I'm working on a project which requires at-least 2,000 ambient light sensors, and instead of using a microcontroler to manage and monitor the whole system I just want to amplify the output of sensors with a transistor array and use it for actuation of solenoids.
As a beginner I'm a bit confused, and I don't know whether I can treat each individual cell(from sensor to solenoid) identical to other.
Even the same component from the same batch from the same manufacturer cannot be expected to be identical.
The solution is to make your product/system able to cope with these small variations. When a certain accuracy is needed despite component variations, often a form of calibration is introduced. This means that each product is exposed to a certain amount of light (for example) and then you program it telling it that this is the switching level. You could achieve the same with a variable resistor (potmeter) which you trim to the required value for that product.
Then every product will "know" that level (you taught it what that level is, the product is calibrated) and then all products should perform the same.
It depends on the tolerances of each part, as guaranteed by the manufacturer. So are all parts "identical"? No. But as long as your circuit design allows for the min/max variation of each part, there's no reason why you can't do what you're suggesting. Note too that many parts come in various "grades", where essentially the same part is available with better tolerances, a wider temperature range, etc. All of this information can be found in the parts' datasheets.
SMD components are no different from any other. Within the parameters of the data sheet, you can expect every individual component to fall in approximately some sort of bell curve (or normal distribution). If the component is specified at +/- 5%, roughly 2/3 of the components might be +/- 2.5%.
Depending on the component, some batches might behave similarly (the compounds mixed into them clearly consist of batches, these may vary from day to day), and the machine calibrations might also drift over hours/days/weeks.
If you have only 10 components in a circuit, there is already a huge potential difference in performance between nominal and each component half-way to the end of it's specified range. You typically find that most examples of your circuit work, but a few percent are 'unlucky' and the tolerances can cause bad performance. A good design might compensate for the expected variations - or you might use an MCU to perform the compensation (generally far cheaper than specifying more precise components).
