[peteeng]

Hi All - I must correct the 6% shrinkage. Although some do shrink that much the usual shrinkage allowance for steel castings is 2%. But that is still a lot of strain. Makers will pay one way or the other. Makers (myself included) will try something in the hope it works then have to pay to get it to work from some sort of "hopeful slush fund". Many builders do not have access to heat treat so other methods have to be found. I'm sure bronze is relatively cheap in USA. My point is with all the known uncertainties round welding just don't go there. Are we happy bolting our rails and bearings into the structure? Yes then be happy bolting the entire structure together. Use epoxy or adhesive to hold it together these are all viable and easy pathways vs welding for hobby and commercial applications. If you've flown on a new Boeing or Airbus lately the wings are glued on with a few screws to boot...Peter

[joeavaerage]

Hi,

I'm sure bronze is relatively cheap in USA.

About $30-$40USD per 2lb spool, and $140USD for 11lb spool.

Craig

[LorenBMS]

One thing that will help is not welding 100% of the joint length. Stitch weld it 1/2”-1” every 6”-12”. Move around on the part welding it up slow, so your are not putting a lot of heat in the part.

For making bolted connects weld nuts can be a great tool to have in your design / fab item in the tool box of construction methods.

[luke peter]

Thanks for the input All.

Pete, how much does a lower rigidity bonding medium effect the overall stiffness of the build? I assume that joint will lack the rigidity that it would have with a homogeneous bond - I say assume because I have done nothing to confirm that. Brazing and bronze mig are not on the cards for me - I don't have the kit to do it.

LorenBMS - I modeled the closure plate with bonded joint assumed on the entire length of the seam, If I just use bracing strips (simulating stitch welds bonding behavior) the c channel flanges were free to deflect a little when the bearing blocks were mid span of the stitches. I don't know how much real world trouble it would have caused, but I keep telling myself it must be as stiff as I can make it within my budget and material constraints.

I am incredibly fortunate to have access to some serious machinery at work, and to have a friend willing to CNC my large components free of charge - this somewhat nullifies the concerns related to warping of the weldments. For posterity - as Pete has repeatedly stated, welding is not suitable for accurate machine building If you cannot perform suitable corrections afterwards - which are expensive usually.

I continue to appreciate the input I am receiving from everyone who has contributed to the thread. Thank you again.

[peteeng]

Hi Luke Peter - Brazes range from E=100-120GPa which is about half the stiffness of steel. However being very small thicknesses plus the meniscus means the stiffness delta to steel is probably not noticable. I'll model it to find out.... Hobby and Maker level builds nearly always fall short of best practices due to low budget, lack of access ot understanding to resources such as heat treat. Or just plain not knowing what the outcome will be if they go down that road. So these forums do provide a lot of info for all builders. Peter

[luke peter]

I would be interested in your results if you are willing to share Pete, also at this point, you may call me Luke without needing to add the surname

Agreed, so many compromises need to be made when you are making a machine out of personal pocket vs when your design is going to pass costs onto a customer, or amortize them into a steady stream of available work for the machine. I bought some tooling for the work I intend to do next weekend, and what is a commonplace purchase for my friend (guhring cutting tools) is a hard sell for me - I settled for somta carbide That being said, the costs for fuel (he is about 700km from me) and tooling are dramatically lower than what I would pay a local shop to do the sort of work I need...

I have always felt forums represent the best peer reviewed knowledge repository for most hobbies, because invariably there are people that know willing to contribute and steer people that dont, either through documentation of their design process, or through direct mentorship and advice. And unlike a professional environment, all contributors have an acute interest in the topic, its expression of passion, not just selling hours and intellect for a paycheck. People are quick to call you out if you are being an idiot, and usually have something constructive to say also.

I compare the average comment quality in the forum to the facebook interest groups I am a part of, and its night and day - simply because commenting is a little less convenient on a forum I think... The persistent record of successes and failures also endures in a way that is searchable and for those willing to read, it represents the collective wisdom of so many like minded people.

[peteeng]

Hi LP - Hi LP - I made a model of a cantilever tube. One is "bonded" ie 100 % efficient elastic connection. The LHS is that version. The middle has a 1mm thick braze connection and the RHS is same as middle version. A 1mm thick braze is too thick but this will highlight the delta. So the middle version is all steel to see if there is an analytical delta due to the geometry which there is. Its 99.63% efficient. The brazed joint has a stiffness of 100GPa and its efficiency is 99.15% but this also includes the geometric efficiency of the model. So 99.63-99.15=0.5% delta for the material braze... Tiny tiny delta... and within all the other uncertainties in a model. So close I'd say ignore the braze and call it 100% efficient same as a weld. Peter

[luke peter]

Thanks for that.

I think it's good for all reading to see - in the case where the design is aimed at minimal deflection (not for structural load bearing), using braze for a metal to metal joint has an insignificant effect on the joint efficiency, and it will help directly relieve the headache associated with re - straightening what was once straight.

[peteeng]

Hi Luke - I used to make fillet brazed steel bicycle frames. These are load bearing joints and they were mountain bikes some 30 years ago., I trusted my life to many crazy frame designs going down hill at ridiculous speeds so I'm confident in brazed structural connections. Peter

[ardenum]

Hi Luke - I used to make fillet brazed steel bicycle frames. These are load bearing joints and they were mountain bikes some 30 years ago., I trusted my life to many crazy frame designs going down hill at ridiculous speeds so I'm confident in brazed structural connections. Peter

would you weld or braze a steel table? The welds are localized at the joints only so I assume the deformation is minimal in regards to the whole structure? Or is it significant enough to braze them instead? Square tubing 60x60 or 80x80 and at least 5mm thick, 600kg mounted on top

[peteeng]

Hi Ard - I have no welding equipment but do have oxy these days. If you want zero to minimal distortion in the table I'd braze. When I made bike frames if I brazed a frame it could be taken out of the jig and placed back in the jig no problem. When I TIG weld frames they could never be put back in the jig (steel or alum). They had to be cold set after welding to correct the distortion. These frames also had machined fits in a lathe so the fits were snug tight. Its up to the Maker/Builder to decide these things in the understanding of the technology used. To me welding is a very crude process. It destroys the grain structure of the metal, it decreases the fatigue resistance of the structure significantly, it distorts the structure significantly, plus other stuff too many to mention. The basic welding faults have not changed in the 40 years I have been involved with it as the physics are the same and unchangable. I still have to deal with welding defects I was dealing with 30 years ago ie we have not stepped far fwd like we have in say composites or bonding. Its a convenient process that's about all. In my current professional commercial work I analyse welded steel structures that crack and fail. I spend considerable time in FE trying to guess where these things will fail and how to correct it. I try to convert steel fabricators to huck bolts vs welding but they are welded to welding. Fatigue is not an issue with cnc structures however but internal stress and deformation is. The structures I design/analyse are totally limited by the welded properties. If huck bolts are used I can raise allowable stress levels considerably maybe 10x fold... So weld away, but be prepared to fix things Peter

[luke peter]

Hi Luke - I used to make fillet brazed steel bicycle frames. These are load bearing joints and they were mountain bikes some 30 years ago., I trusted my life to many crazy frame designs going down hill at ridiculous speeds so I'm confident in brazed structural connections. Peter

Be that as it may, I have not come accross any WPSs that specify a full penetration brazed joint on a pressure vessel, a structure, or a critically loaded roll shaft... It would make post weld heat treat much easier in many cases if the material was never liquified

[luke peter]

Some progress this weekend.

Cleaned up the x axis beams, one will be straightened tomorrow morning at work.


I missed one file when I sent my stuff to the laser cutters, so I had to make little tabs to allow the top plate to bolt onto the Z axis.


I added the relief mount plate to the Z axis, and the belt tension tab to the z axis top plate.


I cut the main 8080 frame elements.


My mitre saw struggled with the first cut, a lot, I respiratored up and added a little tapmatic (Ali cutting fluid) and the rest of the cuts went much more smoothly.

Parts will go for TSR this week, then road trip for machining this weekend.

Slowly coming together...

Luke

[luke peter]

Both beams are straight, the only TSR company left to me in my city were prohibitively expensive. I am going to roll the dice...

Axis of concern is my long axis, (x in my terminology). Will keep cuts as light as possible, and be careful about the order in which I make them. Best case, its all ok, worst case, I leave them in JHB to be relieved there and machined later, and have them couriered back to me.

I prepared an assembly showing the fasteners and the modified intended machining.


For interests sake - fasteners (excluding the ali frame and bed plate mounts)

[luke peter]

I was drilling a bunch of holes last night, preparing for tapping etc. I determined I had overdrilled the z axis spacer blocks by 1.5mm (Clearance holes for m5 bolt drilled 7mm instead of 5.5mm). I traced it back to the model where inventor put M6 coutersunk screws into the assembly as an auto drop instead of m5 - past Luke must have looked at that in the assembly and said, guess I need 6.5/7mm clearance holes...

It triggered an assembly check, I found 32 instances of the auto drop dropping m6 instead of m5 cap screws. For now I am not going to recut the parts, hopefully there is adequate engagement surface on the countersinks for the m5 cap screws to hold ok onto the bearing blocks.

My little drill press (13mm) also decided it didn't want to work anymore last night - I have a spare motor that will suit it on my brass tumbler though. Normally this would not be an issue - but when my available work time is curtailed by 6 hours a day of rolling blackouts every little bit of efficiency helps.

[pippin88]

From what I understand the movement in non-stress relieved parts will usually happen straightaway, not years down the line.
I tried to find some actual literature on timing of movement / release of stress but came up empty (I'm not an engineer and may just have been searching with poor terms)

Machine, release clamps/workholding, reclamp, remachine...

If you machine both sides of a part, the first machined side will likely move when you machine the second side.

[peteeng]

Hi Pippin et al - Most machines are made from "steel" use mild steel. Mild Steel has no significant residual internal reactions going on. This is unlike cast iron and aluminium that have possible internal metallurgical reactions that change the metals volume hence the internal strain. So mild steel once relieved is stable. Mild steel does rust and this is a reaction that has a very big volume change. So coating is important to prevent this.

Internal strain is the result of cold working, hot working, internal metallurgical reactions that change the materials volume and combinations of these processes. All metals are melted, cast into billets then processed to final form. All of this history contributes to its eventual internal strain state. Now 3D printing exists this has its own internal strain issues as well.

Cast iron, various steels and aluminium can be stabilised via thermal treatment. eg if you weld 6061 thats in T4 temper and then heat treat to T6 temper (say 180C for 8 hrs) then you have strengthened the material and stabilised it in one go. Peter

[luke peter]

Hi all.

So - 28 hours of time in the shop later, i have flat and square parts with lots of little tapped holes.

I am exceptionally grateful to my friend Mike from Miltek Engineering SA for the use of his shop, and for the hand holding he gave me during the process.

I wish I had taken more pictures of the setups etc, but it was very busy and I forgot.

Order of work:

Z axis was first, clamped in two vises, and faced - I took of about 1mm to get everything flush, the back plate had bowed inwards during fabrication. no issues with this setup


Then we tackled the y axis - thankfull short enough to cover its whole travel in a single setup.
Clamped between two vises, faced, drilled and reamed for dowels, drilled and tapped for bolts, pocketed, drilled and tapped for the bearing blocks.



We cut some reference surfaces into the top of the top rail and used them to ensure level when flipped for the base plates, clocked the vertical and made the facing cuts. The Y was pretty square already thankfully, so we took around 0.5mm and both sides were flush and flat.


X rails next, more difficult since they are longer than the travel.
Did them as a pair, flattened the first 1330mm of the bottom, flipped them, flattened the first 1330mm of the top, unclamped to check for movement, surfaces were still flat (sighed with relief). clamped up again on the ends and drilled, reamed, drilled and tapped for dowels and rail screws. flipped the pair again and shifted over for the last 100mm (on the flat cut surface) and finished the facing, flipped and repeated. The last sections have a very small perceptible step on two out of the 4 faces, will manually drill and tap the las holes in this section and do some tweaking to get it in line with the rest. The ends that still need some work shown on top of the below pic.


The bed supports were done last, all three clamped onto the bed, and cut flat with the MPG, flipped and repeated. Turned out really well, unfortunately at this point time had run out and I needed to make the 8 hour drive home, so I didnt flatten the bed plate mount surfaces - I will rely on Steve to cut its own bed flat.

Parts out of the shop getting ready to load into my poor hyundai...


Really happy with how it all turned out. No discernable movement in any of the axes after milling, Which is a real blessing.

Regards

Luke

[luke peter]

Finishing decisions.

Powder coaters persist in wanting to charge me by weight - this remains offensive considering the surface area to mass ratio is stupidly small.

So, I will do it the hard way... 2k epoxy primer, filler for aesthetics, then a 2k topcoat - colour TBD.

[Buoyen]

This looks very good. Glad to hear the parts are still straight. I think it helps that you have big cross sections relative to the amount of material removed.