Canon E-TTL

[seaton]

Found this site on E-TTL http://81.216.246.116/e/ettl/ looks fairly simple to implement, have put it the wiki. Can be used as a starting point for those interested in implementing E-TTL

[Rudeofus]

Looks great, but we should also take a look at the wireless ettl protocol. If at some point we decide to fully support ettl via RF, the wireless ettl protocol could give us a good starting point for a command set and timing to implement ...

Unfortunately I haven't seen anything about wireless ettl on the web, at least no protocol description

[Rudeofus]

Here is a first quick and dirty report on the wireless E-TTL protocol. Note, that this covers just a special case, but it should give a decent lead how to proceed. I hope I will soon have time to discover more details. There exists a german web page which also shows some measurements of the E-TTL protocol, however, no attempts are made at discovering the protocol.

1. Communication is based on pulse position modulation, i.e. not the amplitude of the pulses matters (since a flash can't control it anyway), but the position of pulses relative to each other.
2. It is possible to measure the pulses with a simple photo diode, resistor, power supply and digital storage oscilloscope. Since the individual pulses are a few 10s of µs long, nothing particularly fast is needed
3. The first two pulses determine the channel the flash operates in. If they are 100µs apart, we're in channel 1. For channel 2, 3 and 4, they are 150, 200 and 250 µs apart
4. After this some pulses follow which trigger the pre flash, set the flash settings, and finally, trigger the flash
5. The actual trigger flash appears to be just a single pulse (which surprised me)

I was able to trigger a Metz AF-1 (should work with any other E-TTL compatible flash, too) listening on channel 1 with the following pulse train generated by a micro controller sent through an LEDs:
pulse
100µs pause
pulse
600 µs pause
pulse
100 µs pause
pulse
330 µs pause
pulse
600 µs pause
pulse
200 µs pause
pulse

then a few ms pause
then pulse, which fires the flash.

[JonSenior]

Good work that man!

I think the thing that many people liked about the idea of using wireless TTL is that we actually just need to record the output of the master and reproduce at the slave. As long as we are faithful in this, we don't need to understand the protocol.

All that said... more understanding is always a good thing.

Jon

[Rudeofus]

Good work that man!
I think the thing that many people liked about the idea of using wireless TTL is that we actually just need to record the output of the master and reproduce at the slave. As long as we are faithful in this, we don't need to understand the protocol.

In practise we'll have to, since the slaves we want to control will fire pre flashes which we need to distinguish from protocol related pulses.

It's also nice if we could use non-E-TTL capable flashes in wireless E-TTL configurations, at least in manual mode (think of using an ST-E2 with two full auto slave groups, and one 540EZ in manual mode producing the back light)

Or go strobist alltogether, and trigger 4 manual flashes with the small ST-E2 as transmitter and 4 cheap self made E-TTL aware slave controllers.

All that said... more understanding is always a good thing.

Another motivation for decoding the protocol would be that pulse position modulation works well for optical data transmission but would be the worst possible modulation type for RF data transmission. We need to translate those pulses before we push them into an RF transmitter, which means we have to understand at least what kind of pulse sequences can appear so we can encode them.

[Thonord]

That is really cool. JonSenior.
Well done.

Have you identified anything that sets the duration of either pre-flash or flash?
I'm asuming that flash power is a function of time.


Tom

[Thonord]

Outstanding Rudeofus
Got you and JonSenior mixed up
He is Outstanding too

Tom

[Rudeofus]

Have you identified anything that sets the duration of either pre-flash or flash?

These experiments were done on an ancient DCS520 (Canon EOS 1n with Kodak digital sensor). In my experiments I did not even measure a preflash, so I'm afraid that this thing did something TTL, not E-TTL. I hope a friend of mine will bring his EOS 10D, so we can try out some more advanced stuff. Once I get that film out of my EOS 3, we may even try some really advanced tricks (lighting ratios, not sure whether the EOS10D can do this)

I'm asuming that flash power is a function of time.

I'm not sure my setup even did regulate power at all
But PPM implies that your theory is correct. The question I will try to answer soon is how granular the timing of wireless E-TTL is (will have major impact on our design! )