Hi guys, I know this topic has been discussed before but I thought I'd share some testing I did on the shutter delay of my K-5.
Basically I'll reproduce my findings from a webpage I set up:
Shutter speed tests on the Pentax K-5
Testing this [delay] on DSLRs is somewhat tricky, there are many ways to do it but since it involved measuring accurately below 100 ms, usually with 1 ms accuracy it is not a trivial thing to do. I decided to independently confirm the published third party results of the Pentax K-5 of approximately 92 ms without AF. I set the camera to manual focusing, and AF speed tests -- while possible with this setup -- are not currently a priority (I may do AF speed tests at a later time).
A few different ways are used by enthusiasts to evaluate delays, on SLR cameras the acoustic test is fairly straightforward and requires minimal setup: simply point a microphone at the mirror box, record and fire the shutter. Shutter press, mirror-up and shutter curtain should be easily distinguished as distinct peaks. This method is prone to uncertainty since the waveform may not be very clean, it may be difficult to separate the individual components.
A second way considered but not available to me when doing the tests is using a digital counter device, this is a device that takes as input a digital signal, and counts the number of transitions from low to high on its display (usually a VFD type display). VFD display models are ideally suited because of the extremely high operating frequency (or lag, whatever you want to call it) allowing the display to update extremely quickly. This method would be extremely accurate, though some limitations would apply due to shutter rolling. One would simply send the display-reset command to the counter at the same time as the shutter release was pressed (preferably electronicly triggered). If a signal of say 1 kHz were used as a timebase, 1 ms accuracy would easily be possible. However due to the horizontal moving focal plane shutter mechanism it would not work reliably above the flash-sync speed of the camera (1/180 second on the K-5), this would give around 5 ms resolution with absolute certainty.
The method used for this measurement uses an analog Cathode Ray Oscilloscope which has a so-called gate output. This is a digital output that goes high when the beam is tracing a signal.
Let's first examine the oscilloscope, basically it is used for scientific and electronics engineering purposes a lot, it lets you view periodic signals in the time-domain. This means it plots voltage on the y-axis and time on the x-axis. It has a huge advantage in having an old-fashioned CRT display, where the signal is traced by a dot on the screen. It's also pre-calibrated in the speed the dot moves across the screen, so for the purposes of this test it was set to 10 ms per screen graticle. A signal was also fed in to allow closer measurement of the signal, this becomes clear in the pictures.
The scope has a gate-output which was connected via a transistor inverter to the cameras wired remote trigger port (standard 2.5mm stereo jack), the scope was then set to Single mode. In this mode the scope will only "fire" when a signal in is detected once, then it will wait for a reset button to be pushed. This lets the camera and oscilloscope be fired at the exact same instant, pretty cool.
http://longview.be/various/shuttertest/20120507_213926.jpg
This is the actual setup used, the oscilloscope is a HP 1740a.
Results
Three tests were recorded, these were tested a few times and a representative picture was selected as there were some small variances. The first is a wide open shot, the second is stopped down to f/16, the last is from live-view.
For these tests the cameras metering system was activated by half-pressing the shutter first, otherwise a significant delay could occur. Legend: T is the period of the signal shown, one line at the top is half the period. G is the grid-time size.
http://longview.be/various/shuttertest/_IGP3910.jpg
A time of 92 ms was detected here, this matches other findings precisely.
http://longview.be/various/shuttertest/_IGP3918.jpg
A mostly insignificant effect of stopping down, but I have experienced slower burst rates with some Canon EF lenses when they were asked to stop down.
http://longview.be/various/shuttertest/_IGP3921.jpg
Live View is pretty hopeless. Here the starting point is not in calibration, this is a scope calibration error that I forgot to null out when changing the timebase.
So 92 ms is pretty accurate, just though I'd share my findings and methodology.
Similar Threads 
