Strength of master's command flash

When using a flash like the 580 EX II as a master in a wireless system, the commands are sent by the 580 EX II to the slaves using visible light. The ST-E2 has a filter in front of its flash tube, which makes it near infrared, but a normal flash emits normal light, of course.

Since the master flash is on when the picture is taken, the master contributes to the exposure of the image. As long as we want it to, that's fine. The master by default belongs to group A, and will be incorporated into the calculation of group A's flash power, when using E-TTL, or will emit a pulse with the desired strength, if using manual flash mode.

There's also a setting which allows you to shoot with the slaves only, but the master is turned off. This is obviously not 100% true, as the master have to send the command pulses to the slaves, or there will be no slave light either. As this implies that the master must be on when the picture is taken, the master does contribute in some extent to the illumination of the subject.

Canon's manuals don't tell how weak this command flash is, in comparison to anything else, so I decided to find out myself.
I did this by setting the 580 EX II to manual mode, 1/128 power (the lowest setting possible), reflector zoom position to the same as is used by default when the flash is a wireless master and the camera to ISO 100 and f/4. I then exposed a wall at varying distance, until I got a nice histogram "bright edge" in the center of the diagram.
When I found the proper distance, I switched the 580 EX II to master mode, with master flash off and no ratios between groups (A+B+C mode). With this setting I took a picture of the wall again. It was significantly darker than with the 1/128 setting, so I increased the ISO setting until I have brought the "bright edge" of the histogram back to the center of the graph. The ISO setting was now 800.

So from this experiment, one can draw the conclusion that the master command flash only, with "main" master flash off, is about three stops weaker than the lowest manual setting, or about 1/1024 of full power. So it's rather faint, and wouldn't be noticeable in most cases.
But still, a bright wall a meter away, f/4 and ISO 800 it's still powerful enough to expose the subject properly on its own.
In a real picture, you may have to go to ISO 1600 to expose properly with the command flash only, since you probably want the histogram a bit further to the right in such a case. I was only concerned about having it at the same level with both settings, so I could compare them.

I don't have any way to prove it, as I only have one Speedlite flash - a 420EX, but I would strongly suspect that the communication between master and slave takes place while the camera's shutter is still closed.

When cranking up the ISO, you may be seeing image exposure from the ambient light in the area.

I'm still learning this myself, but I believe that Skip is correct, that the flash you're describing is a pre-flash and is not supposed to impact on exposure at all.

Sorry, but you are both wrong.

First, the room was pitch dark when I did this test. The shutter speed was short enough to render the scene black as black can be, if I didn't use the flash.

Second, it's very easy to see that even when the master is turned "off", it still flashes when the shutter is open. I wasn't sure either, so I took a picture of myself, master and slave flash, where both flashes were visible in a mirror. The master flash is lit when the image is taken, no doubt about that.

And if you think about it a bit, you realize that although the pre-flashes could be (and are) used to communicate settings to the slaves, there has to be something that actually triggers them to fire in sync with the camera. That would be very difficult to do reliably, if the master wasn't allowed to fire when the shutter is actually open.

I agree with Anders! Logical thinking, experience, several posters in this forum, and technical references re patents, etc, confirm that the instruction to Fire! is emitted -after- the camera tells the master flash that the shutter is open.

apersson850 wrote in post #8279630
..... And if you think about it a bit, you realize that although the pre-flashes could be (and are) used to communicate settings to the slaves, there has to be something that actually triggers them to fire in sync with the camera. That would be very difficult to do reliably, if the master wasn't allowed to fire when the shutter is actually open.

The logic (computer) in the camera can easily be programmed to deal with the pre-flash for communication with the flash unit(s) well before the shutter is actually tripped. I seriously doubt, even with the testing you've done in the mirror, that the pre-flash (for communication usually regarding exposure calculation) takes place with the shutter open. There could easily be some other reason that the main flash shows some activity while the shutter is open even if it's told to be "off". I suspect that reason may be to pass on the "fire" command to the slave unit.

There are multiple preflashes using wireless ettl, but the final "fire" command from the master happens while the shutter's open. I believe this final flash both contributes to the exposure as well as containing the command, so it's more likely to hit the slave, but that doesn't much help if the preflash wasn't strong enough to get the process started.

Based on the patent information the "fire" control flash is done after the shutter is open when not in hi speed mode. In the high speed mode the fire flash seems to happen before the shutter opens. That makes sense since it has to start producing the continuous light before the sensor is exposed until after it is blocked by the second curtain.

I just did a test with the master flash set "off" and had some exposure (shutter set to 1/250). The set it "on" and had lots of exposure (1/8 power). With the flash "off" and set the shutter to 1/400 I get no exposure. This is consistent with the patent information.

So for no exposure from group A set the flash to high speed and the camera above the highest normal sync speed so high speed is enabled.

SkipD wrote in post #8280673
I suspect that reason may be to pass on the "fire" command to the slave unit.

Yes, exactly. If I was given the task to design such a system, I would send the command flashes to the slaves just prior to shutter release, then send the "trig" pulse when the shutter is open (sync). But that does mean that there is light from the master when the shutter is open, right?

In Nikon's manual for their SB-900 they also state that even with the master set to "off", there will be some light emitted when the shutter is open, and that light may be visible when using high ISO and short distance to the target.
Canon doesn't say anything about this, but that's probably because their manual is much thinner.