D300 built-in flash, Part 1: Blinding speed

Started Dec 25, 2008 | Discussions thread
OP Marianne Oelund Veteran Member • Posts: 6,935
Part 3: TTL range, and Shutter Cycle Timings
1

As an addendum to Part 1, I'll start off here with a comparison of the lowest-power Manual mode flash pulse (1/128 power) to the smallest, fastest pulse which the flash can put out in i-TTL mode. These are shown to the same scale. The small pulse is so tiny that it's starting to be obscured by the flash trigger and cutoff signals. The cursors are placed to show the start and end of the pulses; note that I am excluding the trailing edge after the cutoff peak, since this is probably mostly due to my photodiode's speed limitation (4us time constant).

The scope is computing the area under the pulse curve, between the cursors, which is proportional to the flash energy. You can see that it's 537nV-s for the 1/128 power pulse, and 20.4nV-s for the minimum pulse, which is only about 1/3300 of full flash power. This means that i-TTL mode has a dynamic range of almost 12 stops.

Shutter Cycle Timings

For these measurements, I've added another photodetector to watch the reflex mirror movement. This makes it possible to compare the cycle timings with and without flash. The detector was placed in front of the lens, and picks up light from a lamp shining into the viewfinder port.

Each of the scope traces below has: Ch. 1 (yellow) flash pulses; Ch. 2 (cyan) shutter release signal; Ch. 3 (magenta) reflex mirror. The reflex mirror trace also picks up the flash pulse (when the mirror is up), and in some of the traces, you can even see when the shutter curtains are open, since certain ambient lighting conditions allow the detector to see some light reflecting off of the image sensor's surface.

For all of these traces, a shutter speed of 1/60 sec was used. They are all taken at 20ms/div (there are 10 divisions across the full screen width). If you open these image files with a viewer which can read out cursor position in pixels, then you can make your own timing measurements from these traces. There are 100 pixels per division, so each horizontal pixel represents 0.2ms in time.

i-TTL Cycle

As most readers here already know, the camera's flash emits monitor pulses which it uses to determine the flash power needed for the exposure. The first pulse is at 1/16 power. If the light level returned from this monitor pulse is too high for an accurate measurement (e.g., when the subject is close), then a second monitor pulse at 1/64 power will appear 10ms later. When used, the second pulse will add 10ms to the shutter delay time.

The exposure flash occurs 65ms after the last monitor pulse, for JPEG or 12-bit RAW. When 14-bit RAW is used, this delay will be increased to 107ms.

The delay from shutter release to the first monitor pulse is quite variable. I have measured a minimum of 7ms and maximum of 22ms for this delay. Thus overall shutter delay is:
Maximum, 12-bit mode: 22ms + 10ms + 65ms = 97ms
Minimum, 12-bit mode: 7ms + 65ms = 72ms
Maximum, 14-bit mode: 22ms + 10ms + 107ms = 139ms
Minimum, 14-bit mode: 7ms + 107ms = 114ms

If you are trying to use i-TTL mode with remote control and you need very precise timing, there is clearly a problem with all of the variables. Fortunately, Nikon have given us a workaround: FV lock. When FV lock is used, timings will be the same as for Manual flash mode, but you can still have the wide dynamic range of i-TTL.

Here are the traces for the four combinations of i-TTL timing. Note that on the 12-bit traces, the location of the flash pulse during exposure is under the vertical gray line:

Manual Mode or FV Lock Cycle Timings

Timings for Manual mode, or when FV lock is used, are tightly controlled, with a + -400us variation. For 12-bit RAW or JPEG, delay from shutter release to exposure is 57.8ms to 58.6ms. When using 14-bit RAW, the range is 100ms to 100.8ms. This timing information also applies to RPT mode, regarding the time of the first pulse emitted during exposure.

On these traces, you can clearly see a small dip in the reflex-mirror signal, which is due to reflections off the image sensor while the shutter is fully open. The location of the flash pulse on these traces is under the vertical gray line.

Cycle Timing Without Flash

It turns out that using flash, even in Manual mode, causes a small shutter release delay, which is 6ms. Here are the cycle timing traces for non-flash 12-bit and 14-bit recording, which you may compare to the Manual mode traces above:

Setup used for these tests
For the curious, here is a photo of my D300 undergoing close scrutiny:

Post (hide subjects) Posted by
MOD TOF guy
Keyboard shortcuts:
FForum PPrevious NNext WNext unread UUpvote SSubscribe RReply QQuote BBookmark MMy threads
Color scheme? Blue / Yellow