Second curtain stops early (Canon P)

[retinax]

This Canon P is not only a bit rough and noisy winding, I discovered that the second curtain doesn't travel quite far enough. It just so closes the film gate, but doesn't quite catch up with the first one. When cocking the shutter, the first curtain travels the first 5 or so mm alone before the gap is closed and it takes the second curtain along. I don't think it's supposed to be like that, and I think the second curtain bounce I'm observing comes from this; if the curtain traveled further, this little bit of bounce would happen outside the film gate.
It's not because of friction or too low spring tension. The shutter drum simply doesn't rotate further. Same at all speeds.
Seems to me most likely it's a botched repair in the past. But is there any chance that it might be something that can be adjusted without dismantling the shutter itself?
I'm comfortable tinkering with many other things, but completely taking a fp shutter apart and getting it back together seems a bit daunting.

 

[retinax]

Nobody? Speculation is also welcome, anything! This shutter should be very similar to all Leica type shutters, I don't think the problem is something very specific to this one...

 

[monopix]

Speculation...?

Horizontal FP shutters is not my thing but, if this were a vertical one then I'd say the second curtain brake isn't working. I've seen it where the brake can actually stop the second curtain from fully shutting and you also get bounce. Sounds like what you have. AFAIK, horizontal shutters also have curtain brakes so that's my speculation.

 

[retinax]

Thank you monopix, the shutter brake was running smoothly, maybe a bit weak even but not adjustable from the looks of it.
Actually the problem resolved itself. When I looked at it again yesterday, the second shutter curtain ran all the way. It must have been the oil I had applied finally getting where it was needed. While to my untrained eyes it presented like a problem with the main drum, it was with the second curtain roller: The same problem manifests if I slow the second curtain roller down with a finger, or lower its tension. I thought it was the main drum because once the shutter had come to a slightly early stop, the main drum was locked, couldn't be turned further with a finger. But it does turn further when the curtain pulls on it at the time it's supposed to run. Probably this is obvious to someone who knows fp shutters, I couldn't and still can't see why the main drum locks before completing its motion if there's not enough tension/momentum.

 

[retinax]

Aaaactually you're probably right. Now that the shutter runs much better, I needed to lower the tension a lot to get close to correct speed, and the problem with the second curtain is back. At this lower tension, the brake seems too hard to move. It's clean enough, the resistance comes from the spring. It doesn't seem adjustable... Do I have to make a lighter spring? That doesn't sounds fun....

 

[monopix]

Do I have to make a lighter spring? That doesn't sounds fun....

I've no idea.

I'm surprised someone else hasn't chipped in here. I'm sure there must be others with more experience of similar issues.

 

[retinax]

I've somehow managed to find a setting where everything seems to be working, except for tapering at 1/1000. When the tapering is gone, the shutter is running way too fast. Maybe that means there's still too much dirt in the bearings... I can live with 1/500 as my top speed if I have to. I'll wait a bit to see it it remains stable; more oil migration or old muck stiffening up again might change things...
Is it normal for the shutter spring tension to have much more impact on slit width than on the relative speed of the individual curtains? More tension on first curtain makes shutter slower (wider slit width), more on second makes it faster, much more than each influences the slit width dynamically (affect tapering). This makes it pretty difficult.
The rough and noisy winding I found initially was almost certainly due to way too high tension on the shutter that the previous owner tried to solve the issues with...
Sorry for the confusion, my first time adjusting a fp shutter.

 

[monopix]

Again, I'm no expert on horizontal FP shutters so this is, possibly, the blind leading the blind...

The 'shutter speed' is a combination of the gap between the curtains and the speed at which they travel. You could achieve the same shutter speed with slow moving curtains and a small gap as you might achieve with fast moving curtains and a large gap. The gap is the result of how far the first curtain travels before the second one is released and that travel distance will depend on the curtain speed. Essentially, the size of the gap doesn't matter as long as both curtains are travelling at the same speed as the shutter speed is determined by the delay between the release of the two curtains.

So, if you are seeing a change in shutter speed when you change the tension on the first curtain, for instance, I think what you are actually seeing is more, or less, tapering. If you can, try checking the speed at both sides of the frame, so at the beginning of the curtain travel and at the end of the travel. What you measure at the point of the beginning of the curtain travel will be before any tapering has effect. What you measure at the end of the travel will be affected by any tapering. Doing this might give you a clearer picture of what's happening.

BUT... I believe some shutters work by setting a gap width before the shutter is released. If this is the case with yours then all the above is probably irrelevant.

 

[retinax]

I thought that the slit width is set quite independently of the tension, and the relative tension affects travel speed only, but apparently not so, as I can get more exposure across the whole frame, with some emphasis on the closing side, by tensioning the first curtain spring.

I have a pretty clear picture (actually a lot of pictures) of what's happening in terms of tapering because I use a digital camera to take a 1/2 second exposure of the film gate and through the opening shutter, in a dark room, pointing at an evenly lit target. The poor man's shutter speed tester in the 21st century.

 

[Sarcophilus Harrisii]

The effective shutter speeds are a product of both the slit width and the travel speed of the curtains. You want a nearly constant slit width between the curtains across the whole gate for even exposure. Often a slight increase in the slit as the curtains run will help offset the effects of the curtains accelerating. But to keep things simple let's presume it's a constant slit as it is usually (but technically, incorrectly) referred to.

The thing you have to understand is that it is not really possible to counter an incorrect initial slit width by changing the curtain tensions. That way leads to madness. Sure, you can potentially make the exposure correct at one or two points of the film gate—Eg the middle and end—but it will never be correct right across the gate from start to finish. I don't think you have said which way the exposure tapers (light hitting the film might increase or decrease, as the curtains run depending on the timing and which curtain runs quickest).

Let's look at an example. It's a hypothetical but might well be a case study for any number of different types of horizontal FP shutter I've worked on. The mechanism is a little worn, it's in need of cleaning and lubrication and the first curtain is dragging a bit leading to severe tapering down of the light hitting the film (or maybe even capping off prematurely). So you clean and lubricate the mechanism and it runs a little better. You balance the curtain tensions to get them running at close enough to the same speed. But the exposure is still uneven. It's running slow at the start of the film gate (too much exposure).

Hmm OK. Logically you might think that the first curtain has too great a "head start" on the second, and either reduce tension a little on the first or boost it on the second (or perhaps a little or both). You then find that yes the overexposure at the start of the gate has moderated a bit. But hold on—exposure at the end of the gate is now too low. WTF?

The issue is that fine tuning the curtain tensions will have *much, much* less effect on the initial exposure at the start of the gate, than the delay interval for the second curtain release. AKA the basic slit width dictate by the second curtain timing system.

You can only compensate for it so much with the curtain running speeds (and that "so much" is actually not much, at all).

To get nominal consistent exposure across the whole gate both the curtain velocities and release timing for the various slit exposure speeds have to be in the ballpark. So—assuming your mechanism is basically sound (not badly worn, clean, correctly lubricated) then—if you got both curtains running evenly, but the exposure is still tapering, I'd suggest some attention to the curtain release system is called for.

Of course, reading your comments suggests there may be other factors at play. Eg curtain brake adjustment, that ought to first be dealt with (and likely revisited if needed when the basic times are good to ensure bounce isn't a problem).

The 1/1000 speed is nearly always the most challenging to get right. It runs the tightest slit to yield the shortest effective exposure. A couple of millimetres deviation from optimum slit at 1/125 for the majority of vintage horizontal fp shutters is impossible to detect even with transparency—that same amount, at 1/1000, might just cap the exposure off altogether by the end of the film gate. Proportionally, the smallest variations are always the most problematic at the fastest slit time...

Hope the above does make some sense to you. Ask if desired, but...sleep, now for me.

 

[retinax]

Thank you, Brett!

The issue is that fine tuning the curtain tensions will have *much, much* less effect on the initial exposure at the start of the gate, than the delay interval for the second curtain release. AKA the basic slit width dictate by the second curtain timing system.

That's what I thought previously. But not what I experienced.
At the moment, the first curtain runs away from the second, but only at 1/1000. If I increase tension on the second curtain, 1/1000 as well as the slower times become too fast. That seems ok, points to tension being too high on the 1st curtain. But if I decrease tension on the first curtain, shutter speeds also become too fast! Not what I expect, this happens before the tapering is cured. Effect of the first curtain decreasing exposure is greater, so that I can get less exposure by decreasing tension on both curtains (to different degrees), at nearly no tapering. Also I can get a correct, even 1/500 at different tension settings. In fact, if I increase tension on both rollers parallel from the first (lowest tension) working 1/500 setting, nothing much changes with the 1/500 exposure for a whole turn! The tapering at 1/1000 however goes away at higher tension (which feels and sounds too high), although 1/1000 then remains too short.

Also the problem I described earlier with the second curtain not closing all the way seems to still happens sometimes at the slower speeds only, at high speeds the curtain moves a bit further. That points to a problem with the mechanism that sets the gab at high speeds or some kind of friction between both curtains, the first curtain additionally accelerating the second at high speeds, where they travel together? Something is off!

 

[Sarcophilus Harrisii]

Thank you, Brett!

That's what I thought previously. But not what I experienced.
At the moment, the first curtain runs away from the second, but only at 1/1000. If I increase tension on the second curtain, 1/1000 as well as the slower times become too fast. That seems ok, points to tension being too high on the 1st curtain. But if I decrease tension on the first curtain, shutter speeds also become too fast!

Within limits, increasing the tension of the second curtain spring, or, decreasing that of the first, might give the same result (I.e. a reduction in slit width, due to the distance between the curtains decreasing). In itself this is not so surprising.

Not what I expect, this happens before the tapering is cured. Effect of the first curtain decreasing exposure is greater, so that I can get less exposure by decreasing tension on both curtains (to different degrees), at nearly no tapering.

I've never worked on the Canon so I am merely trying to assist from the perspective of general principles of fp shutter operation, and the time I've spent working on numerous different other types. That said: almost regardless of make, you will nearly always want to get the best overall accuracy at the lowest possible curtain speeds (or spring tensions, the two being directly related).

Also I can get a correct, even 1/500 at different tension settings. In fact, if I increase tension on both rollers parallel from the first (lowest tension) working 1/500 setting, nothing much changes with the 1/500 exposure for a whole turn! The tapering at 1/1000 however goes away at higher tension (which feels and sounds too high), although 1/1000 then remains too short.

Re: the 1/500 being constant over a range of tensions. That's harder to explain, although as Peter (monopix) rightly pointed out, it's entirely possible to achieve a particular fast time with the right combinations of either fast running curtains and a wider slit, or, slower running, and a tighter slit. In the interests of minimising wear, manufacturers have to balance achieving slits that aren't so tight they can be prone to drifting off spec, versus curtain speeds that are not higher than desirable for long service life without excessive wear or service requirements. Hence, the combination of settings with the slowest running speed is probably going to be the closest to factory settings. But (theoretically), if you’ve added a turn of tension to the springs and the 1/500 time seems to be the same, this might be possible, if, for example, the first curtain was gaining on the second and inducing an increase in the slit width, concurrent with faster running speed. Making sense now?

The Canon P has a sync speed of 1/55 according to the manual (I’ve just checked). This is close enough to 1/60 to mean it’s typical of any number of mechanisms from Japan made by Nikon, Pentax, Olympus, Canon, Minolta and others also featuring a maximum 1/1000 shutter speed. Not to mention various European designs with identical maximum times and X sync (or perhaps a 1/50 sync which is not far off). What I'm getting at is it's going to be fairly conventional.

If, in your shoes I was having trouble getting the curtains to run to spec, (perhaps counter-intuitively) I would probably start by trying to get both curtains running consistently at the slowest speed needed to get a flash connected to the PC socket to expose the entire film gate. Remember that (except for unique designs such as pre-war Contax) the vast majority of fp shutters use curtains which run at the same velocity regardless of time setting. If you can get the curtains running such that a flash reliably illuminates the entire gate at the prescribed X sync speed, logically, they should, then, be running rapidly enough to also manage all faster times. Provided the timing mechanism is correctly adjusted that is, but that’s OK. If you’re confident the running speeds are in the ballpark, you can be more certain about what is needed next.

Also the problem I described earlier with the second curtain not closing all the way seems to still happens sometimes at the slower speeds only, at high speeds the curtain moves a bit further. That points to a problem with the mechanism that sets the gab at high speeds or some kind of friction between both curtains, the first curtain additionally accelerating the second at high speeds, where they travel together? Something is off!

You may well be right. Getting a fp shutter to run happily at all its speeds means maintaining a fairly fine balance between mechanical friction, curtain acceleration, curtain velocity and slit adjustment. If any of those things are off, it can throw that balance out badly enough to make it impossible to get all the times correct. Eg you might over-adjust the curtain tension so that the curtain velocity is to spec, but, if the mechanism is dry and dirty, the curtain acceleration rate may still be off enough that accurate exposure at the rated time, across the whole gate, becomes simply impossible to achieve through the designed adjustments. The correct approach is, of course, to clean, lubricate and reset to factory specs, and then adjust.

For clarity—let me reiterate that establishing the optimum slit width for the various fast times from 1/125–1/1000 is not primarily achieved through the curtain tension adjustment. The curtains need to be adjusted so that they will run across the gate in harmony at a particular distance, to maintain the required slit across the gate. But the initial slit width is actually established by the curtain timing mechanism, which is responsible for holding back the second curtain for the appropriate period needed, so that, when it begins running, it is starting at the correct distance (slit) from the first curtain lath.

If the timing mechanism is out of adjustment and isn’t sending the second curtain off at the slit distance the manufacturer has specified, you can certainly play with the tensions, and you’ll probably be able to get the time right at the middle and end of the gate. Maybe even at the beginning and centre of the gate (though, this will be harder if the slit commences way off spec). But, you’ll never be able to get the time accurate across the entire gate, at least, not at all of the different shutter speeds. It’s all got to be right. So, please do not conflate the need to have the curtain timing mechanism correctly adjusted, with getting the curtains running at the best speed. Both are essential and, if either is out of spec, adjusting one cannot substitute for the condition of the other. Hope this helps.
Cheers
Brett

 

[retinax]

Thanks for your patience, Brett!
I'm aware that the gap width should be adjusted with the curtain timing mechanism. I'm very surprised that the shutter tension apparently has such a major impact on gap width. I don't think it makes sense for me to touch the curtain timing mechanism though, because it seems more likely that my problems are caused by dirt and old lubricant as I can hardly get even exposure at all speeds, regardless of the speeds being accurate, I'm dealing with too many variables already and I'm in over my head. Or might said mechanism just need lubrication?
I hadn't thought about the flash sync speed, makes total sense, I'll try testing that. Without a proper shutter speed tester I can only test whether no curtain intrudes into the flash exposure, not if the first on is too fast of the second one too slow or too late. I'll still need to look at 1/500 for that. And then see if I can find a setting where flash sync and 1/500 work, and call it good enough. A proper teardown, clean and adjust is more than I want to try atm, but if there's no way to get it to work with a repeatable 1/500 at reasonable tension, I might rethink.

 

[Sarcophilus Harrisii]

Thanks for your patience, Brett!
I'm aware that the gap width should be adjusted with the curtain timing mechanism. I'm very surprised that the shutter tension apparently has such a major impact on gap width. I don't think it makes sense for me to touch the curtain timing mechanism though, because it seems more likely that my problems are caused by dirt and old lubricant as I can hardly get even exposure at all speeds, regardless of the speeds being accurate, I'm dealing with too many variables already and I'm in over my head. Or might said mechanism just need lubrication?
I hadn't thought about the flash sync speed, makes total sense, I'll try testing that. Without a proper shutter speed tester I can only test whether no curtain intrudes into the flash exposure, not if the first on is too fast of the second one too slow or too late. I'll still need to look at 1/500 for that. And then see if I can find a setting where flash sync and 1/500 work, and call it good enough. A proper teardown, clean and adjust is more than I want to try atm, but if there's no way to get it to work with a repeatable 1/500 at reasonable tension, I might rethink.

Hi,
Certainly, I would not suggest trying to adjust the slit timing, until the mechanism is clean and the curtains are running well together across the gate. You'll be informed this may be necessary, if the timing across the gate is basically even, but the exposure at the gate start is (probably) excessive.

If there is no variation in the amount of exposure across the gate, theoretically it's possible the curtains are simply travelling too slowly. But if the curtain adjusters have 2-4 turns each, you wouldn't expect most designs to need significantly more (there may be exceptions in certain cases, of course). If set somewhere around this you'd think it's less likely. At the same time, if the exposure isn't varying across the gate, it suggests the curtains are actually running well together. In which case, thinking you can use the tension adjustments to achieve the correct amount of exposure probably optimistic. You'll certainly be able to increase or decrease exposure at certain parts of the film gate by doing that. But it's unlikely to then be even at beginning, centre and end. If you need to reduce the exposure by the same amount across the whole gate when the curtains are running well, the correct way to do this is to get the second curtain moving a little earlier. This way, when both curtains cross the gate they'll still provide a consistent exposure across the gate, but, it will be at a reduced slit width--what you need.

Indeed, when you consult many makers service documents adjustment instructions, not surprisingly they presume access to accurate testing equipment. But as far as the high speeds go, the gist of what they will generally say, is: adjust curtain velocity to their specified value using tester, then; inspect and set particular exposure times (1/1000, 1/500, for example) by adjusting curtain release lever position (or something to that effect).

I've mentioned over exposure quite a bit in my posts which might seem strange given older fp shutters often manifest tapering exposure, capping, erratic behaviour. Ie less exposure than desirable. This is true. It's been my experience, however that when a mechanism is cleaned and lubricated and the curtains adjusted to run at the correct velocity, the interim result will be that the fastest exposure speeds are overall, somewhat long. Wear in the timing system I assume.

I recently worked on the shutter of an M3 I own and this was my precise experience. On arrival it was way off. After cleaning and lubricating the mechanism I was pleasantly surprised to find the curtains had not drifted too far apart. From memory I believe the opening curtain needed only a half a turn increase in tension. For a late 1950s build mechanism I felt the springs had held up rather well. That said, the 1/1000 to 1/250 times were all nearly a half a stop long. But I was then able to achieve very accurate times by reducing the slit size via the master adjustment. It's impressive that the M3 provides for individual fine tuning of the 1/1000 time, if essential. It wasn't in my case, 1/1000 fell right into place along with the other speeds, (symptomatic, I'd suggest, of a shutter basically in good repair with only moderate wear). It's now well under +/- 10% of nominal times.

At this point I suggest you'll struggle to get a lot further without some means of more precisely assessing the shutter times. I have a Kyoritsu (Copal) camera tester, it's an older version, but a professional quality model which measures the exposure at start, centre and end of the gate. Hence, I'm able to check the whole gate at the various speed settings. (My comments apropos the M3 above are based on adjusting and measuring its efficiency using the Kyoritsu.)

In the absence of a dedicated tester, a humble CRT television screen can inform a surprising amount of information if a little time is invested in becoming familiar with its use and interpretation. Before I acquired my tester I've personally adjusted fp shutters using one which I've subsequently used to expose colour transparency with excellent results. I don't see one as a substitute to a quality tester because it takes longer and benefits from experience, but, it's still very helpful. You can read about that at Rick's page explaining the principles here.

Using a CRT is basically a variation of the procedure Leitz themselves once used to calibrate their cameras. No, they weren't using CRT screens, rather, a rotating, slotted drum outside a light source, but the principle involved is exactly the same. SK Grimes have an informative page about the Leitz tester and if you want to try it I suggest consulting it, as it adds to Rick's helpful details about how to inspect a shutter this way. Their page is here.

Since these pages were published, many (most?) places around the world have embraced digital television transmissions. And you can't just plug an ancient but working analogue CRT set in and switch it on. It has to actually be tuned in to a transmission. Which it can't do by itself, if the only services in your area are digital. You can get around this easily and cheaply by picking up a digital set top box to connect to the CRT, and you'll then be in business. If the box features the ability to pause the screen picture so much the better. Night scenes or other dark picture areas are of little help with inspecting shutter slits, when I used one often I'd wait for a white screen period and hit pause, (the CRT will still refresh) and this makes things a little easier.
Cheers
Brett

 

[retinax]

I'm happy and a little embarrassed to report that a fair bit of my difficulties went away as I increased the tension to a level that seemed uncomfortable, with the naked shutter crate sitting on my resonating desk in a quiet room at night, but is comparable to my well running Canon 7. I didn't manage do count the turns, but I think it's not much more than 4. Seems like I ran into a sort of low intensity reciprocity failure where the shutter runs, but the adjustments don't at all behave as expected, tension has a major effect on slit width and all the stuff I described.

I now have even exposure, 1/500 is a tiny bit short, and 1/1000 must be around 1/1500. I could live with that just fine, in fact it could be handy, but I'll need to fight my obsessive tendencies that will make me investigate the second curtain timing mechanism... I've found instructions in the Canon 7 service manual, which should be close enough to the P.
I have read about the CRT method. Oddly, some LCDs give a similar pattern (varies with display brightness setting). But that's harder to interpret than the digital camera method, which is plenty precise. I use a digital camera to take a 1/2 second exposure of the film gate and through the opening shutter, in a dark room, pointing at an evenly lit target. I've found a fairly dark exposure easiest to evaluate, I adjusted lighting and ISO accordingly. Of course the higher the contrast in the JPEG settings, the more precisely you can identify tapering, and with very high contrast this can get precise enough to drive you crazy, more precise than needed I think. Just like you suggest for the CRT method, I compared to a known accurate shutter, a vertical one at that, which helps as it means it can't be tapering horizontally like the shutter in question/can help eliminate a horizontal gradient in the target illumination (of course one can just use the shutter in the digital camera as a reference, but I used my phone and didn't trust it not to perform some digital trickery with shutter speeds). Both CRT and digital camera method can't distinguish travel speed and slit width. I've been thinking about one of these simple photoresistor probes that plug into a microphone port and allow analyzing exposure with recording software. But for now I'm fairly happy with the digicam method.

Could a mod please change the thread title to something like "Adjusting shutter tension (Canon P)" to better reflect what we've been talking about, so that future readers can find it and benefit from Brett's explanations and my boneheadedness (and my testing method, which I think is not bad at zero cost)?