SDM speed (or lack thereof)

Last week, I finally took delivery of a slightly used SMC Pentax DA* 50-135mm F2.8 ED [IF] SDM lens. I got it from an eBayer who was switching brands. As I had the 1.20 firmware in my K10D, a video camera and a tripod, I figured I'd tape the lens while focusing back and forth and then measure the focus speed from the video clip.

I setup the video camera looking down on the lens' little focusing window and proceeded to half-pressing the shutter button (via cabled remote) with the lens cap on in the AF-S mode to make the AF system go all the way from infinity to close focus and back again. This way, I'd see the "raw" focusing speed by which the focus motors could drive the lens. I did this ten times for the screw-drive, flashed the firmware and tried it in SDM mode, fully expecting the SDM mode to be slightly faster than the screw-drive.
This was not so.

The SDM mode turns out to be quite a bit slower than screw-drive mode.

The average (there were practically no variations except for those caused by the video frame resolution of 1/25s) was 38 video frames for screw-drive and 49 for SDM. At 25 frames per second, this translates to 1.52s for screw-drive and 1.96 seconds for SDM, a 29% increase in time spent. Also, several frames (11) are used up at the close-focus stop, before the drive goes back to infinity. Correcting for this and only counting actual travel fime, the difference is a staggering 37% slow-down when using SDM.

Furthermore, it has been hypothesized that SDM could accelerate the lens mass faster, but the video clearly shows that the screw-drive get up to speed almost instantly, while SDM not only requires a longer time to accelerate, but also has a slower sustained speed.

The conclusion? Well, either my lens is seriously barfed, my screw-drive is amazingly souped up, OR there's no point whatsoever in getting a Pentax SDM lens for it's focusing speed since it simply does not deliver, compared to the screw-drive.

I have a short (4s, 18MB) video clip of two of the test runs here:

I also did a few more "real-world" focusing tests to see if the SDM would recoup some of it's speed while fine-adjusting focus, but as I mentioned earlier, its slow starts worked against it here too. I don't have any firm figures since this was more of a hit-and-miss affair and it's difficult to maintain the exact same conditions between test runs, but it looks like it's not faster - just much quieter and possibly slightly slower.

So I'd really like the option to turn SDM off in firmware, when speed counts - and turn it back on, when noise counts.

(DPReview discussion here: