Thanks again for your reactions!
I found this video:
and I managed to open the front of the camera and then the lens was easy to unmount. So I saved at least the lens. But the mirror was still locked and the viewfinder was blocked / black. So I opened the bottom plate and when pressing the correct part, I managed to open the mirror so I can finally see through the viewfinder!
I discovered that with slower shutter speeds the mirror / mechanism locks, but with faster shutter speeds it seems to work fine (greater than 1/60s works fine). I have no knowledge about camera repair, so I have to read into it better, but I guess if I'm lucky it could be dried out lubrication. I think the sealing from the back cap has to be replaced as well as it looks bad. I found another video where the sealing is replaced:
Not sure if I can fix the camera, I will try it at least before replacing it.
(I drafted the following several days ago, but vacillated about posting it. It's quite wordy, and I've had a bit of flak in the past about the length of certain posts. That said, the information is based on learned experience and much study, it is accurate, and, if any members are genuinely interested in a better understanding of how their camera works and the real causes of it malfunctioning, well, the detail may assist. If you're disinterested in these things please feel free to skim past the following.)
Apropos your sticking mirror.
A "modern" (it's a relative term) mechanical single lens reflex that features an instant return mirror, should automatically drop the mirror at the end of the exposure, after the closing curtain has completed its run.
It's essential that the components which return your mirror to its viewing position must never begin to drop the mirror before the (second) curtain has closed off the film gate. Otherwise, portions of the image formed through the lens would not be properly recorded on the film.
The typical (and most expedient) way to ensure that the mirror of one's camera is only ever lowered after the second curtain has finished—is to use the second curtain itself, to trip the mirror release. Many manufacturers will do this by fitting an idler gear driven by, and meshed to, the take up shaft of the second curtain. Spring power from the curtain take up roller drives the gear and provides the impetus needed to throw off the mirror release lever. The idler gear is meshed to the curtain roller gear such that the peg or pin on the idler can only ever trip the mirror release, after the curtain has closed off the film gate.
A correctly tensioned and lubricated second curtain has sufficient surplus energy not utilised by driving the curtain across the gate, that some of it can be used to hit the mirror release lever and trip it. By timing the point of mirror release appropriately, an additional, and welcome, consequence of this design, is that using some of the curtain spring energy to trip the mirror release also discourages the second curtain from bouncing back and partially re-opening the gate. It's an elegantly simple solution to the challenges of letting the mirror down at the correct point in the curtain cycle, and only at that point, and, preventing curtain bounce.
Given these points, it will be appreciated that (whilst there can be various reasons why a reflex mirror may not descend, including simple explanations such as sticky damper foam residue clinging to it), in many instances, the root cause may be that, the second curtain is no longer travelling with sufficient velocity, and the residual spring energy needed to successfully throw out the mirror release is not available.
The usual suspects tend to be to blame for this behaviour. It may be old, dried out/evaporated lubricants, or: dust or other contaminants; wear of parts; curtain fabric that's stiffened; possibly, all of the above. Generally, cleaning the mechanism, lubricating it correctly, perhaps some minor correction to curtain tensions, (and, if the curtains are no longer serviceable, replacing these) should usually then see the reflex mirror begin to function properly, without any direct intervention to its pivot or its direct actuating components.
You could say that a sticking reflex mirror is often just a symptom of what ails an SLR. Less frequently will it actually be the root cause. So, get those curtains running to spec, and: bingo. The mirror works again.
This is usually how things play out. Other problems are still possible though, these are not new cameras now. If your mirror release lever is sticking or damaged, or the idler gear that drives it has been tampered with and reinstalled with the incorrect gear mesh, the mirror might not descend at all, or it might descend at an incorrect point in the cycle—even if the shutter itself is in prime condition. (The latter doesn't appear to be an issue with your particular camera, it's just general information, but, I did encounter it some years ago with an immaculate Zeiss Ikon Icarex SLR. It drove me half mad trying to adjust a mirror which resolutely lowered before the shutter had fully cycled—because, after all—nobody would remove the mirror gear, refit it wrongly timed to the curtain gear, and then, put the camera back together and leave it this way? Would they? Apparently, yes—some people would).
Setting aside these, or other, left field glitches which might also yield a stuck mirror, percentages favour a shutter that needs some TLC—it's typical. Sort the shutter and your mirror "spontaneously" works.
The behaviour of your Pentax also offers an additional clue. At the faster speeds the mirror is still cycling fully. It's only sticking at slower speeds. The escapement speeds. This is germane.
At longer exposure durations, when a gear train (escapement) is needed to delay the release of the second curtain (whether by inertia, or mechanical retardation at the longest times) more focal plane shutter mechanisms than not (the classic Exakta is an exception) will use the energy provided by the second shutter curtain take up roller spring to drive the escapement. The curtain is initially unlatched in the same way it would be released at, Eg 1/1000, however with the escapement engaged at medium or slow speeds, it is blocked by the gear train and must run this down, before it is then free to travel across the film gate and shut it.
A consequence of this is that—whilst it is often all but imperceptible to the naked eye—whenever an escapement speed is set, your second curtain is going to be running across the gate at a marginally slower velocity than at the shorter exposure settings. Some of that energy from its tensioned take up roller spring has already been used to drive the escapement. When it has cycled the escapement and is finally unblocked, it will begin moving from a point fractionally closer to the opening edge of the film gate.
At speeds of 1/30 or 1/15 the effect of the escapement on the running speed of the curtain is negligible. The gear train is only operating in inertia mode. So the entire mechanism may still work to spec. When the escapement pallet is engaged (most commonly, at speeds from 1/10 or 1/8 down to a second), however, more energy is needed to drive the gear train that is being friction braked by the pallet. The curtain must first rotate the escapement gears, and while it does this, it will begin to move a millimetre or so, initially (approximately) before it is completely free to run across the gate.
Although the reduction in driving force overall is rather slight—if a mechanism is already in need of servicing and its curtains are dragging a little, then, that additional energy used to cycle the escapement, can make the difference between the curtain spring having sufficient force available to trip the mirror release, or not. Happens more than you might expect. Minolta SRTs (which are an all round excellent quality 35mm SLR) are, if found in original as-manufactured condition, beginning to manifest this behaviour more frequently in recent years. But Pentaxes and plenty of other SLRs will do it, too.
For sake of completeness, note that it's entirely feasible a particular camera might have shutter spindles and timing mechanism in excellent condition, but the escapement itself is sticking and in need of cleaning and possibly lubrication. This may cause similar problems to curtains that have begun to drag a bit. Note, however that in this scenario the second curtain may be more likely to just not want to close at all—as distinct from closing off the gate successfully but not firing the mirror release. Nevertheless, it's conceivable a sticking escapement might discourage proper mirror release with some mechanisms. In either scenario—the solution is obvious. Service the mechanism (including the escapement). Not only will your mirror work correctly, but your shutter accuracy will improve.
It's worth noting that some configurations of camera may let you observe the behaviour of your second curtain when the escapement is in use. It's a matter of removing your lens, if it comes off, and looking into the mount at the front of your film gate. I can't provide a list of types, but if you examine a Leica III series, for example, set to 1 second, and watch the inside wind side of the gate carefully, after you press the release—if all is well, you will see the second curtain creeping its way towards the edge of the gate opening at a snail's pace, for the duration of the one second delay, until the curtain spring has finished cycling the escapement, at which point, the curtain can properly fire. (Obviously, the Leica is a rangefinder and has no reflex mirror to release, however its second curtain powers the gear train, and otherwise operates in a similar way. If an "as found" example will cycle its curtains at the shorter time settings, but the second curtain tends to be recalcitrant as the times lengthen into the escapement settings—the above can be germane, particularly if curtains are still in prime condition).
