Final product

Front view of the final product

We screwed in the 3d printed containers onto the metal frames for easy access and tidyness. The PCB and power bank is located near the back, while the display and controller is at the front.

Side view of the final product

Back view of the final product

Comparison of the dispensing accessory to manual weighing:

Procedure for the comparison of dispensing accessory against one human:

A total of three substrates were tested, namely salt, flour and baking soda. The timing taken, as well as the final weight, was also recorded. An overall procedure is shown below:

Procedure for the comparison of dispensing accessory against many humans:

The procedure from before is the same, just that instead of one person, data from several other people were collected. To ensure that they did not learn and get better by a significant amount, they were asked to only weigh three times for each substrate.

Results and discussion:

As seen in the table below, the dispensing accessory is able to reach within 2 mg of the desired weight most of the time, and >87% of the time, it is within 5 mg of the desired weight. However, the accuracy could be different, especially when weighing solids denser than flour, baking soda and salt.

Overall accuracy of dispensing accessory

The data was analysed and put into box plots, as shown below. The total time taken for the dispensing accessory is longer than the time taken for a human for all solid substrates. This is probably because when the total time taken for the dispensing accessory includes the time taken to prepare the sample dispenser (i.e folding the weighing paper, loading the sample dispenser and installing it).

Comparing dispensing accessory and human total time taken

However, after installing the sample dispenser, the user is able to do other things, unlike the human who has to focus on weighing samples at the analytical balance. As such, we chose to only compare the preparation time for the dispensing accessory with the total time taken for a human to weigh the substrates, as seen below. The time taken for the dispensing accessory and the human is around the same, which potentially makes the dispensing accessory a quality-of-life improvement, as the user does not need to go through the tiresome and vexing process of weighing substrates.

Comparing dispensing accessory preparation time and human total time taken

When comparing the accuracy, we can see that the dispensing accessory is generally closer to the desired weight as compared to the human. However, we also acknowledge that a human would be able to achieve the exact weight desired if given enough time. However, in order for the comparison between the dispensing accessory and the human to be fair, we counted the time and weight readings when the human reaches the desired weight + 5 mg (similar end point as the dispensing accessory).

Comparing dispensing accessory and human accuracy

In addition, the human would have gotten much better at weighing samples after a few replicates, and as such the time taken for the later replicates would get shorter and shorter. This is not a fair comparison, as it is unlikely that a human would weigh a particular substrate ten times in one sitting. To reduce this bias, we also collected more data from other humans, but made them only do three replicates, to minimize the time decrease from the humans getting practice. As seen below, the preparation time for the dispensing accessory is equal or slightly lower as compared to the weighing time of humans. Similar to before, the accuracy of the dispensing accessory is also better than humans.

Comparing dispensing accessory and larger sample of humans

As such, we can conclude that for substrates like flour, baking soda and salt, the dispensing accessory is able to mostly dispense within 5 mg of the desired weight. The preparation time as compared to the weighing time of humans is around the same, or slightly lower, and the dispensing accessory is able to achieve a higher accuracy, at least when setting the completion conditions to be the same. To make a more definitive conclusion, a larger range of substrates have to be used, in addition to a larger sample size.