSE() or TCSe() instruction results in better offset voltage performance.

If RevDiff, RevEx, or MeasOff is disabled ( = False), offset voltage compensation is automatically performed, albeit less effectively, by using measurements from the background calibration. Disabling RevDiff, RevEx, or MeasOff speeds up measurement time; however, the increase in speed comes at the cost of accuracy because of the following:

• RevDiff, RevEx, and MeasOff are more effective.

• Background calibrations are performed only periodically, so more time skew occurs between the background calibration offsets and the measurements to which they are applied.