100 Watt CNC Laser Installation

To: gilbert@haaslti.com
Subject: CO2 processing head accessories.

Gilbert,

I’m following up on the phone call we had earlier for parts for processing head. I got the prices for most items, but I need a final quote, including shipping the parts. I’d like to receive them by Friday.

210CM-28030-X45-FS-DCH bend mirror $95
AC-100 alignment cards $ 65.00
210CM-28030-095-ZS-DAR 3.75in focusing lens $325
Cutting head extension for 3.75 in focusing lens, no part number, $65

Thanks,

Romos gave me some excellent feedback on my post about beam sizes. He points out that the expected beam size can be taken from this table for the G100:

Distance From Laser (mm) vs Beam Diameter (mm)
0 mm distance = 1.9 mm beam diameter
250 mm distance = 2.9 mm beam diameter
500 mm distance = 4.7 mm beam diameter
750 mm distance = 6.7 mm beam diameter
1000 mm distance = 8.7 mm beam diameter
1500 mm distance = 12.9 mm beam diameter
2000 mm distance = 17.2 mm beam diameter

In his case, the focal lens from the laser dinstace is 500 mm.
So, without any beam expander I have...(assume that M^2 = 1.5)

diameter = .013 * 1.5 * (38.1/4.7) = 0.158mm

The distance to my beam expander is 33cm, so using that chart the beam size will be about 3.5mm when it goes into the expander. The beam expander is 3 times the original size so the beam will go to 10.5mm.

Based on the equation:

diameter = .013 * 1.5 * (38.1/10.5) = 0.071mm

This is a great spot size. The problem will be my depth of field. This is based on the formulas shown on this site:

http://www.parallax-tech.com/faq.htm

Depth of field is the distance range that an object can be placed in front of the lens and still get cut. The forumula for depth of field is

DOF = 2.5 x wavelength x ( focal_length / beam_diameter )^2

for the G100 laser it calculates to:

DOF = 0.027 * (focal_length / beam_diameter)^2

I made a chart that gives examples of our these two situations.


The first row is what happens with for Romos. He has a focal length of 38 mm, beam diameter of 4.7, giving a 158 micron spot size, and 1.8 mm depth of field. My current situation I have a 10.5 beam diamter, a 71 micron spot, and a small small depth of field of less than a milimeter. That is probably unacceptable.

I tried to other variations. One would be if I had no beam expansion. My beam diameter would be around 3.5 (close to yours) and a slightly larger spot size. This might work. If you look at the first post it said I have to get at least 280 micron spot size, so it could be enough power density to cut metal.

The last row of the chart is where I use a focal length of 4 inches with the beam expander. If I do that, I get a spot size of 190 micron, and 2.5mm depth of field.

I know that since I may be cutting other materials I will want to have more than one focusing lens anyway, so maybe its time to look into purchasing them.

Taking power measurements requires simultaneous activation of the laser, starting a stop watch, and handling the power meter. I realized it was essential to be able to enable the laser manually -- preferably with using my foot -- so I made this:



I am entertaining the fantasy that this will be the "last time" I have to go into the controller.

The original top on my CNC table was a 1/2 thick plate of aluminum:




I bolted down 24 disc shaped neodymium magnets on this plate with a 4-40 socket head bolts. The idea behind these things is that they'll supply some magnetic force on the sheet metal to hold the sheet in place. The main thing I like about the system is that it is very level.




They work fine. The bolt head projects about 1/8th of an inch above the magnet but because its made of steel it transmits enough magnetic force up the post to adhere the sheet metal.




I get the feeling this isnt the last table I'll be making.

The question I got is if this system will work. One thing I'm wondering is if the beam hitting the aluminum plate will put off undersireable fumes. The other question I have is if there should be more distance between the part and the support underneath. The gap between the sheet metal and the aluminum is around 3/8ths of an inch. I think it'd also be better if I had a grate underneath the part. That would definitely make for better ventilation -- I could put a vacuum system below the part to pull off fumes. Comments are welcome.

I was reality checking some numbers on cutting sheet metal.

Sheet metal cutting requires a power density of 10^6 watts/in^2
(source: Mike Klos @ laser mechanisms)

Converting to millimeters, that's 1550 watts/mm^2.
(using equation: 1in^2 = 645mm^2)

A 100 watt laser can achieve a power density of 1550 watts/mm^2 in a
spot size that is 0.6452mm^2

A spot size that is 0.6452mm^2 has a diameter of of .28mm or 280
micron (using area = pi * (d/2)^2)

280 micron! If I can deliver 100 watts to a spot of 280 micron, I
should be able to cut metal. That's too easy.

Why? Well, how big a diameter can I expect with my optics? The
information on my beam diameter varies. I have read it goes anywhere
from 1.6 to 2.3 mm.

At 1.6mm, if I have a 3x beam expander I get 4.8 mm, which will be 103
micron using a 1.5 inch focal length

(equation: diameter = .013 * M^2 * (fl/D) where M^2 is equal to 1, and
D is diameter of incoming beam. See this site

If I substitute in an M^2 of 1.5, I still get a diameter of 150
micron. So according to calculations I should be able to deliver a
power density needed is 10^6 watts per square inch.

Note: I'd like hear from anyone who could verify that 10^6 watts per inch is the power density I need.

Note: the reason I bought the microscope was to be able to measure in micron -- hopefully I can use it to check my beam diameter

Pretty soon I figure I'm going to want to look at the size of the beam diameter, which should at certain stages get down to 200micron.

After reading this article, I figured it'd be pretty easy if I had a microscope with a reticle in the optics. Seemed like I wouldnt need something much stronger than 100x, so I bought this thing on ebay:


Pretty handsome looking thing, even if it doesnt work. :-)

Okay, situtation doesnt seem too bad -- I took at look at the 90 degree bend and it had a oxidized spot on it. When I take the elbow out of the optics chain, and still pass it through the cut quality enhancer, circular polarizer, and beam expander, I get 105 watts. That sounds pretty good. I got around 110 when I remove the beam expander. I'm thinking I would still like to get the cut quality enhancer and circular polizer cleaned, because I called laser mechanisms and they said I should see more than 3% loss.

still seems like progress. I have to order a new mirror for the 90 degree bend.

I have a lollipop-style laser probe that looks a little bit like the one on the left:


Funny thing is, I gave it a shot for the first time running at full power and I got a reading well over 110 watts.

The probe is set up exactly for a laser of my kind, the dial indicator runs up to 100watts. I used it exactly like the instructions specify. Therefore, I can only make the scientific conclusion that the laser is imbued with supernatural abilities that enable it to operate well beyond its preordained powers.

However, the situation still is not so good. When I mount the optics from laser mechanisms which include a cut quality enhancer and feedback isolator, the power drops to around 100 watts. When I put on the beam expander it goes down to 75 watts. I'm not suggesting that the lasermech equipment has a problem. I mentioned in a previous post that I probably have coated the internal mirrors with some smoke that's causing the drop. As for the 25 loss when I put on the beam expander, that sounds completely ridiculous, and I'm gonna have to look into that as well.

bites.

I had this incredulous feeling tonight "Could this be the last time I need to wrap any more wirewrap wire?"

Background. The PCB I ordered from PCB fab express was completely populated and running. Ordering the board only cost $65. Yes, there were some mistakes on the board, mostly having to do with the fact that the seven segment LED needed a resistor at the anode. When I first hooked up the board the LED was scorching hot. So I ran the Proteus software, the incredibly amazing circuit simulation software. I hung a virtual ampmeter on the anode, and there it was running at 2 and half amps. This was repaired on the physical the circuit by cutting one trace on the board and soldering in a small resistor. If I was going to make a PCB with a AVR chip I would recommend always provisioning a system that allows you to reprogram the AVR when it was already plugged into the board. Pulling a 40 pin chip is too nerve wracking. I also would have put vias next to each pin of all chips. This simplifies soldering new connections when retroactive changes are required. Another thing I would do differently is test the circuit in software more. Put a virtual ampmeter on every part to see if I could find chips that are using an excess of amps.





The strategy of this PCB is that it has an AVR chip right on it, this program allows me to set in two different power settings manually. There is an input to the board that selects these two preset values, and based on the values five outputs are pushed onto another chip on the wire wrap board that handles selecting the pulse width set to the laser.

I completed the wire wrap board, hardly believing it was possible. The completed controller box was closed up and put on the shelf.



If you look at the diagram I created a long time ago you can see there a lot of inputs and outputs to the controller. This creates a problem because you end up with a lot of cables to manage. I have a friend who used to wire phone closets. He suggested doin' it like the phone company and making spools made of posts projecting off the wall. This was very helpful because it keeps the excess wound up and the cables stay off the floor. Its also nice because there will be some excess length to the cables which will be helpful in a situation where some of the hardware gets moved to a new location. The cable expert and I lay over 100 feet of between the controller to the CNC table.



Observe the loops of wire going around these spools in the picture above.

The controller joins some other components on the shelves.



What you're looking at on the bottom shelf going from left to right is the controller, the power supply running the servo motors on the CNC table, and a small color tv. (I like to have it on when no one else is in the house.) The computer that's running linux and EMC is on the top shelf.

Moving over a bit...



this is the desk where most of the electronics has been happening. Eventually I plan to have the electronics bench out of the way, so the keyboard, mouse and monitor were mounted to shelving over the bench.

The following is a video showing the motion control system:


Okay, a recap of the situation.

Real time linux is working.

EMC motion control software is working.

EMC can activate the laser and adjust its power real time.

Motion controller is running UCS servo controller and gecko drives.

Power supply of servo motors is working.

Power supply to laser is working.

Optical limit switches on the CNC table are functioning.

I have 750 watts of cooling, enough to test the laser.

The cat is sleeping in a small cardboard box


What remains...

Optics, optics, optics. Installation, alignment and optimization.

Need more cooling.

Need a bed to support the sheet metal.

Must install gas lines.

It is so satisfying to be done with the electronics -- they are at a point where I will not have to make any additions to allow me to test the laser.

Posted by: owen / 10:17 PM

Okay so I took two weeks off in hawaii. Actually in every spare moment I worked on the next AVR program to over come my pulse width modulation problems.

In the wild chance that the accompanying circuit was made properly, I also am going ahead and ordering a custom printed circuit board from PCB fab express. Picture is included below. What this circuit will do is supply binary input to my other circuit and specify the duration of duty cycle. This circuit has some seven segment LED displays, and some input buttons, to allow the user to predefine two duty cycle settings. When the run-time software (EMC) is operating the laser and CNC table, it will send a binary signal which will switch between the two settings. The previous incarnation of the circuit was trying to have EMC send 5 bits of control that specify the duty cycle range. There's like a pile of previous posts describing different ways that I tried dealing with the problem but it was just getting too cumbersome. I was going to take five bits of input which was tying up all kinds of lines coming out of the motion control card. Better to save some of the lines to control things like solenoids and ventilation.

Knowing what I know now I'm wondering if there was a way to really generalize the whole problem. Could the inputs and outputs between the motion controller card, the laser, and motor drivers be brought to one single platform that then manages everything in software? The wire wrap is great, but I just wonder if management of all I/O could happen in software running on an AVR chip. There are over 100 lines of I/O for the whole controller. With the exception of the circuitry that has to get massaged with line drivers almost everything else is either setting and responding to logic levels, or driving some blinky lights. Yeah, probably could have done this all in AVRs.

But rather than try that I'm trying to hold on to my previous circuit development. I made extensive changes to the previous circuit diagram in response to learning that the output of the laser behaves differently than what I expected from reading the manual over vacation. Serves me right for reading the manaul. The manual has a circuit using line recievers to monitor stats coming from the laser and in some cases the signals are pretty wonky. Right now I'm just going to follow over temp and duty cycle. This requires removing a chunk of the circuitry. I've also made changes that involve allowing the user to override every input that makes my JAM circuit hit the e-stop. I've spent the last 3 days now that I'm home from vacation yanking out old wirewrap wire and installing the new circuit on the prototype board.

I'm really trying to get the thing going in time for Chris Short to come visit. Chris and I are shooting for the Big Day in mid february for when I actually cut metal. I've got a million cables to make, and I havent even really started on getting EMC to behave, but I dont think I have to have an optimized motion control system for the B-Day.

One thing I did was install cygwin. Cygwin is essentially the unix operating system that runs on windows (it is not linux, which doesnt run on a windows desktop). This has been absolutely indespensible because I have a perl program which parses the output of net lists created by the Eagle schematic software, and then tells me where the connections are located on the prototype board. In the past when I did this I would boot up in linux to do the perl work, and then move the file back to windows. This is completely impractical if you're making lots of changes to the program. I also really love that I can run programs like less or grep on a command line which has simplified parsing the output of my program. Cygwin is a major plus. I wouldnt be surprised if there are ways to launch Eagle or AVR compilers on the command line and that might become quite useful in the future.

o.