Switch chattering and low voltage MCUs

[pomk]

It seems that the batch of gateron blues I had was just very bad. I made another keyboard, using the same diodes and whatnot, and cannot observe the same behavior using the very same batteries. This time the board was built using plate mounted black bottom gateron browns, while the previous one was using smd gateron blues.

I should have time next week to do the oscilloscope measurements on the barely working blues, as well as with the perfectly fine browns.

I have however already committed to making changes to the power supply design, just to be on the safe side of things in the long run.

[beltet]

It seems that the batch of gateron blues I had was just very bad. I made another keyboard, using the same diodes and whatnot, and cannot observe the same behavior using the very same batteries. This time the board was built using plate mounted black bottom gateron browns, while the previous one was using smd gateron blues.

I should have time next week to do the oscilloscope measurements on the barely working blues, as well as with the perfectly fine browns.

I have however already committed to making changes to the power supply design, just to be on the safe side of things in the long run.

That's great to hear that you have made some progress!
If you can, with the oscilloscope you are going to use, please take some screenshots of the behavior between the switches.

[pomk]

Of course! Knowledge = power and all that

I also plan on launching a GB next week, or an interest check to be more precise.

[DMA]

pomk how do you save the battery power? Must be very important with BT.
Is just switching off the radio basically enough? Or you do some other stuff also? how many mW is saved by each of the steps?

[pomk]

I use the nrf51 chip from nordic. It has a very low power 'stop execution, wait for interrupts' state it can be forced to go to. In that state, if no bluetooth connection needs to be probed, the power consumption is a bit over 2µA if I remember correctly. Depending on what connection settings are forced by the server (the PC usually), the power consumption for keeping the connection alive is something between 10 and 110 µA. So just as long as you don't take long when you scan the matrix, send a packet and go back to the low power state waiting for the next scan timer going off, the power consumption will be pretty low. Note that not all MCUs have a low power state of that efficiency and you should be careful when choosing which one to pick. Nordic has now another mature chip, nrf52, which can be even more efficient, but I don't think that I'll bother with it for now, too much effort sunken to the previous design.

The bluetooth stack it has is a closed source binary blob, but it handles the radio power on/off automatically with good results.

btw, I opened an IC over at GH, in case you are wondering what this thread was all about really: https://geekhack.org/index.php?topic=86364.0

[beltet]

You seem to be done with your project but I think this debouncing video was quite interesting aswell.
https://youtu.be/Nj-Q8FQxHhU