Strobe & Flash Output Data

mawyatt · Post by **mawyatt** » Fri Jan 15, 2016 8:27 pm

Awhile back I said I would make some measurements on the strobe & flash output timing.

So I snuck out of the lab with a oscilloscope today for use over the weekend. I didn't have any standard "optical detectors" available and decided to use a 2 cent 1N4148 diode as the detector and a 1 cent shunt 75K resistor as the load, along with the scope probe (10M).

Got home and did a couple quick tests with a bright LED flashlight, and solid state physics works, silicon detects light very well!! The 1N4148 works quite well even though most of the diode surface is hidden underneath the leads that butt up against the diode. A 1N918 diode should work even better as it has the beam attachment, but I don't have any of these available.

So I taped the diode and resistor along with the scope probe in the bottom of a small box and preceded to collect some data from a few strobes and flashes.

This is a relative test and nothing is calibrated, but all devices were checked the same, me hand holding & triggering the device about 6 inches above the diode detector. So that's about as scientific as it gets for now!!

Note how the strobes have long durations and the flashes (SD-800 & YN560) have shorter durations as you throttle back the power. Two of the strobes are slaves and not adjustable. I still have my better strobes to test, but that will have to wait until latter.

The ringing you see in the flashes at 1/128 is due to the EM coupling and not optically induced. When these things fire they create electrical havoc with the high current/voltages and fast switching.

Note the two Impact slaves are almost identical in output waveform amplitude and duration. I checked a couple others I have and they all behave the same, very uniform. I would get the cheaper Impact 56 instead of the 80 after reviewing this data, the outputs are almost identical in the test. I might try some other testing later though. The Neewer strobes were not as uniform (didn't show this data) and I had trouble triggering one.

Cheers,

Mike

mawyatt · Post by **mawyatt** » Fri Jan 15, 2016 8:34 pm

I have the results for the SB-800 and Yonguno 560 IV from full to 1/128 power but don't want to overload things. Rik, Chris, Is it OK to add these as well?

Best,

Mike

rjlittlefield · Post by **rjlittlefield** » Fri Jan 15, 2016 8:42 pm

mawyatt wrote:Rik, Chris, Is it OK to add these as well?

Absolutely -- go for it!

And by the way, you've made my day, using a 1N4148 for this job.

--Rik

mawyatt · Post by **mawyatt** » Fri Jan 15, 2016 9:07 pm

We used to build everything with 2N3904, 2N3906 & 1N4148s. If we needed military transistors then we would use 2N2907 & 2N2222 since they were in TO-18 metal packages instead of plastic TO-92.

Best,

Mike

ChrisR · Post by **ChrisR** » Sat Jan 16, 2016 2:34 am

It wouldn't have occurred to me to use a 1N4148

I have a few germanium semiconductors where you can scrape the black paint off to make them interesting, but tbh I'd have dissected a flash slave!.

Pau · Post by **Pau** » Sat Jan 16, 2016 4:00 am

Excellent tests! Thanks for posting them.

For people like me not used to work with oscilloscopes, could you clarify a bit further the flash pulse duration?
I'm pretty sure that it's referenced to the ms/div anotation, and I think that it refers to the major cuadricule divisons in the horizontal axis, not the small ruler divisions, am I right?

ChrisR · Post by **ChrisR** » Sat Jan 16, 2016 4:05 am

I assume that power would be proportional to V², so half power comes at 0.7 x Vpeak.

mjkzz · Post by **mjkzz** » Sat Jan 16, 2016 4:41 am

Nice post! I wish I had a nice scope like yours.

Here is what I did before, with a photo diode wired to microphone jack and used sound card to capture wave, then I wrote a little program to plot the curve. You can use Audacity to do this.

https://www.flickr.com/photos/60552763@ ... ed-public/

A lot of people believe that flashes are fast, but in reality, a speedlite at full power lasts almost 5ms or 1/200 of second.

@ChrisR: I believe amount of energy released by flash tube is proportional to the area underneath the curve in those scope graphs.

ChrisR · Post by **ChrisR** » Sat Jan 16, 2016 4:48 am

energy released by flash tube is proportional to the area underneath the curve

Yes that sounds better. Somebody more awake than me will hopefully integrate the exponential decay...

mawyatt · Post by **mawyatt** » Sat Jan 16, 2016 6:34 am

ChrisR wrote:It wouldn't have occurred to me to use a 1N4148 I have a few germanium semiconductors where you can scrape the black paint off to make them interesting, but tbh I'd have dissected a flash slave!.

Well it didn't occur to me either until I realized I didn't have a photo diode detector. But I did have a few thousand 1N4148 diodes in the lab, so I "borrowed" one

mawyatt · Post by **mawyatt** » Sat Jan 16, 2016 6:42 am

Pau wrote:Excellent tests! Thanks for posting them.

For people like me not used to work with oscilloscopes, could you clarify a bit further the flash pulse duration?
I'm pretty sure that it's referenced to the ms/div anotation, and I think that it refers to the major cuadricule divisons in the horizontal axis, not the small ruler divisions, am I right?

Pau,

The actual "exposure" is the area of the pulse as someone mentioned. The vertical is the amplitude of the 1N4148 diode current due to photo to electron conversion, and the horz. is the time sweep. If you look closely at the trace in the lower mid-right you will see the sweep rate which is per major division. I also indicated the vertical and horz. scale in per major division.

mawyatt · Post by **mawyatt** » Sat Jan 16, 2016 6:53 am

ChrisR wrote:I assume that power would be proportional to V², so half power comes at 0.7 x Vpeak.

Chris,

The "electrical" power yes, or Root Mean Square power is squared function, integrated over the period and then the square root, this is sometimes used as the "heating" effect for power calculations like your 120VAC power is actually 120VAC RMS, or about 180v peak (because its a sine wave the peak is root 2 time RMS value). However this graph represents the conversion of photons to electrons and the further conversion of electrons to voltage.

Since your camera will capture photons in a similar manner as the 1N4148 does, think of this waveform as the optical exposure waveform and the pulse area as the effective exposure.

mawyatt · Post by **mawyatt** » Sat Jan 16, 2016 7:05 am

mjkzz wrote:Nice post! I wish I had a nice scope like yours.

Here is what I did before, with a photo diode wired to microphone jack and used sound card to capture wave, then I wrote a little program to plot the curve. You can use Audacity to do this.

https://www.flickr.com/photos/60552763@ ... ed-public/

A lot of people believe that flashes are fast, but in reality, a speedlite at full power lasts almost 5ms or 1/200 of second.

@ChrisR: I believe amount of energy released by flash tube is proportional to the area underneath the curve in those scope graphs.

Thanks, me too regarding the scope!! It's from my lab, well actually not my lab as that I don't own it

Yes I have have seen many posts about how "fast" speed lights are, and this shows they aren't that fast unless you back well off from full power!

An easy way to estimate the maximum energy from a given flash or strobe is to look at the storage capacitor and the stored voltage. The energy stored is simply 1/2 C V^2. Use this a relative measure since the actual optical output will have lots of things that effect the amount, such as Xeon tube extinguishing voltage, Xeon photon conversion efficiency, Xeon tube switch loss, capacitor series resistance and so on. Point being, if two strobes/flashes have the same energy storage capacitance, same storage voltage they probably have similar optical outputs.

mawyatt · Post by **mawyatt** » Sat Jan 16, 2016 7:07 am

ChrisR wrote:
energy released by flash tube is proportional to the area underneath the curve
Yes that sounds better. Somebody more awake than me will hopefully integrate the exponential decay...

Just get a grad student to do it

mjkzz · Post by **mjkzz** » Sat Jan 16, 2016 7:36 am

Point being, if two strobes/flashes have the same energy storage capacitance, same storage voltage they probably have similar optical outputs.

I tend to disagree with this.

I built a 400Ws IGBT flash myself as DIY project (one year before any Chinese 200ws range products, just to brag) and I have experimented with different flash tubes. Some tubes discharge much faster than others, so you will see shorter tailed graph and they perform much better in terms of consistency of color temp, etc.

The speedlites you are working with probably have same kind of gas mixture inside flash tubes, so they have same characteristics if voltage and capacitance used are the same.

Xeon tube for police/emergency vehicles discharge much slower than those used in photography, I once had a such tube that took a whopping 12ms to discharge same amount of energy as a tube used in Canon 580EX which took about 4-5ms. This is actually good for high speed sync (or pseudo HSS), but that another story.