Luna Pro exposure meter - please confirm my understanding

[chuckroast]

Yes, this device is electrically just a diode in series, connected to the negative side. I read this from
someone's review, who bought the adapter from B&H.

Yes, it drops the battery voltage by *approximately* fixed voltage amount. This drop on the forward biased
diode will slightly depend on the chosen measurement range (more for high range), and is also different
(bigger) for the BATT CHECK function. But the biggest drawback is the output voltage [decline] will
track cell voltage decline, and that curve, unlike for mercury cells, is not flat.

Finally, the adapter's (as any diode's) output is highly temperature dependent. If you can live with these
limitations, this is OK solution, though it is far cheaper to install the same diode inside the meter, which
then will take any two stacked CR44-sized cells, or better - a single CR1/3N cell without any adapters.

Thanks for the answer.

Not sure what you're saying here.

The MR9 adapter will take 1.5V silver oxide batteries that will bias the diode reasonably well without any huge loss of calibration accuracy.

Moreover, the silver oxide batteries will fail more-or-less suddenly just like a mercury cell.

I have worked on many of these meters and have found that either an internal recalibration or a silver oxide/diode
combination works about equally well. Since silver oxide 625 batteries appear to no longer be available,
for those meters I have recal-ed for 1.5 volts, I just use physical MR9 style adapters without a diode and stick silver oxide 386s or SR44s in them, depending on the application.

There is also someone selling small voltage regulators that can be installed internally in the meter to hold the internal VCC at 2.7 volts, but that does involve tearing them open and some soldering.

Let us keep in mind that these meters were accurate to about 1/2 stop best case and were decidedly nonlinear from the top of the low range to the low end of the high range.

All that is to say that - at least for monochrome, and possibly for color film too - it's close enough. As always, repeatability is more important than absolute accuracy.

 

[336v]

The extreme ends of the scales are somewhat nonlinear on these meters. The top of low range will not perfectly match the bottom of the high range even when correctly calibrated.

It does on mine - amount of light deflecting pointer to "12" on low scale, deflects it to exactly 12 on the high scale after I've calibrated it.
12 is the only common readout for both scales.

Non-linearity within each scale is taken into account by offsetting tick marks on the scales - you will notice that beyond "6" tick marks no longer
coincide, one on top is skipped, and the high range bottom scale looks squished. The same deflection angle has 9 gaps between tick marks
on top, but 10 on the bottom. This graphic "fix" is not strictly ideal, but more than accurate enough for practical photography.

Gossen knew this when designed the thing.

 

[BrianShaw]

...

All that is to say that - at least for monochrome, and possibly for color film too - it's close enough. As always, repeatability is more important than absolute accuracy.

It is true for color, including color positive film. I think of it this way: Gossen (and probably all other light meter makers) had experienced and knowledable EEs on their design and manufacturing staffs, plus these meters wouldn't have lasted in teh market as long as they have if they weren't up to the vast majority of photographic tasks. Indeed, they knew what they were doing.

 

[Bill Burk]

Learn Camera Repair site is run by generous people who faithfully gathered manuals. For a long time they were free but they recently changed to a pay-as-you-go site. This is to distinguish them from scam sites that turn up in every Google search result.

I put a 2.7 v regulator from Filip Dee in my Luna Pro and it immediately acted as if calibrated. So I didn’t touch the dials inside.

 

[336v]

Instead of the SR44/357 battery I use the SR-43/386, where the discharge curve seems quite flat and the meter performance seems quite stable. The minor height difference between the two cells doesn't seem to matter. Even with the less-flat discharge curve of the 357, I'm not sure it matters much given the "allowable variance" of most photographic applications.

From the Renata spec:
View attachment 398977

Yes, if this curve to be trusted, the SR43 is quite a bit more superior to SR44, so for these battery non-linearity
of the curve can be taken out of discussion. Only the diode's limitations remain. Good find!

 

[Dan Fromm]

Yes, thanks, I've seen this one before, but looked closely now. The adapter looks like 3D-printed container,
one would think Gossen frankly could come up with injection-molded one.
On the electric side, It has no description of how it works, based on the images, the positive end of SR43
stack touches the metal cap of the meter just like original battery did.

This means if it reduces voltage, it does something on the negative side, most
likely integrates a series diode in the casing. The reason I'm saying this is in order to have active LDO
regulator you need a reference ground, e.g. solid negative connection to the meter, which this adapter
does not have - bottom (neg) side is insulated. That is, unless a -2.7V negative output regulator with ultra low
quiescent current is deployed here, but AFAIK, such a thing does not exist - the best device I saw has 3 uA
quiescent current, which will eat fresh SR43 in about 4.1 years. But it's possible Gossen still went with
such solution here - they don't say on this website.

Does anyone have this adapter to confirm if this is a passive or active device?

I have two of the long-discontinued original. Injection molded. 3-d printing may be less expensive and might allow production to order, reducing the risk of being stuck with unsaleable inventory. AFAIK the originals are passive. I've always thought they had a Zener diode. I used one in a LunaPro (= LunaSix III) and the other in a Canon AZ 814. Both worked. Both are now in a Horseman Optical Exposure Computer and still do the necessary.

 

[Chan Tran]

I found that using 2 resistors one in parallel and one in series to the CdS cell and playing with the values of these 2 resistors I can get the current relatively linear to LV (which is logarithmic). It's kind of an S curve.

 

[336v]

I found that using 2 resistors one in parallel and one in series to the CdS cell and playing with the values of these 2 resistors I can get the current relatively linear to LV (which is logarithmic). It's kind of an S curve.

Yes, you can do it, there are ways to re-shape the galv. response electrically, but keep in mind that based on the circuit any addition of either series or parallel resistors to CdS or existing trim pots
will either decrease sensitivity or increase power consumption (or both). The shape of CdS resistance to light response (which is log function, e.g. very straight line on a log scale, not S-curve)
is taken into account when graphic tick marks were silk-screened on the meter's scale plate. The S-curve you're referring to is caused by the decrease of sensitivity of the galvanometer itself near extremes
of deflection, it is the most sensitive in the middle, thus you see black swatches stretched between about 1/3 and 2/3 of deflection angle and squished toward the ends. This got nothing to do with CdS's Lux/R curve. Just keep in mind that loading the CdS with extra resistors, while will linearize needle deflection per each EV step, will invalidate this designed-in "graphical compensation", so overall accuracy near the ends of scale(s) will certainly be worse.