Sync Speed and Mirrorless Cameras

a7All About the Gear host Doug Kaye goes in-depth on a listener question:

“I have a question for the gearheads in the crowd. Why does the Sony A7 have a sync speed of only 1/250 of a second? I gather it must still have a curtain shutter but I can’t understand why any modern CMOS sensor needs a mechanical shutter. Any ideas?”

I wish it was a simple question with a short answer! Actually, there are two questions in there, and while they’re related, they’re not the same. The first is, “Why is the a7’s sync speed no faster than 1/20 second?” The second is, “Why does the a7 have a mechanical shutter at all?”

First, remember that the sync speed is based on the amount of time 100% of the sensor is exposed to light simultaneously and electronically “enabled” to capture all those little photons. With a mechanical focal-plane shutter, which sweeps across the sensor, this window of opportunity only exists at relatively slow shutter speeds like 1/250 or slower. (At higher speeds, the shutter is just a slit that swipes across the sensor, so there’s no time when all the pixels are exposed simultaneously.)

Now think of the sensor’s photosites as buckets in the rain. They can be exposed (allowed to fill), they can be read & erased (measured & emptied) and they can be covered over by something that keeps them from collecting light (or rain) such as a mechanical shutter. But remember, it takes a finite amount of time to read/erase (measure/empty) them. Unlike a CCD sensor, which can be read all at once, the data from a CMOS sensor are read by rows or groups of rows. The amount of time it takes is directly proportional to the number of rows, so it takes more time to read/empty a sensor with more pixels. Most importantly, you either have to measure/empty the rain in the buckets on-the-fly or cover them over and keep them from filling unevenly to give yourself more time.

A mirrorless camera is always in the equivalent of Live View mode so the sensor has to serve dual purposes. It’s a video camera while you’re composing and focusing. When you press the shutter release, the camera must first erase all the pixels — empty the buckets — to start fresh.

When you shoot with an all-electronic shutter, the camera erases (empties) all the rows from top to bottom, allows them to collect light (rain), then reads (measures) them a fraction of a second later but in the same sequence and at the same rate as it erased (emptied) them. The time that passes between the initial erase and the reading of each row is the electronic shutter speed. This row-by-row enabling and reading of the data is why we get “rolling shutter” from these cameras, but it means that each pixel or bucket is exposed for the same amount of time, just not *at* the same time.

To answer the first question, think about a flash exposure. The duration of the flash is extremely short, so all the pixels need to be “on” at the moment the strobe fires. You have to erase/enable *all* the pixels before you fire the strobe and (more challenging) you have to stop collecting light and then read all the data, which you can’t begin until the strobe has fired. Whereas for a non-flash exposure you can be reading from the upper rows of pixels at the same time you’re already reading from the lower rows, you can’t do that when syncing with a flash.

In the case of a mechanical shutter, it’s a matter of how quickly you can open it !00%, close it, and still have all the pixels exposed for the same amount of time. The limits are mechanical. In the case of an electronic shutter, you’re limited by how quickly you can erase and read the data.

As to the second question, an a7 (24MP) has 4,000 rows of pixels and the Sony engineers wanted to achieve a shutter speed of 1/8000 second, which is pretty much state-of-the-art for full-frame (i.e., large-sensor) cameras. The sensor/processor combination can’t erase then read that many rows fast enough to achieve a shutter speed of 1/8000 second electronically, so the engineers put in a mechanical rear-curtain shutter that stops the exposure while the camera reads the data. If they were willing to settle for a slower shutter speed, they could have eliminated the rear-curtain shutter. One Sony engineer said in an interview that the shutter was, in fact, their greatest challenge in the design of this camera.

Note that the a7r has 4,912 rows of pixels, which turned out to be too many to erase at the rate necessary to achieve the equivalent of 1/8000 second. That is why the a7r also has a mechanical front-curtain shutter and makes the same double-whack sound you get when you shoot a still image on a DSLR from Live View mode.

The a7 has an optional mechanical front-curtain shutter, which needs to be enabled when using 3rd-party longer focal-length lenses at high shutter speeds. This phenomenon I have not yet been able to understand, let alone explain. For example, using a Leica 135mm f/4 lens at 1/8000 second and the electronic front-curtain, nearly 50% of the top of the image is cut off. The portion of the image that’s affected diminishes as you go to slower shutter speeds, but doesn’t disappear entirely until 1/1000 second. With a 90mm Leica prime, the problem is only in the upper 10% of the image at 1/8000 and is gone at 1/4000. I have no idea why focal length should make a difference. It shows up even with Sony’s own 70-200mm f/4, although just in the upper-left corner. With the mechanical front-curtain shutter, the problem disappears with all lenses, but you do get that double-whack sound and the camera isn’t quite as responsive.

If anyone can explain the above phenomenon, please let me know!

[authorbox authorid=”3012″ title=”Author”]