Thanks all who replied; all answers are appreciated, though most were not to the question at hand. I was just asking if the top and bottom scales are suppose to match or not. The answer consensus seems to be yes. Logically, if the light intensity hitting the meter is so many LUX, it must show identical number on whichever scale you choose, so you set [the same] exposure based on whichever scale you choose. Not the case with my meter.
I wasn't asking if occurring mismatch between readings is due to the wrong battery or whatever else the reason might be. All I wanted to know if mismatch is by design or something is not right.
My understanding is that since top scale ends with 12 and bottom scale just starts with 12, you switch to bottom one only if pointer on top scale overshoots beyond12, e.g. it's too bright for that. Similarly, if for low light the needle on top scale shows *anything* below 12 (like 11), the bottom scale must show dead zero since it starts from 12 on, and the light intensity is only 11, so it's not suppose to register ANY movement on the bottom scale. Do I understand this right? However, it is not really possible to make top scale over-range. When the light is very intense such that it shows near 22 on the bottom scale, it is *barely* beyond 12 on the top scale, which is full 10 stops off! I think there is a movement limit implemented restricting over-range and slamming pointer to the right when top scale is chosen and the light is too bright, but it starts way too close to 12, like "soft saturation" and non-linearity starts beyond 10 on top scale. Basically, sensitivity between 10 and 12 is so compressed, that it cannot be trusted. 10 on top corresponds to 11.5 on bottom, 11 on top - 13-1/2 on bottom, and near 12 on top is anywhere between 16 and 20 on bottom; as i said I can't quite reach 12 on top no matter how bright the light is, as if a Zener diode limits galvanometer movement beyond 12. My question was if that's by design or my unit settings got drifted that far and this can be fixed by (re)calibration. And yes, 'course I don't switch between scales in one second and make a judgement, I waited long enough to settle (up to 15 sec for top scale). The battery I use is two 357 Silver Oxide button cells with a Si diode in series dropping total voltage to exact 2.7V value, which makes battery test pointing smack in the middle of the red mark (see my cells above the meter on the attached photo - the diode is soldered electrically between them). But being an electrical engineer, I've also tested the meter with variable power supply, dialing above and below 2.7V and see how far I can deviate and whether voltage value makes a difference when changing scales. For my particular meter the pointer is at the beginning of red zone (17) at 2.64V and at the end of it (17-2/3) at 2.95V, so this must be acceptable spread of the battery voltage. There is no internal voltage stabilization circuit, so these 2/3 of the stop "length" will be your nominal error when the battery is within "normal" red limits. This also tells me that the internal circuit is purely passive one. Another firm conclusion is you can use ANY chemistry battery (alkaline included) AS LONG AS you manage to keep its voltage within 2.95V...2.64V limits while it's being discharged. As with any analog multimeter, the voltage affects the top scale the most. Changing it from 2.7 to 3.3V changes readings just half-stop (12 to 12-1/2), but at the top with light intensity resulting in 10 at 2.7V it will read 11 at 3V (full stop difference) and 12 at 3.3V (two stops difference), but that's consistent, and otherwise no adverse effects on the meter whatsoever. Meaning you can feed it with two 1.5V cells in series, no problems, the battery test points just to the end of red mark at that voltage. A single 3.3V lithium cell would not be a problem as well if you're willing to make mental adjustments - it reads 1 stop higher at the mid scale and 2 stops at the end, sort of "expanded" scale, but that's all. The point of my long rant is that the battery voltage being 2.700 V sharp is not critical at all, and those who think otherwise unnessesarily restrict themselves in choices of alternative power sources for their meters.