Some cool flash photo images:
No Hands
Image by GoneBush
Ali and Julien on the No Hands boulder Problem. Arapiles, Victoria.
furry Flash
Image by Cryptonaut
Muppet Flash? Flashy Monster? There were a whole bunch of fuzzy superheroes, but this is the only one I got a photo of.
Highspeed RFM12B remote flash control design by fotoopa
Image by fotoopa
Highspeed RFM12B remote flash control design.
The drawing give an overview of the many delay details. For highspeed flash control some of the flashes need to be wireless and others are direct connected. Wireless need some time to transfer the commands to the remote units. I tested a fast wireless system using the RFM12B chips from Hope RF. The best timing reached is only 361 us delay between command and flash activating. Normal this short delay have no influence on the flash lighting. Most cameras works on 1/250 sec giving sufficient time. But for highspeed capture of very fast moving objects all the flashes need to fire as short as possible and at the same time. If some flash is delayed compared to the others then the time between the flashes gives extra blur into the picture. Especially at very high speed and short exposure the flash durations are short. For lowerpower setting this times become below the 100 us. in this case a delay from the wireless system near 400 us is to long.
Because the camera give the startpuls for the flashes, the flash on the camera will start be to early. The only good solution is to send this X-sync start signal to your controller that drive the wireless system and to extra delay this no remote flash. If there is a big difference betweem the powersettings of multiple flashes, the best result can be reach by calculating the central position of the highst power flash and adjust all the timings of the other flashes on this central position. The drawing show this ideal central position calcultated from the powersetting of 1/16 in this example. The non remote flash need to be driven in this case via your controller. This controller know all the timing relations between all the flashes.
Now I have also tested the normal shutterdelay of the D300 camera. Surprise the shutter delay for 12 bit and 14 bit RAW are different. Oke, search on Google show that this problem is not new, but for me it was a little surprise. So I tested the 2 settings and I read 54 ms for the 12bit and 96 ms for the 14 bit. Both are a little longer then the value given by Nikon but this are the reality in my case. (All timings are measured with the Intronics logic analyser). The drive signals for the camera fire commands are generated by a Terasic DE1 FPGA board.
For my highspeed laser detection there I have only 6.5 ms shutterdelay for the second shutter. There I can calculate even very precies the needed delays to fire the flashes optimal very early in the flash window (NR:8). This optimal delay is 6.5 ms and can be used for all powers becaus I can calculate the delays from this fire start (NR:3) because I have early a start signal. For normal camera use, the start signal is furnished via the X-sync signal so the flashes are drived a little later see (NR:4). But the only difference is a bit shorter shutterdelay. Shorther give less displace for the flying object.
Conclusion is that even for normal camera use, you have the possibility to fire all the flashes optimal even if they are remote, wireless or direct on your controller. The wireless delay of 361 us is tested here at 10m, 2 walls, 868 Mhz and 115 Kbaud data rate.
I hope with this information that most people have a good idea of the preciese and specific problems using highspeed flash photography.
See also the RFM12B minimal delay measure:
www.flickr.com/photos/fotoopa_hs/3762346539/
I build now an Freeduino AVR controller (Atmega168) as wireless receiver to drive 4 Nikon flashes in TTL mode. Each flash can separately set to a powerlevel. The ideal central timing is calculated from this powerlevels as show in detail 8.
Many parts of the software are ready now. Into a few days I hope to test the first version.
See new info:
www.flickr.com/photos/fotoopa_hs/3798474398/
Update 2009 Aug 18:
The hardware is ready on the breadboard together with the STK600 developement board. LCD display connected via the 2 x 74HC595 ICs works and also the leds. Data is diplayed on the LCD. The 3 local flash outputs on this master are also ready. The next days I will continue to put the RFM12B module as transmitter. Thereafter the tests will be done together with the slavemodules.
Update 2009 Aug 31:
The SC-27 TTL cables from Nikon just arrived this morning. A lot of extra other components will furnish tomorrow. Extra LCD displays, registors, connectors,swithes, plugs etc. End this month 10 extra Arduino boards and shields will be furnished. Time to plan how to build all this together. In the meantime I test the software on both sides (Master and Slave).
Update 2009 Sep 02:
Many components just arrived today. Extra LCD displays, connectors, switshes chips etc... I work on the software and most parts are working now. Into the next days I will connect the flashes via the SC-27 cables and test the flashes. 4 flashes on the slave and 3 flashes on the master are supported. The LCD display show the different powersettings, flash types and used channels. Once all this works, I have to build this into a nice box. At the end of this month, 10 extra Arduino boards and shields must be delivered and also 10 extra Atmega328P chips.
On an extra Arduino board I will drive 8 highpower leds just near the flashes to see into the dark. So just wait a few weeks for all this new tools....
Update 2009 Sep 04:
Master software is running now, LCD display show all this values. Powerlevel can be change on every flash but also on a groep flashes. Results stay on the LCD display. Every flash can be enabled or disabled. The groep offset value have now 6 steps. This mean, the steps are added to all the flashes levels into the groep. 2 different groeps are supported, one Master groep (4 flashes) and one Slave groep ( 3 flashes). But the number of flashes into each groep is a parameter into the software code. The number in now limited by the I/O pins on the Atmega168 or Atmega328p. Now the step change is near 0.3 EV. On each channel you can select the type flash connected at the output. So if you have a SB29S ringflash you just need to change the type for the connected channel on the display. All flashes are connected via a small type phone connector 3.5mm stereo on one side. The other side is the SC-27 Nikon connector. I just divide every SC-27 cable in 2 parts. So now I have 6 cable to connect 6 Nikon flashes.
Now I have to build this hole Master into a nice small box. The display works on 5V, the other chips are running on 3.3V. Power will be delivered via a small 9V NIMH battery.
The next task is now to test the previous build Slave unit via the RFM12B modules. Software on both sides are ready for test now.
