I
got (almost) all I need to build the actual meter (save the voltmeter, it
should take quite longer to come, not really sure why). I got the LDR sensor,
so the first thing I did was to build some kind of enclosure and light diffusor
around it. I just recycled a piece of black plastics for the base and the
middle of the (Fuji) film canister cap for the diffusor. I drilled two holes in
the base for the sensor leads and painted the inside of the cap with white
opaque nail polish (taking care to make an even layer). When the nail polish
dried, I glued the sensor on the base and then glued the white-painted cap
onto. I know, it is not exactly a dome-shaped diffusor like in commercial
meters, but probably (hopefully) will do the job more or less in the same way.
It is more like a »hybrid incident light adapter« between the dome-shaped and
the flat diffusor (the ones used to asses the contrast ratio). See photos
below.
|
Starting materials for the sensor: light-dependent resistor (LDR), black plastics for the base, white nail polish and a Fuji film canister cap. |
|
|
LDR glued to the base, and the cut mid-section of canister cap painted inside. |
|
The sensor assembled on the breadbord (don't mind the resistors nearby, they remained from a previous project). |
|
The
so-prepared sensor was ready for testing! Unfortunately I came home quite late,
so I catched the last sun rays. There wasn't a 15 EV intensity anymore, but
only about 13.5 EV. Then, I measured the
response-resistance down to about 4 EV at different values. I then plotted the dependence of LDR
resistance against light intensity (EV). The outcome was quite a nice
exponential curve (as it should be) with a very good correlationship.
|
The testing rig: multimeter measuring sensor's resistance and the Minolta exposure meter for getting the actual EV value. |
I
then used the obtained formula of the curve equation to calculated the
predicted resistance at a given EV value (also for the points I did not
measure). Then, the calculated resistance values served to calculate an
appropriate series of voltage values to be obtained between Rx and R1 (voltage
drop across R1; see the previous
post). For that purpose I used the first part of the formula:
UOUT1=
(UZ2 * R1)/(Rx+R1)
Where
UZ2 is the voltage of the Zener diode (supplying the voltage to Rx
and R1), Rx is the value of the LDR and R1 is the chosen resistor value.
Now,
I must confess, I wasn't really picky about the Uz and R1 values, but I tried
to match them to what I have at hand (and/or combining various values), but
anyway, I wanted to get satisfactory results, at least. So for Uz I chose a
Zener diode with voltage drop of 3 V and for R1 I chose the value of 3200 ohms
(3k+2x100 ohm).
I got this, quite a linear curve:
The
curve equation now tells me that if I want to get the output of about 10 mV/EV I
first need to add (offset) 1290 mV to this (voltage) signal and then divide it
by a factor of about 28.4. Very luckily to me,
1290 mV is quite exactly the voltage drop of 2 regular diodes connected
in series(cca 1.3V)! This is not necessarily the case, but luckily for me, it
was. Otherwise, I would need to use another Zener diode and a trimmer to adjust
the offset voltage, in a slightly different circuit arrangement. Using a
different LDR and light diffuser would
certainly yield different values and curves. For the voltage divider I didn't
use exactly the factor of 28.4, since the calculation gave too much shift from the theoretical
values, especially at high EVs (where the meter is used mostly). Given my
resistor choices, I opted to use 1267 ohms for R2 (1k+220+47 ohm) and 47 ohms
for R3. This gives a ratio of 1:29.57. By applying this ratio and the voltage
bias of 1300 mV (two diodes), it gave me the following (calculated) measurement
error at different values:
At
first, it doesn't look like nice. But, we seldom use a meter below 6 EV (very
dim light), and the error of around 0.3 EV is totally acceptable in practice
for cameras and meters alike. Only between 11 EV and 13 EV the error is quite
large, but as long as we know how much the error is, we can always correct for
it. But clearly, all this is still theory only. The practical measurements will
tell how good or bad the meter is.
Anyway,
at least I came up with the final version of the circuit, with resistor and
diode values to test, and hopefully, solder into the circuit board. See below
(this is only the signal part of the circuit). But let me stress once again: this circuit is (should be) suitable for MY very own case of sensor, not necessarily (or likely) yours!
|
The signal part of the circuit I came up with. |
The
diodes D1 and D2 are just ordinary small-signal diodes. The Zener diode, as said,
has a drop of 3 volts, while the
resistor Rz has been set arbitrarily at 4.7k, just to get a current somewhat
higher than 1 mA, (at high EV values current can approach about 1 mA in this
configuration). Voltage drop across Rz is 7.7 V (12-3-1.3), divided by 4700
yields about 1.6 mA. The photo below shows a more general (and probably also more appropriate) case of a signal circuit-using another Zener diode and a trimmer potentiometer to adjust the bias voltage. The latter is probably the largest source of measurement error in such a meter:
|
A more generic signal circuit. |
During
the weekend I'll test the sensor and circuit on the protoboard (»breadboard«),
and I am quite anxious to get the results, which I'll promptly report to you. If time will permit, I'll also get in the final construction till
next time.
Silver
regards
Mitja
CORRIGENDUM: While the circuit in the penultimate photo (without the trimmer pot.) is in principle OK, your restless editor forgot for a moment a basic aspect of Ohm's law, and a vital coefficient....Therefore, the correct values for R2 and R3 are 1.267 M ohm and 47 k ohm, respectively. I apologize for that.