Multiblitz IIIb

[Chris S.]

Interesting thread. Mike's depiction of measuring flash duration with an O-scope is the canonical method, and no-doubt the best. One nice thing is that the readout can depict the light's decay--very useful.

Since I don't have an oscilloscope, for fun I got to noodling on how I could do a test with things I have. Since I have a camera, my thought was to photograph something moving fast at a known rate of speed, using only the flash for illumination, and measure the amount of movement in the photograph. How about my drill press? After a bit of calculation, I see that this is not practical: Drill press tops out at 3200 RPM. This means that a single rotation takes about 19,000 μs--way too slow to measure Iconiclastica's 10 μs flash.

Then I thought about a guitar string. The standard high E is tuned for 329.23 Hz. So one complete back and forth traverse of the vibrating string takes about 3,000 μs--still way too slow to show much blur with a 10 μs flash.

The bullet from a .22 caliber rifle might work. Muzzle velocity ranges around 1500 feet/second, and I have a friend who has the ballistics equipment to precisely measure the speed of a given bullet/rifle combination. Such a bullet would travel about 2mm in 10 μs. Properly photographed, any zone of blur on such a bullet is measurable. But for this, I'd need to buy or build a motion actuated flash trigger--and I'd much rather put the money into an O-scope.

What this drove home to me is that a 10μs flash is very fast indeed. OK, not as fast as the air-gap equipment Harold Edgerton used, but impressively fast nonetheless. Nice piece of kit you have, Iconoclastica.

--Chris S.

PS--If I've messed up a decimal or something in my calculations, please correct me.

[iconoclastica]

I was thinking along the same lines, Chris, but this being an orderly country, flying bullets are hard to come by. Perhaps I could substitute it with an arrow shot, but then, the triggering would be a project in itself. I thinkt it will be easier to find someone with a oscilloscope and try Mike's method.

[rjlittlefield]

The bullet from a .22 caliber rifle might work. Muzzle velocity ranges around 1500 feet/second, and I have a friend who has the ballistics equipment to precisely measure the speed of a given bullet/rifle combination. Such a bullet would travel about 2mm in 10 µs
...
If I've messed up a decimal or something in my calculations, please correct me.

Not a decimal, but I get it as off by a little over 2X.

1500 feet/second = 18000 inches per second = 457,200 mm/second = 4.572 mm per 10 microseconds.

BTW, Google now answers the query "convert 1500 feet per second to mm per microsecond" with the number 0.4572 --- definitely a useful method for crosschecking units conversion!

this being an orderly country, flying bullets are hard to come by

Thank goodness, no bullets are required.

Ordinary power woodworking equipment such as a table saw usually has blade tip speed around 100-150 feet per second, so 0.30-0.45 mm per 10 microseconds. That should be easily seen and measured with a macro lens.

The oscilloscope method is great, except that due to saturation of the photodiode, it's prone to clamping the top of the pulse. That flat top shown on Mike's trace probably reflects a light pulse that peaked a lot higher and fell off exponentially, so its effective pulse time could have been a lot less than the 1 ms indicated. If you use this method, I suggest playing around with dimming the light until you clearly don't get saturation, and then see what the pulse shape looks iike.

--Rik

[mawyatt]

The oscilloscope method is great, except that due to saturation of the photodiode, it's prone to clamping the top of the pulse. That flat top shown on Mike's trace probably reflects a light pulse that peaked a lot higher and fell off exponentially, so its effective pulse time could have been a lot less than the 1 ms indicated. If you use this method, I suggest playing around with dimming the light until you clearly don't get saturation, and then see what the pulse shape looks like.

Rik,

Don't think the diode saturating will have much effect on the overall pulse width, certainly would effect the peak amplitude though. A few of my strobes/flashes have somewhat flat top such as shown, others do not, so the flat topping may not only be due to the diode saturating but just the natural optical output and/or the setup and diode loading.

An easy simple test would be just move the diode or/and strobe further away, if the waveform looks about the same but lower in amplitude then likely the diode isn't saturating.

A really good test would be to use the light meter that has the pulse width output (as well as the total integrated power) for comparison.

This would be a good test for someone to do if they have one of these meters and a oscilloscope.

BTW I don't know if these used the same diode, load or distance from strobe, what strobe fixture was used and so on This is just for illustrative purposes to show the different waveform types using the diode technique. Obviously a well controlled setup and test would be in order to show actual comparisons.

Best,


AD600 Modern IGBT Strobe

Modern Speedlight

Old Conventional Strobe

[Chris S.]

Not a decimal, but I get it as off by a little over 2X.

Rik, thanks for checking my math! Looking over my workings from last night, I can't read my chickenscratch; perhaps I couldn't read it last night, either.

BTW, Google now answers the query "convert 1500 feet per second to mm per microsecond" with the number 0.4572 --- definitely a useful method for crosschecking units conversion!

Google has gotten clever! Trying it just now: "convert 1500 feet per second to furlongs per fortnight" returns 2749085. "Convert 1500 feet per second to cubits per lunar month" returns 2.5514e+9. I may never use my chickenscratch for unit conversion again.

this being an orderly country, flying bullets are hard to come by

Thank goodness, no bullets are required.

Ordinary power woodworking equipment such as a table saw usually has blade tip speed around 100-150 feet per second, so 0.30-0.45 mm per 10 microseconds. That should be easily seen and measured with a macro lens.

Personal reactions can be odd things. For me, the idea of running a table saw in a dark room is hackles-raising. Firing a bench-mounted, low-caliber rifle into a bullet trap, in a dark room, seems perfectly comfortable.

--Chris S.

[rjlittlefield]

Rik,

Don't think the diode saturating will have much effect on the overall pulse width, certainly would effect the peak amplitude though. A few of my strobes/flashes have somewhat flat top such as shown, others do not, so the flat topping may not only be due to the diode saturating but just the natural optical output and/or the setup and diode loading.

An easy simple test would be just move the diode or/and strobe further away, if the waveform looks about the same but lower in amplitude then likely the diode isn't saturating.

Don't think the diode saturating will have much effect on the overall pulse width, certainly would effect the peak amplitude though. A few of my strobes/flashes have somewhat flat top such as shown, others do not, so the flat topping may not only be due to the diode saturating but just the natural optical output and/or the setup and diode loading.

An easy simple test would be just move the diode or/and strobe further away, if the waveform looks about the same but lower in amplitude then likely the diode isn't saturating.

I agree completely.

Here's a test that I ran just now, using some random glass diode and a 100K resistor, flashed with a Canon 580EX II at 1/16 power, changing only the distance.





It seems pretty clear to me that one or the other of these, maybe both, will not give the right idea about what the light pulse looks like.

I'm really most fond of an operational test: how well does the pulse freeze motion?

So, if I were seriously interested in knowing that for my flash unit, I'd use the oscilloscope, diode, and a bright continuous light to measure the interval between teeth on my table saw, then I'd photograph the blade with flash and see what I got for blur lengths.

--Rik

[iconoclastica]

Tonight we measured the pulse length with an oscilloscope. An interesting excercition, really. This is it:



So, not 10µs, but rather half a millisecond. That is, 90% of the curve's surface falls within 500µs. It looks like the diode got saturated, but when I extrapolate the exponential tail back over the flat top, it doesn't make that much difference (450µs).