D300 built-in flash, Part 1: Blinding speed

Started Dec 25, 2008 | Discussions
David314 Veteran Member • Posts: 4,790
so you agree full power > 1msec?

Kluso wrote:

http://forums.dpreview.com/forums/read.asp?forum=1034&message=30551677
I do not wish to hijack the OP's excellent thread with which I find
no fault.

so you agree with the results that full power flash is greater than 1msec and is a decaying exponential function?

????

troysg Regular Member • Posts: 100
I think this is an example of measurebating at its best....

But I love it!!!!! I have o-scope envy......

ianz28 Veteran Member • Posts: 5,382
Re: I think this is an example of measurebating at its best....

I think this is one of the most impressive posts I've ever seen on dpreview. Your test seems extremely thorough and is very well written/organized. Thank you.

I'm humbled (mentally) and only understood like 20% of it.

I appreciate the effort and you (the op) definitely get added to the shortlist of 99% credible posters. I imagine you are an electrical engineer or something along those lines?

 ianz28's gear list:ianz28's gear list
Nikon D200 Nikon D5000 Nikon D7200 Nikon AF-S DX Nikkor 35mm F1.8G Nikon AF Nikkor 85mm f/1.8D +10 more
OP Marianne Oelund Veteran Member • Posts: 7,274
Timing measurements

David314 wrote:

i just read how you used a D3 to capture these images using flashes
to measure the relative timing?

Yes, the timing is the delay between the D300's flash trigger pulse, and the SB800 pulse peak which provided the light for the photo. The measurements were consistent down to 10-20us; photos with the same timing also showed the same shutter curtain positions, so I have high confidence in the results.

Did you try using the sync output? just put a pull up on it and you
would see the start stop pulse

I used this for the D3 shutter series back in September. For that setup, I didn't use optodetectors at all - just the PC terminal and a magnetic pickup coil on top of the flash that was illuminating the subject. Either approach is equally convenient; I used the optical sensors this time, because they were already set up for the pulse measurements.

also, it would be nice to see the flash in the shutter image - i
assume that was cropped out?

The photos were taken with a macro lens at high magnification, and light from the D300's flash wasn't visible. The thumbnails I posted were downsampled 6:1, so the originals are much larger (which explains the shallow DOF - only the shutter curtains are in good focus).

Although it would have been interesting to see, showing the flash light vs shutter position in the photos would have required firing it at high power every time. Since I took over 160 frames for the test, I didn't want to subject it to such heavy use and heating.

David314 Veteran Member • Posts: 4,790
sync signal in relation to shutter

Marianne Oelund wrote:

Since I took over 160 frames for the test,
I didn't want to subject it to such heavy use and heating.

perhaps i missed it but did you ever figure out where the shutter is in relation to the pc sync signal start and stop?

given that, i could relate the flash back to the shutter position

that's always been a question on my mind

David

OP Marianne Oelund Veteran Member • Posts: 7,274
Flash, PC sync and shutter timing

David314 wrote:

perhaps i missed it but did you ever figure out where the shutter is
in relation to the pc sync signal start and stop?

given that, i could relate the flash back to the shutter position

that's always been a question on my mind

The PC sync signal start edge is located at exactly the same point in the shutter cycle as the trigger pulse for the built-in flash, within about 1us. The PC sync ends 7.1ms later. However, the PC sync pulse is only issued when the built-in flash is off; you can only have one or the other.

I put together a setup this weekend, which provides inputs for the scope, corresponding to light shining on the sensor's bottom and top edges. This is an alternative way of measuring the shutter/flash timing, and is somewhat more direct. (The only difficulty is aligning the optical sensors to accurate edge positions, which took me a couple of hours of tweaking!)

This setup uses a halogen lamp shining into the D300's mirror box. The light reflected off the image sensor when the shutter is open is projected through a 200mm lens onto a pair of optotransistors. The 200mm lens is in symmetrical configuration (400mm from the D300's sensor, and 400mm from the optotransistors) so it projects an actual-size image of the D300's sensor. The optotransistors were shielded in foil, with a very small opening at the top (less than 1mm wide) for light, in order to restrict their sensitivity to a small range of shutter travel, right at the image sensor edges.
Here is the layout:

On the following scope traces, Chan. 1 is the lower-edge light sensor and Chan. 2 is the upper-edge sensor. Chan. 3 is the flash pulse, but I have added a second chan. 3 trace from a separate sweep to show the full-power pulse (at the bottom) in addition to the 1/128 power pulse. Time base is 1ms per division (10ms across the full screen width). You can make precise time measurements from these images, since the scale is exactly 100 pixels horizontal to 1ms.
The sequence is read as follows:

The rising edge of Chan. 1 is the front curtain uncovering the bottom edge of the image sensor. This edge starts off a little slow because the curtain has just started moving and is accelerating.

The rising edge of Chan. 2 is where the front curtain has fully opened (i.e. it's reached the top of the image sensor). This is a more abrupt edge since the curtain is moving quite fast at this point. You will note that the flash triggers very quickly after the top of the image sensor is fully exposed.

The falling edge of Chan. 1 is the rear curtain starting to cover the lower edge of the sensor, and the falling edge of Chan. 2 is the rear curtain fully closing. These falling edges have a somewhat slow decay since I used a rather large load resistor (22Kohms) for the optotransistors.

The a and b cursors mark the time period where the shutter is fully open. Flash pulses constrained to this range will illuminate the sensor evenly, but pulses which extend outside of this range will cause uneven illumination due to curtain shadowing.

This first trace shows the sequence at 1/250sec shutter speed. Cursor b shows where the rear curtain is starting to close; note that the full-power flash pulse extends beyond this point about 400us:

Here is the sequence for 1/320sec shutter speed. In this case, the camera has restricted the full-power flash pulse even more than at 1/250 shutter, but it still exceeds the rear curtain starting to close by about 400us:

rhlpetrus Forum Pro • Posts: 25,769
nice thread, thanks, tagging for later use n/t
-- hide signature --

Renato.
http://www.flickr.com/photos/rhlpedrosa/
OnExposure member
http://www.onexposure.net/

Good shooting and good luck
(after Ed Murrow)

 rhlpetrus's gear list:rhlpetrus's gear list
Leica D-Lux (Typ 109) Nikon D7000 Nikon 1 V1 Nikon AF-S DX Nikkor 12-24mm f/4G ED-IF Nikon AF Nikkor 35mm f/2D +4 more
David314 Veteran Member • Posts: 4,790
Re: Flash, PC sync and shutter timing

Marianne Oelund wrote:

The PC sync signal start edge is located at exactly the same point in
the shutter cycle as the trigger pulse for the built-in flash, within
about 1us. The PC sync ends 7.1ms later. However, the PC sync pulse
is only issued when the built-in flash is off; you can only have one
or the other.

SB800?

the reason i ask is you can see the start and stop of the trigger pulse using the sync signal very clearly

I put together a setup this weekend, which provides inputs for the
scope, corresponding to light shining on the sensor's bottom and top
edges. This is an alternative way of measuring the shutter/flash
timing, and is somewhat more direct. (The only difficulty is
aligning the optical sensors to accurate edge positions, which took
me a couple of hours of tweaking!)

yeah, only a couple of hours?

and please take the following as honest feedback and inquiry

your cursor A position is showing the top of the sensor uncovering 200usec before the flash is firing - that seems like a lot of time

so again to me - as practiced skeptic I would look for confirmation of this

and that confirmation would be seeing the shutter open and the flash not going off - and on the other end, the flash going off and the shutter partially closed (although as you previously pointed out there is other evidence of this)

so - if i could see a picture of the shutter, the flash, and maybe an LED being driven by the sync signal all in the same frame - that would show quite graphically what is going on and would remove any issue with sensing the edge and tweaking the photo transistor window position

i suppose i should walk the talk on this one but since my second camera is a D40, not sure it is up to the task

thanks for the amazing work

David

OP Marianne Oelund Veteran Member • Posts: 7,274
Obtaining photo evidence

David314 wrote:

SB800?

the reason i ask is you can see the start and stop of the trigger
pulse using the sync signal very clearly

I'll have a look with the SB800 this evening, after dark (the setup requires a dark room, of course).

your cursor A position is showing the top of the sensor uncovering
200usec before the flash is firing - that seems like a lot of time

It's 160us. But if you look at the point where the phototransistor is fully "on" then the delay is virtually zero. I can't see how Nikon could mass-produce a camera like this and tighten the timing tolerances any further. There must be some allowance for shutter transit time variations, and the camera needs to work over a fairly wide temperature range. We are talking about a tiny fraction of the transit time, and shutter movement of less than 1mm when the shutter speed is many meters per second.

so again to me - as practiced skeptic I would look for confirmation
of this

and that confirmation would be seeing the shutter open and the flash
not going off - and on the other end, the flash going off and the
shutter partially closed (although as you previously pointed out
there is other evidence of this)

so - if i could see a picture of the shutter, the flash, and maybe an
LED being driven by the sync signal all in the same frame - that
would show quite graphically what is going on and would remove any
issue with sensing the edge and tweaking the photo transistor window
position

Precise confirmation would be nice, but there are some difficulties. Remember that the photographing camera is an SLR also, and has moving shutter curtains which take 2.5-3ms to traverse its sensor. One couldn't achieve the photo you want with shutter speeds of 1/250 or 1/320 since the sensor is exposed for a relatively long time, so max shutter speed would be needed. In that case, only a narrow slot of the sensor is uncovered at any single instant; in order to pick up the flash light, one would need to construct a light bar which runs down the full height of the image. Then there is a small problem of the shutter "race" since both camera's shutters are moving at the same time, and at slightly different rates (the D3 is a bit slower). It would be difficult, if even possible, to reconstruct the timing from the result.

Probably the best photo evidence would simply be to have the photographing camera on Bulb in a dark room, and use the D300's own flash light, via a mirror, to take the photo. This gives only a single time point (showing the shutter curtain position when the flash fires) and wouldn't provide an actual measurement of the delay, but would just be confirmation that the sensor isn't shaded at that moment. Repeating this at flash full power might be able to show how far the rear curtain moves over the sensor by the end of the flash pulse, but it would be rather faint in the photo and probably no more valuable as data, than the image recorded by the D300 (which we already have). Personally, I'd rather spend my time going on to look at other camera models.

All things considered, I think the measurements already done are more trustworthy and meaningful, especially the shutter photo sequence with timings. I suspect you haven't yet fully utilized the information available from those, as they really do contain the answer you're seeking (Hint: Do a little interpolation).

OP Marianne Oelund Veteran Member • Posts: 7,274
SB800 traces
1

These were taken with the same setup as yesterday, except I've increased the photo-transistor loading to square up the falling edges a bit.

As before, Chan. 1 and 2 are the lower-edge and upper-edge phototransistors and chan. 3 is the SB800 flash pulse. Added is chan. 4 which is the PC sync signal. When using the SB800, there is a 10us delay from the PC sync start edge to the flash trigger pulse. Time base is again 1ms/div, for 10ms across the full screen width.
At 1/250sec shutter speed:

At 1/320sec shutter speed:

OP Marianne Oelund Veteran Member • Posts: 7,274
Closest Possible Look

David314 wrote:

your cursor A position is showing the top of the sensor uncovering
200usec before the flash is firing - that seems like a lot of time

so again to me - as practiced skeptic I would look for confirmation
of this

and that confirmation would be seeing the shutter open and the flash
not going off - and on the other end, the flash going off and the
shutter partially closed (although as you previously pointed out
there is other evidence of this)

I've put together a 5-frame composite of new shutter photos from this evening. These are like those shown in Part 2, but with more detail, and each one has its accompanying oscillograph so you can see the timing measurement yourself. The D300 shutter speed for these was 1/250sec.

The scope traces show the PC sync output (which as determined before is simultaneous with the built-in flash trigger, or 10us before the SB800 trigger) on chan. 4 (green) and the flash pulse which illuminated the photo on chan. 3 (magenta). Time base for the main scope window is 400us/div, except 100us/div on the third frame where the timing is extremely close.

The first two frames show the front curtain approaching its fully-open position, at 438us and 242us before the PC sync pulse. The third frame is at 9.7us before the PC sync pulse (how's that for manual coordination?). By interpolating with the 2nd and 3rd frames, we can calculate that the front curtain is at the top row of sensor pixels 195us before the PC sync pulse.

Frame 4 shows the rear curtain starting to move, and frame 5 shows the rear curtain just at the lowest row of sensor pixels, 1250us after the PC sync signal's starting edge. Again, this is at 1/250 shutter; for the 1/320 shutter value we subtract 875us from that time, so the rear curtain will start to cover the image sensor at only 375us after the PC sync start. That's only enough time for a 1/2.5 power pulse from the built-in flash, or 1/4 power -2/3ev from the SB800.

David314 Veteran Member • Posts: 4,790
no sign of the shutter

Marianne Oelund wrote:

Probably the best photo evidence would simply be to have the
photographing camera on Bulb in a dark room, and use the D300's own
flash light, via a mirror, to take the photo.

SB800 triggered by D300 1/64 power D40 at F10 1 sec exposure taking picture

no sign of the shutter

All things considered, I think the measurements already done are more
trustworthy and meaningful, especially the shutter photo sequence
with timings. I suspect you haven't yet fully utilized the
information available from those, as they really do contain the
answer you're seeking (Hint: Do a little interpolation).

without the flash in the sequence pictures you can't relate the shutter position in the pictures back to the scope traces

David314 Veteran Member • Posts: 4,790
Re: Closest Possible Look

Marianne Oelund wrote:

The first two frames show the front curtain approaching its
fully-open position, at 438us and 242us before the PC sync pulse.

frame two is almost all the way open and so you have verified your timing

well done

David

OP Marianne Oelund Veteran Member • Posts: 7,274
Look closer

David314 wrote:

SB800 triggered by D300 1/64 power D40 at F10 1 sec exposure taking
picture

no sign of the shutter

You caught it.

Look carefully at the upper edge of the sensor, and compare to frame 3 of my latest sequence (which is only about 30us earlier than your photo). See the gray edge hanging over the very top of the sensor? Compare to frame 4 of my sequence to see how it differs when the shutter curtain is completely retracted.

without the flash in the sequence pictures you can't relate the
shutter position in the pictures back to the scope traces

All of the information you need is there, because the time values included with the photos tell you the exact moment of the photo, relative to the flash trigger (which is the same time as the PC sync start edge).

Referring to the sequence shown in my Part 2 post, the key frame is frame 8, which is at -0.20ms (-200us) before the flash trigger, and shows that the front curtain is almost exactly at the top row of pixels on the sensor. [If you cut/paste frame 8 and place it next to frame 9, you can see how closely the curtain position matches the top pixel row position.] Thus it tells you that the last row of pixels is uncovered almost exactly 200us before the flash is triggered, in agreement with yesterday's composite sequence and scope traces with the phototransistor signals.

David314 Veteran Member • Posts: 4,790
Re: Look closer

Marianne Oelund wrote:

David314 wrote:

SB800 triggered by D300 1/64 power D40 at F10 1 sec exposure taking
picture

no sign of the shutter

You caught it.

well, i guess it is there - you pictures were so cropped i didn't have a frame of reference and didn't know what i was looking for

Referring to the sequence shown in my Part 2 post, the key frame is
frame 8, which is at -0.20ms (-200us) before the flash trigger, and
shows that the front curtain is almost exactly at the top row of
pixels on the sensor.

i maybe slow but not entirely dumb - i finally figured out how you related time of the flash back to the shutter position in your original series by using the flash from the camera taking the picture and the sync signal from the camera you are taking a picture of

got it thanks!

David

bob elkind Veteran Member • Posts: 5,815
followup Q for Marianne - 'disable' onboard flash

Here's the problem:

Shooting through glass (aquarium, window), which is partially reflective. If using commander mode to control off-camera speedlight (SB-900 in this case), is it possible to set the camera so that the onboard flash is not lit while shutter is open ? I've tried all the obvious settings, but I've been unsuccessful in avoiding reflected onboard flash appearing in the image.

The whole reason for using the off-camera speedlight is to move the flash off-axis, out of the reflection/image zone.

Any help? Thx!

BTW, I'm running on a D700, but the answer probably applies equally to the D300.

-- hide signature --

Bob Elkind
Family,in/outdoor sports, landscape, wildlife
photo galleries at http://eteam.zenfolio.com
my relationship with my camera is strictly photonic

 bob elkind's gear list:bob elkind's gear list
Nikon D700 Canon EOS 500D Nikon D600 Canon EF-S 18-55mm f/3.5-5.6 Nikon AF Nikkor 50mm f/1.8D +5 more
MMuddler Veteran Member • Posts: 4,526
'Mary-Anne'! What's your game now? Can anybody play?"

What fun! Been years since I played with a Tektronix scope and fired flash tubes with intentions of measuring output waveforms. I'll have to read your series of articles. couple of years ago i bought some components to rig up an audio and photo trigger. Thought i'd try some milk drop and balloon burst shots. Just don't think I'll set up the rifle and apple shot .. a la strobe alley.... but was fascinated by the TV ad that shows a bullet travelling thru a bottle and blasting out the rear and morphing into a car. That's a thin slice of time!

Do you have some fun projects at the end of your articles? Guess I have to read them. Take me longer than a microsecond!
--
Marabou Muddler
http://MMuddler.smugmug.com/Sports

RobertMartinu Contributing Member • Posts: 696
Re: followup Q for Marianne - 'disable' onboard flash

bob elkind wrote:

The whole reason for using the off-camera speedlight is to move the
flash off-axis, out of the reflection/image zone.

The macro kit that uses the onboard flash as master has the same iusse with residual light from the main flash. For that reason it comes with a sheat of plastic that allows the IR-part to go through but blocks the visible part.

That tells us two things:
You can't do it with a setting or something.

You need either an IR-only window (perhaps salvaged from a dead remote control) or use a piece of white cardboard to bounce the flash while blocking the direct line of sight. The latter would probably fail outdoors/in bright light, but dimly lit interiors should allow the slaves to get their telegramms.

menameisatsushi Regular Member • Posts: 268
WOW! Very informative and interesting

Technically very interesting and would be usefull when pushing the limit of flash photography.

Thank you very much.

Atu

OP Marianne Oelund Veteran Member • Posts: 7,274
Final flash pulse
1

bob elkind wrote:

Shooting through glass (aquarium, window), which is partially
reflective. If using commander mode to control off-camera speedlight
(SB-900 in this case), is it possible to set the camera so that the
onboard flash is not lit while shutter is open ? I've tried all
the obvious settings, but I've been unsuccessful in avoiding
reflected onboard flash appearing in the image.

Bob:

The flash during exposure is needed, as this is what synchronizes the remote flash to the shutter. Thus there is no setting which can avoid it.

Even if an SU-800 is used instead of the built-in flash, there would likely still be enough visible red light to cause a problem.
Solutions would include:
1. Flash extension cord or radio remote

2. Shield/reflector - it may be possible to fashion a piece of foil which blocks light from shining forward from the built-in flash, but leaves an opening to the side to send light to the remote SB-900.

3. IR filter - similar idea to the shield. Again, it would need to be carefully arranged so that light bouncing off its shiny surface, or leaking around the edges, wouldn't then reflect back off the camera and into your photo.

Keyboard shortcuts:
FForum MMy threads