Gossen Lunasix 3, although built around the early ’70, is a very capable hand-held meter even by today standards. It beats my Canon 20D built-in meter in terms of actual low-light performance by 5 stops. Unfortunately it belongs to one of these designs, that relied on mercury battery voltage staying constant throughout their life and mercury batteries were banned few years ago. This article describes, how to deal with this problem.
There is a battery chemistry available that can replace the now obsolete mercury batteries. Silver oxide cells have equally stable voltage output and are based on the same principle, as mercury batteries, the main difference being use of silver in place of mercury, which makes these batteries more expensive to manufacture. Since the amount of silver in these button batteries is tiny, the manufacturing cost difference isn’t really relevant to the sale price, but it shows why the mercury was used in the first place – cents sum up to dollars fast.
The Gossen Lunasix is a purely passive circuit, i.e. there are only resistors and the galvanometer in it, no diodes, no transistors, no fancy electronic circuitry. Light is detected using a CdS (Cadmium Sulphide) LDR (light dependent resistor) that directly influences amount of current allowed into the galvanometer. For details on such meter design see my article here (although the relevant PDF isn’t finished yet, the part covering mercury cell based meters is ready).
Being a purely passive, resistive, linear circuit it has no principal problem with a slightly higher battery voltage, besides showing wrong exposure values of course. As CdS cells are not made to very tight tolerances CdS based meters have to provide for a relatively wide adjustment range. Because of this, in most cases it is possible to readjust the meter for 1.55V silver oxide cells with no further hardware modifications. This, however, could require moving some of the adjustment pots towards their extremes and limit the adjustment precision, so as long, as you can solder two wires together, putting in a voltage-dropping diode is a good idea. It will also prevent any surprises with batteries put the wrong way around (the original 625 cells were mechanically keyed, the replacement SR44 aren’t, this isn’t really a problem as Lunasix won’t be damaged by reversed voltage polarity).
Disassembling the meter
- A flat screwdriver with a narrow blade (2-3mm).
- A sewing needle or a sharp pin might prove useful.
- A hair dryer or a nice sunny spot.
As with most Gossen devices there’s going to be some aluminum plate involved 🙂 In this case it’s the back plate of the meter. It has a hole in it, through which the serial number can be seen. Don’t get fooled: the plastic feet are part of the housing, they don’t hold anything in place and cannot be removed!
The plate is glued in, you have to pry it carefully off taking care not to bend it too much. It helps to warm the glue up, so either use your hair-dryer or put the meter belly-up in the sun (in either case remember to remove batteries first!). Don’t overdo the heating thing, or you will deform the meter housing, or, worse, damage the movement inside. Get it pleasantly warm to the touch, but not very hot! Use a needle or a sharp screwdriver blade to lift the plate through the serial number opening at the back.
Under the plate you will find six screws (see photo to the right, above). Remove four: all that are outermost at the corners (in one corner there are two screws, remove only the one closer to the corner). The upper half of the case will lift off. Inspect carefully the front part where the sensor is.
There will be, from the outside towards the inside:
- A black plate with the incident metering dome.
- A small piece of glass window (it makes the impression of being IR-cut filter too, but I’m not sure)
- A sliding metal plate with silverish ND6 filter mounted in it
The last element in the list you have to be extremely careful about: it will likely spring out if you try switching the meter on. If you lose this part or damage it, it will be extremely hard to replace or repair.
Putting a diode in
Tools and parts needed
- Soldering iron, some solder and a little flux (or colophony).
- Side cutter.
- Pretty much any physically small diode you can lay your hands on. Do not use germanium diodes, they have too little voltage drop and you would need two of them (remember, this meter ran on 2 cells, so we have now 2*(1.55-1.35) = 0.4V to lose.
Inspect the left-top side of the meter. You will notice back of the battery compartment and two wires coming from it. The negative pole goes to the corner of the board, while the positive ends somewhere around the center. Solder the diode into the negative wire, the small bar on the diode housing shall point towards the battery compartment (see photo to the right).
Insert batteries and activate the battery check (no need to reassemble the meter at this point). The needle should stop now within the red field or very close to it. If your meter was dead-on with mercury batteries and the meter needle on battery check stops now within the red field you are most likely done. If your meter was dead-on but the needle doesn’t show good battery voltage now, you might try another diode type. It will cost you less work than recalibrating the meter.
If the battery check passes on current settings, put the top cover again on (look for the small glass window, if it tilts inwards you won’t be able to put the cover on) and test the meter against a good known one. Chances are that it is either dead-on now, or off by a small amount, like 1/3 of a stop. If it’s off by a little and the offset is close to constant across the range, note this offset and compensate with film speed setting. Resist the urge to recalibrate, as it is not as trivial task as it sounds and getting the meter back to factory-quality calibration requires either proper tools or some patience.
However, all parts age, so if your meter is old it could be necessary to recalibrate anyway. Or it wasn’t all that well calibrated before. Anyway, it won’t improve with the diode in. You might also have decided to try recalibrating for silver-oxide batteries without using a diode.
There are two approaches to calibration: you can either do it directly, by exposing the meter to different light levels and succesively adjusting the trimm-pots, or indirectly, by recording the LDR characteristics and calibrating with substitute resistors. The first approach is easier when you either have an adjustable lightbox, or several white walls with different levels of illumination accessible at all times. The second approach is easier for a casual tinkerer, it requires more steps and could take more time, but makes calibration without a good reference set possible.
There’s a neutral density filter of 6 stops attenuation installed in front of the LDR. When you switch the low-light range on, this filter is removed from in front of the sensor for the period of measurement. This means, that the amount of light actually reaching the LDR for EV6 on low-range and for EV12 on the high range is the same, as in the second case there’s an attenuation of 6EV.
The corresponding scale points that result in the same effective illumination of the CdS sensor are listed here:
|EV without/with filter||Scale L||Scale H|
As you can see, for calibrating the point 20 on the high-light scale you either need a lightbox outputting EV15, or you cheat, remove the filter, set the lightbox to EV9 and adjust the meter as if the filter was there, i.e. to read 20 on the highlight scale. Equally, producing an accurate EV0 for calibrating low-light scale might turn tricky, so you can instead cheat, put the filter in (decoupled from its drive spring) and measure EV6 instead. The net result is, you need only EV5 to EV10 reference range to adjust this meter.
Recalibration by a direct approach
- A flat screwdriver with a narrow blade (2-3mm).
- An adjustable lightbox or equivalent.
- Unless your lightbox is calibrated in EV, you will need a known, good, accurate light meter. If you use a meter built into some camera, make sure it isn’t in ‘matrix’ metering. Center-weighted is best.
Towards the bottom of the PCB you will find four trimm-pots available through holes in the pcb (see photo). The left pair is responsible for low light range (1-12, no ND filter), the right pair – for the high light range (12-22, ND6 filter in the light path). The two pots closer to the center are for adjusting offset, while the two pots that on the outside adjust the slope (more or less, they are actually cross-linked). Adjustments done for the low-light range do not influence the high-light range, nor the other way around.
Calibration is an iterative process:
- For low light: set your lightbox to EV5 and adjust the slope pot for low-lights till the dial read `10′. Set your lightbox to EV0 and adjust the offset to make dial read `5′. Repeat, if this time dial reads more than `10′, reduce slope, else increase slope, and so on.
- For high light: use EV14 (19 on the dial) and EV8 (13) instead, and tweak the right two pots.
- Remember you can cheat: you can set EV6 and put the ND in, when on low light range, to get EV0, and you can set EV6 without ND on high lights to get EV12.
Next step is to completely assemble the meter and check the calibration throughout the range. Especially if you have `cheated’ on EV settings! If you use a dSLR internal meter for reference, don’t rely on camera indicated exposure, but rather make a test exposure of a uniform (preferably white) wall on meter-indicated parameters and evaluate histogram. It should be a centered peak. Remember, that Canon dSLR usually have about ASA125 equivalent on “ISO100” setting.
You will likely find the meter to be off around the bottom of the scale. If it reads good above 13 on high, and above 4 on low, just accept it. On my meter there’s 1/6 of a stop mismatch about `12′ on the scales, i.e. 12 on low-light is about 11 5/6 on high. While you can get these to match perfectly, it will throw the other end of scale off by much more. Remember, that the upper scale end is more accurate and around cross-over between the scales just take the reading from the lower range as more reliable. The lowest end of low-light will anyway need correcting for Schwarzschild effect, so getting it more accurate than 1/3 stop makes no real sense anyway (even though it is dead on on my meter).
Recalibration using substitute resistances
As you must have already noticed, calibration is a repetitive process that converges on the proper value in several steps. While many meters can be easily calibrated in one-shot by computing necessary resistor values and just setting them, Gossen meters make this approach extremely difficult by using a non-linear scale. Working the scale nonlinearities back into their original formulas would likely lead to too many assumptions and thus inaccuracies, so I consider it a void attempt in this case. However, repetitive process like this requires you to have a necessary set of reference luminances at hand, e.g. an adjustable lightbox. For many a hobby tinkerer this is not the case.
An observant reader might have noticed, that the CdS cell does not change during calibration. All we really do with the adjustable lightbox is making the cell assume its characteristic resistances for a given set of illumination levels in sequence, repeatedly. Since CdS cells will vary from piece to piece, I can’t tell you what your cell’s resistance at EV10 or EV5 is, but it will be the same at EV10 each time. You can actually measure it with a multimeter. If you collect enough data, you can pick a few fixed resistors and put them in place of the LDR cell during calibration. That’s what I did, and how I did it is described below, note, however, that this method requires some idea about basic electronics and may not actually be easier for you than the direct approach.
- A digital multimeter. It has to be a digital one, as much as I love analogue meters they are not accurate enough for this task (with a few noble exceptions, but if you have one of these you’d know it).
- A known, good, accurate light meter. If you use a meter built into some camera, make sure it isn’t in ‘matrix’ metering. Normal, center-weighted is best.
- A spreadsheet program (Excel, Gnumerics, OpenOffice Calc etc.) or (hardcore) a programmable calculator with statistic functions.
- This file: LDR calibration sheet.web. It’s an Excel 97-2000 xls sheet, I believe the other spreadsheet programs should be able to import it. Notice me if this is not the case, I’ll try my best to provide a solution.
- Some thin insulated wire and a few resistors evenly spaced in the range of 1kΩ – 220kΩ
First, disassemble the meter and unsolder the negative battery wire as before. But don’t put a diode just yet in. Have a closer look under the galvanometer dial. At the same pad, as the battery wire, arrives another one, in a yellow sleeve (or at least it was yellow in my meter). Slightly deeper there’s another one just like that. These are the LDR leads. You need to unsolder at least one of them (if you prefer to do a minimum of disassembly, unsolder just the one going to the negative battery pole, it’s closer and easier to reach). Once one leg is free, connect two thin, flexible wires to the LDR – one to each leg – and route them out of the meter. Close the meter as far as it goes, making sure that the one leg of the LDR does not make contact anywhere.
Now connect your multimeter to these two wires and start collecting data, use the sheet referenced before to evaluate it. Remember, that each evenly lit, preferably white or gray surface you target, can provide you with two readings: one with and one without the ND filter. The ND filter is very close to 6 stops, but just in case: mark your measured values taken with and without ND filter to track them back later, shall discrepancies show. Take a route around your house, in and out. A wall in a windowless toilet can be as dark as EV-3, a whitewashed wall outside in the sun will hit EV17, try your best to fill at least the gray marked range in the spreadsheet. You don’t need values every 1/3 stop, but try to avoid gaps exceeding 2 stops. Remember to remove the dummy data from the column “R measured” in the spreadsheet before use (there will be a lot of DIV/0 errors until you feed some numbers in, that’s OK)! Also, remember, that by default the ND filter is there, you need to press and hold the wiper towards the ‘low-light’ range to remove it (it pops back in as soon as you let go).
Once you have collected some data, put it in the provided sheet, in the column ‘R measured’. Compare them with the reference data I have put in for you – you may be few stops off, that’s ok, but be suspicious if you get readings 5 or more stops off. As soon as you input few points you should get some line plotted in the plot area. If you did your homework right, it will be almost parallel to the sample data, and it will be not too far from it either. Your points, marked as red crosses, should not stray too far from the green line, if they do, either you did something wrong, or the CdS cell is cooked, or I screwed the XLS. I have put a lot of notes into the XLS, look for the red cornered fields and read them, they might explain what hasn’t been made clear here.
If the data seems OK, take few fixed resistors (or precision potentiometers), read from the column “R computed” to what reading do their values (measured values!) correspond, and note it down.
Now install the diode into your meter, reattach the battery wire, but DON’T re-install the LDR yet. Disconnect the thin wire you previously attached to the free leg of the LDR and attach it to the corner pad on the meter’s PCB, where the LDR should be connected. Now put your reference resistors between the wires, activate the meter and set the trimmpots as described before, i.e.:
- take a resistor that has a value appropriate for something close to EV5, connect it to the two thin wires, activate the meter and adjust the slope pot for low-lights till the dial reads what stands in the Excel table as appropriate for your resistor.
- take a resistor that has a value appropriate for something close to EV0… etc., you get the drift.
You can’t cheat this time and you have to check the meter after you reassemble it with LDR in. It might still need a slight adjustment, but it will be tiny.
Final step – battery check
After all is done and the meter reads right on check, that the battery check shows ‘red strip’ for fresh 1.55V batteries. Very fresh silver oxides will be 1.6V, so better take ones that are good, but not right-out-of-the-box. Adjust the fifth trimmpot so, that the battery check reads “red”. Remember the exact needle position to tell any voltage drift in the future.