I notices that when milling a board who’s NC code was generated by Copper CAM, there is an offset between the drilled holes and the etching of the tracks.
Once I drill the holes, I’ll enter an offset of X-1 and then zero Mach3. That shifts the subsequent steps (etching, cutout) by 1mm to the left.
On closer examination, 1mm may be too much. I will try 0.8mm on the next board to see how it turns out.

UPDATE:

It turns out that my mill is not perfectly perpendicular to the board. The error comes out of the fact that the end of the tool for the engraver mill and the drill is much different and when a small rotation is added to the mill head a linear difference is observed.
I adjusted the head rotation and now both the milled and drill points line up perfectly.

Looking into how I’m going to get audio into the different rooms of this house where there is no structured wiring in behind the walls to take advantage of.

http://www.avsforum.com/t/1341706/any-alternate-product-just-like-sonos

http://www.musicmultiroom.com/

UPDATE: pricing out some components, and I think I will go the DIY route with a squeezebox backbone. I have a squeezebox classic and radio already. In each room that I just want some music that doesn’t need to be played at loud volumes I can put a squeezebox radio (they go for about CND$100). In the living room I can use a squeezebox classic or squeezebox receiver with some quality speakers and a small class T amp to drive them. As a reference, the Sonos Play:5 is $450. For that kind of money I can put together much better components (speakers, amp) that will be much louder.

For compact class t amps, read this: http://madgoat.hubpages.com/hub/High-quality-inexpensive-mini-power-amplifiers

More on what small amp to use with a squeezebox: http://forums.slimdevices.com/showthread.php?91027-Stylish-tiny-amp-for-Touch/page1

100W Class T amp on ebay (w/ power supply): http://www.ebay.ca/itm/2-X-100W-Class-D-Audio-Amplifier-Combo-Kit-w-24V-350W-Power-Supply-TK2050-/350604100288?pt=US_Home_Audio_Amplifiers_Preamps&hash=item51a1a202c0#ht_4095wt_1080

A lot of the class D or class T amps seems to be based on the Tripath integrate amp chips. TA2020 is a 30W amp(@ 4 Ω), TK2050 is a 100W amp (@ 4 Ω) .

This also looks like an interesting option. In wall keypad to control with amp built in: http://www.pyleaudio.com/sku/PVC8U/In-Wall-Background-Music-Amplifier-with-USB-And-SD-Card-Player-With-35mm-Input-Jack-For-iPodiPohne-and-Volume-Control

Made some progress today.

Measured the wavelength of a couple of Canon AF assist lights.

AF assist light from a Canon 540EZ flash. The wavelength is about 725nm.	AF assist light from a Canon 550EX flash. The wavelength is about 700-715nm.

Now to look for some LEDs in that wavelength.
The other requirement is that the light source needs to be able to handle several hundred mA. (so the 20mA regular LEDs are out of the question)
I did find a 1W 730nm IR emitter at DigiKey that may do the trick. I’ll have to order one and try it out.

http://arduino.cc/en/Main/Standalone

Work on the new AF Assistant has gotten to a point where I thought I would post some progress.

First off, the name is slightly changing. I am calling this one the AF Assistant IR. This will be version 1.x
As the name implies I am working on using an IR lightsource (well, near IR) from a Canon 550ex. I have mounted the LED module in a case which sits under the camera. It mounts to the 1/4-20 threaded hole on the bottom of the camera. The unit is still triggered by the remote shutter release port on the side of the camera just like the AF Assistant v3.1

Fig 1. Pattern at 3 ft.

I have chosen to align the light so that the projected pattern of the module is horizontal (as opposed to the vertical line the laser  AF Assistant put out). Due to the fact that the pattern is not just a line, and that the module is closer to the lens axis then a hotshoe mounted device, parallax errors in alignments are quite small. It seems that (in the first alignment of the LED module), the center focus point can pick up  on the projected pattern from 3ft to infinity. That’s more then enough for me. I can probably tweak it to get a closer distance, but I’ll play with that later.

Fig 2. Pattern at 30ft.

Several advantages are already apparent by using the LED module.

- The pattern that this puts out is quite wide. So wide in fact that with a 50mm lens on a full frame sensor (of a 5d mk2) the side (left or right) focusing points are able to use the pattern.
- The light the LED module puts out is quite dim. Figure 1 (at 3tf) was taken at ISO6400 , 1/30, f1.4. Figure 2 (at nearly 30ft) was taken at ISO25600, 1/13, f1.4. In a totally dark room, at 30ft, I can just barely see (with the naked eye) the pattern on the wall 30 ft away.

For both the 3ft test, and 30ft test I was focusing on a flat wall with no features (in full light, the camera can not focus on the wall, since there’s nothing there to focus on) and with the AF Assistant IR light, I can focus on the blank wall at 30ft away even with the side focusing points. Quite impressive.

Some internal details

Fig 3. LTC3490 LED driver in action

I have moved away from the 1 battery setup of the previous version. The LED is drawing up about 250mA (which is 10 times what the laser was using) so I figured I’d move up to 2 x AA cells.
I have also moved away from thru hole for most components. Surface mount technology is what I’ll be focusing on from now on.

The LED driver IC is a LTC3490. It can drive up to 350mA though  I cannot drive this LED that high. The LED module heats up too much for my comfort. At about 230-250mA it still gets warm but I can keep my finger on it indeterminately. Anything above that current and I can’t keep my finger on it for too long. There is a trimmer pot on the board so I can adjust the current to the LED but I may take it out in the production version.

So far all the tests are looking very positive. The other surface mount IC is a MAX4289 op-amp. It’s able to run on low voltage (down to 0.7V). Most other op-amps need at least 3 volts to run, which I may not have if I’m running two NiMH batteries at 1.2V a piece.

Fig 4. Fully populated board

I’ve only been playing with this setup for a couple of days, and will be testing it at music festival next weekend.

UPDATE:

I took some measurements of the wavelength of light from a couple of AF assist lights form Canon flashes. Read about them here.

LED Drivers

Single or multi cell LED driver solution 800mA current

SINGLE OR MULTI CELL LED DRIVER SOLUTION 800mA current

350mA LED STEP-DOWN CONVERTER

500mA LED STEP-DOWN CONVERTER

Up to 1.5 A Constant Current
Switching Regulator for LEDs

Up to 1.5 A Constant
Current Switching
Regulator for LEDs with
ON/OFF Function

LOW-POWER DC/DC BOOST CONVERTER (250mA and 400mA versions)

Single Cell LED Driver, Variable Current (max 350mA)

http://forum.allaboutcircuits.com/

So adding a capacitor and resistor in series, adds a delay.

T=RxC

So for a 0 .1 sec delay, I need a 100μF capacitor and a 1kΩ resistor.
For a 0.5 sec delay, I need a 100μF Capacitor and a 5kΩ resistor.

Process to get PCB boards made.

- Design in Eagle, and export from the CAM module in Eagle layers 16, 17, 18, 20, as GERBER_RS274X
- Also export layers 44 and 45  as EXCELLON
- Import gerber file and excellon fine in CircuitCam
- Isolate as required (using only small and standard tools), and batch export DXF ;this outputs 4 files (small tool path DXF, Standard tool path DXF, Drill DXF and border DXF)
- Import DXF files in CNC USB Controller software, and export “Toolpath to G-code”
- Each DXF file will have a .nc file to be cut with the appropriate tool.

Work has been progressing on a newer version of the AF Assistant. The next incremental version (v3.2) will be made using surface mount components, and extra safety precautions have been added to protect the camera from any unwanted electrical signals.
Both inputs from the camera are isolated using buffers, and the ground is protected against reverse voltage.

Testing on this new design should start soon, and once complete all new AF Assistants will be built using this process.