Got my 3rd K5 this week (first one had sensor stains, second one front focus issues, 3rd is stain free but still has FF).
It appears to be a bit better in low light, but still has significant front focus at EV 2-3.
Here is a test sequence, 3rd K5 on top, 2nd K5 in the middle, K20 last, EV 11 on the left to EV 2 on the right:
FA 50 @ f2.8, ISO 100, AF and 2 sec MLU/Shutter release tripped by IR remote. No sunshine today, so I used my SSC P7 led flashlight to illuminate the target and placed ND filters over the light to reduce it 8 and 10 stops.
Note that "white" leds are not really white as they have very little response in the red spectrum. In this case, the camera saw more and more green as the led light was reduced, which is another nail in the IR color shift myth. The SSC P7 LED white led is nominally 6400K and I measured 6150 at full brightness, 5400 at the lowest light level, so this test does not isolate color vesus light level, not that I believe this makes any difference, as I do not.
All AF adjusts were reset, so you can see some differences in starting focus point.
The good news is that while the 3rd K5 (top) still has a bad shift at the lowest light level, it holds up much better to just above the steep drop-off at EV 2-3.
It may be that I can dial AF adjustments that are biased towards the back a bit in bright light and still get good focus at least to EV 2-3 or so.
The K20D works just as I expected. It tends to FF as well, but is much less certain in getting a lock in low light, hunting a bit more and actually locking into a somewhat worse FF position in the middle image than in the lowest light. It is much better at the lowest light level than either K5.
I had the FA 50 dialed to a bit of BF in daylight on the K20D, which keeps it within the DOF as the ight is reduced.
So, I have tested my cameras with reduced daylight, reduced tungsten light, reduced led light that shifted green and also with different colors, and the only consistent result is that the camera shifts to FF as the light is reduced and has a steep FF shift somewhere around EV 2-4.
There may well be K5s that do not FF this badly at EV 3 and below, but I am on my 3rd body and they all had a very similar problem in low light. I would really like to see a test from a K5 at EV 2-3 F2.8 or so that does not FF like this. At least then I would have some hope that the K5 can focus somewhere near the advertised specification of -1 EV.
No more testing for me, however, as I am convinced that there is a consistent AF problem with the K5 and I have little faith that Pentax will do any more about this than they have with SDM failures, or KR FF in low light issues, or K10D AF in low light issues.
I guess that I need to decide if it s time to seriously consider moving to another brand, as much as I do not want to do so for many reasons.
Ray
EDIT: 2nd K5 has latest firmware (middle images) 3rd K5 does not (top images).