"I lost an order to a cheap counting scale. You know why? Because its repeatability was better than ours." "It was a lot easier to demonstrate the accuracy of a counting scale with FY. All I have to do is to put a 10 gram weight as a sample, then put 1kg and show the result. It consistently shows 1000 pieces without fail." Once in a while I receive questions like these from the field. I have come to realize there exists some complexity about a counting scale after having heard such stories several times. When you have two scales; one scale that reads in 0.1 gram and the other that reads down to 0.01 gram. Let's call the former one Scale A and the latter, Scale B. Suppose you conduct five measurements with each scale, and have gotten the results as below.

Scale A: 10.1 g, 10.1 g, 10.1 g, 10.1 g, 10.1 g

Scale B: 10.11g, 10.11g, 10.12g, 10.13g, 10.12g

Which scale do you think is more accurate? Scale A has consistently shown 10.1g, and it repeats, yet nobody would say Scale A is more accurate than Scale B. With Scales B it is safe to say the weight of the object is 10.118g in average and its accuracy is +/- 0.02g. What can you say about the results of Scale A? All you can say is the weight of the object can be between 10.15 gram and 10.24g at best because of the limitation of its resolution. Nobody is going to be fooled to believe Scale A is more accurate than Scale B because of its repeatability. Simple, isn't it!

However, in the case of a counting scale, people tend to be fooled and get confused about accuracy with repeatability. One reason comes from the fact the measured weight of a sample or unit weight becomes a denominator, when the scale calculates total pieces, which makes it a little difficult realize how the resolution of a scale affects the accuracy of counting, that is, people tend to forget the fact a calculated unit weight carries the same inaccuracy arising from the limitation in a scale's resolution as was obvious in the case of simple weighing.

Let's say the above measurements represent a sample weight of 10 pieces. Then put 1kg weight and convert it to counts by using the unit weight registered as denominator.

With Scale A the counts will be 990 pieces for all five measurements, while Scale B shows 989, 989, 988, 987, 988. Now people tend to think Scale A is more accurate because it shows the same results consistently, while Scale B shows a variance of +/-1 count while its average shows 988 pieces.

Now I hope you can see where the pitfall is. As in the case of simple weighing, the conversion to counting carries inherent inaccuracy resulting from the accuracy limitation in weighing of a sample. Therefore, Scale A's counting results should be interpreted as 990 pieces within the accuracy of +/-4 pieces, since you will get 985 when the sample weight of 10.1g was obtained by rounding up 10.05g, and you will get 977 when it was obtained by rounding down 10.14g.

Therefore, the counting results which Scale A gives, though it shows consistently the same results, are only accurate within its limitation of weighing accuracy or resolution, which is +/-4 pieces in counting. However, even with this logical explanation, we face the fact customers are not totally convinced, since the repeatability is there to see with Scale A, while Scale B requires your explanation why it may not repeat.

To resolve this dilemma, I recommend that you conduct the following demonstration. Prepare a piece of paper. Your name card is fine, and cut it to a small piece so it weighs around 0.03 gram. Then repeat the above experiments. This time you add this piece of paper to the sample weight. With Scale A the reading still will be 10.1g, which you should have tested ahead of time and make sure it be the case, while Scale B will show 10.14 or 10.16g. Proceed with the counting operations, then Scale A still shows consistently 990 pcs or the same results as before, while Scale B responded to the fact it has heavier unit weight and shows different counting results. Then ask your customer which counting scale is more accurate. If your customer still thinks Scale A is more accurate, then you may simply offer a SK scale with a pocket calculator, and move on to other customers.

You can win more orders for our counting scales if you can count on your intelligence.

You may address any comments concerning this editorial by email to Mr. Eto

Index of Mr. Eto's other articles