Hey CHDKers,

I should have titled this REVISITED: fudgey's "Accurate USB remote sync..." but I don't know who besides the usual suspects cared about the original topic, yet it's been read over 5000 times...

Background: I use a Sunpak Auto 433D (ubiquitous '80s GN 120 @ ISO 100 hotshoe flash) mounted on a Sunpak DIGI-ADAPT flash-activated slave. It triggers fast, ~1µs, and it has never missed a 1/500th sync with the SX100. But it (all light-activated slaves actually) has two big limitations: Since it's a phototransistor slave, it needs to detect a flash firing. It doesn't have to be your flash firing, either. And the reverse of that, if no objects are close to bounce enough light from the weak in-camera flash back at the sensor, it won't trigger, rendering it useless.

I want it to become the main strobe on my stereo SD1100 rig. (image) I need it hard-wired, so I can disable the flashes in the 1100s and have a single point light source as well as extend the runtime of those NB-4L batteries, with enough flash power that red-eye reduction is never needed. (unless your target is 100' away.)

I know I have to build a delay circuit to do this, perhaps something like Damir Vran?i?-type continuous shooting, synchronised flash PCB. [[EDIT: More on that, next post down]]

Looking for sync test results from those who still have CRTs or an oscilloscope over at http://tech.groups.yahoo.com/group/StereoDataMaker/files/Synchronisation%20Tests/, I see no one's done SX100s nor SD1100s. Blast. My oscilloscope is currently inaccessable. Where am I gonna find a CRT!? I'd have better luck finding a Canon 110ED.

OK, the slow way then. I wired up the simplest circuit (image) I could: A single PNP transistor with a pull-up resistor, with a small base resistor as a current limiter, and wired the flash sync across the emitter-collector. Since the flash trigger voltage for this particular unit never rises over 3.750VDC, I'm skipping the optocoupler. But first, a small hurdle: no sync socket. I could have used a female hot shoe with a cord, but the plastic already had a hole for one with a sticker over it (ahh, mass production) so it was a quick mod. I used a RadioShack 3/32" jack/plug pair. (surgical procedure image)

For a timer I use one of the Arduino Duemilanove boards, I send two pins high (+5V) for a few seconds, then drop the voltage. I started incrementing delay between camera pin and flash pin going low, and immediately started seeing what fudgey's original thread was about... the USB triggering delay is variable.

The delay seems very random at first, but after a ~800 shots, I think I have the delay dialed in for the (or at least this) SX100... the best case center-point seems to be 40ms, +/- 1500µs. It will take a few thousand shots to drill down and average out, but I don't think the accuracy can get much better unless there's a way to make the camera triggering more repeatable. Like microsecond repeatable.

Still, wow! The SX100 doesn't have that much shot variance. 1/1000th is 99.8% (image), 1/2500th is nearly 100% too (image), but many are only partially lit, perhaps 1/3rd are full flash. Even overriding Tv to 1/5000th catches shots (image) -- it's erratic -- but it does catch them. This is with the flash in 1/16th power mode in manual (1/10000th duration), so whether or not it's actually ever truly sync'ing up at 1/2500th or not, it's still true that a 100µs flash burst is being caught by a 400µs shutter window on a $200 P&S pocket camera with an old '80s flash unit.
Cool. 😁

Next step is to test the SD1100, actually both to compare the unit-to-unit delay, since I need to do this anyway to make a good stereo sync... this is going to take a lot of time. Next weekend, hopefully.

Question to the functioning brains: Do you think there's any (worthwhile) chance to improve the shot delay variance with CHDK, either globally or per-camera?

Well, anyway, thanks for reading.

Attached: Token high-speed sync off-camera flash shot.