adrian's domain

Updated version available here. Version 3.1 is complete.

It’s done!

Version 3.0 of my AF assist tool is complete.

Taking a step back, I should explain why I needed this tool.There are really 2 reasons I built this:

1. When photographing concerts or clubs, or any other venue where light is at a premium, the AF sensor in the camera can only do so much to give accurate focus. I am using off-camera flashes most of the time now so I am not relying on the ambient light to paint my scene; that’s what the flashes are for, but I still need to focus. No remote flash trigger to date has had a AF assist light built into it. I use a Pocket Wizard Mini TT1 trigger and Flex TT5 receivers to trigger my flashes, along with an AC3 zone controller (which allows power control of 3 flash groups). So with a flash trigger in the camera’s hotshoe I loose my AF assist light, which normally I would get from a regular flash.

2. DSLR cameras (that I’ve looked at) don’t do AF assist in AI-Servo mode. They don’t allow you to track a subject in the dark, by continuously adding the light from the AF assist lamp on flashes to help the AF sensor in the camera continuously maintain focus. You get One-Shot focus, and once focus has been acquired, the light goes of. But if I want to track a singer as the move on a stage, like one would track a moving person or a race car during the day, you’re stuck. You can’t. This AF assist solves that problem but allowing full control of the AF assist light to be projected as long as you’re focusing (focusing is done by 1/2 pressing the shutter). Read more…

I think I have a schematic I am happy with to drive either an LED or laser module.

The 3.3V from the camera comes from the Remote Shutter connection on the side of the camera. One of the pins puts out 3.3v continuously, and when the shutter button is 1/2 pressed, the 3.3v drops to 0v. The op-amp is a buffer to separate the camera from the circuit, and to provide no load on the camera connection.

The pnp transistor turns on when the voltage is 0Volts at the base, and turns off when the voltage starts to increase. The potentiometer injects a little voltage into the transistor allowing for some control of the brightness.

Will build this on the weekend and will comment on how well it worked out.

The op-amp is a low power op-amp (NJU7016D) which will work down to 1V. To get a constant 3V supply to the circuit I’ll use a step up DC-DC converter (893-1150-1) which will take anything from 2-6V and convert it to 3V. So I can use rechargable Ni-MH batteries without a problem. I suppose I could use Ni-MH batteries but the light output will not be as bright. Will need to test this.

Read more…

I decided to take apart the AF module from the 550EX flash mainly to see if I could use my own LED in the AF module lens.

The unit is comprised of a lens and holder assembly, and an LED assembly. The LED assembly has the LED (which I believe to be a near IR LED) glued to the back of a lens.

The pattern the AF assist module projects is created right on the LED itself, unlike the 540EZ module. The pattern created by the 540EZ module is drawn on a small piece of plastic that lives inside the module. On the 5503EX module, the pattern is right on the LED.

So it looks like there’s no replacing the LED in this module.

Using the same laser pointer as in my previous post AF Assist tool v3.0 – laser pointer trials, I decided to stick a cap from a dollar store laser. You know the caps. The ones that draw hearts, or arrow, etc.

I tried a few, and one really stood out as sharp pattern, and coverage across the largest area of the frame.

Frame coverage at 50mm, on a crop body (x1.6). Not a very large coverage, but the contrast it produces is very defined. Even at 40ft the image is crisp. All focus points have no problem locking on even on f2.8 lens (not just the center cross type AF point)

I will have to investigate other laser pointer caps to see if I can find something that has a larger area coverage and provide lots of contrasty details for the AF sensors on the entire frame to use to acquire focus.

There is this diffraction lens I can get  but I don’t think it will project much larger then the bow and arrow I already tried.

Ideally something like this
Frankfurt Laser Company – Grid
or this
Frankfurt Laser Company – Ring/Concentric
would be perfect, but again, it all depends on the size of the image it projects.

On the other hand the pattern can’t be TOO large because at distance it looses it’s contract effectiveness as the pattern gets to spread out.

This pattern will get too thinly spread out. It’s great if you’re only a few feet from the subject but much more then that and the space between dots/patterns gets too large and the AF sensors/points can too easily fall between the patterns.

Did some more experimenting.

The problem I am battling with (one of them at least) is that using the AF module out of the Canon 550EX flash uses way too much current to be used almost continuously. In my last post (AF Assist tool v3.0 – more info) I measured 400 mA current draw. That will kill a set of 2500mAh AA rechargeable batteries in 5hrs ish. Not to mention the beefier power electronics that are needed to handle nearly 1/2 amp continuously. (though as I write this, I’m beginning to think that maybe 5hrs may not be too bad. Assuming that the tool will be used 50% of the time, that puts the usage time at about 10hrs. That’s a day’s worth of shooting. Not outrageous.)

What if I take the AF assist module from the 540EZ flash, and shoot a defocused laser beam through it? I found a Ø 4mm laser module from DealExtreme and stuck it where the LED would go on the 540EZ AF assist module. The laser runs on 3v, and draws 25mA.

The pattern of the bare unfocused laser is just a splotch of light. When the AF assist module is added to the mix, a clean sharp pattern emerges.

The images shown above are the actual frame coverage at 50mm on a crop body (x1.6) Canon 40D. The pattern on the right has no problem getting the camera to focus a f2.8 lens in pitch darkness at 40ft away even on the external focus points (that aren’t cross type like the center one). A f2.0 or f1.4 lens should be able to focus with a much less intense laser power.

The pattern of just the defocused laser (on the left) will focus with the center point only, and it’s hit and miss, and takes longer to focus.

I may be onto something…

Light source

I just received an AF assist module out of a Canon 550EX. Bought it here on Ebay from seller procamerarepair. The ebay item title is ‘CANON SPEEDLITE 550EX AF LED ASS’Y NEW REPAIR PART OEM’

It’s the same part as the AF assist lights in my Canon 420EX flash

So I measured both the voltage and current at the LED in my 420EX flash. The voltage across the LED is 2V. The power drawn is 400mA (yes, 0.4Amps; pretty crazy, I know). The Supply voltage (with the LED disconnected) is 5.5V. That means the resistor that’s in line with the LED must be 8.5Ω.

I powered the ebay LED with 3V and with a 3.6Ω and it was drawing 0.25A and it seemed just as bright as the flash AF assist LED. I had them both going at the same time.

http://www.everlight.com/datasheets/Shuen1W_datasheet.pdf

Because of the high current draw, I am betting these are IR emitters rather then visible light LEDs. I have seen what a 30mA red LED looks like, and it’s just as bright as these LEDs. So for the extra 10x the power draw, I am betting on them being IR emitters.

I found a similar power IR emitter at Digikey. 400mA, voltage drop of 2.2v. 730nm, which should be a deep red, considering 660nm is the other LEDs I bought, and they’re pretty bright red. Will look for more, but will order this one next time I put an order in. They’re not very cheap at $4.16 a pop.

I don’t really need to buy an LED for this. It’s more for me to figure out roughly what wavelength these AF assist LEDs are in commercial flashes.

Remore trigger port characteristics

I measured the 3 pin connector on the side of the camera where the remote trigger goes. One pin is ground, one pin corresponds to 1/2 shutter press, and one pin corresponds to full shutter press. I don’t know if the full shutter press also engages the 1/2 shutter pin.

What happens on those pins? When the shutter button is not pushed, there’s 3.3v between both pins and ground. When the shutter button is pushed 1/2 or full, the respective pins go to 0V (ground). So if I’m going to use an op-amp (which I am) to pick up and use the the signal, it will have to be an inverting comparator. When there’s a voltage on the op-amp input I need 0V to be seen at the output and when there’s 0V at the op-amp input I need to see full voltage to drive the LED.

Speaking of op-amp, I will need to source an op-amp that can handle 1/2 Amp. The LM358 I have I think can handle 30mA, which is fine for the laser pointer.

Going forward

I think I will pursue both the laser pointer (line laser) as well as the AF assist emitter from ebay (the 550ex emitter)

From my last post(AF Assist tool v3.0 – more testing)I discovered that the LEDs I have found are just not cutting the mustard. Today it just happens that I receive a red cross hair laser I ordered from Deal Extreme a few weeks back. I had all but forgotten about it.

So I stuck 3 volts onto it, and put it through the paces like I did the LEDs in the last post. The laser is bright. At 40ft (in pitch black), it’s very visible. Too bright in fact. So I stuck some resistors in series with the 3v battery to see if I can dim it. At 60Ω, the laser is dim. So dim in fact that in a brightly lit room it’s hard to see at a few feet away on a white wall. Yet when I turn the light off, and point at the wall 40ft away, the camera can still focus (with the center focus point) on the flat featureless wall. (I will post some pictures soon)
Perfect!

Not having had much success so far with LEDs and this laser module arriving in the mail I think I will rebuild my tool using this cross hair laser. Really I just need one line. The perpendicular line is of not much use. I really want this line laser from DealExtreme. I just ordered a few.

Ideally it would be nice to have several lines (close together) to allow for a little misalignment of the laser line and the focusing points be projected out of the same laser, but I haven’t yet found a diffraction lens that will do that.

Since contract is really what makes the difference in helping the AF sensor focus (as opposed to more light), the laser line is perfect since the light it produces remains sharp (distinct edges) up to a very long distance. And the line gives the enter focus point something to focus on regardless of the subject distance, whereas with projecting a spot, the will be some distances where the center focus point will not line up with spot of light.

As a start, I will add a potentiometer in series with the battery to adjust the brightness (a 100Ω pot should do it). Eventually I want to go with full PWM dimming control, so I don’t needlessly drain the battery.

Now if I can only find a laser of a higher wavelength, so that the light is nearer IR. The current laser is 650nm. Something in the 750nm range would be interesting to experiment with. If anyone knows of a source of inexpensive, low power (under 5mW) near IR laser modules (700nm – 800 nm), please let me know.

UPDATE:

So you can get any pattern projected… Frankfurt Laser does a variety of patterns that may work very well. The Concentric circles I think would be awesome since it could cover most/all AF points very nicely

I tried out the 540ez lens with both the LED that came with the assembly as well as the LED I bought from digikey. The digikey LED was brighter then the led that came with the 540ez lens. Brightness was judged with the LED installed in the housing, and the projected pattern was the object of judgment.

However, when it came to what the camera saw, it was a different story. At 40ft (the length of my basement; can’t go any longer indoors) the camera was pointed at a blank wall, and all the lights were turned off. The LED that came with the 540ez lens assembly was able to get a focus confirmation out of my 40D whereas installing the Digikey LED in the same lens assembly, did not provide AF confirmation. This was very interesting because the same contrast producing pattern was projected, and even though to the human eye (my eye) the digikey LED seemed brighter, the camera saw something else. At 40ft, I was could not visually  see the pattern projected by the 540ez led (yet focus was confirmed by the camera). With the Digikey LED, I could see the projected pattern (just barely) but the camera was not able to use that information to focus.

The Digikey LED is of 66onm wavelength. I can only guess that the led that came with the 540ez flash is higher then that (nearer to infrared).

I also tested a few other flashes to see if I could achieve AF confirmation on the same blank wall in total darkness, 40ft away.

Canon 420ex – AF confirmation achieved (pattern just visible 40ft away)
Nissin Di622 mark II – AF confirmation achieved (pattern not visible 40ft away)
Younguo ST-E2 – AF confirmation achieved (pattern clearly visible 40ft away)

Looking at the flash quoted AF assist distance I compiled for a few flashes seems to be understated?

The Plan:

Near-infrared is starts at 780nm and goes up from there. So I’ll have to find some LEDs that are between high 600nm and 800nm
OR
Find some brighter deep red LEDs.

UPDATE:

For the next version of the autofocus tool I’m currently working on I need to test several wavelengths of IR light .
3 x IR LED 770nm (p/n: 15P2280)
3 x IR LED 830nm (p/n: 93K0966)
3 x IR LED 850nm (p/n: 97M1058)

So I got some parts from DigiKey today; enough to  do some more testing. Took out the ebay purchase I’ve had for a few weeks, and got to work.
I picked up this AF assist lens from a Canon 540EZ flash, complete with LEDs.

CANON AF LED UNIT SPEEDLITE 540EZ FLASH NEW

I hooked up an LED to the 540EZ AF lens along with a resistor (200 ohm) and a 9v battery.

This is what I got in terms of a pattern. Shot on the left was from 7.5ft, then 15ft, then 30ft. At 30ft, I could barely, just barely make out the pattern on the flat white door onto which I was projecting the pattern.

The camera exposure was:  ISO:3200, aparture: f1.4, exposure: 1/40

Again I must stress, there was VERY little light on that door from 30ft away, however the AF sensor on the camera locked in an instant. Maybe took a fraction of a second longer then from the 15ft distance, but it was hard to say.

I can hardly believe that such little amount of light can produce such a focus lock on a flat color (white) door in total darkness.

I took out my 420EX flash and fired the AF assist light (was not able to capture it since the flash was on the camera, and the light goes off once AF is achieved) and the light is much brighter. Maybe 1-2 stops brighter from 30ft. Maybe I need to use a lower value resistor to get more power out of the LED that came with the 540ez lens, or just replace it with a brighter LED.

Either way, it looks like adding contrast is MUCH more effective then just adding light to help the AF sensor focus and the idea of using a flash AF assist lens is looking much more promising then the laser pointer idea.

The 540ez lens module is kind of bulky. I would prefer it to be a little slimmer. The AF assist lights in my 420ex is made of 2 side by side lenses which are a better size. Will need to scope out some of those lenses.

In the search for a focus solution, I am going to abandon the laser pointer light source in favor of an AF assist light from a proper flash unit.

Doing some research here and looked at a whole bunch of flashes, and some of the flash manuals have a range in which the AF illuminator/assist light is effective. The listed number is the max range listed in the official user manual (in meters and feet). Read more…