[rawr]

I recently picked up a Tamron 28-75 f2.8, and am puzzled by it's behaviour.

On my K-x the AF is accurate and the lens is sharp wide open. However, on my K-5 the same lens front-focusses very noticeably, so I have to dial-in [-10] AF adjustment for it to get sharp results (and even [-10] doesn't seem completely adequate).

When using LiveView AF on the K-5 the 28-75 shows that it can indeed deliver sharp results even wide-open, so I know the lens isn't simply a dud optically.

However I also have a Tamron 17-50 f2.8. When I mount it on the both the K-x and the K-5, it is sharp as a tack and there is no issue with the AF on either body. The K-5 needs no adjustment and the K-x shoots sharp too.

So why should there be a AF adjust difference between bodies with the 28-75 (which I can understand), but no difference between bodies with the 17-50?

I can't figure it out.

[Eulogy]

That is strange...perhaps clean the contacts and the glass? Past that, I'm going to say this is one of those mystery cases.

[rawr]

Hmmm. I'll give the contacts a good clean and try again.

I may also do a test with more AF lens pairs on both the K-x and K-5 just to see what happens.

[thechumpen]

That's odd, although I do recall seeing somewhere that AF accuracy is as much to do with the camera as the lens. I can though tell you from everything I have read (as I have one as well) that front focus on the 28-75 is pretty common, and is more pronounced at 75 than 28.

You mention -10 isn't enough. What you could do is use PK_Tether to set a global adjustment to your camera. Obviously this will then mean your other lenses become slightly out and need adjustment but as you have a K-5 you can store an adjustment for each lens. This should allow you to resolve the issue.

[bdery]

It's not odd at all.

Anything you purchase is manufactured with certain tolerances. The screen of a Galaxy phone (or iPhone, whatever) is, say, 4,,3 inches in diagonal, but that's ALWAYS ± something. There is always an uncertainty with any measurement, calibration, etc. You get my drift.

Now Pentax's AF system (or any brand's) is specified with tolerances and uncertainties too. It will be accurate within a certain range .

Now the lens also has uncertainties and tolerances to its AF system. Sme same applies.

Then there's the tolerances for the screw in the AF drive connection. There's the uncertainty of the positionning of the lens on the flange. There's a tolerance on the angle of the mirror and the position of the AF sensor. Then there's... you still here?

It's quite rare, but entirely possible, that in some circunstances all these tolerances, which should ideally average to zero (since they should statistically be as often positive and negative offsets) are instead all adding up and creating bad operating conditions. That's what's happening here.

Understanding this, serious companies like Pentax give you the possibility to correct for these rare occurences by offseting the values to correct for extreme cases.

I might also add that Tamron and Sigma do not license the mount technology and standards from the manufacturers, they reverse-engineer everything. That might lead to some occasional errors. More so than with OEM? I have no proof but it seems possible.

a serious company will set its manufacturing tolerances ad "2-sigma" meaning essentially two standard deviations from the average. That means devices will comply 95,45% of the time. That's pretty strict. But there is still about 5% of devices wich will not entirely comply.

[rawr]

a serious company will set its manufacturing tolerances ad "2-sigma" meaning essentially two standard deviations from the average. That means devices will comply 95,45% of the time. That's pretty strict. But there is still about 5% of devices wich will not entirely comply.

Sounds like Deming Award stuff - Ricoh won one a while ago btw. I don't know if Tamron ever did though...

I agree that this is likely to be a tolerances interaction between body and lenses that is probably not worth asking a Tamron repair centre to do anything about. With the AF adjust on the K-5, the AF performance of the 28-75 is now adequate at f2.8 and at f4 it is a non-issue.

And any manufacturer adjustments to the lens to get it perfectly tuned for the K-5 would then probably throw it out of whack on the K-x, which I don't want to do, since unlike the K-5 the K-x has no per-lens AF adjust feature.

Incidentally, even though I have a number of AF lenses, this is the first time I have had a lens which had an AF issue that was easily, visually noticeable and which required any AF adjust. It was quite a new experience, so generally tolerances seem pretty tight nowadays.

[bdery]

Incidentally, even though I have a number of AF lenses, this is the first time I have had a lens which had an AF issue that was easily, visually noticeable and which required any AF adjust. It was quite a new experience, so generally tolerances seem pretty tight nowadays.

I have had many AF lenses and only one needed AF adjustment with my camera too. It was an older FA zoom lens. AF adjustment in the K20D took care of it.

[rawr]

I took the lens for a test drive last Saturday using the -10 AF adjustment, and things turned out OK. Funnily using f3.2 seemed to be a 'sweet spot' giving better results than wide open or even stopped down further.

There was only a few times where the AF failed and I had to go on manual, which arose where the subject was some distance away under what looked like red gelled tungsten lighting, which may be real evidence of the 'tungsten AF front-focussing' bug that people have commented on and which has been cured by the new AF system in the K-5 II and K-30.

But overall I am happy with how the lens performed after adjustment. I am still not entirely confident about using it for critical shots, but it certainly seems a very sharp lens and while using the AF the keeper rate was very high. Some pix related:


Nimbin Aquarius Folk Concert


Nimbin Aquarius Folk Concert


Nimbin Aquarius Folk Concert


Nimbin Aquarius Folk Concert


Nimbin Aquarius Folk Concert

[bdery]

Funnily using f3.2 seemed to be a 'sweet spot' giving better results than wide open or even stopped down further.

It's entirely possible that the lens is optimal at f3.2. Photozone charts could inform you for this, or use "MTF" as your program line in P mode and see what the camera selects (it will be the sharpest aperture).

Also, since you were obviously shooting in low light, it might simply be that at smaller apertures you were inducing camera shake.

[isb.deep]

Nice shots Raws, especially under challenging lighting. I have the same lens on a K-5 II body and have to use a -4 adjustment to get decent results, I have not yet done extensive tests to see if this is the best setting. I also seem to get significantly sharper results at F3.2, so maybe it is really the sweet spot for this lens.

I have a question though: my camera sees anything in focal range of 50-65 mm as 75 mm. So, when I shoot with the lens at say, 50 mm then EXIF shows the shot taken at 75 mm. Anything similar with your lens ? It doesn't bother me much except that maybe the SR is thrown off if the camera is reading FL incorrectly.

[rawr]

Anything similar with your lens ?

I just looked at the focal length filter results out of Lightroom from the shots I took at the gig, and there were lots of shots listed at 50, 55 and 60mm in the EXIF, for example, so no problems so far.

Eg this is at 55mm and the EXIF shows 55mm:


Nimbin Aquarius Folk Concert

And yes, the lighting was dim and challenging so I had to go to ISO 5000-6400 a lot, as the shot above shows, to keep a good shutter speed.

[rawr]

or use "MTF" as your program line in P mode and see what the camera selects (it will be the sharpest aperture).

I must try that one day.

But as a non-Pentax lens, I wonder if that would work. The manual says it only applies to DA, DAL, DFA, FAJ or FA lenses, which implies that the camera firmware must use a database of Pentax lenses MTF curves to implement MTF priority.

It's probably logical to assume that 3rd party lenses might not have been tested by Pentax and had their MTF properties entered into the firmware. But I will test it out. The 28-75 has been around for a LONG time so it is certainly possible Pentax may have MTF data on it.

[bdery]

But as a non-Pentax lens, I wonder if that would work. The manual says it only applies to DA, DAL, DFA, FAJ or FA lenses, which implies that the camera firmware must use a database of Pentax lenses MTF curves to implement MTF priority.

That information is in the lens.

The manual is written by Pentax, and they prefer that you buy their own lenses The manual of the AF540 flash also states that PTTL only works with Pentax lenses...

[isb.deep]

I just looked at the focal length filter results out of Lightroom from the shots I took at the gig, and there were lots of shots listed at 50, 55 and 60mm in the EXIF, for example, so no problems so far.

Eg this is at 55mm and the EXIF shows 55mm:


Nimbin Aquarius Folk Concert

And yes, the lighting was dim and challenging so I had to go to ISO 5000-6400 a lot, as the shot above shows, to keep a good shutter speed.

Thanks for the reply. I guess its just my copy then. Shows the same thing in the K-5 as well as the K-01. I used Pk-tether too and sure enough, it jumps to 75 from 35 and then comes back to 63. Wiggling the zoom at around the 50 mm mark does bring 50 mm in the tethered settings display but it jumps back to 75 at the slightest movement of the zoom ring. Which explains why my Lightroom library has only 3 shots from this lens at 50 mm out of approx. 800

Will send this in to Tamron for repairs, which seems to be the only solution. I searched the net and found a couple of forum posts of Canon users having a similar issue with this lens, so mine is not entirely an isolated instance. Maybe other users of this lens can check their copies too. Thanks again for looking it up.

[rawr]

That information is in the lens.

Interesting. You learn something everyday.

But how? Is it standard in modern AF lenses, in a chip somewhere? Focal length, aperture etc I can understand being transferred to the body by the lens, but the MTF performance profile?