Considering the broad interest this thread seems to have captured - I thought the following might be appropriate to throw in....
For the casual reader, it should be pointed out that there are two very distinct classes of tests that can be performed on the Batteroo sleeves.
The first is the characterisation of the sleeves themselves.
This is where power is provided by sources such as laboratory power supplies and dummy loads are used to draw power, just like a keyboard, clock or toy train would. With these types of tests, there are several benefits:
- there is great control over the parameters available, so you can set very specific conditions and you can see how the Batteroo sleeve performs on each one. You can have input voltage levels like 1.0, 1.1, 1.2, etc. or 1.00, 1.01, 1.02, 1.03, etc. It just depends on how detailed you want your data.
- these conditions are stable. You might want to see exactly how a sleeve performs at a point where a battery is dropping very quickly. With lab equipment, you can set up any number of such points and keep them stable while the measurements are taken and recorded
- these conditions are repeatable. You can do them any number of times and there should be no variation. If there is, you will need to look at your setup to work out why. Using batteries for this type of testing makes repeatability with certainty practically impossible.
- because this type of testing is stable and repeatable, you can test a number of different sleeves to see how consistently they perform when compared to each other.
- lab style testing allows very precise measurements to be taken - particularly in the proper use of the measuring equipment, so that voltage drops along wires and through meters are not being included in the measurements being taken. (These can be quite significant, especially at 1.5V battery levels, so avoiding this trap for young players is vital for accurate results.)
All of this will end up with a mass of numbers. Numbers which can be used to create graphs and charts and all sorts of observations that electrical engineers drool over. They do this because these graphs will show exactly what the Batteroo sleeves are capable of. They will be able to tell you if the sleeve can deliver enough current to power a particular toy train. They will be able to answer a multitude of questions from those tests about how the Batteroo sleeve will perform in various situations.
As much as people might go "Well, goody for you." the fact is, that if a chart generated from testing a sleeve says it won't work in 'X' conditions ... then it won't work - and if it will, it will. Just remember, these are the same engineering practices that put man on the moon, a laser in your pocket and the internet you are looking at right now.
The second class of testing is in real world applications.
Yes! This is what it's all about - and you might be tempted to say "Why carry on with all that rubbish you just talked about when this is what matters?" That is an understandable view - but I like to think of it like this: All the characterisation testing produces a kind of a map of the terrain and in a real world application, you are going to follow a path across that terrain. That is often very helpful in planning the trip and assessing the problems if something went wrong on the trip.
The big challenge, however, is that with this kind of testing - especially for battery powered equipment - is that there is a big unknown ... the actual condition of each battery. Even brand new out of the pack, there is no guarantee they will all be the same. Fortunately, they do tend to be fairly consistent - but even then, if we get a weirdly different test result somewhere along the line, we cannot exclude the battery from suspicion. This challenge is made even more complex because we can be talking about batteries that have already served for a time, making their condition even more of a mystery.
While some real world tests that have been carefuly constructed (to provide as many consistent parameters as possible) will be able to produce some very useful results, they will not cover each and every possible situation. That will take tens of thousands of tests out among the public to start painting such a picture.
But even then, test results may not be really accurate if, for example, a critical parameter was not noted during these tests. "What sort of parameter?!!" you might demand. I'll offer this as one: Temperature. It has a rather significant effect on battery chemistry - as I'm sure a lot of people already know.