Thank you again for the very thoughtful and useful comments. I really appreciate them. I will try my best to answer everything:
- Yes the mechanical switches are the weakest points for dust and rain, but everything else is also not perfect (touch button, display bezel, and a little bit the case itself). Making a waterproof version is definitely planned for the future, but to prevent costs from exploding I had no chance to do that in the first version. Waterproof will mean a more expensive case, probably an integrated battery with waterproof charging connector, a custom membrane switch panel or similar, and waterproof display bezel. The per-unit cost certainly goes up, but more importantly for me the initial investment becomes much higher for custom parts. I am happy to risk my time developing this, but I don't want to also risk also some tens of k$ at the very beginning. If people buy it I will invest in improvements.
- There are a lot of options for doing on/off and other things with a tactile switch. Here is one example:
&ab_channel=EEVblog ...in fact an earlier prototype of mine used something comparable to that, tactile switches only, but I was not very happy with it. It was a lot of extra components on the board, and for the user it meant pressing the button the correct amount of time - just one more thing to think about. The mechanical slider switches are a few cents more but I like the simplicity. Many people who gave feedback said simplicity is most important. The 6150, for example, can be a little complicated to operate (I also have experience with it).
- The buzzer is independent from the microcontroller. No real alarms or battery warnings, but if the count rate gets high enough the buzzer becomes paralyzed (always on) and this gives a different sound, a bit like an alarm. It is far easier to keep the timing/counting of the microcontroller reliable and accurate when the buzzer operation is separated. For battery life I plan to add a meter which is shown on the screen.
- There is no flashing LED to indicate each count. I think if a person can look at LEDs flashing they can look at the screen. I plan that in case of sharp change in count rate the screen reacts in 1 second or less, I cannot imagine the brain can process LED flashes much faster than that, or that it is necessary.
- The screen is a significant part of the power consumption but not the majority. Still, having a "screen off" option is a very good idea, I think. Originally I had in mind only power on/off and sound on/off (zero other options), but now with my "display change" button recently added one can change units etc., so maybe it makes sense that one "display option" is actually "no display". I am thinking now that the display change button goes between a list of modes: uSv->mrem->cpm->analog meter->cumulative dose->OFF->then repeat from beginning
- This "two color" display was an accident. I saw "two color" and thought: OK why not, I can try that and see if I like it. What I did not know is that it is not realy "two colors", i.e. two colors for every pixel... each pixel has one color, at the top it's yellow and the rest blue. As I said before I changed 0% of the firmware before I took the twitter video, I just plugged in the bigger display and by chance it displayed something in a reasonable (not optimal) way... at the same time, just by chance, the meter at the top was yellow and the rest blue - it looked like I did it on purpose, but it was just chance. I will be using a single color display later to keep things simpler. I don't think the two colors affect the power consumption significantly.
- Yes making the unit letters smaller is important and I will do that. In the old smaller screen it was hard to make any letters extra small because everything was already small. In fact, at that time I was thinking to write uSv/hr permanently on the case and have on the display just a bigger number (at that time no option to change units), but I switched screen size before I did that.
- I think the analog meter is a very good suggestion, as you saw I had something like it on the top already, I think having one display option with a large one is a good idea and I will try it.
- Even with DIY/arduino solutions there is nothing close to $100 except using a traditional Geiger tube. The core functionality of my device is 5x more sensitive than a Geiger tube. If someone does not think that's better they can buy something different. I think it's better. All other features are in my opinion superficial (except data logging which can be important, but all evidence I have so far suggests a small percentage of people care about that).
- If someone wants to do "the same" they cannot do it with a glass Geiger tube - they would need 5 tubes or more on one device
- I agree "harder to break" is not the most important point - still, they do sometimes break. I have seen this in some amazon reviews (sometimes even from the shipping process).
- The Radiascan 701 does seem to have some nice features. It also appears to be around $400. If size is the most important feature, and SBM-20 is sensitive enough, then there is also the MIRA detector (AKA Soeks 112), it is incredibly small, almost the size of an SBM-20, and around $150. Mine still fits in a pocket, just a big tighter.