To kill a laserbird. (1 of X)

China is cheap, we all know that.

China also doesn’t know what quality control is.

Well, they know, they just don’t perform that kind of process in most of their cheap production. So, what can happen if you buy a cheap china laser tube and power supply? Well, mostly two things (three, if you count absolute failure, four if you also count pieces falling off.):

  1. Power supply doesn’t work well for engraving because of massive power “ringing”:
    (yeah, in this particular case it produces a beaitiful 3D effect, but that’s beside the point)
  2. Beam shape is not a roundish circle but dual kidney shaped, having 0 power in the center of the cut:

So, what did happen when I bought the cheapass laser and supply? Obviously, all of the above.

The interesting  part is that the original tube the machine came with (alongside the almost absolutely crap moshiboard) worked fine and had no ringing. I had grown into engraving reliably when the happening, so when I found I could no longer do it…it was a bit of a let down.

Since the beam had virtually zero power in the beam’s center, I was forced to pump it up so it could get through, wich meant it would more readily charr/melt the edges and also slowed down cutting speed to a crawl (I think at the end of it, I was doing like 125mm/min at full military overpower to cut 2mm MDF, and it couldn’t even handle 4mm…yeah)


This is a laser tube on steroids and no water filter in the circuit.

As the tube aged, it required more oomph, and once I went over the laser rating, performance cascaded downhill. (laser rating was 15mA normal, 17mA Max. and I was pushing about 23mA through it. I got away with that because a  beer chilling unit, I could freeze the tube if I wanted. BTW, if you do freeze the tube, from 18 to 8ºC the power goes down noticeably, then goes back up @ 0ºC XD! )

Finally I decided it was time to change the tube, and the supply…and since I was at it, maybe upgrade a bit.


First step: Make a hole in the box. It was square but as an afterthought, it should had been round and lower.

You can see previous mods there:
3 power output connectors with a relay board controller, air assist input, with solenoid valve. And oh, yes, also a ginormous laser tube wich doesn’t really fit as much as protrude


Oh, hai, bro.

The original tube was (chinese) rated at 40W but sources say that’s peak power for a 55/750mm tube, and it’s really 35W. Working as it was, I probably had an effective power of 20-25W at best . The new one tough is a Lightobject 45W rated, 50W peak power laser.


  • Power: 45W (50W peak)
  • Triggering Voltage: 20KV
  • Operating Voltage: 15KV
  • Current: 18mA
  • Life Span: 1500~1800 hrs
  • Length: 1000mm/ 39.4″
  • Diameter: 55mm/ 2.2″
  • Water Cooling required
  • Water temperature: 20~25’C(68-77F)

I also bought a matching (more on that later) power supply:


Fun thing is that since the tube was 45W, the power supply had to be their 40/60W one. What with that? Well, the 60W tube is just longer (1250mm instead of 1m) so, sometime in the future, I could just make a longer cover and change the tube…half a meter long cover, tough…that’s a bit of an overhang, ain’t it?

Anyhow, here are some more shots:


I was going to make the cover in metal, that’s the reason behind the squarish hole, but I didn’t had the bending press at hand, nor was going to buy one, yet. While I was shopping for supplies, I saw a PVC tube in the shop, and just ran with it (not literally, of course


7 screws for the 7 dwarf lords to hold the cover. I would have preferred a bit more margin in there, but the square deed was already done.


Almost everything has sleeves, especially around power cabling.


The laser in the workshop. Yes, I have lost a bit of table…but it will be worth.

I have some bits to wire up, but all the important parts are in place.

Next installment: Controls, wiring, cutting performance.

Mistery circuit winner.

So, who won the mistery circuit? and what was it?

Luftek, you made the best analysis of it, so the circuit goes to you. You only had a bit left wich would have given away the purpose for it: “Q1”
That transistor sits in parallel with 555’s main capacitor…so…shorts it out whenever it’s told to. And if you short the 555’s capacitor, what happens? Yep, it starts charging again, efectively retriggering it.

Laser watchdog

So, what does the circuit actually do?

The circuit is an EDGE detector plus retriggerable 555, so, it can detect the change in state (not the state itself) of an input, and, given enough frequency (0,1 Hz or less) it mantains the output ON. If the water sensor (or other) fails to give life signals for some seconds, it will shut down the whole laser machine. (provided you wire it that way, of course)

Laser watchdog 001
Capacitor voltage vs switch/sensor input.

Laser watchdog 002
The edge detector gives away a short 2ms pulse. (switch press was 150ms)

Edge detection circuit:

Laser watchdog 003

Internet says that just a NOT + AND, will work, however the 555 required an inverted pulse for it’s activation, hence, the NAND. R3 as pulldown so input doesn’t float, and C3, because otherwise, the propagation delay from U2 as so short, the circuit didn’t work. (yep, 20uF is a LOT, I know)

But, why bother to make the edge detection rather than just retrigger the 555 from the sensor? Safety (heh, XD) reasons. Imagine the worst case scenario:

Water pump stops and the simple flow sensor also stops in the ON position, you have no way to know that the pump has stopped if you retriger by level (+5V, for example). The edge detector waits for the rising level shift, so, pump stopped = laser shutdown.

I hope the winner has some use for the circuit, and that someone can use the edge detection info.

Squeezing a K40 laser engraver, part 2.

Excerpt from “It’s been a long while II”, because I’m lazy:

Using moshisoft, whatever version, is just a pain in the ass for other than engraving (remember, you bought a ENGRAVING machine, it didn’t say anything about being a real cutter). Engraving tough, is quite nice and fast. You just upload a bmp, and BANG there you go.

Frankly, it’s engraving properties are quite good for me, but I won’t miss all of them:

  1. It can’t engrave more than 200×200 (machine will go haywire). (why would you want to?)
  2. Can’t reliably cut DXF’s, it will ALWAYS move the final vertex from each separate piece, to the “center”.
    When doing simple pieces, is just annoying, but keep in mind, it does not snap to grid or vertex, so placing them again won’t have any precision. Just don’t think about tons of pieces.
  3. Software key (it has a hardware key) might reset amidst job, for no reason whatsoever.
  4. Program may change language for no reason, and try to display one you don’t have an interpreter for. (luckily, I remembered where where the options I needed, but THAT was insane…)

Software VS. Ports VS. Price:

At this stage, you might be wondering “What options do I have”? Well, I think you might have too many. XD

Since I’m not much of a Linux guy, I was left with the two middle options, and you can guess wich one I picked up, can you?

Kcam, the little software that works (most of the time*).

It might not be the most advanced software and has some limitations wich I would change, but otherwise, it works for controlling a laser.

142 - Laser softwareties
Not my screenshot.

It is interesting to comment some things, tough:

It doesn’t have a specific pin to control the laser so you must use a bit of a trick, with it. You enable “Z single step”, and (me at least) I used the direction pin of the Z axis as laser control. I had set it to activate the laser on 5V from pin 4 (I just like the settings to be in accordance with their function (5v-ON, 0V-OFF)

However, and this is quite IMPORTANT, the parallel port has it’s own mind when you switch on the computer, and you must observe how the pins are set, as you don’t want your laser to start firing for no reason, do you?.
Mine, had pin 4 ON all the time, until you powered it off through software, so I had to negate the activation of the laser. I changed the pin to one that doesn’t change state on power-up.

142j - Pport

*I’m using an old version, so I haven’t had any experience with the latest updates, it might be better it might be worse.

Version 40047 (last one to work in Win2k) specifics:

  • It can’t understand partial circles (includes chamfering), so it will draw a line between start and endpoints. (complete circles are OK)
    Solution: -Explode- all the drawing and then -join polyline-, it will convert curves to linear segments, and you’re good to go. I use DoublecadXT, wich doesn’t allow me to change how many segments I want when converting, so on big circles, it does show the individual segments. You have two half solutions there (apart of augmenting the segment count in your program of preference).
    -> For circles, draw polygons with a high facet count, treat them as circles.
    -> For curved lines, draw splines by fit points, and then explode, it will give nice and smooth curves.
  • If you change engraving/cutting speed sometimes it doesn’t acknowledge the speed change, so you must compulsively compile (XD!) before starting a job.

Aaand this is all I have to say about Software…you know, I’m more of a hardware guy.

Next post (whenever that is) in the series:


Because not all rulers are the same, and why your first piece should be one.


The linear Trickster (Image engraving for modified K40’s)

Unless you have spent a tad more in an Epilog laser, wich then makes me wonder why are you reading this (apart from having a laugh at my problems), many people say that you loose the engraving option once you change the hardware to a cheap one, however that is not entirely true. You still have a pair of tricks… (OR you can use Kcam’s built in BMP converter wich I just discovered. XD!)

Vector images:

For vector images, fill solid the areas to be engraved, then convert those polygons to horizontal lines with apropriate spacing (like the ones in CAD drawings). -Explode- , remove unwanted lines (probably the original vectors that made the outline) and there you go, a raster engraving of your drawing. As each laser is different, you must explore the spacing in order to achieve best results, for me it worked at around 0,1/0,2mm.

142k - vectorialThat was a cookie mold for Eurosteamcon 2013

Pros: It’s free. It works.

Cons: depending on your program, it might create artifacts, partial or repeated lines, or not fill the whole area at once, requiring of you to repeat the process in smaller areas.

B/W Images:

As far as I know, there is only propietary software to do that. Moshidraw did it, Epilog does that, and they work pretty well so far. (do you recognize the movie? XD!)

So, is there no hope? It is, if you have tons of patience and don’t mind somewhat convoluted ways.

You know, Altium does have a small little script to create logos. It does convert black and white BMP’s into  horizontal vector lines in whatever layer you choose. It also does it in an horizontal fashion, just the right vector for the lightest part of the laser to do the work (you WANT to engrave with the X axis so the Y, wich is heavier, doesn’t do much work). The drawbacks is that resolution is dependent on image size. So, it’s not the same converting a 100px image than a 1000px image.

142a - Rasterizing
Now we have a rasterized image just like we were going to do a PCB.

142b - RasterizingCloseup of previous image

Next thing is to convert this to DXF. Altium doesn’t do that directly, but can convert first to gerber, then to DXF. However, if you try very big images, it might give you a “Film too small” and won’t convert from the PCB to gerber, and anyhow, if you have to engrave more than say, solid fill of 150x150mm you’re just wasting your time. It can be done, but well…it will take AGES, seriously.

The following are 100, 500 and 1000px images to vector. At 4mils per pixel conversion, it gives us 10, 50 and 100mm image height at 0,1mm (0,1016) line separation. Denser than that and the laser will just waste your time and material.

142g - Rasterizing

Let’s get a closer look at each one:

100px (10mm)

142h - 100px
Meh, useless.

500px (50mm)

142h - 500px
A bit better.

1000px (100mm)

142h - 1000px
That’s more like it.

But, okay, now you have downconverted the images to lines, how will it look? Let’s print the DXF’s so we get a slight idea:

142i - rasterizing

Frankly, not bad at all. Even the 500px one looks acceptably good, doesn’t it? Now onto the laser, let’s see what we get with the 500px one:

142L - vectorial

Yeah, that’s probably the most extreme example you will try with this technique. Anything that is less than 3 pixels wide, will be lost, so be careful. As an afterthought, the laser has about 0,25mm width, wich is more or less 3 pixels wide at 0,1mm/px…so, if we engrave the big 1000px one, it should look better, shouldn’t it?

142m - vectorialSpraypainted black and rubbed with a rag to increase contrast in the wood.

Meh…probably I’m just picking up a graphic too complex. Just look at the hair… Let’s try a simpler drawing:

142o - vectorial
I like this picture a lot, erotic or not.

Not bad at all. I think the material I’m doing the tests isn’t the best choice…that, or my new laser tube doesn’t work well. I must investigate further. Anyhow, the method works more or less.

142p - vectorial
Same engraving on balsa wood. Looks nice, doesn’t it?

NOTE: the vectors are not optimized on the X axis, so the head will travel back and forth a bit, not like it will wander around like a mad dog, but you will nottice it. It will induce vibrations, so put a weight or secure somehow the smaller pieces you want to do.

Pros: It somewhat works, just don’t pick an image like the one I tried.

Cons: It’s very convoluted, I wish there was a program to do this right away and output DXF’s for us to engrave.

90º (Quickbuild number 4)

Do you remember the little circular saw that could?

Ever wondered what it was for?

Well, the time has cometh to re veal the plan. There is this RGB board at work, that, as better as we can, must have the led board at 90 degrees against the control board. The saw is actually used to cut the pins before zipping the led board (because that is what we do…with so many header pins you are no longer inserting them, or introducing them, this is a zipper in all what matters). So, those boards must sit at 90º, but, how the hell do you ensure that that angle is right all along the 50cm boards?

Picture Neo in the Creator software:

…You’ll need levers, lots of them…

134a - 90 degrees
Yeah, one is broken…I underdimensioned it, but the rest are holding.

Let’s take a closer look:

134b - 90 degreesThe power of the rubber band!

Someone said, gimme a lever and I’ll move the world…interestingly to mention, he could have also said: “gimme a lever and I’ll hold the world still

So, what’s up with those? Let me show you the dxf of the piece and you’ll get it in no time:

134 - 90 degrees
By the way, the eccentric lever is longer now.

Clearer now?

With the long lever, you push down on several leds at a time (to distribute pressure) and then, holding pressure on those, you move the eccentric to apply pressure on the control board, forcing the two to rest against a 90º angle. Do it at enough places and you’ll have a more than acceptable right angle. Finally place a rubber band to hold pressure, and repeat with the rest, then solder.