Years ago I picked up a second-hand Sekonic L-358 and have used it ever since with great pleasure. One of those things you just turn on and it works – provided you feed it a new battery once every few years or so. Recently, I got a 5 degree ‘NP Finder’ for it, which turns this incident light meter into a proper spot meter. However, today I ran into a disconcerting problem: totally wrong meter readings. What’s up with that? And can it be fixed?
The backstory to this is that I decided a few months ago that I really needed a proper spot meter. So I scouted around a bit, didn’t quite find what I was looking for, so I turned to the DIY option. I was actually underway with the conceptual groundwork and research when I realized that the Sekonic L-358 I had been using for incident metering actually had a detachable white dome. There must be a reason for this, right? And I also vaguely recalled having read something about accessories for this device many years ago.
So I went online and googled a bit to find that there have indeed existed three spot meter accessories for the L-358: a 1 degree, 5 degree and 10 degree type. They’re either plain unobtainium, since this type of meter is already quite old by now, or they’re massively overpriced. Until I hit upon a shop that listed a 5 degree type for a very decent price. In fact, I suspected it was some kind of fraud, but the shop seemed legit and I decided to just take a gamble on it. Lo and behold, two days later the item arrived – in mint condition, no less! Awesome.
So I’ve used it a couple of times lately, including today. But something happened today that I didn’t like one bit. Here it is:
I developed these in one go in a Jobo 2500 tank, using rotary processing, in Pyrocat HD. Not that it matters, because the problem should be very apparent and not related to development. Indeed, it’s the absolutely massive difference in exposure between these negatives. Do I really suck this badly at metering?
Surely, I make mistakes all the time, but this is kind of extreme. The left-most negative is just about what I intended it to be: darn dense. Lots of shadow detail, lots of density in the highlights. I wanted to use these for carbon transfer and that process does like some meat to the bones of a negative. The three negatives to the right are…well, just about useless. It’s not the development – see the negative on the left. So it’s an exposure issue. The shutter used here isn’t the problem. I use it all the time and most of the exposures were bulb anyway, around 2 seconds, so I could actually hear that the exposure times were more or less correct.
User error? Perhaps, but (1) this was pretty flat light, so even if I metered a bit odd, I would have been out perhaps one stop. This looks much worse. Besides, (2) these scenes are pretty straightforward and I metered quite carefully as well. And this level of ineptitude is really very rare, even for me.
Which leaves the meter. The least likely suspect under normal circumstances, right?
I googled a bit, but sadly the combination of L-358 and NP finder is either too rare to have generated much web content, or it’s so trouble-free that nobody ever posted problems about it. I could find one or two mentions of metering problems, but they were either clearly due to user error, or remained unexplained.
One thing did strike me – out in the woods, I noticed that the meter would reset its reading to zero regularly. This is something it normally doesn’t do: it just keeps displaying the last reading until it shuts off or a new reading is taken.
And something else bugged me as well. I had never looked into (or needed to) how the meter actually recognizes which accessory is mounted. After all, the white dome used for incident readings will pass a lot more light to the sensor underneath than any spot accessory would do. Since the spot accessory doesn’t appear to have any electrical contacts, there’s no amplification electronics inside it, and thus, the meter itself needs to adjust the amplification to the fitted metering accessory. By this time, I had a very strong feeling that this somehow had something to do with the problem. So let’s explore.
Well, sure enough, there’s a provision that allows the meter to know which accessory is fitted. See the image above: when taking off the metering accessory, there turn out to be two little pins on opposite sides of the center turret that can be depressed by the accessory. ‘Can’, because the trick, of course, is that depending on the accessory, either of the pins, both or neither can be depressed. This gives a total of 4 possibilities, which coincides nicely with the existence of three spot meter accessories (1, 5 and 10 degrees) and the default white globe for incident metering.
The white incident globe happens to depress none of the pins, and the 5 degree accessory I’ve got seems to depress both of them. Note the ends of both accessories that slide into the swiveling head:
Two notches on the mount of the NP finder depress both pins, and no notches are present on the incident globe. So that’s how it works: the type of accessory is encoded physically on the mount, and the little pins form a sensing mechanism that the meter can somehow read out. Let’s dig a little deeper.
Removing three screws on the back of the swiveling head gains entry. Inside it, there’s some wires. The thick grey one is a shielded cable that goes right to the photodiode in the middle of the head. The green and the red ones go to spring clips (the ones fitted with screws), and the brown one goes to a central U-shaped bracket that the spring clips can contact with. It’s these spring clips that are essential here. Let’s have a look at one of them from a slightly different angle:
The spring clip with the green wire can contact the U-shaped bracket if the bracket is being pushed back through the little black cylindrical object underneath it. You guessed it: this is the housing where one of the little pins sits that we say earlier. Depressing these pins makes the green resp. red wire contact with the U-bracket and hence the brown wire. The wires go into the main body of the meter.
Actually, at this point, it seems we’re already onto something – and I’d wager to say, we’re just about done fixing the problem. Contacts like these just ask for grime and dirt to collect on their surfaces. While everything looked pretty clean in there, sure enough I could see a little dirt on the U-bracket underneath the spring clip with the green wire. I scratched it off as well as I could, and I think this should fix the problem.
So what happens? It turns out if the green-wired clip doesn’t contact the brown-wired bracket, the meter is tricked into thinking a different kind of accessory is fitted. My guess is that it thinks the 1 degree spot meter is fitted. Since the 1 degree spot meter will allow only very little bit light through (it only sees a 1 degree angle, after all), it’ll have to be amplified a little more than the light that makes it through the wider 5 degree spot.
I trialed it a bit, and it seems that if the meter takes a reading through the 5 degree accessory while not having the green-wired clip touching the bracket, it reads two to two-and-a-half stops too fast. I.e. a reading I took of my desk that should have been 1 second was instead displayed as 1/4 second. To my eye, the two-stop differences matches very well with the degree of underexposure I had on some of those negatives I showed earlier. Hah! So my guess is that this little contact is the culprit.
Wiggling the contact also makes the meter reset its reading. This makes perfect sense: making and breaking the contact makes the meter believe the accessory was changed, and then it’s useful to reset the meter reading so the user doesn’t mistake a previously taken reading for one they made with a different accessory fitted at a later moment.
There’s a bit of an unfortunate engineering choice at work, here. The NP finder accessory is pretty bulky with slightly heavy optics. This puts quite a bit of strain on the mount, and it tends to wiggle a bit. This wiggle is sufficient to sometimes break the contact, making the meter believe a different kind of accessory is fitted. I guess it’s best to firmly hold the accessory against the swiveling head when taking a reading, so that it contacts properly with the little pins inside the head.
What other problems could we find along the same route? After all, it’s quite possible that cleaning the contacts doesn’t solve the problem. They weren’t all that dirty after all (although it does seem to make a difference now that they’re clean), and I might as well investigate a little further what other problems could arise in this direction.
Firstly, there’s the wires themselves. Running wires through a swiveling head is essentially asking for trouble. There’s a huge risk that in due course, they’ll eventually break due to repeated rotation of the head. Secondly, wires means joints, so they’re electrically attached on both ends. In this case, they’re soldered to the metal clips in the head – and the solder joints looked pristine enough at this end. How about the other end? Let’s check briefly since we’re in the neighborhood anyway.
Removing some more screws gives easy access to the internals – and there’s lots of those. The complexity is surprising, in fact, with two large PCB’s and a smaller one. The latter, vertically fixed near the head, seems to be for shielding mostly/only. The one in the top of the picture above seems to house most of the analog circuitry; my guess is that they kept this separate from the other PCB with the main controller, power supply and digital stuff for noise reasons.
Some informal Googling shows that some of the chips on the analog PCB near the incoming signal cable from the photodiode are MJU7014 opamps, which sport a bias current down to 1pA, which makes them a sensible choice for this application. Sadly, the green, brown and red wires that I was hunting down go to the logic board below, so I pulled off the analog board, which is only held fast by a single connector that releases fairly easily.
We can now see the other end of the red, green and brown wires. They are soldered to the logic PCB near the rotary encoder of the edge dial (the blue box in the photo above). We can also see a fancy microcontroller, but I couldn’t quite make out what type it is. Well, it’s of no consequence anyway.
The important thing for now is that the solder contacts to the wires look OK and the wires themselves also do. I measured them with a continuity tester to verify that they showed no signs of breakage, and indeed they appear to be in good order. My conclusion remains that a dirty contact really is/was the main issue – for now. And in case anything along similar lines happens in the future, I’ve got a bit of an idea where to go looking.
For now, fingers crossed that cleaning and a better understanding of proper use (don’t wiggle the spot meter accessory and hold it firmly against the head while taking a measurement) will keep this problem at bay.