This is NOT an official factory adjustment. Do it on Your on risk.
The shutter used in the DSLR cameras works on a “simple” principle:
In the front of the sensor we have 2 curtains (on the figure represented by two plates). Both of them are “cocked” before the exposure, which means that a mechanism with a spring stores energy, with he help of which we can bring the curtains in a relative uniform linear moving.
The 1st covers the sensor, the 2nd is out of the sensor area.
In the moment of the exposure the 1st starts to go down with a uniform speed.
After a certain time reaches the bottom and stops.
The second will be started with a delay. After this delay the 2nd curtain will travel the sensors window too, and will stop at the end .
In reality: it’s not possible to speed up / brake down the curtain in 0 time. The curtain is made of many blades, which are brought in moving through a rotor and a spring. The moving has a probably very complicated speed/time or speed/distance diagram.
This would be no problem. If both of the curtain would have the same speed diagrams, any point of the sensor will be reached by the 1st curtain’s upper edge in the same time from the 1st curtains start, as the 2nd curtain’s lower edge after the 2nd curtain’s start.
The time difference for that point will be exactly the delay of curtain start.
For our examples I will consider constant speeds, and I simplified other parameters too.
A particular case is, when the 2nd curtain starts exactly in the moment of stopping of the 1st in the lower position.
This is used for DSLR cameras as flash synchronisation speed, as in this case the whole sensor is visible.
Normally at 1/250 seconds (the full opening time is in reality a little bit more).
You can fire the flash: the flash duration is less then 1/1000.
If we need a shorter exposition time, we have to start the second curtain delayed exactly with the desired exposition time.
The sensor will be visible only in a narrow stripe.
(Or for longer times then the synchron speed, the shutter will wait some time with both curtains stopped)
What happens when the 2nd curtain is quicker?
In this case the original “geometrical delay” of the 2nd curtain to the 1st will be smaller and smaller – in some cases can be zero.
It means: normal exposition values at start, but no exposition, no picture at the end.
What happens when the 1st curtain has a small extra delay by starting, caused for example by mechanical problems (which means, not seen by the electronics)?
(You can imagine, that a real shutter has mechanical parts with friction, with deformable parts, wear, etc.)
I this case can happen, that the gap between the curtains will be not formed at all – so no picture at all.
The above mentioned two problems can happen together too.
At which shutter speed will we observe it first? Of course at shorter exposition times.
If You change a shutter, normally you should make a calibration of the shutter – that means the small factory tolerances of the different shutters can be compensated in the software – as I know, always only for the high speeds (short expo times). Yes, If You have the factory tools and software.
I’ve got a 10d camera with the symptom, that it is not possible to use shorter exposition time than 1/500.
At 1/640 the picture was already black.
I disassembled the camera, I took out the shutter. The 1st curtain has a little bit sticky moving, and a little bit less tension, then the 2nd.
I disassembled the shutter itself, took out the moving blades and the separator blades – nothing special to see. Assembled again.
The two springs are coiled around two cylinders, which are tensioned. Each of them has a small toothed wheel, stopped in one direction by an elastic spline. You can tension it more with a small screwdriver (the spline lets You to rotate), or You can tension it less with 2 screwdrivers: with one You have to fix the cylinder, with the second to open the spline, and then to let the cylinder to rotate tooth by tooth. Be careful, if You don’t hold it, it will rotate to zero tension.
Ok, this is simple. But how much tension is needed?
If You tension it too much, the mechanism will kill itself. If You tension it less then optimal, the friction forces will have too much influence on the movement. It would be great to measure the force. Khmmmm. Its not a force, in the springit’s there is a torque. But if we hold against the rotation on a certain point, we will need a force, which is proportional with the torque and with the radius of the point.
In this case Canon was very helpful: on the shutter the moving parts have some points which can be used for this reason:
The point in the middle of the small brass part. And each shutter has this! (Is it for factory adjustment?…)
(sorry about the quality, photo made with a phone)
For measuring I used a digital kitchen scale (0-2 kg). Not the best measuring device, but very cheap…
For the measuring I made a small “measuring tip”. For general purposes I usually have to steal my sons Lego bricks, this time it was a Duplo. A screw and the smallest tip of my soldering station – and that’s all.
First step is the zeroing of the scale – with the measuring tip on it.
Secondly I took a good shutter from a 300d (which actually is the same), and I measured the force holding the shutter in a repeatable position, keeping the measuring tip perpendicular on the moving lever.
Yes, it’s no very “scientific” – but works. It’s interesting the relatively big hysteresis: if You push in the spring, the force is more, the after, when You retract it (this is the effect of the inner friction). Anyway, You can determine two values, which has to be adjusted on the wrong shutter too. If I will have too much time, I will make a normal device. Once…
That’s all. After adjusting and mounting in the camera, there was possible to use the 1/4000 exposition time too.
Be careful! Overtensioning will reduce the lifetime of the shutter!
(any text, figure, photo from this article is made by me
– excepting that ones which are special mentioned –
case You use them somewhere I appreciate, if you mention the source )
I’ve got some Photos from Bert, here they are:
Don’t forget: by this type of springs normally we speak about torque (M), not a force (F). When we measure a force instead the torque, we have to make the measuring at the same radius (R), to get comparable results! (M=FxR).
The real radius is always perpendicular on the force – so it’s importan in which position You hold the pin (by me) or the thread (by Bert)