Basic Canon flash experiment = unexpected results

[bill.akstens]

<p>Just for kicks I wanted to try manually set fill flash with a 580EXII mounted on a 30D. For the first part of this experiment I metered for proper ambient exposure in a dim room, say f4 1/60, and set these settings with the camera in Manual exposure mode. With the flash off, I get a nice ambient exposure as expected. Now I add ETTL flash set with flash exposure comp of -2, and get results as expected - a nice touch of fill flash. Next I set the flash in manual output mode dialed for 1/128 power, the lowest setting available, figuring it should be about the same as the previous shot with ETTL at -2. This shot turns out grossly overexposed by about 4 stops. What's going on here?<br>

The only explanation I can think of is that 1/128 power is not really the minimum output power of the flash, even though that's as low as you can dial it in M mode. In ETTL mode maybe it can dial itself down as low as 1/1000 or 1/2000 power? I would have thought that 1/128 power would just add a kiss of flash, thus acting as a fill, but it didn't work out that way. Any ideas why that is?</p>

[bobatkins]

<p>I'm guessing it wasn't firing at 1/128 power. I don't know why, but I doubt that in auto mode the power can get much lower than that. If it did you probably wouldn't even be able to see it in most situations.</p>

[picturesque]

<p>1/128 isn't the lowest the flash can go in ETTL. Also, while perhaps not the case here, if ETTL flash metering is set to evaluative, automatic fill flash reduction may occur.</p>

<p>As for 1/128th being way overexposed for fill, this would depend on the ISO, aperture, and focal length you were using. Look at the calculator on the LCD panel. It should have told you what subject distance would have been correct for 1/128th. If you wanted two stops below the ambient, you would have set it for a subject distance that was 2 full steps more than the actual. You might have run out of partial power steps.</p>

[daniel flather]

<p>I have used the 580ex in a homemade softbox. As an experiment I took a photo of the softbox, the flash fired, the photo was "properly" exposed. I think the flash in ETTL can go below 1/128 power. Maybe we'll have to try some other tests.</p>

[chrisjb]

<p>I`d agree nadine with the auto fill reduction, try a repeat and turn off the auto fill reduction feature. problably then as the `m` flash. you`ve metered correctly then introduced more light, be interesting the result :)</p>

[jcolwell]

<p>I suggest that if you want to perform an experiment to examine the E-TTL II flash protocol and how it behaves, then you need an independent measure of light output - perhaps a flash meter.  This would let you compare settings with output, rather than trying to interpret the output through an image that introduces many new variables.  </br>

<p>With respect, your comment about adding "a kiss of flash" is not relevant, unless you define who you are kissing and what is the intent. ;)</p>

[bobatkins]

<p>I'm not sure that even E-TTL can get down much below 1/128 power. As I recall, at 1/128 power the flash duration is something like 1/15000s. Regular commercial strobes aren't supposed to be able to go much shorter than about 1/30000s, so it might just get down to maybe 1/256 power in E-TTL.</p>

<p>I've not seen specific flash times published for the 580EX II though. I suspect if the 58EX II was capable of reliable operation at 1/256 power, Canon would have added a 1/256 setting to the menu. The problem with very low powers is quenching the flash tube fast enough to keep the power down.</p>

[bill.akstens]

<p>OP checking back in. Thanks for the thoughts above. During my initial test I was at ISO 1600, focusing about 4 ft away with a 50mm lens. When I looked at the 580exII flash coverage bars it did show coverage for a much greater distance (thanks Nadine). When I dialed down the ISO toward 100 the flash coverage was reduced, as expected, until it eventually landed in the correct range, giving a good exposure.<br>

I have to conclude that during my initial test 1/128 power was simply way too much flash, given my ISO and aperture settings. I'm initially surprised at this, since I'd expect a flash in manual mode to be able to be dialed down low enough to just barely be detectable. I guess I could bounce or diffuse the 1/128 output to reduce it more.<br>

It is my understanding that auto fill flash reduction would not come in to play in a fully manual flash mode.<br>

I'm going to predict that in ETTL mode the flash can go lower than 1/128. I'll see if I can test more tonight...</p>

[picturesque]

<p>Auto fill flash reduction would not come into play in manual flash mode--that is true, but in ETTL, even with camera manual mode, it might have been in play if the ambient was above 10EV, and you had evaluative flash metering set (custom function 14).</p>

<p>I agree that probably the limit of flash quenching is 1/256th. I've never seen anything less than that. This is why I like my Metz flashes--most of the better models can go down that low in manual flash mode. I do think the 580EX can go lower than 1/128th, though. I don't know how you can actually meter a test though, since flash meters have trouble metering really short duration flashes to begin with. Maybe do a test comparing an ETTL shot (nice, medium gray valued target) to what would be a correctly exposed shot at 1/256th, using the guide number to calculate.</p>