Skip to content

Archive for March, 2016


Kossel Fan Verdict – Success

Thus far I have six versions of the fan structure for the Kossel. Each one has been an improvement over the previous version so, much like software I was starting to think it would never be done.

Today, I put that concept to rest. With frogs. And not little frogs, 80mm tall frogs…

See, printing critters is the tough part when it comes to 3D printing. There’s always some overhang or angle that is not going to be printer friendly. At which point you have to decide if you want supports turned on in the slicer software, or you change orientation of the print, or, you just go for broke and hope.

So we start with these two bored looking guys. Who resemble me while I’m waiting on a print job. You can find the original here if you want to print it out yourself.

IMG 1277

The green frog is printed with MG Chemicals green PLA (no, not Soylent green! People!), the white frog is printed with MG Chemicals white PLA. They do have somewhat different flow characteristics. As does most PLA. But this is not so much about the PLA quality, it’s about a fan.

My fan design, below, is designed to replace the standard fan shroud in the Kossel FVM kit. It is fairly sturdy, uses the same axial 40x10mm fans as the cooling fan in the present shroud. The fans are held in by some 3mm hex head screws that are screwed right into the plastic. I generally drill them out with a 2.5mm bit so they thread in easier.

IMG 1284

If you look closely at the bottom of the new shroud you’ll see a couple of circles. Those are adjustable louvers for aiming the exhausting air. From underneath they look like this:

IMG 1285

They are a SNUG fit too!

I made them adjustable originally so I could “aim” the airflow, as it turned out, I really didn’t need to do that. Just make the slot so it pointed at the tip of the nozzle. Hence, set once, forget them.

Testing – Green FROG both fans 50% speed | White Frog – NO FANS

The SAME print file was used for the frogs. Any difference you see in the print quality not fan affected will be the different flow rates of the PLA themselves. And again, not the point here.

The frogs arms are the problem areas. See the angle the arms go up? Yep. The white frog has problems with laying hot PLA on top of the HOT PLA.

IMG 1279

A better look at the back of the same arm (sorry it’s not vertical, white doesn’t photograph easily).

IMG 1281

Pretty easy to see that the heat was causing some issue, or, maybe it’s just a 97 year old frog. Whatever.

The other arm now. Check the elbow and cheeks (I don’t know if frogs have elbows so don’t start with me)…

IMG 1283


IMG 1282

You can see even the green frog had some problems with the under arm area (bad deodorant?) but the white with no fan was much worse. Not to mention the mess where the frogs hand is at the side of his mouth (check his thumb).

For the green frog, maybe more fan would have helped with that one arm, or not. It’s over a 3 hour print, I don’t want to keep doing it. It’s suffice to know that, yes, fans really help.

So maybe you want to make one for yourself. If that’d be the case, here is the link to my Kossel fan. Feel free to use it and I hope it works as good for you as it has for me. Oh, I am driving it with a stepper driver PCB and an Arduino MINI for speed control.


I built a Kossel

Yep. That’s right. I built a Kossel. For anyone that knows me, this news will not come as any surprise since they know I built one in June 2015. It took me three weeks. With many hours every day. A number of good stiff drinks. Long quiet walks on the beach.

Although advertised as a “kit”, in all honesty, it was a box of parts with a link to an online manual that didn’t reflect the parts that came in the box. Which made the assembly and calibration one of the most frustrating things I’d ever experienced. Apart from wanting to heave the whole thing in the garbage bin (not sure which parts of it would be adequate for the recycling bin), I started to find the humour in building it.

IMG 0119

As I remember the build, two things were obvious. First, this wasn’t a kit by any stretch of the imagination. Second, although I’d label myself a “maker”, what I knew about 3D printers on a scale of 0 to 10, was running at about -5 (that’s right MINUS 5). Mechanical I know some of, electronics I know more of.

My stubborn and “want to know what makes it tick” attitude determined that I would indeed learn the thing inside, and out. Thus I didn’t give up. I got it working the absolute best that it can work. I’ve even bling’d it out with my own add ons.

So fast forward to today. Ten months down the road. Over 1100 hours of printing time. And the printer, as of this minute, it printing something else for me. That’s right, in spite of it all, it’s working like a champ. Who’da thunk.

Back to the topic, I built a Kossel. Actually, I’ve built TWO of them now.

I printed out a complete set of parts for the FVMakers “kit” in MG Chemicals GOLD PLA (MG has some VERY nice deep and vibrant colours in PLA!). Rather than just build the thing, I decided to do a photo build of it. Not just “this part goes here” but building it the way a real kit would be built. I.e. you build in subsections and then assemble the subsections.

The fully assembled and calibrated Midas is for sale from FVMakers (if you’re interested ask them about it by name) so I didn’t built it for myself. I built it for someone get as many hours of enjoyment from as I get from my own.

In two evenings the complete mechanical sections were assembled. Two more nights for wiring and some bling and it was done. Bling? Yeah. Kossel with power supplies hanging off somewhere or graphic displays laying on the desk are what I call sloppy. So, make it neat, make it compact, bling it on.

With the gold PLA, I decided the easiest way to differentiate it from my other Slic3r/Firmware settings was to “name” it, therefore it became “Midas”. It printed it’s own name as a matter of fact.

Kossel Midas

A HUGE advantage of building a second one is that you know where the weak areas are. And you pay attention to them to make damn sure you get it right the first time. The first stumbling block is getting the basic PLA parts printed accurately to start with. I printed these on my own printer and although I admit it’s not 100% accurate, it’s within .07mm dimensionally.

As I was building Midas I never looked at a build manual. I just put together the way it made sense to me that it should be assembled. As in logical. A much more enjoyable build. Besides I’ve had mine apart and together so many times that was just easier knowing what part had to be installed before some other part.

Secondly because of my own and others input, this kit has been upgraded in a number of key areas. Let me point out that those areas are extremely critical. Take the bottom frame corners. Taller and more robust. Take the extruder, extremely efficient compared to the original one. The carriage holders for the Traxxas rods, fabulous. Plus more.

In the process of building the Midas I paid attention to the areas where any inaccuracy is going to come back and bite you in the butt later on. Like the Traxxas rods. I have a custom jig that I made that does nothing more than make sure the Traxxas rods are equal in length to each other. Each rod is a perfect 301.92mm.

Why is this important? Those rods hold the effector, that’s the plate the hot end hangs from like a bunch of grapes and ensures that one rod isn’t stressing another rod. You want that effector plate to glide around like the marker on an Ouija board.

The hot end should fit snugly in the effector plate and you shouldn’t be able to rotate the hot end once it’s clamped in.

The bed rests, where the aluminum plate and Borosilicate glass sit, should be sitting flat on the 2020 beam. Peak underneath to make sure that things are sitting level on those pads.

Getting away from assembly for a second, let me talk about tools. Yeah. If all you own is a flat bladed screw driver, forget building one of these. Or plan on a number of evenings at a maker space where they do have the tools.

And what will you need? Metric hex wrenches, normal and ball end (2 & 3mm), long ones too! Metric tap for some of the fittings, metric drill bits (1.5mm through to 5mm), DVM (digital voltmeter), alligator lead, jeweller screw driver, soldering iron, 63/37 leaded solder (cause it’s better than unleaded in my opinion), crimper for the Dupont connectors, wire strippers/cutters, extra wire (16 gauge) and good lighting so you can see what you’re doing.

Interestingly enough, in spite of the fact that Canada went metric in the mid-1970’s, you’ll find it next to impossible to buy metric drill bit sets. Instead they sell the non-metric sets. You can buy taps and dies in metric. If you need those things, eBay or a maker space are your best options.

Back into the assembly portion, one of the most common displays that comes with the 3D kits is termed the “Full Graphic Display”. Essentially a 12864 graphic monotone LCD with a daughter board to connect to the Arduino MEGA 2560. They refer to this as a “smart controller”, all evidence to the contrary.

With this graphic display, you can’t just write at some point on the display and update that specific area. Nope. You have to refresh the whole display each time. So you’re writing a lot of pixels out. Of course it doesn’t look like that’s what you’re doing because in the background that’s what the u8glib Arduino library is doing for you. And it only goes so fast.

Herein lies the underpinnings of the rub. The computer is sending GCODE, the Arduino is figuring out where in space it has to go and extrude something, then it has to update the display so you know it did it. Now to be honest I have no idea why it updates the X and Y positions. They don’t mean squat to me, the Z is the only one I’m interested in.

To continue, we need to spin the graphic fan icon of course. Which is pretty much useless because there’s no feedback loop from the fan to the controller to even tell you have the fan connected. Hence, “smart controller”, yeah, not so much.

When the printer gets to a tight area, specifically a bunch of tight curves, it stalls. There’s just so much that the little Arduino MEGA can do, so, rather than just skip the display, it stutters the hot end for you. Yeah. Handy as heck.

This is why the rage a while back was to add a SmoothieBoard or some heavy weight processing power to keep it munching through the data. Which of course, introduced a number of its own problems. No such thing as a free lunch.

To remedy this problem there are three avenues open.

One, toss the full graphic “smart” controller and put in a non-graphic controller like the 2004 LCD.

800px Smart Adapter 9

The second, although not quite as foolproof, print from an SD card. Even then it can stall slightly on tight curves.

The third is to modify the firmware. Look for the XYZJerk values, normally set at 20 or so and drop them down to 5 or slightly less. If you drop them to 1, you WILL solve the problem, you will also entering a waiting game for a print or the completion of the next ice age. Yeah, it slows it down that much. The whole printer is spasticated.

In the case of both the 2004LCD and the 12864 “full graphics smart controller”, they are current pigs. if you’ve ever looked at one of these displays and notice it flickering, there’s a reason for it. And it’s like this…

To power the Arduino, that 360 watt 12V power supply feeds 12VDC to the Arduino Mega. On the MEGA is a 5V regulator that accepts the 12V coming in and drops it down to 5V so the MEGA and any attached devices can use it. The MEGA uses a modest bit of current, but when you tack on the display it goes up. The problem with getting the incoming 12V down to 5V is that anything above about 7Vdc is dissipated in heat. And the hotter these regulators get, the less efficient they become, until they overheat and go south.

The Arduino developers will tell you to run them with 9V DC adapters if you must and don’t advise using 12V. Seems the REPRAP PCB designers missed the boat on that one. No surprise, it’s open sore.

The instant you plug in a computer via USB the MEGA automatically switches from that 5V regulator to use the 5V coming in the USB bus. And the display stops flickering. So…want the MEGA to last a long time and only want to print from an SD card? Plug the Mega into a USB powered hub, use a 5V USB adapter (like you charge a phone with), or leave it plugged into a computer. Mine happens to be connected to my Astroprint box which is supplying the 5V to it.

The more you print and dive into the workings of 3D printers, the more you can see weak areas in them. You can spend a lot of time innovating little things that could be better.

I’d hazard a guess a lot of people jump into 3D and drop out of it just as fast when the problems rear their ugly heads. If you’re in that leaky boat, one oar, and a 200 foot waterfall dead ahead, check for a local makerspace. In general these folks will go out of their way to help you get running again, and probably educate you in the interim. Probably save your sanity….

My Kossel update….today @ 1143 hours of printing time total, my temp for the hotend was all over the place. I wiggled the thermistor wire and it came off in my hand. Seems all that shaking around on those little thermistor wires took its toll. Fortunately I had a spare so within the hour it was back printing. Hence, spare parts are a must…


3D PLA Finishing

If there’s a down side to 3D printing it’s the absence of colours. Basically the colour of the filament you have is the end colour of your print. True, that is some cases this works out fine. Like if you have white filament and you’re printing ghosts. Other times, not so much.

There’s been several attempts at solutions for this dilemma and they range from a custom multi-filament feeder Mosaic Palette to running filament through a felt tip marker to colour it. Course you just can’t change colours all that easy on the fly with it.

Appearing is a new extruder that has a single output but takes three feeds in. So tricolour if you will. My guess is that since the head looks like a plumb bob it’s going to have a fair bit of mass. Not something you want to swing around on the delta style printer.

So what’s some of the alternatives? Well, one is painting the end print. By hand or air brush. I tried this a while ago, sprayed a base coat of primer on a print and than hand painted it.

The paints I used were the typical Testers Acrylic’s model paint:

IMG 1238

Even with the under coating of the primer, you can see coverage is no exactly what I’d call great. Sure, you can say it’s got some “texture” but really coverage is minimal and depends on the primer colour.

IMG 1018

As stated I’ve tried the Dollar Store paint, Testers, Createx, and a number of others. Okay, but not great.

Back at the Dollar store the other day looking for some solar stuff to scavenge for parts, I happened across these…

IMG 1237

Same sized bottles as the Testers, but in scads of colours. Nail Polish. Like the wife uses. As I recall it’s pretty thick stuff and covers anything. So, I grabbed a few packs of the stuff (cause it’s dirt cheap).

First thing was, no base coat, just paint a sharkz head and see what it does. Single coat.

IMG 1239

You know, it actually covered pretty darn good. But what about trying to cover something on top of dark PLA…I used some gold glitter nail polish:

IMG 1240

Single coat and I wasn’t too careful about putting it on so I didn’t really try to fill in all the holes, just slathered it on. I’m thinking an additional one or more coats and it would be solid.

Then I went back and found my Minion. Repainted him with the colours from the nail polish…

IMG 1235

That’s two coats of nail polish colour. The light colours will not cover the dark coloured nail polish so keep that in mind. Nail polish always comes with a little brush and it works for the most part. If you need to get into small areas, you’ll end up using a custom brush size.

Put it on thick because it will shrink down thinner as the solvent in it evaporates.

The stuff dries reasonably fast…BUT, it stinks. You need GOOD ventilation or about about half an hour you’ll see nice little moire patterns in front of your eyes and be mumbling to yourself…ohh…look at the pretty butterfly…


Kossel Fanboy Conclusion

In part 2 of the blog I showed the new fan arrangement on my Kossel. I also pointed out that just turning on the fans and running them flat out is not the best way to use them.

Thus, for the final entry in the Fanboy series (at this point any way; months of testing my change this at some point) is that we need some form of control for the fans. Because these fans are not really part of the Kossel, you just can’t plug them into D9 on the RAMPS board let the firmware control them. Why not? With two fans running flat out you can over cool and since the firmware has no feedback to tell it that, this job is better left to your eyes so you can make adjustments to compensate.

Fan controls can be as simple or as complex as you want. For simple, just using a variable fan control off eBay. These are called PWM (pulse width modulation) motor speed controls. Cost is a couple of dollars.

$ 57

The slight disadvantage is that you might need two of them. I haven’t tried to run two fans at the same time with one of these types of controllers. When you have two of them, if you’re trying to balance the output from each one, well, close is the best you can hope for.

Something else I should mention here, since it bit me, is that not all fans will work with a PWM signal to control the speed. I have a number of 4010 (40mm sq x 10mm deep) fans and one of them flatly refuses to work via PWM. The other fans, and they are all two wire fans like the Kossel hot end cooling fan, work fine from PWM.

I started off with a couple pieces heavy servo silicon wires and to connect the fans to the controller I was going to make.

IMG 1050

The positive of the fans share a common connection and the negative of the motors is the PWM signal. Silicon wires make for a very supple cable.

For the controller I had all sorts of ideas and ending up using a board that was actually designed for those cheap stepper motors. These things are all over eBay and very inexpensive. The controller board itself is usually only a buck.

5V Stepper Motor 28BYJ 48 AND Drive Test Module Board ULN2003 5 Line 4 Phase | eBay 2016 03 07 11 42 35

I left the servo wire end on the cable so it could plug directly into the controller board. The ULN will drive 7 servos, up to 500ma each. The Kossel fans barely draw 100ma so easily within spec.

IMG 1052

Making things small comes after the prototype phase so a standard UNO was pressed into service as the test platform.

IMG 1056

In trying to keep things small, the goal was use an Arduino Pro MINI, OLED display, couple of pots (50K) for speed control setting and the ULN2003 driver board.

Adafruit’s SSD1306 library for the OLED provided all the routines for display.

IMG 1058

We have a left and right fan, and complete independent control of each. It takes about 40% PWM to get the fans starting to turn but once going I found you can run them down to 15% and they still keep turning, albeit very slowly. Like a gentle summer breeze.

The Arduino takes 5V and the fans take 12V. The ULN2003 board is wired to take 12V, then a dc-dc buck converter was used to stop the 12 down to 5 to feed the Vcc pin of the Arduino. The source of 12V was going to come from the 20A 12V power supply that runs the Kossel.

IMG 1061

I quickly designed up a temp housing for the bits and pieces and then mounted it on the top of the Kossel where my power supply lives.

IMG 1073

So there you have it. The ultimate adjustable fan control. All that remains is the eleventy-leventy years of testing…


Kossel Fanboy Part 2

When we last left our hero he was hanging onto the side of a …. wait a minute, it wasn’t THAT long ago…

Or then again, maybe it was. I know everyone has had days like I had yesterday. First off, a print on the Kossel failed. Imagine that huh. Yeah, I know, not a news flash there. As it happened it took me a while to find the problem. I wasn’t extruding enough filament and no amount of fiddling with the extruder tension or heat would cure it.

Keep in mind that I’m using David’s new UBER designed extruder.

IMG 0917

This new design is a work of art if I may say so. You can finally get some descent tension, it’s using the larger 1/4 NTP pneumatic fitting (which makes threading the extruder EASY!). Notice the knob? Yeah, beats taking the thing apart to feed in filament or shove it up from the bottom. If you print one (ask David for the design), print it at 50% or higher infill.

I didn’t.

This of course explains the problem. That lever arm with the bearing managed to break between the bearing and the spring. Of course I didn’t find it right away. Plus I didn’t have a spare extruder I could lay my mitts on to print a new one. Into the junk box to find one and, find one I did. Ages old now (in Kossel printer years), but I swapped all the fittings and printed off some new lever arms. With mucho infill…

Finally started to print and decided to bump up the flow rate in Slic3r. It’s in the filament tab and the instructions aren’t all that clear on the number range, but I typed in 103 (103 percent) for a little boost of filament. Sent it to the Astrobox (my wireless print controller). This resulted in the nozzle trying to etch the glass once it slammed into the bed. Turned it off, reset it. Re-sliced, uploaded. Same etch result. Ugh.

So with the power off, I’m busy trying to figure out what went wrong. First off, if you have to shut off everything when it’s all heated up and going, well bad things can happen. The filament broke off inside the hot end. the nozzle plugged, I couldn’t get the old filament OUT of the stupid fitting and the only good thing was that I didn’t burn myself on the hot end.

After clearing all the mess up and getting it back together, I thought I’d found a bug in Slic3r. Nope. Operator error. To verify that, I actually opened the GCODE files and looked at the “E” values. Apparently when you want to increase the flow, say by 3%, it’s entered as 1.03 not as 103.

But that’s more or less how the day started out and ended…

None the less, I did manage to design a new fan shroud. When I showed David what my previous designs looked like, I believe the same bolt of lightning hit us at the same time. Adjustable nozzles.

Mels Fan Adjustable

The nozzle in the duct is tapered so you can more or less aim it. I added a small piece to the back to sort of clamp it in place since I didn’t know if it was going to shake itself to pieces or not. It doesn’t…

Mounted on the hot end, it’s fairly small, it blows a whack of air out, and it doesn’t rattle or anything.

Mels Fan On Kossel

I put the arrows on the direction so I can “kind” of aim it. It’s snug enough in the housing that I need some pliers to twist it and it doesn’t move after that.

These are the same style fans as the Kossel kit comes with. Standard 12V fans.

Right now if you’re thinking that all the Kossel cooling problems are solved, you’d be wrong. See it works a litle different than one might assume. In spite of all the info about filament, when you blow air on it, you cool it and cause instant shrinkage. With PLA it’s about 2-4%. ABS is more. What you want is enough heat to give you good layer adhesion but at the same time, enough cooling so overhangs and details aren’t cooked.

In my testing I found that if I ran ONE of the 12V fans at about 9Vdc, it actually did a pretty decent job. But it was limiting in other ways because only one fan was running.

As they used to say in those old movie serials, “Stay tuned for our exciting conclusion in the next episode”. Which as I recall never happened because they were a SERIES.

Kind of applicable to a Kossel isn’t it…

But Part 3 is coming. Soon.


Kossel Fanboy Part 1

Anyone with a 3D printer knows what happens when you print hot filament on top of hot filament. For those that don’t know, you get sags, warps, and it can be one butt ugly print when it’s done.

Of course there’s sometimes ways around this conundrum. For example, printing multiples of the same wee part. This gives the objects a brief time to cool down before they get hit with more hot filament again. Some try to slow the printer down so it takes longer between layers or it just travels slower.

Hence, what we have here is a box full of bandaids. The solution is cooling. Easy huh. This begs the question of “So why don’t the companies put cooling fans on the hot ends and be done with it”.

For the simple reason, it’s just not that easy to do it right. Therefore better to turn a blind eye to it, tell everyone to print multiples of small objects, slow things down…yadda yadda yadda… Or as W.C. Fields put it, “Go away sonny, ya bother me.”

Check out Thingieverse or Goggle “hot end fan” and you’ll find ALL manner of examples. Some are truly innovative, others would serve better as a “how not to”. The majority of the ones that I’ve spotted all seem to be at least partially melted by the hot end. Probably a rite of passage. Others look like the hot end has a bad case of fan hemorrhoids that no amount of Prep H is going to help.

I’ve had a 120mm case fan mounted on the side of my printed bed for a couple of months and it kinda works. The fan has an adjustable iris so I can control the amount of air it is spewing out. To a point, it works not too badly. PROVIDED the area that needs cooling is facing the fan. Yep. If the print is flipped with the area needed cooling not exposed to the fan, ah yeah, not so effective.

The quick and dirty solution is mount another 120mm case fan on the other side of the print bed.

Since I have a delta (Kossel) printer and the bed doesn’t move this was an option I considered. But darn it, those fans aren’t small and they do get in the way at times.

One of the guys at my makerspace designed some fans with magnetic mounts that fit on the side of the head and if anything should hit them, the magnets lets the fans and shroud fall where it may. Obviously if it was going to hit something, it’s already too close to the print bed and I’m not big on magnets for attachments that are going to be exposed to some rapid movements.

I found a basic design on Thingieverse that I kind of liked and that was the basis for my trip down windy avenue….

A shroud evolution looks something like this (note that my final design is not in the line up, it’s coming in part 2 of this blog):

IMG 1060

On the far left we have the PEEK cooling fan for the hot end. There is no cooling for the filament at all.

Next is a series of designs that replaces original cooling arrangement and allows two fans to be mounted for filament cooling.

The first one in the series (next to the original) is for a 40 x 40 x 20 blower type fan. Blower fans have the advantage over axial fans in that they handle back pressure just peachy and maintain a fairly constant flow while doing it.

The original cooling fan fits on the back of those new designs.

Mels Peek Cooling Fan

I keep the fans about 5mm above the print bed.

Mels Fan Clearance

Axial fans on the other hand, they move way more air than a blower fan, but they want a fairly unrestricted airflow path to do so. Back pressure is not their friend and generally if you have too much, it can shorten the life of the fan. Keep in mind we’re talking about a $3 fan.

I’d ordered some blower fans from eBay but since the slow boat is going to be another month before I see them, I decided to try out some axial fans. That’s the next design in the series. I never modified the slots for the blower fan so they are small and offset from each other. In my testing with a fan on one side only, it worked quite well for the most part.

The last design is a better slot arrangement and internally more directional. You don’t want to cool the heater block or the nozzle, you want to cool the filament.

Mels Fan

With two fans in the last design, running flat out, I was starting to see why you can’t just add a fan and say it’s done. On some prints, there was too much cooling and while the print didn’t delaminate, it wasn’t great either.

Remember when you started printing and you wondered why some prints came out great while others didn’t? Then over time you got so you could recognize design fault areas before you printed them? Cooling is more or less the same thing. You know where you’re going to need a goodly amount and where some will be fine. Ergo, variable.

So this is how things start and why at some point you’ll feel like Edison in that you discover yet another way it won’t work….stay tuned for part 2.