AC versus DC: The Truth

The purpose of this paper is to truthfully answer various questions one might have with regard to resistance bridges and readouts after reading some of the sales hype arguing that AC is better than DC or vice versa. It is generally assumed that we are discussing resistance bridges for thermometry applications and the advantages or disadvantages of one type of bridge or technique over the other.

1.         Are AC bridges more stable and accurate than DC bridges?

No. AC bridges and DC bridges are both true ratiometric devices using a ratio transformer to compare resistances. With both types, the measured ratio (excluding random noise) depends only on the ratio of the windings of the transformer. Accuracy and stability with both designs result from the fact that transformer windings always have an integer quantity. With either bridge, neither aging of components nor varying environmental conditions cause drift or loss of accuracy.

2.         Are DC bridges more susceptible to thermoelectric EMF?

They can be, but generally are not. DC bridges are not affected by constant EMFs, since the current is reversed between each measurement. They might, however, be affected by EMFs that change significantly over the time it takes to make a measurement. The result will be to increase the noise in the measurements, not cause a steady error in the readings. Since most modern DC bridges reverse the current every ten seconds or less, EMFs are not usually a problem. More care is taken with DC bridges to reduce thermoelectric EMF by using good quality copper connectors and covering the connectors with a draft shield. As a consequence, the noise caused by thermoelectric EMF in DC bridges is generally immeasurable.

Worst case conditions for thermoelectric EMF noise occur when using 0.25Ω SPRTs at high temperatures such as at the freezing point of silver. Using a typical current of 10 mA, it is necessary to resolve voltages on the order of 1 nV for accurate temperature measurements. Even with AC bridges, noise from thermoelectric EMF is noticeable at this level because the noise is spread over a wide frequency range, including the narrow band used by AC bridges. The best solution in this case is to use an SPRT that is carefully designed to minimize thermoelectric EMF.

3.         Does Peltier heating cause errors when using DC bridges?

No. The Peltier heating power can easily be calculated by multiplying the thermoelectric EMF by the excitation current. SPRTs have a thermoelectric EMF that's typically less than 10 µV. With an excitation current of 1 mA you can see that the Peltier heating power is less than 10 nW. Compare this to the Joule heating power, which is 25 µW for an SPRT at 25Ω with 1 mA of current. The Peltier heating is several orders of magnitude less than the self heating and is immeasurable. Peltier heating is only an issue with germanium thermometers since these thermometers have a relatively large Peltier coefficient. Still, many metrologists use DC with germanium thermometers anyway since the error is small and DC offers some advantages over AC in this application.

4.         Are AC bridges more immune to interference from AC power lines?

No. In fact, the opposite is true. AC bridges are more sensitive to interference from nearby power lines, especially if sub harmonics and asynchronous noise are present in the supply. For this reason, labs that use AC bridges for fixed-point measurements prefer DC-heated furnaces, which are more expensive. It is difficult to achieve better than 50 µK of noise when using an AC bridge with an AC furnace. The type of furnace makes no difference with DC bridges. A DC bridge can still achieve less than 15 µK of noise with an AC furnace.

5.         Do AC bridges have lower noise than DC bridges?

This depends. Many labs that use AC bridges with 25Ω SPRTs see noise on the order of 10 µK. They generally use DC-heated furnaces and take care to eliminate problems from interference and reactance. Other labs have not had as much success with AC bridges, seeing noise closer to 50 µK. We have gotten better results in our lab with DC bridges, which give us noise within 15 µK, compared to about 50 µK for our AC bridge. Again, this may only be because we're using AC furnaces and haven't taken as much care to reduce interference and reactance.

6.         Are AC bridges faster than DC bridges?

Yes. Most people agree that AC bridges are able to settle and make measurements more quickly than DC bridges. The most accurate AC bridges generally settle within about 20 seconds and produce a new measurement about every ten seconds with the typical settings. They can measure as fast as two seconds with poorer resolution. The fastest and most accurate DC bridge requires up to two minutes to settle initially but then produces measurements at about the same rate—eight to ten seconds. The interval can be reduced to as low as four seconds with poorer resolution. For time-critical measurements and process control applications, AC bridges may offer some advantage. However, you can see that neither type of bridge is very fast. For applications where speed is important, a variety of lower cost DC readouts are available that offer extremely short settling times and measurement intervals. These DC instruments are much faster than any AC devices and are better suited for secondary-level PRT calibration, process control applications with multiple sensors, and time-critical mesurements.

7.         Do DC bridges need to warm up longer than AC bridges?

Perhaps. For DC bridges, the recommended warm-up time for full accuracy is about ten minutes. However, this is generally not an issue since neither type of bridge is considered portable. They are usually kept in one place and left powered on so they are always ready to use.

8.         Do AC bridges exhibit errors due to reactance effects in the thermometer, lead wires, or resistors?

Not usually. Several steps are taken to eliminate reactance errors with AC bridges. First, most SPRTs and standard resistors are designed to minimize inductance. Second, AC bridges are used with low-reactance coaxial cables. Third, the excitation current used with modern AC bridges has a fairly low frequency of 25 or 30 Hz. Fourth, AC bridges use special quadrature balancing circuitry to cancel reactance effects. As a result, reactance is not usually a problem in primary-level thermometry. It can be a significant issue, however, under any of the following conditions: the measured resistances are very low or very high, the thermometers or resistors are of lower quality and have high inductance, or the lead wires are very long or are of poor quality and have excessive inductance or capacitance. For these cases, DC bridges and readouts will give better results.

9.         Does AC exhibit errors due to dielectric losses or induced currents?

Possibly. Any frequency dependent effect that dissipates power, even though very small, can produce errors when AC excitation current is used. Two such effects are dielectric losses in insulating materials within the SPRT, lead wires, or resistors and induced currents in conductive materials in close proximity to the wires. Most SPRTs and resistors intended for use with AC bridges are constructed to minimize these errors and particular types of wire are used for the interconnections. Also, as before, the AC frequency is very low, which minimizes the effects. Furthermore, measurements with SPRTs are based on resistance ratios, not absolute resistances, so constant errors such as these in the resistors and lead wires tend to cancel out. Thus, extremely accurate temperature measurements can still be made with AC bridges. Studies have shown that the AC losses are most apparent with SPRTs at very high temperatures but they are still generally considered insignificant with most SPRTs. The errors can be significant, however, when AC bridges are used with lesser-quality PRTs, resistors, and connecting wires. It's not unusual to see errors from AC losses as large as 0.1°C with some secondary and industrial level PRTs. DC should be used for secondary and industrial level thermometry applications as well as resistor calibration.

As mentioned above, AC bridges are used with special AC/DC resistors. Compared with DC resistors, the AC/DC resisistors are less common, often more expensive, and may present problems with regard to calibration and traceability.

10.       Why do many national thermometry labs use AC bridges?

The earliest automatic bridges used AC. Good, automatic DC bridges weren't available until many years later. Many labs were originally set up with AC bridges. They obtain excellent results using them, and have had no reason to switch. Some metrologists still prefer AC bridges because measurements can be obtained more quickly. Others have had less success with AC bridges and have been able to achieve desired results with less difficulty using DC bridges. Some have chosen DC bridges because of cost constraints and to avoid the calibration and traceability issues of AC standard resistors. Resistance calibration labs use DC bridges almost exclusively because they can be used with such a wide variety of resistors, can measure over a very large range of resistances, and can operate over a larger range of excitation currents. Secondary-level calibration labs tend to use DC thermometer readouts or current source/digital voltmeter systems.

Conclusion

In summary, for some applications, one type of bridge may be better suited than the other while for other applications either type may be used. The table that follows summarizes the advantages and disadvantages of the two types of resistance bridges showing '+' if the bridge is a good choice for a particular consideration, 'o' if the issue is moot, or '–' if the issue is more likely to be a problem with this type of bridge.