We are often asked to provide an analog output signal for a wide operating range project. While we are happy to accommodate this request, there are certain realities that must be considered. When you select an analog signal, you are setting the span to the project’s full scale value and the voltage or current level produced at each flow rate will then have a fixed magnitude of resolution. No matter how long you hold a flow rate and observe the analog level, you cannot improve on the resolution or reduce the uncertainty of the information.
In contrast, if you are using a pulse generating system, you can extend the observation period to increase the certainty. For example, if you are collecting a 100Hz signal and wait for 1 second, you can calculate your flow rate to 100 pulses +/-1. This is based on the fact that you may have just missed the next pulse, or caught the lead edge of the next pulse. This +/-1 count of the right hand digit cannot be overlooked. Now, if you wait for 10 seconds and collect 1000 pulses, your +/-1 count is a 0.1% uncertainty. If you would like to further reduce the uncertainty, you could wait for 100 seconds and be able to calculate the rate with a 0.01% uncertainty.
Returning to the fixed uncertainty that exists in an analog signal. When you set 10.00 volts equal to 100% of the flow rate, the uncertainty in the right hand digit is equal to 0.1% at the full flow rate ( 1 part in 1000). When you drop to 10% of full flow, the output is 1.00 volts and your uncertainty is now 1%. At 1% of full flow, the uncertainty rises to 10%. The meter may be accurately parsing up the flow and spinning at exactly the right speed, but the output resolution cannot report the flow beyond the right hand digit.
Valve testing and real time data gathering may require that you pass rapidly through a range of flow rates. Stopping at each flow rate and collecting a statistically valid sample may not be practical. So there is a place for analog signals in flow testing, but their answers can only be as good as the resolution of the data. Please keep this in mind when setting pass/fail criteria.