Resolution and Accuracy
The terms 'resolution' and 'accuracy' are similar in many people's minds, but they mean distinct things. It is important to keep the details in mind when reading and comparing specifications.
Resolution is the ability to 'resolve' differences; that is, to draw a distinction between two things. High resolution means being able to resolve small differences. In a digital system, resolution means the smallest increment or step that can be taken or seen. In an analog system, it means the smallest step or difference that can be reliably observed. Some examples of resolution in FTA systems:
- Contact angle. Resolution = 0.01 degree.
- Interfacial tension. Resolution = 0.01 mN/m.
- Pump control. Resolution = 1 step.
The first two are measurements and the last is a control. However, it does not convey much information to simply say 'one step' without converting that to a volume or a distance. Taking the FTA200 syringe pump as an example, the pump takes approximately 90,000 steps over its full scale range. If you fit a 10 milliliter syringe, this resolution amounts to
10000/90000 = 0.11 microliters per step
You can see that the resolution of the system depends on both the number of steps and the syringe fitted. See FTA200 Pump Resolution for more details of practical pump resolutions.
The accuracy of a system refers to how much the sytem, whether in measurement or control, deviates from the truth. To be meaningful, accuracy must really refer to 'worst case accuracy'.
As a trivial example, a stopped, unworking clock is still correct twice a day! We must describe this unworking clock by its worst case accuracy, which would be +/- 6 hours. Sometimes the '+/-' is dropped and the extent of the inaccuracy is reported: for this clock, it would be 12 hours.
Beyond describing the worst case, we must also consider resolution. The following is very important:
The accuracy of a system can never exceed its resolution!
This follows from the worst case requirement. Conside this example. Say we take our FTA200 pump with a 10 ml syringe. It has a resolution of 0.11 microliters with the 10 ml syringe. Let us say we start with 0 dispense and the pump motor at rest. We ask for exactly 0.11 microliters dispense. Presumably the pump can do this accurately as the requested volume happens to exactly match the resolution. In this case, you might falsely think the pump was infinitely accurate (zero error). But this is not true, because we chose a best case rather than a worst case. What if I had asked for
1/2 of 0.11 ul
3/2 of 0.11 ul ?
Then I would have had an error of +/- 1/2 of 0.11 ul or +/- 1/2 step. We say the accuracy of the pump is +/- 1/2 step or it has an inaccuracy of 1 step, as you prefer.
Note the resolution and accuracy, expressed in volume units, improve as you fit smaller syringes. Say, 2.78 nanoliters with a 250 microliter syringe, but they stay the same expressed in steps: 1.
Repeatability, Precision, Noise Level
Repeatability is the variance in repeated measurements on an unchanging sample. If measured statistically, it is sometimes referred to as the noise level in the measurement. In general, high repeatability means a low noise level.
Precision is used in different ways by different people, so we try to avoid the term. However, it is generally regarded the same as accuracy.
By design, measurement resolution is normally set better than accuracy. In the case of contact angle measurements, the resolution might be 0.01 degree but the absolute accuracy 1 degree. Why have resolution better than accuracy? One reason is that often accuracy is limited by the availability of calibration standards to show or prove the accuracy. If the best available standard is +/- 0.5 degrees, you can not claim an accuracy better than that. However, the ability to measure small differences between samples, whatever the absolute number, is useful and for this reason a higher resolution is provided. In general, FTA instruments can measure repeatedly the same sample with a variance of a few units of resolution. This is called the noise level of the measurement. If we report a resolution of 0.01 degree, we expect to measure a static sample with a repeatability of, say, +/-0.02 or 0.03 degrees.