Large Kossel all metal delta build log

The magnets on the effector and carriages are cup shaped, I've added several more flat magnets underneath them to increase the force required to disconnect them. If they were flat the effector would rotate about all over the place.

I get you you mean angle the cup shaped magnets towards the arms. That's worth a try.

The different arrangement with the magnets on the rods is fundamentally the same just the other way up where it differs for the better is having some low friction cups to act as guides to stabilise the joint. Why not do away with magnets and just have metal ball and delrin socket joints?

Edited 2 time(s). Last edit at 09/27/2016 01:31PM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Adding a string and spring tensioner for each leg would be a simple solution. It looks like there is enough room on the effector and carriages to add these.

That's also well worth considering, the cups in the magnets are fairly shallow, and when the rods are fairly low angle to the effector they can almost slip out of the cups, I wonder if a string/spring tensioner would help or whether they would slip out anyway due to the angle. The flying extruder is essentially an elastic link pulling the effector towards the carriages.

Its also worth wondering if my rods might be a bit short. They would not move into this low angle position if they were slightly longer.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Quote
DjDemonD
That's also well worth considering, the cups in the magnets are fairly shallow, and when the rods are fairly low angle to the effector they can almost slip out of the cups, I wonder if a string/spring tensioner would help or whether they would slip out anyway due to the angle. The flying extruder is essentially an elastic link pulling the effector towards the carriages.

Its also worth wondering if my rods might be a bit short. They would not move into this low angle position if they were slightly longer.

Yes I think the Spring tensioner would just exacerbate the issue in your case I think the Magnet cups need to be angled a bit to give that extra grip at low arm angles

I could remount them using printed parts fitted to the effector but I intended no printed structural or functional parts. Might be worth it as a proof of concept then maybe get a new effector machined/printed.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Here's a section view of my effector:


The upper magnet is more than enough to hold the arms in place. The lower disk magnet just provides a nice socket for the ball plus a little insurance magnetism.

Another option that a lot of people use is a manually deployed limit switch. This one is even easier to set up, you make a bracket that magnetically attaches to your effector, with a small limit switch on the bottom. The limit switch just a normal endstop type switch, positioned so that it is exactly underneath the nozzle. It's permanently wired up to your board.

When you want to level the machine, you take the Z sensor off its hanger, and clip it to the effector. You then do the probing, and unclip the sensor assembly to put it back on its hanger.


This one might be easier to set up if you don't have an optical endstop handy.

At least this way you can test to see if effector tilt or reflectivity issues are causing your level problems.


I don't like IR sensors due to reflectivity and effector tilt issues (I've done lots of experiments myself on this).
I don't like nozzle sensing schemes due to difficulty of sensing (piezo sensors often have issues, accelerometers even more so, hinged nozzles cause issues in the print), as well as the height being wrong if there's a bit of ooze on the nozzle.
I don't like inductive sensors due to not sensing the actual height of the material, so if you put a different print bed on there then your Z offset will be wrong.

The only way that makes sense for me is direct physicial contact, and even then the contact must be directly underneath the nozzle if you're using a delta printer.

Edited 3 time(s). Last edit at 09/27/2016 09:44PM by nebbian.

Well thanks for all the suggestions. For the time being the magnets are holding using much reduced speed/accel/jerk settings for calibration, then back to normal for printing.
The calibration is now giving me a deviation of 0.015, which I am fairly sure is about as close as I am going to get and the bed is now level across the surface.

So for the time being it looks as though I've got it calibrated.

Now to print some stuff.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Sometimes you feel as though dark forces are at work. Began printing with pla and the hot end fan (brand new e3d one) stopped working and now the hot end is jammed. 3d printing definitely is no easier if you spend more money on a printer.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Now this is a new one. So I couldn't figure out this failed fan for the life of me, it was intermittent - no wire break anywhere. Removed the JST conenctor and soldered the fan to the cable just to be sure.

Turns out it was the proximity of all these extra neodymium magnets that was slowing/stopping the fan from turning, they must have been pulling harder on the rotor in the fan motor than the coils around it.

My apologies to e3d their new fan is working perfectly.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Plus tried a novel approach to unjamming the nozzle. Removed the nozzle and heatbreak, with them attached to one another, set temp for 230 deg C and held it in an adjustable spanner which would have been impossible without the silicone cover. With it insulating the heater block it suddenly becomes feasible. I then pushed some esun cleaning filament through, then pulled it back out. All the pla was gone. 10 minutes work, most of that waiting for things to cool.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Decided that the bed needs to be clamped. 330mm x 5mm tooling plate with a 300mm silicone heater stuck to the bottom is not heavy enough not to move about a bit, especially during calibration.



Added the stls for the modified bed mounts and clamps to thingiverse here.

I did not reprint the bed mounts just retrofitted the clamps by heat pressing some m4 nuts.

Calibration now at 0.02 deviation and quite reproducible.

Finally printing some stuff.

Edited 1 time(s). Last edit at 09/28/2016 05:02PM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Quote
DjDemonD

Turns out it was the proximity of all these extra neodymium magnets that was slowing/stopping the fan from turning, they must have been pulling harder on the rotor in the fan motor than the coils around it.

Damn. That's some nice detective work to figure that one out.

More persistence than skill. Took ages for it to click, I thought I was over tightening the bolts and bending the fan case, as it slowed right down when I reattached it. I even ordered an expensive replacement fan before going back to it still puzzled.

In the process I've been looking at a lot of 30mm fans, I had no idea there are fans aimed at RC modelers that push 5.5 cf/m at 8v at 22k rpm! That would keep pla from jamming even if they do sound like jet engines.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Your cup problem at low angle is probably because you have no angle on your ball/cup at your effector, redesign them with a 45 degree and make sure there some clearance for the arm to not touch the effector at low angle and it wont happen anymore. Best case scenario is the when in middle of the bed the ball is facing the cup straight at both effector and at the carrier, giving you the max angle in every direction.

I got some N42 ( high temps) magnet on mine 3/8 X 3/8 8lbs and have not ran into that problem so far and my cup are not magnet themself its a nylon cup with the magnet inside and there a gap between the magnet and the ball so I dont have the full pull force for sure but the angle of 45 help them a lot and there no dead zone where the arm cant go and pop out . I dont think a spring will fix your problem if your ball/arm hit the border of the cup or effector, spring or not it will detach.

Looking at your graphic make me wonder WHY the bottom magnet under your balls. maybe they not helping as much as your think, what happen when your cup magnet get at low angle get near the magnet under the ball.. I guess pull happen and your magnet in the cup want to bypass the ball to go directly on the bottom magnet, that probably help to pop them out. with an angle of 45 they will be farther away from each other but with that angle on your graphic I cant stop thinking maybe they affect each other at low angle. Without making a redesign maybe you can try to remove the magnet under the ball see if that help.


With a very minimal gap between the ball and magnet and the right magnet and there really no need for double magnet. I use a multimeter on mine to adjust them, they pretty tight in place so I knock a little on them then with the multimeter I check to make sure the connection is loss and im set.

Thanks. Its definitely in the pipeline for V1.5 to re-orientate the magnets in the effector.
The second magnet was just a quick-fix attempt to see if it improved the holding force, and I'd agree with you that it probably doesn't. It is causing fan problems having the additional magnets.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Hi nebbian,

I am a little intrigued by your statement from a couple of days ago which I have quoted below.

Quote
nebbian
.....
I don't like nozzle sensing schemes due to difficulty of sensing (piezo sensors often have issues, accelerometers even more so, hinged nozzles cause issues in the print), as well as the height being wrong if there's a bit of ooze on the nozzle.
.....

It may be that I am too close to the subject as I put in quite a bit of work on piezo sensing - being blind to the flaws of you own creation is something that nobody is imune to. Knowing the foregoing, I woule be genuinely interested in what issues you see in piezo disk sensing.

Mike

Quote
leadinglights

It may be that I am too close to the subject as I put in quite a bit of work on piezo sensing - being blind to the flaws of you own creation is something that nobody is imune to. Knowing the foregoing, I woule be genuinely interested in what issues you see in piezo disk sensing.

Mike

I haven't tried it myself, but I have seen many, many people with piezo sensing issues. If you have gotten it to work really well, then congratulations!

Some issues I've seen others report include:
1) Sensing outside the triangle formed by the sensors causing one sensor to read negative or not at all
2) Difficulty summing the signals from the sensors
3) Bed needing to be suspended and therefore not being solidly mounted, causing issues when the nozzle starts pushing on already deposited plastic that has warped slightly
4) Complexity of bed mounting arrangement
5) Ooze on the nozzle causing the reported height to be different to the actual height

All of these can be overcome of course. I'm not saying that it's a bad system, but I am saying that I consider a good mechanical probe to be easier to get reliable, than a piezo sensor based system.

@nebbian, I think I see, I believe you may be confusing piezoelectric disks with FSRs (Force Sensitive Resistors)

Two of the issues that you flag up are partly true - I will come to points 1 & 2
Point 3. The bed does need to be mounted so that the force goes through the piezo disks but I find less than 0.02mm of movement in any axis during printing. A spring mounted bed will give problems if the print has started to warp but I think that is true with almost any bed with 3 point adjustment.
Point 4. On a Delta printer - guilty as charged although it can be overcome. On a Cartesian printer I think that the mounting can be simpler than for any other type of force sensor.
Point 5. I think that this is a red herring. Even using paper under the nozzle the ooze should be cleaned from the nozzle. Using a mechanical switch such as a Bltouch still requires that a sensor to nozzle distance must be obtained - again with a clean nozzle.

Back to points 1 & 2. Unlike FSRs, Piezoelectric disks generate a charge (and hence voltage) which is proportional to the force on them. The force on the points of the triangle sum to the applied force even if one of them is negative. In addition, simply tying the outputs of the three sensors will give an output proportional to the applied force. It took a while to work out that this was true and a lot of experimentation to prove it, but it does work.

One of my printers is totally trouble free but it is set to trigger high at about 100 grams. The other one is a delta set at 10 grams which has demonstrated that piezo sensors will also trigger from radiated heat from the build stage - I am still trying to optimise this one.

Mike

When I first build my delta I wanted a sensor to help with calibration too, and I did a manual calibration with paper using the DC42 webpage calculator ( if you use reprap firmware) and the calibration is perfect so far. I did it in march, move the printer 2 time and the printer still perfect, no need to redo the calibration for all those months...

So for me automatic calibration or do it every print is just a gimmick to sell stuff needed only on poorly made printer (mostly plastic frame one) those sensor are only a workaround if you require a calibration every print there something wrong with your printer and you should spend time finding why and fix it instead of spending time on installing those switch/sensor.

This is only my opinion, I bet those who sell the sensor disagree with me but this is the experience I have so far, a solid frame build very carefully should not need those.

I'm not sure a sensor is needed with a solidly built printer, once everything is a known quantity and its reliability has been proven. Its a method of avoiding having to do it the slow way (and to the uninitiated the complicated way) and therefore it makes something like a delta printer more accessible to those who aren't engineers or mathematicians.

I've taken the sensor off my corexy as once the bed was levelled it was nearly always level, and the ABL was not applying much, if any, compensation. So why bother?

Sensor do save time. If I want to quickly check if my bed is level I hit a button and 10 seconds later 3 values appear. Sure I could move the head around whilst looking at the nozzle or position at 3 points and go up and down with my piece of paper but that takes longer, a lot longer.

The problem is we have no perfect sensor, so we try to save time by using them and spend time making them work and figuring out why they don't give 100% reliable height readings all the time.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Before I leave this thread which I have inadvertently taken off on a tangent, I want to quickly reply to some points raised.

@GroupB, I move my printers on a weekly basis, use them in a variety of temperatures, change the nozzles and even hotends frequently for different sizes and materials and change the bed often. The printers are very solid but I do see the ABL coming into use when it warns me that I have a bit of waste filament trapped between the bed heater and the build stage. It also automatically corrects for new nozzle height.

@DjDemonD, we still have no perfect sensor but I have found that piezoelectric disks are hugely better than FSRs. Since they may cost less than 17 pence each rather than the £4 to £11 each for FSRs I think I am happy.

Mike

I like the idea, I've always been quite happy with fsrs so the idea of something cheaper and better certainly appeals.

This thread was more or less done anyway.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

I'm also using Leadinglights piezo sensor circuit on my delta with zero problems and very high accuracy. I swap glass beds frequently and find it very useful. I know that with a sufficiently rigid printer no ABL should be needed but for printers that are less than perfect it's a lifesaver.

Moriquendi

Its finally finished. Got my favourite print surface installed.



Calibration now quite consistent at 0.02 deviation.

This is how its printing:



So I can now say I am happy, 3D printing nirvana has been reached, at least for a few weeks.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

That looks a really nice printer!

---

Large delta printer [miscsolutions.wordpress.com], E3D tool changer, Robotdigg SCARA printer, Crane Quad and Ormerod

Disclosure: I design Duet electronics and work on RepRapFirmware, [duet3d.com].

I copied a lot of your ideas, so thank you.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

It had all been going well when my PSU died. It was this one [www.ebay.co.uk]. Clearly not up to the job of running at max power within my electronics compartment. Quite limited for space so I'm going to try another similar unit but 6A rather than 5. This time no adjustable voltage and no 5v rail, which I didn't use anyway. [www.ebay.co.uk]

I might half open the case and print a case with lots of air holes and fit a fan to it.

Also my cardboard insulation between the bed and electronics compartment was slightly scorched in the middle so I wrapped it in aluminium foil.

Edited 1 time(s). Last edit at 10/13/2016 06:12PM by DjDemonD.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

Okay so back printing after bringing some 24v power over from my other printer. Just have to remember not to run them both at once, I am not sure a 10A 24v supply can power the heated bed on my corexy and the large delta.

Decided to print some petg on this printer. Made the mistake of not running a good amount through the hotend after swapping from ABS, as a result the nozzle jammed. So fixed that, then the IR sensor stopped working, as I had broken the solder joint on one of the connector legs. Resoldered it.

Put it all back together and the IR sensor wouldn't stay in my fan shroud, so I modified that a little to allow the sensor to sit little bit better in the fan shroud.

Decided to attach the insane 30mm fan sold for RC modelling its called a WTF fan, which apparently stands for Wild Turbo Fan. Designed for 6-8v (which is not problem as I am using an adjustable buck converter to step down from 24v) I am actually running it at about 3.5v. At 8v it probably could take off, its more like a drone motor than a fan. So I will be quite surprised if there is any heat creep now. It will probably only last 5 minutes, but it is a level above any other 30mm I've seen before it pushes a serious amount of air. Implemented the thermostatic fan option on the duetwifi, as I can;t stand to have it on unless the hot end is hot. It also makes a serious amount of noise.



I now have the new power supply which worryingly looks smaller than the old one, but is claimed to be 6A instead of 5A. The mosfets on it are better attached to the heatsinks. I decided to print a top cover for it with lots of ventilation.

I have also redesigned my insulation between the electronics compartment and the bed, the previous version was triple walled cardboard with foil bubble wrap on the bottom. It was too thick so was touching the silicone heater. This version is thinner stiff-cardboard, and I had some cork left over from insulating the bed mounts, so I put a sheet of cork over it.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions

So I've ironed out almost all of the issues. Many hot end rebuilds despite using genuine e3d v6. Upgraded hot end fan. Changed the thermistor on the heated bed as the built in one started giving erroneous readings. Tuned the heated bed thermistor so now I can use pid. Shoehorned a thermal fuse in under the bed heater.

The only remaining issue is the k800 magnets keep coming off despite backing off the accelerations and jerk values. So it might be time to change to Hayden's magnets but whilst they are probably much better I'm not sure there is a metal effector or carriages to suit these magnets. So it means compromising on the all metal design philosophy. I'm happy to try them with a printed effector and carriages. Then maybe get some parts machined/printed in metal. It would be interesting to see if it makes a difference.

---

Simon Khoury

Co-founder of [www.precisionpiezo.co.uk] Accurate, repeatable, versatile Z-Probes
Published:Inventions