The distinctive blue-and-white enclosure of the Chinese-made K40 laser cutter has become a common sight in workshops and hackerspaces, as they represent the cheapest route to a working cutter that can be found. It’s fair to say though that they are not a particularly good or safe machine when shipped, and [Archie Roques] has put together a blog post detailing the modifications to make something better of a stock K40 performed at Norwich Hackspace.
After checking that their K40 worked, and hooking up suitable cooling and ventilation for it, the first task facing the Norwich crew was to install a set of interlocks. (A stock K40 doesn’t shut off the laser when you open the lid!) A switch under the lid saw to that, along with an Arduino Nano clone to aggregate this, a key switch, and an emergency stop button. A new front panel was created to hold this, complete a temperature display and retro ammeter to replace the modern original.
Norwich’s laser cutter has further to go. For example, while we secretly approve of their adjustable bed formed from a pile of beer mats, we concede that their plans to make something more practical have merit. The K40 may not be the best in the world, indeed it’s probable we should be calling it an engraver rather than a cutter, but if that means that a small hackerspace can have a cutter and then make it useful without breaking the bank, it’s good to see how it’s done.
This isn’t the first K40 enhancement we’ve featured. Norwich might like to look at this improved controller, or even extend their cutter’s bed. Meanwhile if [Archie]’s name rings a bell, it might be because of his Raspberry Pi laptop.
26 thoughts on “Hacking A K40 Laser Cutter”
If you like this, you’ll very likely want to take a look at the other guides here : http://smoothieware.org/laser-cutter-guide#machine-specific-guides , several of which are for K40s
i am surprised that the machine made it past the food and drug administration (who’s jurisdiction that lasers fall under) and into the united states.
maybe just a better than nothing safety switch is enough to allow the machine into the united states.
Why would the US FDA care about a laser in England?
What makes you think this slapped together in China and sold via eBay hardware was ever tested or approved for official sale in the US?
Norwich is in the UK.
And in the US: http://www.norwichct.org/
That norwich doesn’t count.
The K40 is sold in the USA, so the location of Norwich doesn’t really matter here, does it?
(I suspect the importers of the K40 just haven’t been caught yet)
What makes you think this machine is in the United Sates? :-D
Sates, States, woops, lol
Buyers of the K40 should regard it as a kit of pre assembled parts from which to start turning it into a descent machine, if you want good quality, then you must pay for it.
Forgot to add, K40 whisperer is a good alternative to stock software, and keeps supplied controller
HAD covered K40 whisperer here;
You only get one pair of eyeballs… Please, NEVER use Arduino (or any uC, really) in a safety circuit. Especially a clone, your eyesight is worth more than a few dollars/etc. This is a low power laser, but your eyesight is worth the cost of doing the safety modifications properly.
You would be better off having no interlock at all than having one that makes you think you are safe when you are not. There is a reason the Machinery Directive exists, because of dangerous “safety” like this.
Keep it simple. Use direct hardwiring between the interlock switches (wired in series with each other, normally open with lid open) and a beefy N.O. relay that will disconnect power to the laser any time the lid is open or improperly seated. You can wire the E-stop button contacts (you should have two, ideally) in series with the interlock switches as well. For the level of hazard, that will be adequate, simple, and inexpensive (as long as you use proper switches). Use brand-name switches that have legitimate spec sheets so you can calculate reliability and lifespan based on your worst-case lid cycles. One-Hung-Low switch specs should be assumed to be total bull. I would also add two LED lights. Amber on the N.O. or power supply output to indicate laser is powered, and some other color for the NC to indicate interlock is open. That way you have a clear confirmation that the safety interlock is working properly.
My very first thought when they described the safety features was “what if the arduino crashes…” Bad software based interlocks are how people get killed.
From the actual linked article though, it sounds like the emergency stop button is actually wired to the mains input, not through the arduino. The HAD description got it wrong.
It’s also good to have a keyswitch that is “as safe as conveniently possible”.
Basically when you turn off the switch it locks the table from moving and the laser from firing, but without actually cutting power to anything that takes time to reboot.
We have a big 2kw laser at work that has a keyswitch that controls the table and laser. But it shuts off the laser fully which is a 15 second shut down and a 15 second initialization process, so nobody uses it. It’s not worth waiting 30 seconds every time you need to climb in to grab a part real quick, so nobody ever uses it.
And to clarify on how this can be done:
-design your motor drivers to have separate control and motor power lines, so you can cut off motor power but allow the driver to still operate and hold position.
-Lock down a shutter and lockoff anything other than preionization current through the laser.
-In a pinch, have a relay that shunts off a resistor inline with the power supply. The resistor should be sized so that the controller section is able to function, but the voltage drops too low if it tries to drive a motor. It’s possible this could destroy the motor driver if your safety circuit fails, but if your safeties fail then they aren’t safeties.
If your safeties make it nearly impossible to use the machine when they’re in use, then they aren’t safeties. IIRC it was on the Practical Machinist forum where someone told about a CNC machine at his workplace. The operator station was positioned so that the operator could not see the turret and workpiece through the small door window. that made it impossible to safely jog tooling to do things like setting tool offsets and starting positions.
So they made a special wedge to jam the door switch so setup could be done with the door open, and made it easy to remove quickly if OSHA inspectors came around. Better design of the machine, a much larger window and better positioning of the screen and keyboard, would have eliminated the need to override the door safety just to be able to do setup operations.
I only work with fiber lasers at the moment, so even at 2K they are physically very small. It’s surprising how much you can do with just 2-3K. Most of our cutting is done below 4K. For welding we have been up to 8KW CW but I think that customer just wanted a big laser:) They didn’t need that much for the application at that time. There is a 20K CW on a demo robot that I can’t wait to see in action :)
If you provide a defeat like that it will definitely get used more than intended. A lot of drives offer STO and separate power/logic also, like you said. Our solution was to use locking interlocks on the doors. I am getting a significantly better SPLC into the next panel design, so that will change for the better with encoder monitoring modules. If you need to keep stages enabled while allowing access, that is the slickest way to do it safely. Expensive, but worth it.
Well said, I came here to make a similar comment.
Ripping out the arduino and simply wiring everything they’ve done in series would be simple and 10x safer.
I use laser safety glasses with the K40, even if the lid is closed! Never mind the lack of in interlock, how about the big holes in the bottom of the case?
Those holes are so it can mark its territory.
I somewhat agree. It’s somewhat appropriate here, since most people don’t own a K40 and don’t know anything about them. I do, and these are the most basic upgrades that pretty much everyone does. Yes, it seems like K40s are usually a steaming pile of poo when they arrive. But if you’re a tinkerer, you can make it work pretty well. Ours get used a lot and is really nice to have. We’ve spent around twice what the unit costs to add smoothie board, ammeter, air assist, compressor, adequate ventilation, honeycomb bed, etc… Still far cheaper than a glowforge!
Our laser (not a K40) was a mis-manufactured and ill-conceived mess.
As shipped, the interlocks were wired in PARALLEL rather than SERIES.
(Door Open Switch, Coolant Flow Switch, Temperature Fault Switch)
Consequently, with the PARALLEL wiring this meant that ALL faults would have to appear simultaneously to fault out the machine.
(Door Open AND Coolant Pump Fail AND Temperature too Hot).
Clearly someone unclear on the concept.
Really want my K40 to cut through some hardass hardwood so i don’t have to use my xcarve for everything. Anyone have any recommendations into lens upgrades with a longer laser range to make thick depth cuts?
Please be kind and respectful to help make the comments section excellent. (Comment Policy)