15 Oct 2017
Some more work done on the code, check video.
Now the Grimlight has functionality to change lighting effect. The light is switched on and off by holding a magnet close for a short time. If a magnet is present for about 10 seconds on the other hand it changes effect. Currently three effects are implemented, flickering, steady and pulsing.
The controller remembers current state while in sleep, so it is possible to configure with the effect you want only once (if battery is changed state is lost thought).
08 Oct 2017
tl;dr. Changes done to increase stand-by time, should be about one and a half year before the battery is drained now.
Upon measuring the power consumption behaviour in sleep-mode it looked like there were some room for improvements after changing the code to deal with the issue with the turn on time for the sensor.
In that implementation the behaviour was to wake from sleep, turn on the sensor, do nothing for about 100 ms, check the sensor and finally go back to sleep if no magnet is present. The power consumption during these 100 ms not spent sleeping had large impact on current draw. The average consumption at 3 V was 11.5 µA during one wake-check-sleep cycle.
Changing the code so the behaviour was, wake, activate sensor and then go back to sleep for 64 ms to make sure the sensor has time to be activated improved matters some but not much. On a hunch we tried removing the decoupling capacitor (100 nF) for the hall sensor, removing this had a huge impact. It seems that during each cycle the capacitor was charged and discharged wasting power.
Removing the capacitor didn’t seem to affect stability. That change together with putting the MCU in sleep instead of just waiting brought the stand-by power consumption down to about 1.5 µA!
Also, see the Sample gallery for new images!
07 Sep 2017
We have had an issue with units automaticly waking from sleep when using code that switch off the Hall sensor when in sleep to save power. Today I took some time to investigate this instead of just using a workaround (not disabling the sensor while sleeping).
I had an idea that it might be that it takes some time for the sensor to output the correct value after getting power. This turned out to be correct but not in the way I expected. When the circuit is fed power at over about 3.1 V there is no delay for reading correct state of sensor. When the voltage drops to 3.1 V or under on the other hand it takes about 70 ms for it to output correct value.
In the below figures, C1 is the level at the Vs pin of the Hall sensor and C2 is at the output. Decreasing the voltage down all the way to 2.4 V (minimum working voltage for the sensor) there is no change in time to turn on.
21 Aug 2017
The PCBs for the final prototypes have arrived and been assembled, here are some picutres. Some footprints for parts need minor adjustment but otherwise this should be the final version. The circuit measures 21 mm in diameter and 2.5 mm at the thickest part (this will be reduced by 0.2 mm for the production version due to changing from 0.6 mm thick PCB to 0.4 mm thick).
17 Aug 2017
We are starting to spread the word about Grimlight on various forums and sites. We will update our homepage adding better quality photos and videos the coming week and we need your help in spreading the word!
The final set of prototypes will hopefully be in our hands soon and the dimensions will be smaller than ever. We look forward to sharing ideas and pictures with modelers, wargamers, larpers and anyone in need of our unique miniature lighting solution.
Would you like to get your hands on one of our prototypes? Anyone who signs up for our newsletter before October 1st will have the chance to get a prototype. We will draw a lucky winner out of all our newsletter subscribers on October 1st who will have a prototype sent to them by mail.