The standard type, with a white to vermillion round head. I have this project nagging in my head. My battery of power banks is very tightly packed and protected that while I can pull one out of the stack I need a lot of ef-for-T to LIFT any of them and see in the LCD screen how much it was filled when I leave before completely recharging them. When the power bank finishes recharging, the LCD display goes off; while the bank is charging it displays a percentage. So the idea is simple: use a sensor (I think this is patentable). But I do not want a one-to-one correspondence between sensor and LCD segments, nor need a fancy camera like sensor because it is more than one bank (incidentally, the top one has its LCD display broken, disconnected). Then the logic becomes a little bit devious... For one figure charges, below 10%, I can take the bank to be as good as empty, so I only have to check whether it turned on or if it is off, like when the transformer self unplugs. When it is charged all the way up to 100% I have only one meaningful digit, so I can ignore the other digits and only check for light on or off; the LCD remains on for a while after reaching 100% then goes off. I only have to detect the middle figure to be done. I assume maximum glow is reached at 88%, which gives me a figure on the 80s if I do not detect the first figure. 80% or above is OK to say the bank is charged. I suspect 8 and 0 may get diffusely confused, but it does not matter because there is no 0digit% output and the second position can only be zero if charge is at 100%. Then 1 and 7 may also be confusable but it does not matter because if it is 7x% it will soon reach 80% so I ll know in a few minutes if the charge was low in the 10%s because 2x% will not glow as much as 80%, or if it was in the 7x% range and already reached an OK 80%. 3 and 5 may be confusable among themselves, but 4 and 6 may be distinguishable if the sensor can discern one segment differences; also 2 and 4 may be confusable. In all these cases I only have to wait for so many minutes to detect the change in glow and determine on what percentage range the power bank is. I have to go empiric on this one, but for 20%/40% I can say the bank is in middle low charge, for 30%/50% I can say it is in the middle range charge, for 60% i can say it is in the high charge range and 70% is again a matter of waiting for a few minutes. Not much, since partial charges will fall on average on 5% so I have to wait only for a 5% change to determine the range and then increasing/decreases in glow can only happen in one way for specific x0% digits. In general I would not know exactly the charge level but this depends on whether the sensor sensibility can or cannot detect specific digits and I can safely ignore single digits because in terms of powering they make little difference, the real important distinction is betwen completely empty and some charge (I would at least be able to turn on a spent device in the worst power emergency).
The real problem is then not the circuit design nor the programming logic, but the physical installation of the sensor. I may use from two to four photoresistors, one or two for each digit, making a total of six to twelve arduino photoresistors. It would be straight if there was enough space above each power bank, but when stacked the standard photoresistor is just too bulky and may even break th LCD! Thinning them down to a slim wafer seems OK with some padding, which is needed anyway to control the glow input, but I am NOT SURE if the standard photoresistors would lose sensibility or even stop working altogether if I thin them out!! The effort is not simple, they are way too small to hold with bare hands and the power tool may be too forceful for a small clamp, but anyway... I know of no other photoresistor brand or type and I do have a few of these at hand. So I want to know if anyone has had the experience of if there are a priori consideration why this is or is not possible before doing the effiort.