fredmiranda.com

Moderated by: Fred Miranda

FM Forum Rules
Macro World Resource

FM Forums | Macro & Still Life | Join Upload & Sell

Archive 2008 · Speedlight flash pulse widths

JimH.
Offline
• •
Upload & Sell: Off

p.1 #1 · Speedlight flash pulse widths

There was some discussion on a different forum in a thread devoted to the MP-E 65 macro lens about what kinds of flash pulse widths we actually get from our speedlights. It's often important to have a very fast flash pulse to help freeze camera shake or subject motion blur. Especially for us hand-held macro shooters

I'm posting this whole thing in this forum too because I figure everyone here will have some interest in it too.

As long as the flash is providing all or most of the light for a shot, the length of the flash pulse is, effectively, the "virtual shutter speed" for the shot.

I wondered what the actual pulse widths were from the Canon speedlights because I felt that I was not getting as much motion-freezing with them as I did with old thyristor flashes in the past.

So I set up a test device to capture the waveforms of the optical outputs of these flashes. It consists of a fast, blue enhanced PIN photodiode detector arranged to operate in the photoconductive mode with a fairly low resistance load to enhance the speed response of the circuit. The signal was then captured with an analog storage oscilloscope.

A test of the system using a simple optical pulse generator consisting of an IR LED driven by an oscillator through a small high-speed power MOSFET shows that the pulse response of the detector has a rise/fall time of 100 nanoseconds or better. This turns out to be far faster than needed to accurately resolve the light pulses from our flash units.

Another aspect of this discussion has been how best to quantify the pulse widths from these flash units.

Standard practice for optical pulse width measurement is to use the "full width at half maximum (FWHM)" method.

But another way of looking at it would be to try to figure out what part of the light pulse will contribute to a possibly visible blur in the image. How far below the peak brightness of the pulse does the light intensity need to drop to before it's no longer creating a visible addition to the image?

Particularly at higher power settings, these flashes tend to emit what looks like a standard "tail pulse" with the intensity rising rapidly and then dropping off slowly as we see the flash capacitor discharging with a typical RC discharge curve. If we take the FWHM of that pulse, I believe we'll have an overly optimistic (too short) estimate of the potential for image blur.

I was also curious as to what the output waveforms looked like for the high-speed synch mode on the 580EX.

Here are the optical power waveforms. Please note that the time periods shown in the upper right corners of the O-Scope traces are the time per major division. There are ten major divisions across the 'scope screen:

http://farm3.static.flickr.com/2208/2267782190_589c831e1f_o.jpg
1. MT-24EX Flash, One head only firing at 1/64 power.
2. MT-24EX Flash, One head only firing at 1/32 power.
3. MT-24EX Flash, One head only firing at 1/16 power.
4. MT-24EX Flash, One head only firing at 1/8 power.
5. MT-24EX Flash, One head only firing at 1/4 power.
6. MT-24EX Flash, One head only firing at 1/2 power.
7. MT-24EX Flash, One head only firing at full power.
8. MT-24EX Flash, both heads firing. A = 1/16 power, B = 1/8 power
9. MT-24EX Flash, both heads firing. A = 1/16 power, B = 1/4 power

http://farm3.static.flickr.com/2288/2267782238_2576edb455_o.jpg
1. 580EX Flash, Firing at 1/128 power.
2. 580EX Flash, Firing at 1/64 power.
3. 580EX Flash, Firing at 1/32 power.
4. 580EX Flash, Firing at 1/16 power.
5. 580EX Flash, Firing at 1/8 power.
6. 580EX Flash, Firing at 1/4 power.
7. 580EX Flash, Firing at 1/2 power.
8. 580EX Flash, Firing at Full power.

http://farm3.static.flickr.com/2076/2266993735_b6d9b307b2_o.jpg
1. 580EX Flash, Firing at 1/32 power, high speed sync mode, 1/1000th shutter speed.
2. 580EX Flash, Firing at 1/16 power, high speed sync mode, 1/1000th shutter speed.
3. 580EX Flash, Firing at 1/16 power, high speed sync mode, 1/2000th shutter speed.
4. 580EX Flash, Firing at 1/16 power, high speed sync mode, 1/4000th shutter speed.
5. 580EX Flash, Firing at 1/16 power, high speed sync mode, 1/8000th shutter speed.
6. 580EX Flash, Firing at 1/16 power, high speed sync mode, 1/4000th shutter speed, partial trace at higher sweep speed to show detail of waveform.
7. 580EX Flash, Firing at 1/16 power, high speed sync mode, 1/4000th shutter speed, partial trace at an even higher sweep speed to show better detail of waveform.

The full sized images are here:

http://www.flickr.com/photos/98036178@N00/sets/72157603894160312/

EDIT:
I guess this forum does not like the "@" sign in the URL. If you copy and paste it into your browser, it'll work.

Edited on Feb 15, 2008 at 06:13 PM

Feb 15, 2008 at 06:07 PM

LordV
Offline
• • • • • • •
Upload & Sell: Off

p.1 #2 · Speedlight flash pulse widths

Interesting stuff Jim- I'm confused though on the time top right- it that the time of the full sreen width ? because it seems to indicate flash durations of approx 1/2000th sec for full power and 1/200,000 sec for 1/64th power ?

Brian V.

Feb 16, 2008 at 02:20 AM

frans_vdm
Offline
• •
Upload & Sell: Off

p.1 #3 · Speedlight flash pulse widths

Nice stuff Jim!
I believe that most of the actual flashes have the same kind of characteristics. If I read the manual from the Nikon SB-80-DX then this results are given:

Flash duration (approx.)
1/1050 sec. at M 1/1 (full) output
1/1100 sec. at M 1/2 output
1/2700 sec. at M 1/4 output
1/5900 sec. at M 1/8 output
1/10900 sec. at M 1/16 output
1/17800 sec. at M 1/32 output
1/32300 sec. at M 1/64 output
1/41600 sec. at M 1/128 output

Global the same range. I've found that between 1/128 and 1/32 power of the SB-80-DX the motion-freezing is sufficient. The light pulsshape for the short durations is not the ideal block, it's more a triangle form. This is the reason why I've an order on 22 highpower leds. I hope to pulse this leds at 100-200 usec but with an ideal bloc pulsform. Disadvantage will be the lower light but I hope to compensate it with shorter distance, higher ISO on the camera and a very narrow lens (4 degrees) on the leds. I will give the results into a few weeks.

Highspeed mode on the flashes are inadequate for highspeed photography considering the lightpulses are distributed over a long time.
Frans.

Feb 16, 2008 at 04:42 AM

bsteels
Offline
• • •
Upload & Sell: On

p.1 #4 · Speedlight flash pulse widths

Jim - great work. Any chance you can publish your data in tables like Frans has given? I think the Canon numbers are very close to what Frans has stated for Nikon, from estimates I've seen published. - I guess I can work it out myself too

I think that both heads on the MT-24EX should not affect duration either, but just give more light. Does this jive? Looks like combined waveforms are present when both heads are set to different power.

The 580 seems a bit faster at 1/64 as compared with the MT-24EX. I get a FWHM of about 35 micro sec vs. 50 micro sec. Interesting...

Edited on Feb 16, 2008 at 12:01 PM

Feb 16, 2008 at 11:54 AM

Dalantech
Offline
• • • • • •
Upload & Sell: Off

p.1 #5 · Speedlight flash pulse widths

Thanks Jim! Cool to see you post the results here (I post as JohnK at the other forum).

bsteels wrote:
I think that both heads on the MT-24EX should not affect duration either, but just give more light. Does this jive? Looks like combined waveforms are present when both heads are set to different power.

Actually it would Brad. With both heads firing the duration would be half as short as a single flash head trying to produce the light for the same exposure. When shooting water drops I resorted to using two flash heads so I could reduce the power needed from each flash and shorten the duration. I'm sure that's also why Frans uses up to 4 flash heads...

Feb 16, 2008 at 03:39 PM

bsteels
Offline
• • •
Upload & Sell: On

p.1 #6 · Speedlight flash pulse widths

Dalantech wrote:
Thanks Jim! Cool to see you post the results here (I post as JohnK at the other forum).

Actually it would Brad. With both heads firing the duration would be half as short as a single flash head trying to produce the light for the same exposure. When shooting water drops I resorted to using two flash heads so I could reduce the power needed from each flash and shorten the duration. I'm sure that's also why Frans uses up to 4 flash heads...

Yes, absolutely agree John. In my statement I meant just for pure duration measurements as Jim carried out, when the flash heads are set to manual power settings. Should have been explicit.

For example, if you add more flashes to a given setup, all at a manual setting of 1/64, the duration should be the same, but the aperture can be stopped down more since more total light will be available. What's neat is when Jim used two flash heads at different settings, the waveforms are adding together.

Edited on Feb 16, 2008 at 03:46 PM

Feb 16, 2008 at 03:43 PM

Dalantech
Offline
• • • • • •
Upload & Sell: Off

p.1 #7 · Speedlight flash pulse widths

Sorry Brad, I misunderstood. Yes, setting two flash heads to the same power will give you a smaller aperture (or a lower ISO setting).

With the MT-24EX's flash heads set to different power levels the resulting flash duration mimics what you'd get in E-TTL mode if you set the controller for ratio control -one flash head simply fires for a longer duration than the other. The "stopping power" of the flash would be determined by the flash head that has the longest duration.

Feb 16, 2008 at 04:31 PM

JimH.
Offline
• •
Upload & Sell: Off

p.1 #8 · Speedlight flash pulse widths

LordV wrote:
Interesting stuff Jim- I'm confused though on the time top right- it that the time of the full sreen width ? because it seems to indicate flash durations of approx 1/2000th sec for full power and 1/200,000 sec for 1/64th power ?

Brian V.

Hello, Brian.

The times shown in the upper right corners are the time "per major division" for the oscilloscope sweep. The display is 10 major divisions wide. This seems to be standard for oscilloscope displays so one gets used to thinking in terms of "XX microseconds per division" and such. But it's not obvious to someone just looking at such a capture without having spent a lot of time using an oscilloscope.

So for the full power shot with one head of the MT-24 EX, we have this:
http://farm3.static.flickr.com/2163/2260672522_b5ea5cef47_o.jpg
The divisions are difficult to see in these shots. I should have turned the 'scope "graticule illumination" up to make the divisions easier to see. But at least on the full-size versions of the photos, you can see them (barely).

In that shot, the full width of the 'scope screen is 10 milliseconds (1/100th of a second).

Here's one of the MT-24E with one head firing at 1/64th power:
http://farm3.static.flickr.com/2140/2260671906_917e20a9cb_o.jpg
In that shot, each major division is 20 microseconds in width. That makes the full 'scope screen 200 microseconds in width.

As Franz notes, at the extremes of flash power (very small or very large), the pulse shapes are far from being the rectangular shapes we might envision. And that makes it extremely difficult for me to know how best to state an actual "pulse duration" for each setting.

The most used way of stating optical pulse widths seems to be to measure the "Full Width at Half Maximum" (FWHM). That means that we measure the width of the pulse at 1/2 the peak amplitude.

But another standard that is sometimes used is to measure the width at the 10% point.

For us, I'd think that the latter of those two might be best because the whole question for us becomes one of wanting the flash to eliminate motion blur and until the light power drops significantly, it will still be contributing to the formation of visible detail in the image. At 10%, we're down about three stops which seems like a reasonable cut-off. Half power is only one stop down, so we'll still have significant image detail being formed.

Here's what I came up with based on those two methods for the MT-24EX and the 580EX from looking at what I captured:

Power levels shown on the left, pulse widths shown on right:

MT-24EX using "above 10%" rule.
1/64 = 90 uS (1/11,000th)
1/32 = 110 uS (1/9100th)
1/16 = 200 uS (1/5000th)
1/8 = 350 uS (1/2850th)
1/4 = 600 uS (1/1660th)
1/2 = 1500 uS (1/660th)
1/1 = 5000 uS (/200th)
1/1 both heads = 3000 uS (1/330th)

MT-24EX using FWHM:
1/64 = 45 uS (1/22,000th)
1/32 = 75 uS (1/13,000th)
1/16 = 135 uS (1/7,400th)
1/8 = 250 uS (1/4000th)
1/4 = 500 uS (1/2000th)
1/2 = 1400 uS (1/710th)
1/1 = 1500 uS (/660th)
1/1 both heads = 900 uS (1/1100th)

580EX using "above 10%" rule.
1/128 = 40 uS (1/25,000th)
1/64 = 65 uS (1/15,000th)
1/32 = 100 uS (1/10,000th)
1/16 = 160 uS (1/6250th)
1/8 = 250 uS (1/4000th)
1/4 = 550 uS (1/1800th)
1/2 = 1250 uS (1/800th)
1/1 = 4500 uS (1/220th)

580EX using FWHM:
1/128 = 22 uS (1/45,000th)
1/64 = 37 uS (1/27,000th)
1/32 = 65 uS (1/15,000th)
1/16 = 115 uS (1/8700th)
1/8 = 210 uS (1/4800th)
1/4 = 480 uS (1/2100th)
1/2 = 1200 uS (1/830th)
1/1 = 1500 uS (1/670th)

The fact that in Franz's table from the Nikon flash, they show the 1/2 power time being so similar to the 1/1 power time makes me think that Nikon is using the FWHM method for measuring things. We get the same sort of measurement for the Canon speedlites if we use FWHM as the way of expressing the widths.

I probably ought to play with the flashes set to ETTL mode just to see for sure what they actually do.

I'm confused by what I'm seeing comparing manual power settings for the MT-24EX with both heads used (ratio mode) to one head used.

If I set the flash to emit only from one head, and set it to 1/4 power, we get this:
http://farm3.static.flickr.com/2276/2259879915_f55fc4ddc0_o.jpg
That shows a pulse width of about 500 microseconds for 1/4 power for one head only. (Using FWHM).

But, if I shoot with both heads, at a setting of A=1/16th and B=1/4, we do NOT get the 500 microsecond pulse width from B. Instead, it appears to be about half of that value.
http://farm3.static.flickr.com/2220/2259880279_d9019bc398_o.jpg

I would have expected that we'd want to get a total power delivered to the scene of 0.3125 (1/16 + 1/4) with the flash set up as above. To get that, I'd want the B head to still deliver the same 1/4 power pulse and the 1/16th power pulse from the A head would simply be added to it (and seen as the higher amplitude but shorter duration beginning of the pulse). The total pulse width should be the same for A=0 B=1/4 as it is for A=1/16 B=1/4, right?

But that's not what I'm seeing.

In my tests, it seems that the total power delivered may be being adjusted to be equal or something.

But one thing to keep in mind is that when I made these tests, I tried to set things up to keep the pulses high enough to be easily measured without letting them go off the top of the trace.

According to my notes, when I went to using both heads of the MT-24EX, I added an additional filter to keep the pulse heights on-scale. So we cannot read the "areas under the curves" to compare the actual power delivered in the one-head versus two-head tests. The two-head measurements have a different effective vertical scale than the two-head measurements.

Obviously, the peak power delivered when both heads are firing (to a target equidistant from both heads, and at the same angle to them both) will be double what you'd get with only one head firing.

I may try shooting some where I do NOT compensate for brightness so that we can better judge and calculate "the areas under the curves" to compare power levels as well as timing.

My main concern in these tests was just to try to evaluate pulse widths for the purposes of eliminating camera shake and for freezing fast objects. So I was trying to optimize the waveforms for us to be able to view the effective pulse widths.

And finally, I would like to play a bit with things using ETTL. But to do so requires that the time delay between the pre-flash and the "real flash" be somewhat constant and repeatable because I'll need to use the "delayed trigger" feature of the oscilloscope to be able to capture the "real" pulses since the pre-flash will obviously trigger the 'scope.

But just thinking about it, my guess is that we won't see any differences in the wave shapes for ETTL shooting because the camera and flash must pre-calculate what flash pulse powers to use based on the pre-flash, and then commit to blindly using that power level for the real flash pulse. So effectively, we should not see any sort of changes in the pulse shapes by shooting in ETTL.

I may have a chance to play with some of this later this weekend.

Edited by JimH. on Feb 16, 2008 at 04:17 PM GMT

Edited on Feb 16, 2008 at 06:17 PM

Feb 16, 2008 at 05:57 PM

bsteels
Offline
• • •
Upload & Sell: On

p.1 #9 · Speedlight flash pulse widths

Great info - Thanks Jim.

Feb 16, 2008 at 06:13 PM

Dalantech
Offline
• • • • • •
Upload & Sell: Off

p.1 #10 · Speedlight flash pulse widths

The shape of the wave form can make a difference in the flash's stopping power. Check out this page at the HiViz web site.

Feb 17, 2008 at 01:24 AM

LordV
Offline
• • • • • • •
Upload & Sell: Off

p.1 #11 · Speedlight flash pulse widths

Ah thanks for the explanation Jim- means my guestimmates were 10 fold out

Always had trouble with oscilloscopes at school

Interesting the highspeed sync mode reslts- perhaps slightly counter- intuitive at first sight but not when you think about it- would explain why I got blurrier shots trying high speed sync at 1/500th sec on water drop shots than I did on standard 1/200th Xsync - the flash pulse would have been longer.

Brian V.

Feb 17, 2008 at 01:40 AM

frans_vdm
Offline
• •
Upload & Sell: Off

p.1 #12 · Speedlight flash pulse widths

Dalantech wrote:
I'm sure that's also why Frans uses up to 4 flash heads...

Yes correct! even to 6 flashes. I have the 6 flashes in 2 groeps but normal they are driven the same time by my controller. I hope to connect this week my highpower leds. Especially the possibility to drives this leds multiple times in a short periode can given special effects on highspeed pictures. Thats why I like to drive the leds 3 or 4 times into the same picture with a interval of a few msec. I have just to set the camera at a lower speed ( as an example: 1/50 sec give me 20 msec interval for the multiple flashes). I am curious to see the results during the development of the waterfigure into the same picture.

An another way to measure the flashes is to setup the flashe at a fix distance
aim at a withe paper and take a picture. The camera settings give you the results of the light capture at 1/3 stop resolution. I have used this methode multiple times for photopainting techniques to know how long you may stay with a light on the same surface during the paintperiode.
Frans.

Feb 17, 2008 at 04:59 AM

JimH.
Offline
• •
Upload & Sell: Off

p.1 #13 · Speedlight flash pulse widths

Frans:

Which LEDs are you going to use? I've been looking at some of the very bright multi-LED chips used in high-power flashlights for this very use. There are quite a variety of high-power LED modules available these days.

One of the strange things about many of them are their output spectra. Particularly for your uses, with such wonderful colors, I wonder what type of LED you've chosen to experiment with.

Feb 18, 2008 at 08:37 AM

frans_vdm
Offline
• •
Upload & Sell: Off

p.1 #14 · Speedlight flash pulse widths

Jim,
I just received this morning all my highpower leds and lenses. The types are the P4 from Seoul:

Seoul datasheets

I have 4 version:
The P4-Pur white W42182 Rank U total 10 leds
The P4-blue B42182 total 4 leds
The P4-red R42182 total 4 leds
The P4-green G42182 total 4 leds
The white verson can be pulsed at 100 usec 1/10 duty at max 1800 mA but I will start at 1000 mA first to see the results. The blue and green may have max 1000 mA and the red 800 mA. But I have also different lens for this leds:
Narrow beam lens
Medium beam lens
Wide beam lens
Elliptical beam lens
All this lenses are from Frean.
This lenses are complet with holder to mount on the star type leds. Now I'm making the drawings to use all this leds and to measure the lights. I just have buy all the voltage regulators, connectors, power resistors etc. to make a separate little box connected to my central controller to drive this leds in 5 groups, 2 groups of each 5 white leds, and one group of each color: red, green and blue. Making a combination of all this leds must give some freedom for the lightspectrum ( I hope).
I'm only be worried on the lightpower compare to the standard flashes. The working distance will be 10 cm, very close to my figures.
As soon as I 've tested this I will give the results. But first I have to make the mechanical parts for cooling and to fit them on my setuptable.
Frans.

Feb 18, 2008 at 11:10 AM


FM Forums \| Macro & Still Life \| Join Upload & Sell