Randall Spangler

I'll be interested to see if you find the same discontinuity in the TTL response curve that I'm seeing. I'm using Cree C503B red LED, driven at max current. 40us preflash trigger pulse, ~50ms between preflash and main flash, and then variable-length main flash trigger pulse. 13us = ~GN2.8 smoothly up to 60us = ~GN5.6. (output flash pulse 36us...70us) Then between 60-70us it jumps directly up to GN8+0.5EV. There is no input pulse length which generates an output flash pulse length between 70-160us. Above that it's a smooth curve from GN8+0.5EV up to GN22 at 1000us and GN22+0.3EV at 1600us. (output flash pulse 160...1400us) Unfortunately, GN5.6 - GN8+0.5EV is kind of a big gap, right in the range where TTL exposures are likely to fall. I don't know if that's something they can fix with another firmware update. Why do strobe manufacturers not simply have a straight linear profile for us LED-based trigger board folks? Something like: 100us input pulse = GN2.8 150us = GN4 200us = GN5.6 250us = GN8 300us = GN11 350us = GN16 400us = GN22 450us = GN32 That would be really easy for us to generate, and would allow plenty of accuracy on the strobes themselves (much more so than the exponential lengths they try to capture now, which are really short for low power and really long for high power). And it's not like the strobes don't already have different profiles built into them for number and timing of preflash pulses; this would just be one more profile. If we standardize on that profile, then ANY strobe would work with ANY trigger board. No need for us to profile them every time a new strobe comes out. That'd be good for the strobe manufacturers too; more potential customers for their new strobes without needing to wait for us TTL board folks to profile them.

So, my second YS-D3 died last month in Roatan. And is getting fixed at Backscatter now. New main circuit board, for about half the price of a new strobe. Though sad it went after only about 7500 shots. And I got my hands on a YS-D3 Mark II, which is... different. It is indeed more sensitive to LEDs than the original; I can trigger it with a 3 us pulse, rather than 35 us. TTL has a wider usable range, with a 13 us pulse corresponding to about 1/128 power and 1400 us to about 1/2+0.3EV. And the preflash pulse is properly brighter, 1/32 instead of 1/128 on the original. Which would make this a great TTL strobe, except... there's a weird discontinuity in the flash's TTL response curve for input pulses in the 60-70 us range, where output power jumps suddenly from 1/32 to 1/16+0.5EV. I'm wondering if the two photodiodes inside the YS-D3 (see my previous post) respond to different input ranges. Maybe the first one works for >70us, and that's the only one which was functional on the original YS-D3 for LED triggers. The second one is higher-gain and saturates at around 70us, but responds to shorter dim LED pulses. That would explain the discontinuity. The original YS-D1 had a max preflash input pulse length of around 65us, which is suspiciously at almost the exact same place. But there'd still need to be a firmware bug in the YS-D3. I'll see whether I can get a working profile. That weird output jump will be a bit of a problem. I shoot RAW and the a7RIII's pretty tolerant of +/- 0.5EV mis-exposure, but something right in the middle will end up 0.7EV mis-exposed. I'll also be curious if my RMA'd strobe behaves like my original YS-D3, or like a Mark II. Or worst-case, something in between; that'd leave me with 3 different YS-D3 variants, and no two that I could pair together for TTL. Still very annoying, since the YS-D2J didn't have any of these problems. It just had the problem of overheating and shattering its tubes. Sigh.

I'm hoping that a 4-shot burst at 8 fps while moving the camera towards the subject will work. Inhale slightly, exhale while dropping down slightly towards the subject and use focus peaking to anticipate when to fire the camera. Same thing I do now, but right now I only get one shot at a time with my YS-D2Js. I'm also hoping those bursts will help for black water dives. I'm using fixed flash power there anyway since TTL doesn't work very well on transparent critters in a black ocean. But AF doesn't work really well on little squirmy mostly transparent things either. Everything underwater moves enough I suspect longer bursts will result in too much scene movement to stack well. Above water, a tripod mount on a rail with a drive screw works pretty well. I agree that's not going to work underwater. (If I were made of money, a Sony a1 at 20 fps would probably work even better...)

Victory! I figured out how to convince the Sony a7RIII to match the actual preflash power from the YS-D3. Now I'm getting consistent TTL performance from GN2.8 - GN22, even in bright sunlight. TTL also works for 3 fps continuous shooting. In manual power mode, with the trigger board set not to fire a preflash, I can get 8 fps bursts (~8 shots at GN8, ~4 shots at GN11). Can't wait to try to use that with focus stacking in Photoshop on super-macro critters.

I got my firmware-upgraded strobes back a couple weeks ago, and finally had time to profile them today. The good news is the strobes are now very consistent triggered with a LED. In hundreds of test shots, I've never seen them not fire. And the odd double-triggering I was seeing with short pulses is gone as well. The response curves seem better adjusted to work with P&S cameras. I haven't played with them and my RX100, but I have a feeling they're going to behave better. The strobe response curves to LEDs are shifted even further than they were before. So it takes longer LED pulses to trigger a given GN pulse via TTL. That's not too horrible; I can get a reasonable range of output up to GN22 with the strobe set to +1EV. Though I don't seem to be able to reliably trigger the strobe with LEDs below GN2.8. But there's one new annoying behavior. The preflash used to be fixed at just a bit less than 1/16 power. Now it's capped at GN2.8 (1/128 power!). I think they did that to stop the strobe from totally saturating P&S cameras; that matches their release notes. But the Sony a7RIII would really like to fire a 1/16 preflash when using a 90mm macro lens stopped down. Oh, one other interesting tidbit: On most cameras, 1/3EV stops are translated to 0.375 and 0.625. The +/- 0.3 and 0.7 settings on the strobe appear to be +/- 0.25 and 0.75. 1/8EV difference shouldn't matter in practice, but it threw me for a while when I was trying to line up the profiles at different DS-TTL settings. - Randall

That's excellent news! Copying it here, to aid in searchability: When cameras open up to a wider aperture (lower f-stop), they often fire a weaker preflash. That would have trouble triggering the original YS-D3 firmware. The original firmware also always fires a 1/32 power preflash regardless of sync pulse and only syncs down to about 1/32 power, so if the camera needed less than 1/32, it would end up with an overexposed main flash. My strobes are 180700757 and 180700791, so need the update. I'll head down to Backscatter this coming weekend to see about that. I'm guessing they need to be shipped back to Sea&Sea. I have a pretty complete profile for the original firmware, so I'll be able to compare how the new firmware differs when I get them back. Fingers crossed.

I checked my notes from when I still had a working YS-D1. Here's some rough notes on the relative sensitivity of YS-D2 and YS-D1 when triggering in manual mode. Not directly comparable to the numbers I've quoted before because this was with an early dev board with somewhat different resistors. Red LEDs YS-D2, S&S cable = 1 us YS-D1, S&S cable = 3 us YS-D2, cheap cable = 4 us YS-D1, cheap cable = 10 us White LEDs YS-D2, S&S cable = 2 us YS-D1, S&S cable = 6 us YS-D2, cheap cable = 4 us YS-D1, cheap cable = 18 us IR LEDs YS-D2, S&S cable = 1 us YS-D1, S&S cable = 7 us YS-D2, cheap cable = 3 us YS-D1, cheap cable = 29 us Green LEDs YS-D2, S&S cable = 3 us YS-D1, S&S cable = 10 us YS-D2, cheap cable = 6 us YS-D1, cheap cable = 31 us For the same current limit on my trigger board (~80 mA), red LEDs were most efficient in triggering the strobes. And YS-D2 was actually more sensitive than YS-D1. I also couldn't reliably trigger a 1/256 power TTL pulse on YS-D1, but can on YS-D2. They do have somewhat different profiles, though, so a profile optimized for YS-D1 may not work as well on YS-D2. In particular, the lookup table for preflash intensity is different that that for the main flash. And on the YS-D2, it's capped at a fairly short number (~75 usec). Longer than that, and the YS-D2 will fire a full intensity preflash, and then the main flash won't go off at all (because it's recharging). From the user's point of view, it's a nice bright flash, but the photos are still dark. For comparison, YS-D3 seems to have given up on measuring the preflash, and always fires a 1/32 power (GN5.6) preflash regardless of the preflash trigger pulse duration. I wonder if that was a firmware compensation for the sensitivity problems in the photosensor circuit; GN5.6 is right where I really have trouble triggering it reliably at all.

Interesting. I wonder how they implement that? Different lookup table is easy. But that's duration, not sensitivity. Switching different resistors onto the photodiode would work. The photodiode produces current as a function of input intensity. Higher intensity means more current. If the circuit has too much capacitance, then that extra charge takes a while to die out and messes up the duration math. For comparison, my photodiode that I use to measure strobe duration is actually just a LED with resistor soldered across the leads. Not extremely sensitive, but it doesn't need to be when it's right in the path of the flash. Unfortunately, outside of Retra, I don't know of any user-upgradeable flash firmware. I haven't torn apart a YS-D2 or YS-D3, but the YS-D1 uses a Renesas R5F212B7 SNFP 8-bit CPU. There's a debug port on the board (edit: double-checked the board; it's labeled "DEBAG" ), so I suppose I could solder up a connector and see if I can read out the firmware. I'm not quite willing to do that with a strobe I might have to RMA, though, so the YS-D3 is off-limits (for now). It seems like we should be able to come up with a way of comparing trigger brightness. Something like: Use a multi-stranded cable in the camera housing Hold the strobe end 1" away from a piece of white copy paper Change aperture/ISO of the camera until the center of the illuminated spot is around half-brightness as shows by the camera JPEG. That wouldn't be exact, but it'd get it within a stop of brightness, which should be enough to make some comparisons. I could also compare with the light from my RX100's flash.

Yeah, as far as I can tell, that's what strobes are doing in their fancy adjustable TTL modes. Slave TTL mode does directly mimic the pulse length, with no delay. Which would be great for trigger boards that were calibrated for that. But it also seems to need a brighter sync pulse than LEDs can generate. I took apart my old YS-D1 after its tube finally died. There are actually two photodiodes in the sync port. I wonder if it uses one of them for DS-TTL, and the other for slave TTL directly to the triggering circuit. I haven't traced the circuits to reverse-engineer them yet. YS-D2J and YS-D3 seem to have the exact same photosensor board, though different than the YS-D1. See attached photo. Wow, that's hard to light and get a decent macro shot. Also note the two photodiodes. (Note for anyone contemplating disassembling strobes: They have BIG capacitors charged up to 300V+, and can retain that charge for a long time. The only difference between a strobe and a defibrillator is where you put your fingers. If you don't know how to safely shunt those capacitors to discharge them, don't open a strobe unless it's been sitting on the shelf for a year. Please be careful.)

If your ports face forward, you can measure the relative brightness by putting a piece of regular paper in front of the sync ports and take a picture pointing into a mirror. The light shining through the paper should be equal brightness from both ports. If both are overexposed / saturated in the picture, stop down, or add a second sheet of paper. If your ports face up, tape a piece of cardstock onto your camera and fold it so it's at about a 45-degree angle, so the ports shine onto the cardstock where you can see it in the mirror. You can do the same thing to test sync cables; point them at a piece of paper and take a picture of it. - Randall

I can generate arbitrary 1 us - 1 sec pulse lengths from my trigger board when I have it hooked to USB, accurate to 0.2 us, and I'm using a photodiode and 100 MHz scope to measure the strobe output. I'm also using about the brightest red LEDs money can buy. Red is the best wavelength for the combination of optical fiber and photodiode sensitivity (silicon photodiodes are twice as sensitive at 620 nm red than at green or blue). They're so bright you can't even look at them directly. I can't use a strobe in the housing, because there isn't enough space for me to put one that would cycle fast enough to use YS-D3 burst mode. About the only way to get more power to the other end of the fiber optic cable would be to switch to laser diodes. And then I'm worried I'd burn out the photodiodes (or my eyeballs). I'm using Sea&Sea's own multi-fiber cables, in good shape. And I'm still not getting consistent results from the YS-D3. With slave TTL, either (unless I move the fiber optic cable from the slave port on my YS-D2J to directly in front of the tube, where it makes a shadow on the subject). So I don't think it's the trigger board. I think it's the strobes. Either a design oversight or a manufacturing defect. I'm happy to share more detailed results and scope traces, or to look at anyone else's strobes (of any model) for comparison. It only takes about half a day to profile a strobe. - Randall

So, hey, if it IS a problem with early strobes, maybe we all have low-serial-number strobes. But if someone's having trouble with higher-serial-number strobes too, then it's a design defect rather than a manufacturing one. My already-dead YS-D3 is 180700757. My still working-ish YS-D3 is 180700791. If you've got a YS-D3 that isn't working with LED trigger TTL and your serial number is higher, what is it? (Silver sticker in the battery compartment).

Sigh, it's not just LEDs. Long story short, slave TTL mode doesn't work reliably, even triggered from a YS-D2J. I connected my YS-D2J to my trigger board, and then hooked my YS-D3 to the slave port of my YS-D2J using a Sea&Sea sync cable. Then held down the focus light button to switch my YS-D3 to slave TTL mode (cyan; backwards from YS-D2J colors...) Now I should be able to use DS-TTL on the YS-D2J, and the YS-D3 should follow along by matching the strobe pulse lengths. The actual strobe output lengths are pretty similar between the two (1/4 power is 302 us on YS-D2J and 341 on YS-D3; 1/32 power is 78 us on YS-D2J and 72 us on YS-D3; 1/128 power is 42 us on YS-D2J and 37 us on YS-D3). But... epic failure. Sometimes the YS-D3 doesn't fire at all, sometimes it fires at full power. It doesn't seem to be able to match the right pulse lengths, even though it's now being triggered by a real xenon tube. Ok, let's try the YS-D2 in manual mode, and see if the YS-D3 can follow along. At least then I'd only have to turn *one* knob to adjust power on both strobes. Also very weirdly, when I change the strobe power on the YS-D2J, the YS-D3 seems to get the first shot wrong, but subsequent shots are better. That seems like maybe a firmware bug? But it kinda works, at least at GN22 - GN8. At GN5.6, it doesn't trigger reliably. And at GN4 on the YS-D2J, the YS-D3 stops triggering at all. There's no longer enough light coming in the sync cable from a xenon tube flash. That's probably why the first experiment failed; the preflash on the YS-D2J (roughly GN5.6) isn't bright enough to trigger the YS-D3 reliably. But if I put both strobes on manual, the YS-D3 will trigger all the way down to GN1 on the YS-D2J (1/1024 power). And it'll trigger at GN2 on the YS-D2J even if I use an old and much-abused crappy and dim single-fiber sync cable. So the YS-D3 can see adequate light coming in on the sync port. But all its TTL implementations are broken. Really hoping this is a defect in the first batches of strobes.

I do fine with manual for macro photography, since I also usually drop to manual focus and use focus peaking to get exactly what I want in focus. Nudibranchs are polite enough to stick around while I fiddle with my strobe knobs. And with snoot work or black water dives, manual is your only option anyway. A mispositioned snoot means TTL will dump a full-power flash (heating up the strobe), and black water is, well, black. But shooting damsels and flasher wrasse is much easier with TTL. The subject-to-camera distance can change very rapidly. Or if I'm shooting blennies and I look up and a ray is swimming by, I can just take the shot with TTL. Or if I'm in current, and I need one hand to keep position with a tickle stick. Manual power means adjusting strobe power is a slower two-handed operation. It's great if you can set up your camera and then wander around looking for the shot that fits your setup. But not as good for spontaneous shots.

I'm really not sure what's going on at Sea&Sea to ship a strobe with such problems being triggered by LED boards - mine, Pavel's, etc. The whole point of having a fast cycle time in the strobe is to be able to fire shots quickly. And the only way to do that is with LED or electrical sync. Triggering with a small strobe in a P&S camera means waiting for that strobe to recharge, which takes 3-5 sec. Pretty much any external strobe will beat that. (Which is also why the YS-D2 overheating problems aren't really a problem with P&S cameras; you just can't drive the strobe hard enough fast enough to melt it. But with a LED trigger, you can...) The DS-TTL implementation in the YS-D2J is remarkably solid and extremely reliable, and basically unchanged from YS-D1 a decade ago. So I don't know why they would have changed that part of the strobe circuitry. I'm really starting to wonder if Sea&Sea got a bad batch of photosensors, or put the wrong reel of some other component in the pick-and-place machine. I can't imagine they'd design a high-end strobe that can't be used in TTL with high-end cameras with LED trigger boards. But then why wasn't this discovered in QA testing before the strobes were shipped? My hope right now is that I need to RMA my one YS-D3 which has already failed (!). Sea&Sea has always been good about sending replacements. And maybe the replacement will behave differently, showing that the initial production run really did have some defect they've since fixed. In which case, I'll RMA the other one. Otherwise, I'll end up using the YS-D3s with my RX100, and stick with my YS-D2Js for my a7RIII. And maybe we'll see a YS-D3J in a year or two with fixed DS-TTL for LED trigger boards?