New low cost DSP laser controller board

[sshwarts]

With the AWC-608 and I'm assuming the other DSP controllers as well, the pulse width and speed are coupled (I've conformed this with my scope). They don't implement PPI the way that Dirk had done it in his work <http://www.buildlog.net/blog/2011/12/getting-more-power-and-cutting-accuracy-out-of-your-home-built-laser-system/> or the way Lasersaur or Smoothie based solutions are doing it. I'm not saying thats good or bad, just different.

The way others are doing it, is that the laser pulse time is a constant (say 2 or 3 ms) and it's then fired (PPI) times for every inch of travel regardless of the speed--it's decoupled.

Scott

[educa]

Yes and that was what I was initially asking. How long 1 pulse is in duration. I am aware of Dirk and Ben's ppi solution because I'm using it. There I believe the pulse is 3ms. So now I was looking at this dsp system and wondering if that also have around the same pulse duration

[BenJackson]

I can put a scope on it later this week and let you know. I'm taking baby steps with stuff like the limits and the safeties because the documentation gives me no confidence. Given that I have a pointing laser that drops right into the beam path when I open the door, I want to be damn sure the laser can't come on when the door is open! For example, the door limit says, "If the port signal comes to be too low of a level, processing job will be suspended; when the port
signal comes to a higher level, suspended job will resume." Those inputs are sandwiched between 24V and GND terminals. Should they swing to 24V? That doesn't seem super likely, but maybe. Of course they're not wired to anything in the connection diagram.

Re: my earlier comment about Z: I lowered the bed all the way to get better access while re-installing my laser tube. Due to the current E5 limitations I had to reset the E5 3 times to get it all the way to the bottom (repeatedly hitting the soft limit due to having to set the axis up at 10x scale). I can probably live with that.

[educa]

Ben, you use a dsp board nowadays instead of your linuxcnc ppi system? I run on your linuxcnc config with my 80w cutter and am very happy with it. If you went with the dsp solution can you tell me why ? Kind regards, Bart

[BenJackson]

I'm just getting my laser out of storage where it's been since I moved 3 years ago. Instead of setting up the dedicated linuxcnc system again, I thought I'd try something different. I always liked the CAD/CAM workflow for laser cutting mechanical parts (e.g. being able to compensate for laser kerf, and work in serious file formats). The workflow for more artistic stuff was not as good, and not accessible to my wife.

I also never got around to making raster engraving work on the FPGA card (5i25).

[educa]

Nice. I am currently trying to achieve PPI cutting on 2 arduino's. 1 will run grbl for X and Y and then I can attach laser fire to the Z. Using that technique I can easily write a preprocessor which calculates how many steps Z should take. For example if I do a G1 from position (0 , 0) to (1 , 1) in inches, then the distance traveled is 1.4142 inch. If I want to shoot 400PPI then I just have to make sure that Z needs 566 STEP pulses. If I then connect Z step pin to another arduino which based on interrupt holds the shooting signal high for 3ms, then I have a very cheap PPI shooting code which is completely speed and axis coupled . I don't dare to run that same code on the first arduino running grbl because I think grbl is allready maxed out. + 1 arduino nano costs about $2

This would of course only be for gcode vector cutting. For real engraving, I'll try to find another solution, but I really wonder if you can really engrave very nice details on a 40w co2 . On my 80W its even more difficult because of the double laser power and the difficulty to nicely dose it for example for grayscale. Therefore I'm thinking of also mounting a 600mW 405nm diode to my gantry for the very small engraving I sometimes need. With such a diode I can achieve dots of 0.1mm in 256 tones grayscale. Thats 256DPI real grayscale, but quite a bit slower than a co2 tube of course.

[BenJackson]

If you're going to switch to a 3D printer style controller, you should look at the "E" axis. That's the "axis" that's used to dispense material. It was practically designed for PPI. I'd be worried that combined XYZ moves would not be handled properly in a 3D printer controller (they're not really common), but the E axis is meant to do pretty much exactly what you describe (and with pretty much the same purpose: moving the computational load to the gcode processor and off the arduino).

You could use something like a 555 oneshot to make the 3ms pulse. I feel obligated to say that in case my friend Alan (winner of a contest to make something ridiculously complicated out of only 555s) finds out.

I think you *can* get nice details on 40 and 60w cutters because I've seen Epilog do it. But you really need to be able to sling that final X carriage around. This is why I had given up on direct parallel port drive on my own cutter (and gotten the FPGA board). My machine just hit the limit of how fast any axis could go because of how fast it could make pulses.

[BenJackson]

Ok, this got off topic.

More work with the E5 board:

The E5 will move the gantry and fire the laser while OPEN PROTECT is asserted. This is especially bad for me because I actually have an aiming laser that drops down in front of the main laser when the door is open. Summary of behavior with OPEN PROTECT asserted:

• Will not fire when LASER firing button is pressed on the front panel (good!)
• Will move the gantry and fire the laser for about 120ms when "Start" is pressed (each time) on the front panel (bad!!)
• Will move the gantry and fire the laser for about 120ms when "Start" and then "Continue" is pressed (each time) in the GUI (bad!!)

The latter two depend on the job. I made a cutting rectangle at the origin, so laser firing happens immediately on start. If your job has some travel before start you'd get some warning.

By the way, the IN1 input (the open protect input) does have an internal 24V pullup, so the disconnected state is the dangerous "door closed" state.

The WATER PROTECT seems to work correctly. It will not fire the laser (or do anything, really) when WATER PROTECT is not pulled low. This also means that disconnected is the safe state. I am going to have to use this for my door because it actually works. The downside to that is that the display says "Water protect!" when engaged and makes a loud beep. It also requires explicit clearing with "ESC" on the front panel, unlike the door, which has no panel indication and does not require explicit clear.

On setting laser max power: The "max power" setting in the parameters won't go above 98%. The setting is not a scale factor (it is not setting the point of 100% power for this system's laser), it's a cap which is applied to all settings. So if you set it to 60%, any setting from 60..100% will just cut at 60% power. You don't see this enforcement happen. Your job says 100% (or whatever) and it will cut at 60%. The front panel lets you enter any value, but does show the capped value after you try to lock it in. I added a trim pot so I could keep the scale at 100% and use the trimpot to control max laser output.

Even looking at the scope while operating, I don't really understand the "MIN/MAX" power (is it straight and cornering power?) job parameters.

[educa]

Ben , how do you measure a WATER PROTECT to send it to the controller input ? Some rpm/flow sensor connected to give pulses somehow ?

[BenJackson]

I don't have an actual flow sensor. Yesterday I was just testing all of the inputs/outputs of the E5 before trusting it to actually fire the laser.

As I recall from my original build, I didn't find a flow sensor option I liked. Thinking about it now, the best plan might be to put a temperature sensor on one end or other of the laser. Temperature is what you really care about, even if water is flowing. The chiller (or your 5 gallon bucket!) might fail and get warm, or flow might stop, or ambient temperature might get too high.