The Tamron AF 17-50mm f/2.8 SP XR Di II LD Aspherical IF lens has received some good reviews (see review by PhotoZone.de) and appear to be a nice low-cost alternative to more expensive offerings from larger camera and lens manufacturers like Canon and Nikon. However there have been quite a few complains (including myself) of the lens overexposing in flash photography, especially in Nikon TTL-BL mode. Here are some threads I found on the Internet.

My copy over exposes upto ~1.5 stops. The following photos illustrate the issue.

Tamron 17-50mm f/2.8

Tamron 17-50mm f/2.8
Tamron 17-50mm f/2.8 with -1.5EV exposure compensation.

Tamron 17-50mm f/2.8 with -1.5EV
Nikkor 50mm f/1.8.

Nikkor 50mm f/1.8
Nikkor 18-70 f/3.5-4.5G AF-S DX.

Nikkor 18-70 f/3.5-4.5G

All test shots were taken at 50mm, f/7.1, and 1/60s using the same camera and settings. The photo taken with the Tamron lens is over-exposed compared to other lenses used in this test. The top-right photo is the top-left photo with -1.5EV exposure compensation in post-processing. I also did tests with both Nikon D200 and D70. Same over-exposure issue happened to both cameras.

The interesting observations are: It works fine in regular TTL mode, no over-exposure issue at all. You can force the camera into regular TTL mode by simply selecting center weighted or spot metering mode. It also works if you use the FV-lock function when the camera and flash are kept in TTL-BL mode. By “works” I mean that the shots will look similar to the ones taken with other lens at the same focal length, aperture, and shutter speed.

I sent my lens to Tamron for repair in the summer. It took them almost two months to work on it. At the end, Tamron sent me an replacement that has the same issue. I wrote an email to their service manager but got no reply.

Just recently, I wrote an article about EXIF information embedded in the photos. While I was examining the information extracted from Nikon NEF raw files using ExifTool, I found the that Nikon DSLR cameras have the focus distance recorded in the photos. This is nothing new. All D and G type lenses from Nikon can report the focus distance data to the camera. You might have not seen it because not many software or EXIF reader reports the data. What surprised me was the focus distance data from the test shots above. As shown in the following table, the Tamron lens sent an incorrect distance!

Lens Focus distance
Tamron 17-50mm f/2.8 XR Di II 3.35m
Nikkor 50mm f/1.8D 1.78m
Nikkor 18-70 f/3.5-4.5G 1.68m

It is now very clear to me why I got the over-exposed photos in TTL-BL mode. The TTL-BL mode on digital SLR bodies such as D200 and D70 is the so-called “3D Balanced Matrix TTL” or “3D Multi-Sensor Balanced Fill-Flash”, in which the focus distance data from the lens and the metering data (ambient and pre-flash) from the camera are combined to determine a proper flash output. The Tamron lens in question sends an incorrect focus distance that is much longer than it actually is to the camera. The camera is then fooled to ask the flash unit to put out more power to compensate for the longer distance, causing the over-exposure. In regular TTL mode, or when FV-lock function is used, the flash amount is determined mainly by measuring the reflected pre-flash intensity without involving the distance information. Therefore no over-exposure occurs.

This issue is quite common and since Tamron cannot do anything to fix it, I suspect it might be a design flaw. The lens does have a very short focus path: a quarter turn gets the lens from the shortest focus distance to infinity. It could be the inaccuracy of the distance information that is causing the problem.

The technical details behind Nikon’s 3D Matrix TTL-BL flash is quite complex. I will be writing an article about it in near future. For now, I am pretty sure this is the cause of the issue. If you have the same issue, you may want to check the focus distance data recorded in the photo and compare it with the data from a lens that doesn’t have the same issue by shooting the same scene at the same settings.

Update on November 20, 2007:

I have seen some follow-up posts on an old DPReview thread after one of my users posted the link to this article there.


Related Posts

Comments