Full crosbraces for Prusa Mendel

Re: Full crosbraces for Prusa Mendel

droftarts — Mon, 06 Feb 2012 16:03:54 -0500

After reading this thread and thinking about the issue, I've had a go at producing a new frame vertex which aims to stabilise the square section of the Prusa frame. It does this by strengthening the joint from the triangular part to the square part, at the expense of the (as I see it) overly-strong constraint of the triangle, which, as someone posted earlier, is inherently strong. There is also provision in the vertex for full cross-bracing, or using a sheet material to constrain the squares of the frame. The scad file is parametric, so feel free to play around with width and depth of the vertices if you feel that the size I have come up with is compromised in another way.
[www.thingiverse.com]

Re: Full crosbraces for Prusa Mendel

Justblair — Fri, 20 Jan 2012 18:18:55 -0500

martinprice2004 Wrote:
-------------------------------------------------------
> Yes sounds a workable solution. Perhaps theres a
> way to set it up so the laser movement is
> amplified across the scale.

I am not sure we will need to amplify it. Observation says that the difference is considerable. It should be easy to see it using a mm scale.

Re: Full crosbraces for Prusa Mendel

martinprice2004 — Fri, 20 Jan 2012 18:10:55 -0500

Yes sounds a workable solution. Perhaps theres a way to set it up so the laser movement is amplified across the scale. The improvements will likely be in the fractions of a millimeter in X so we it needs to be something quite accurate.

Accurate measuring devices are getting quite cheap nowadays, so don't be put off getting a bit professional, for example dial test indicators (dti's) start at around £10 and digital micrometers around about £8. Theres many to choose from on ebay. A dti is probably best for this sort of measurement. Either buy one with an adjustable stand or just bolt it to a simple wooden frame and zero it on the end of the z axis motor mount.

Re: Full crosbraces for Prusa Mendel

Justblair — Fri, 20 Jan 2012 17:37:03 -0500

I was thinking about this today and I have an idea for measuring improvements in stiffness using a line laser, ruler and a luggage scale.

Mount the line laser securely on the top of one of the z-axis steppers, pointing at a parallel wall.

Secure the ruler horizontally so that the line laser intersects the 0 marker.

Attach the luggage scale (or more accurate if you have it) to the z-axis smooth rod at the top of the printer.

Pull till you get a decent amount of shift in the frame

Read the scales.

Apply the upgrade (in this case the cross brace)

Now apply the same amount of force on the luggage scale (ie get the reading to be the same as before.

Read the amount of deflection as indicated by the line laser and the ruler.

I am sure that someone far more knowledgeable will point out the flaws in this method, but it is quick, uses household items and if an upgrade is worthwhile it should at least tell us if the x-axis (which is the weakest) has been stiffened.

If scales are out the question we could also use a weight, string and pulley to apply the lateral force.

Re: Full crosbraces for Prusa Mendel

martinprice2004 — Fri, 20 Jan 2012 16:24:13 -0500

Nothing is ever as simple as in the text book.

Theres a slight confusion here that corner to corner rods are the best solution, they MAY be better than the one in the original post, but they may be worse. There are several reasons for this.

1) It assumes steel is the same in tension and compression. This is approximately true for steel beams, but for thin rods. Euler Crippling Loads come into play, making the effective strength/stiffness better in tension than compression, this makes the simple corner to corner analysis not quite so clear as the mechanism then works in 3 dimensions imparting side loads on all the members. Fastening the cross point at the centre would probably be an improvement.

2) The corner to corner assumption is more valid when analysing simple pin jointed structures with thick struts. What we have here is not a pin jointed structure, so that assumption is not true. The bars are not subject to simple tension or compression, but also side loading (not present in pin jointed structures). The only real analysis which truly reveals which is best would be FEA.

3) The Thingiverse structure shown in the first post has substantial width centre blocks, which stiffen the assembly considerably to resist bending in the rods, A simple cross structure may not have this.

4) The forces applied to the mechanism are complex and vary considerably in service. We are not looking at a texbook strut assembly with point loading.

We need to keep an open mind as we may be surprised what works best on a reprap.

Re: Full crosbraces for Prusa Mendel

Buback — Thu, 19 Jan 2012 12:02:50 -0500

If you can in someway secure it on it's side, you can hang a know weight like a kg on the top vertex and then measure the deflection. I think that'd give us a more quantitative measurement we can compare other machines against.

Re: Full crosbraces for Prusa Mendel

Lanthan — Thu, 19 Jan 2012 08:37:03 -0500

martinprice2004 Wrote:
>
> The basic principle is shown here and should be
> familiar to a lot of people.
>
> A simple bench setup could be made to test this,
> perhaps clamping the base and pulling the top of
> the machine with a known load in X.

This is a very sound proposal.
By just firmly pressing vertically with one finger on the end of a rod on the top, I can induce a noticeable displacement of the lower Z bar end on the same side, maybe half a millimeter to one millimeter...

Re: Full crosbraces for Prusa Mendel

martinprice2004 — Wed, 18 Jan 2012 16:22:09 -0500

In the automotive field, a good test of a structure is tortional stiffness, where you fasten one end of the structure then apply a twisting load to the other to measure angular displacement (its often quoted as a measure of how good a chassis is). Perhaps this might also be a simpler way of getting a feel for how various designs perform against one another.

The basic principle is shown here and should be familiar to a lot of people.

A simple bench setup could be made to test this, perhaps clamping the base and pulling the top of the machine with a known load in X.

Re: Full crosbraces for Prusa Mendel

Lanthan — Wed, 18 Jan 2012 11:50:14 -0500

martinprice2004 Wrote:
-------------------------------------------------------
> Resonance? Try attaching a large mass somewhere
> to change the natural frequency. Or better still a
> large mass suspended in rubber attached to the
> frame. As an added bonus it apparently works quite
> well if your printing during an earthquake.

Yeah. Tuning. But first we need to determine what those natural frequencies are (hence the other thread and first attempts)

>
> How about putting 10mm od/ 8mm id tube around the
> main frame members with a washer and bolt either
> end. It would be a relatively easy upgrade to see
> if it improves things. You could pre tension the
> frame struts.

Very interesting idea!
However, if the main problem is at the plastic/metal interface rather than with the metal rods, as we are suspecting it, that might not help that much.

Just found this paper. Those guys get experimental data by building a tetrahedron and submitting it to hammer blows. It is possible that this practice would satisfy many a Mendel builder ;) Besides, good starter for anyone whishing to do some FEA on the frame.
http://sem-proceedings.com/22i/sem.org-IMAC-XXII-Conf-s20p04-Modal-Analysis-Tetrahedral-Machining-Structure.pdf

Re: Full crosbraces for Prusa Mendel

martinprice2004 — Wed, 18 Jan 2012 10:48:18 -0500

Resonance? Try attaching a large mass somewhere to change the natural frequency. Or better still a large mass suspended in rubber attached to the frame. As an added bonus it apparently works quite well if your printing during an earthquake.

How about putting 10mm od/ 8mm id tube around the main frame members with a washer and bolt either end. It would be a relatively easy upgrade to see if it improves things. You could pre tension the frame struts.

Re: Full crosbraces for Prusa Mendel

Lanthan — Wed, 18 Jan 2012 06:33:51 -0500

Indeed. For the majority committed to the Prusa Mendel frame (although coming upgrade options will permit a very hign rate of parts reuse) a couple of meters of threaded rod and three to six printed parts will allow stiffening the base and one of the sides.

What I am seeing in the spectra (more coming soon) is that high freqency resonances in the rods themselves aren't that much of a problem as I initially thought, most of the disturbances and energy are concentrated in the very low to low frequency range, up to 500 Hz, among them the large vertex rocking one can see and feel. I suppose too, play between the thin plastic parts and the rods.

Benchmarking should resort to the accelerometer rather than to a mike. I am having a look at the ADXL335 which features a very decent refresh rate.
ADXL335
Though, I wonder if the Mendel frame vibrations warrants plus or minus 3g ?

Re: Full crosbraces for Prusa Mendel

Justblair — Tue, 17 Jan 2012 17:38:17 -0500

I have the printer running tonight printing a wades extruder in 100% fill. Lots of backwards and forwards acceleration and deceleration.

With the braces attached the Prusa looks and sounds a lot more composed. There is still a little movement in the x-plane. But we are talking a couple of mm at the top of the vertex now. This is a fraction of what I was getting before.
in
More impressive is the damping effect. Minus the bracket I reckon that the Prusa may be prone to oscillation, especially when filling into tight shapes. The bracket really brings this right down. The mechanism I think is that the rods on the lower half of the printer are bending and absorbing the shock. Pre-bracket the printer is rocking about due to the slight play in the plastic brackets

So what I am seeing is that there is a double benefit. In the first instance the movement in the X plane is reduced, Secondly oscillation is reduced considerably. I agree that getting a completely solid printer may be impossible on the mendel format, but for the legions out there who have now committed to the design, the brackets are going to be a worthwhile addition. I still am unsure if the quality is better, but with less noise and vibration I will wager that the printer is going to stay in alignment for longer and wear should be reduced.

Re: Full crosbraces for Prusa Mendel

Lanthan — Mon, 16 Jan 2012 05:56:35 -0500

@Justblair: THX for the feedback and comments!

Same findings here: it helps, but it is not the end of the road.

Should I keep that frame, I'll crossbrace the lower side - or a couple of parallell bars supporting Z motors where they belong, at the base of the machine.

I'd like to help getting the guesswork, hear-say and promotional snake oil out of the evaluation of machine frames (and other components). This might foster real improvements...

So I started some preliminary work on acoustic analysis of different frames Have a look at this thread, your comments would be much appreciated: [forums.reprap.org]
Next one to be auditioned (with same firmeware & gocde) is a cupcake. Has always been some sort of musical instrument, since it is built much like a resonance box(and I suspect this is a weak point of boxed frames)

Also: who could provide us with a calculation of the fundamental frequencies of the rods (stainless steel) with the lengths involved in a Prusa Mendell? I started to look this up, I think I found a couple of sweet articles, but things advance faster with several people investigating some points.

Re: Full crosbraces for Prusa Mendel

Justblair — Sun, 15 Jan 2012 17:33:43 -0500

I have been reading with interest the debate as to how effective the cross braces can be.

I have just completed a set for my Prusa and while there is still movement in the frame, the cross brace makes a valiant effort in removing lateral movement.

I have not used the upgraded frame in anger yet, It will be interesting to see if the additional stiffness improves the print at all.

Re: Full crosbraces for Prusa Mendel

Emmanuel — Fri, 13 Jan 2012 14:14:47 -0500

On the aluminium extrusion supplier side, I just received an order from KJN (discovered on the wiki they sells to individuals, prices seems correct), and I'm starting to love that over the threaded rods...

At £2.90/m, most of the cost was the t-nut (but their basic t-nut can be inserted in the slot at anytime).
I also ordered an handle to carry the reprap ^^ [reprap.org]

Re: Full crosbraces for Prusa Mendel

Buback — Thu, 12 Jan 2012 12:41:08 -0500

Ah ok. Yeah the acrylic will look slick, like it's floating :-)

BTW, if anyone is interested in the place i found for dibond, it's here:
[www.858graphics.com]

But i just noticed they have a $100 minimum :-(

Re: Full crosbraces for Prusa Mendel

nophead — Wed, 11 Jan 2012 19:28:30 -0500

I have used Dibond for the Y carriage as it is stiff and light, but I may use the piece cut from the middle of the frame for that.

I don't think it is stiff enough for the frame though and the sheets need to be thick enough to screw into or tap. The upright sheets could have nuts on the back but its tricky to place them without clashing with the buttresses.

Re: Full crosbraces for Prusa Mendel

Buback — Wed, 11 Jan 2012 18:56:57 -0500

You've used dibond before, Nophead, why not for the frame? too thin?

granted, I haven't had much luck finding it, myself. I found one printer (as in a purveyor of printed signs) that sells small custom signs for around $10, but most suppliers are selling 4x8 ft sheets (or large sheets at least) for $150-250.

What's nice with getting it form a printer is you can send them an image file with all the holes marked out and cut lines highlighted.

Re: Full crosbraces for Prusa Mendel

nophead — Wed, 11 Jan 2012 18:36:14 -0500

Yes the Prusa bar clamps are a pain to adjust. I replaced them with Mendel bar clamps on mine and just fitted two of the four bolts.

Re: Full crosbraces for Prusa Mendel

yellow_fish — Wed, 11 Jan 2012 17:46:26 -0500

Fair enough, I retract the comment about limits of plastic. Although I still feel that interlayer bonding can be improved.

If anyone has anything they want modelling, give me the data to model and I will do my best to model it, for example frames ect, especially for vibration as that appears to be more of an issue than deflection due to static loads.

As an aside, if anyone can give me a rough set of loads that are produced during printing I will try and do a topology optimisation and find an optimal structure. Obviously the more complete and accurate the data the better I can perform the optimisation. I would calculate the load myself but I simply do not have the time right now, but I need to work out how to run optimisations in ABAQUS from script and this would be a relatively simple task to attempt it on.

@ Nophead:

I've been following your progress on the mendel90 and the design does seem very good, a vast improvement over the mendel. The orientation of the material (MDF/acrylic) does have a huge effect on its strength and your design exploits that, meaning you can get away with using minimal material. Time will tell how acrylic stand up to the vibration of a fast printing reprap, Although MDF is another alternative I think acrylic is the better option as it is stiffer, although due to that stiffness it is also less forgiving, especially if its badly machined and has jagged edged to propagate fractures. I have no idea if this will be an issue in practice and look forward to seeing the fruits of all your hard work! If it does work as well as expected then it will be a vastly cheaper and simpler machine. I say simpler because putting all the nuts on a mendal, although easy, is tedious and time consuming. It's also not that easy to adjust when you clamping something which deforms as you clamp therefore changing its position, like the smooth rod for the y axis, or the z axis for that matter.

Re: Full crosbraces for Prusa Mendel

TomCeeMe — Wed, 11 Jan 2012 16:30:31 -0500

now where are those .scad / .stl files Nophead?? :D

just kidding, I know a lot of time is involved with it!

/tom

Re: Full crosbraces for Prusa Mendel

Lanthan — Wed, 11 Jan 2012 15:50:54 -0500

yellow_fish Wrote:

> It
> does appear that we are reaching the limits of the
> development of frames using printed parts if we
> want to print with great accuracy at high speed,
> whilst having a small and light machine compared
> to the build area.

Printed plastic might still have some role in a frame if we value it specifically for dampening purposes.
More data and modelization would be useful for those purposes.

At the other extreme: would you consider a frame made with arc-welded iron profiles? the inverter/welder comes with an initial cost, but the base material is quite inxpensive... needs a good rig to pre-position the profiles with enough precision...

Re: Full crosbraces for Prusa Mendel

nophead — Wed, 11 Jan 2012 15:41:31 -0500

The Sells Mendel has much thicker vertex brackets but it still isn't stiff enough, so I don't think you can blame the plastic brackets. It is simply the geometry that is wrong.

As stated before triangles are strong even when made with soft brackets, they are just not strong in the direction they need to be as deployed in Mendel.

I use plastic brackets in my Mendel90 and they seem plenty strong enough even though they only have 3mm walls or are 4mm thick. They are strong strong in the direction they need to be. I could make them a lot stronger, but having built the prototype I am happy with its stiffness.

As far as aesthetics are concerned I agree MDF does not look pretty, but I think acrylic sheets with big radii on the corners will look nice in a domestic setting.

Re: Full crosbraces for Prusa Mendel

Buback — Wed, 11 Jan 2012 15:35:31 -0500

The printed tnut is really just to prevent the nut from spinning. I also recommend using a washer with the tnut, but don't have a depressed cavity because it won't print. You would certainly have a problem if you removed the screw, because the washer would slide out, and getting them back into place would be a major chore. I haven't assembled all of the printonian yet because i got sidetracked with the prism.

I wouldn't use m3 with extrusions in a printer, though, except to attach accessories or sensors.

Perhaps M5 are the way to go, but as you say, the plastic is the weak element, so why bother. I'd rather use plastic instead of manufactured parts, though, since extrusion already feels like cheating a bit already, as do threaded rods.

Re: Full crosbraces for Prusa Mendel

yellow_fish — Wed, 11 Jan 2012 14:30:31 -0500

The Rexroth only sell M6 and M8 t nuts for their 10mm groves. This reflects how strong the 10mm grooves are and the pressure they can take. Using smaller screws seems a little pointless to me as you can't fully exploit the strength of the construction system.

No offence meant by this but to me is seems a bit stupid to print T nuts which are not as strong as steel (or even Al) for use in a structure where your trying to maximise the rigidity. Its fine to use M4 for the 20x20 but personally I would want a stiffer frame than that and actually ensure that any issues due to lack of rigidity of the frame during high speed printing is removed. To achieve that I would optimise the brackets for uses with extruded plastic (which is an orthotropic material, like CF, where the strength is in shear and compression but not so much between layers) and then use steel T-Nuts.

If, on the other hand I just wanted to simplify the build and reduce build time by using extrusion, as opposed to increasing rigidity, then I would be quite happy to use printed T nuts. It does appear that we are reaching the limits of the development of frames using printed parts if we want to print with great accuracy at high speed, whilst having a small and light machine compared to the build area.

One more thing, nut and washer is not a good combination for T-Nuts which may be placed in the middle of a length of extrusion as you cannot tighten them. Any T-nut cannot need access from the underside.

A potential improvement would be to use the captive nut printed version and then add recess on the top and bottom for washers.these washers spread the load across a larger area of plastic making it take much higher loads and stop the plastic making direct contact with the Al. I hope that made sense, but if not I'll knock up a quick solidworks for you to see what I mean. The only advantage this method has over brought ones is a very significantly reduced cost.

This seems like a very negative post, and I guess that's because I personally feel that I would prefer better print quality over excessive use of printed parts in the frame. I also feel that the quality of printing is not really there yet on my machine to print structural elements that can compare to any isotropic materials such as moulded ABS. I find the that interlayer bonding is significantly weaker than the in plane strength and when I have some time I plan on scientifically measuring the structural properties different infill patterns provide and attempting to improve the interlayer bonding to create an isotropic material.

My basic point ,I guess, is that the prusa is weak mainly in the x axis and wobbles. This is not so much the rods deflecting as the plastic deflecting. Since we know that he plastic is the weak point of the Prusa, surely we want to ensure that it is not the weak point of the next gen, and therefore should use it very carefully in critical points like joints, where it kinda failed before.

Re: Full crosbraces for Prusa Mendel

Buback — Wed, 11 Jan 2012 12:42:29 -0500

m3 plus fender washer (9mm diameter) fits perfectly, except for the spinning nut, but i've made some m3 tnuts that print nicely and prevent spin (up to some unknown torque. i haven't investigated at all). I would only use m3 in situations where weight (edit: or space) is an issue, though. I used mostly m3 when designing the Printonian truss, but used m4 with the Prism. I think the mendelmax uses m5.

Re: Full crosbraces for Prusa Mendel

yellow_fish — Wed, 11 Jan 2012 12:30:56 -0500

bit more digging on rs and found that the 20x20 is only £15 for 3m, £5/m seems pretty decent. The fixings are very expensive but brackets are not hard to make, especially if you have a CNC mill! the groove for the 20x20 is 6mm, M3 square nuts a 5.5mm and m4 are 7mm. The M3 wouldn't be a perfect fit but it's diagonal length is 7mm so would lock into the groove without rotating. Personally I don't think this would give a strong enough joint although that is defined more by the bracket and how many screws than the size of screw itself.

Do bare in mind though that the 40x40 is >10x stronger than the 20x20 and only 3x times heavier. plus the 40x40 has 10mm grooves which seem to be the standard from the way things are going.

Also just remembered about a product call helicoil. Not sure of a supplier but they can give a very good thread into aluminium thats much more durable that just tapped Al. Essentially they give a steel thread set into the aluminium. Not sure if its a good idea but this could be used for threading plastic, it would be interesting if someone got any for threading Al, for them to try threading extruded plastic with it.

Re: Full crosbraces for Prusa Mendel

Dale Dunn — Wed, 11 Jan 2012 12:24:38 -0500

Buback Wrote:
-------------------------------------------------------
...
> But i totally agree that misumi doesn't seem to
> want to do retail work and so seems to have an
> ordering system designed to disuade normal people.

Definitely not a retail outfit, so they're geared to be business friendly. I've had them ship to my house without any mention of who I work for. I got the follow-up call too. I don't remember what I put in the field for business name. I might have just said "hobby"

Re: Full crosbraces for Prusa Mendel

Lanthan — Wed, 11 Jan 2012 11:37:31 -0500

@Buback: Certainly! I could order with the university address or something like that, arrangements are always possible I could see with ppl from IT dept for example, but I tend to not to buy from companies that practice this sort of discriminatory policies, namely refusing to sell to individuals etc. >:D< That's too much "the old system" the Internet has helped us beat back up to some point... there is always a risk of seeing it come back full blown. It is still too much in the mindset, alive and lurking. (sorry for the fear & loathing rant)

@DeuxVis:
En France
[cgi.ebay.fr]

In Germany
[cgi.ebay.fr]

@yellow-fish: THX for the RS-online link!

@all: you guys are always welcome to my threads B)

Yeah I love the Prism, but I think I'll build some Nophead-inspired contraption first. As for the aesthetics, I currently have the cnc router (provisorily) stationed in the living room, so I guess anything will go ;-)
Meanwhile, been re-designing clampable x-ends for flanged bushings (sigh)

Re: Full crosbraces for Prusa Mendel

Buback — Wed, 11 Jan 2012 10:43:53 -0500

yes Sorry Lathan w all the thread-jacking. great job with the crossbraces. :)-D