Impressively powerful floodlight, thanks for the review.
“providing up to 3,000 lumens of output with three hours of run-time”
This is not really true, is it? It steps down to 1800lumens for the majority of the 3 hours.
While I like the detailed runtime tests/graphs and it is nice to know the details of how the light steps down, the vast majority of users would not use the light like this. Almost always the light would be used in relatively short bursts. For this reason I would like to see how bright the higher modes are at different points in the battery discharge. For example, with 50% discharged cells how bright would turbo or turbo s modes be? This is not a criticism of your review in particular, just an observation that most reviews lack this sort of information.
Lack of information? I don’t see it at criticism. In fact, I have this test on its way.
However, consider the following:
- how do you identify the 50% level of a battery?
- if you can identify that voltage, how would it be compared to the ones of other batteries?
- the result you get from your test, done at RT, can be the same when the cells are 50%… but have been charged quicker? or you are in the freezing snow?
all the above, how will it compare to other batteries? If you have 3500mAh that are able to get turbo only above 50% charge, would you rather run 2100mAh that are able to get turbo only above 20% charge?
You just CAN’T have enough information.
After the runtime at mid mode, I’ll run the “50%” test.
You are welcome.
Also, the on-off test is very very tricky.
Simple to do, but how do you do it?
at Room temperature? with a fan?
how well does the light in between these conditions?
Yes, the users does on-off many times, but how many times do they on-off at RT, on a table? or in front of a fan? in winter? or summer?
what will happen in winter, where you get extra heat removal, but lower battery efficiency? And vice versa?
with a temperature sensor the interpretation that you can get from the data, is very “situational”.
What I would do is do a runtime test, but in ~5min (or 2min) intervals, allowing enough time for the light to cool between intervals. This would simulate more closely actual usage. Of course, you are right it might perform differently with different cells.
I don’t like the 5 minutes, I prefere to do it closer (30”).
Again, I like to make the “worst” situation, so you should at least expect something better.
Consider also this. I can make the test , but my test are run with the light sitting on a table.
If you keep your hand on the light, you actually make it cooler.
Yesterday 9/21 was on my calendar (unspecific to location) said it was the Autumn Equinox, but in my location it will not occur until 9/25; four days later.
A good example of relativity in a flashlight test, as Budda has just written, is that there is a difference between holding a light with your hand and having it mounted.
All these variables have an impact on test results. This is also one of the reasons that car mileage tests are done in laboratories. The side effect is that unrealistic scenarios where ridiculous low acceleration values are used etc. The outcome is a set of unrepresentative data. It’s hard to come up with standard test methods. I’m very happy with the runtime graphs produced here, and the only additional interesting thing I would like to know (asked Maukka yesterday) is whether the full 9000 lumens can be achieved with semi depleted cells (answer: no, only about 4500 lumens). Anyhow, thanks a lot Budda for the hard work!
