Define what the error compares
A delivery comparison needs two quantities from the same test: the volume commanded and the volume actually collected. The signed difference says whether the collection is above or below that command. A percentage expresses that difference relative to a chosen reference. Without a clear event boundary and denominator, a precise-looking error value may compare unrelated numbers and mislead a calibration decision.
This educational article concerns pump-delivery arithmetic. It does not select a supplement amount, define a universally acceptable accuracy or supply a firmware correction factor. The Aquarium Dosing Pump Calibration Calculator includes an optional same-test command field so it can make this comparison explicitly. A timed flow observation remains useful without that field, but it cannot invent a command that was never issued.
Use the commanded volume as the reference
Let M be measured collection and P the positive programmed volume for that event. The signed difference is M − P, and relative error is (M − P) ÷ P × 100 percent. For an invented command of 20 mL and measured collection of 18 mL, the difference is −2 mL and the relative error is −10 percent. The negative sign means the collected amount is below the command.
Dividing by the measured 18 mL instead would give about −11.11 percent, which is a different comparison. It is not the error definition used by this calculator. Keep the denominator visible in the worksheet rather than describing either result merely as percent off. The value should identify its reference so another person can reproduce it without guessing the intended convention.
Interpret positive, zero and negative results
For the same 20 mL command, a 22 mL collection gives a +2 mL difference and +10 percent relative error. A 20 mL collection gives zero difference under the measurement's resolution. That zero does not prove exact delivery to unlimited decimal places. The collection device and procedure still determine what differences could have been observed, even when the entered numbers happen to match.
A zero or negative programmed volume is not a valid denominator for this comparison and should not be entered as a command. A blank optional field means no same-test comparison is available; it does not mean zero error. Avoid replacing a negative sign with an absolute value unless you explicitly need a separate magnitude-only metric. The direction can matter when explaining an observation to support.
| Collected volume | Measured minus commanded | Relative error (command denominator) |
|---|---|---|
| 18 mL | −2 mL | −10% |
| 20 mL | 0 mL | 0% |
| 22 mL | +2 mL | +10% |
These values describe collection arithmetic, not an acceptable tolerance, chemical makeup dose or firmware calibration factor.
Separate relative error from a hypothetical correction ratio
An observed −10 percent delivery error does not imply that adding 10 percent to every command will perfectly correct it. Under a purely hypothetical constant proportional response of 0.9 times the command, recovering an intended amount would require dividing by 0.9, a factor of about 1.11111. Multiplying by 1.10 would produce only 0.99 of the original target in that simplified model.
This algebra explains why percentage shortcuts can fail; it is not a device-setting instruction. Real controllers use model-specific calibration procedures, and a single collection does not establish constant proportional behavior. The calculator deliberately reports the observed difference without outputting a universal firmware multiplier. Follow the exact device instructions rather than writing an inverse ratio into an app field whose meaning has not been established.
Check flow and command error as separate outputs
If the hypothetical 18 mL collection took 60 seconds, its observed flow is 0.3 mL per second or 18 mL per minute. A chosen 6 mL volume projects 20 seconds at that rate. The flow uses measured volume and time; the −10 percent command comparison uses measured and programmed volumes. Both can describe the same test while answering distinct questions.
The chosen desired volume is not the programmed test volume unless the actual event makes them the same. Entering a daily supplement amount into the optional command field for a separate calibration cycle would create a meaningless comparison. Preserve the three concepts in the record: measured collection, actual test command and independently selected future delivery quantity. A shared milliliter unit does not make them interchangeable.
Assess repeatability before calling the difference a stable bias
Suppose three comparable 20 mL-command trials collect 18, 20 and 22 mL. Their mean is 20 mL, so the mean signed error is zero, but the individual errors are −10, 0 and +10 percent. The range is 4 mL. Reporting only zero mean error would hide variability that a single proportional correction would not resolve.
By contrast, three collections all near 18 mL suggest a different observational pattern, though they still do not identify the cause or establish a formal accuracy claim. Keep raw values, measurement conditions and device instructions together. Do not combine different channels, liquids or tubing configurations into one average. Repeated numbers are informative only when the tests genuinely address the same defined delivery question.
Keep measurement limits and safety decisions separate
A measuring container with coarse graduations may not support the small percentage difference the calculation displays. Rounding the inputs to convenient whole milliliters can also hide or create apparent agreement. Use appropriate equipment and the manufacturer procedure, and report the result at a sensible precision. More decimal places in the percentage do not reduce uncertainty in the collected volume.
No error percentage in this article establishes whether a chemical addition is safe or whether a pump is acceptable for a particular application. Product concentration and aquarium needs are separate. Do not add a calculated shortfall to an occupied tank or attempt to remove a calculated excess without appropriate guidance. A delivery comparison describes a measurement event; it is not a chemical correction plan.
Make the comparison reproducible
A support-ready record includes date, model, channel, operation type, programmed test volume, measured volume, elapsed time and collection method. Note the permitted line preparation and whether the liquid was collected outside the aquarium using the prescribed procedure. Include any repeat tests individually. This makes the signed result an auditable observation instead of an isolated percentage copied from a result card.
Use the calculator to verify the arithmetic, then follow the model-specific calibration process when appropriate. Recheck under the permitted procedure after any supported adjustment and retain both records. The reliable endpoint is not a universal correction factor; it is a clearly defined command-to-measurement comparison whose limits are understood, with chemical dosing decisions and equipment suitability assessed separately.