Where is it all going? Claude – CAD – product

[Mike HurleyParticipant @mikehurley60381]

Just finished viewing the video of ‘what’s coming in FreeCad 26.3.’

One point that struck me was a post by ‘Blowlamp’  about Claude AI and Freecad. His key comment was “It seems that I don’t need to learn FreeCad after all”.

After watching the video that Martin had attached, I really wonder how true it is? Essentially now, from project to completed item(s) once you add CNC or  3D printing into the project stream, it seems that human interaction is now somewhat insignificant in many respects.

So, are we soon stopping calling ourselves Model Engineers and be referred to as machine minders / loaders?

While all of the wonderful tech now available is fascinating in its own right, does it mean that the hobby has simply come to a point where, like many things, the practical side is becoming legacy?

Not to undervalue the superb work done by many contributors to this forum. However, the dividing line between how much is actually done by the individual, compared to what the technology contributes seems to be reducing logarithmically in favour of tech.

It makes me think of Electronics – when I was younger, I spent many, many hours in a cold shed soldering together various things – from valve radios (Does anyone remember the widely advertised ‘HAC’  – Hear All Continents – kit) through the marvels of OC71 transistors etc and then Sinclair calculator kits etc etc. It was my Hobby, and I loved it! But now, as a pure hobby it’s somewhat irrelevant. Yes, some people still do useful bits, often cleverly using modern processors and of that ilk,  but mainly to directly fulfil a quite specific need. The technology has overtaken any requirement to just ‘make’ in many cases.

I’m finally thinking that the hobby we have all enjoyed over the years is finally becoming a dodo. In life, you have to accept that things will change. Whether you think that’s a good or a bad thing is down to personal choice.

Sorry if I’ve bored readers to tears, just one of those things I wanted to get off my chest!

Mike

 

 

 

 

 

[noel shelleyParticipant @noelshelley55608]

Dear Mike, I’m of the same opinion as you in many ways. Kits of the late 90s became ready builts of of the 2010s. If the stories about AI have any truth and I fear they have then the survival of the human race could end quite quickly.

Your reference to the OC 71 brings back memories of 1959 and a small radio I made, things progressed and by 1966 I had made a 7 valve  communications  set. Do you know why the OC71 was painted black ? Never mind, we do our best and plod on ! Best wishes. Noel.

[BazyleParticipant @bazyle]

A decade ago some club members bought ‘fully machined’ kits for locos but found they needed a lot more work to actually fit. Kits are better now but it reminds us that AI and CNC are not perfect.
For a long time MEs have been using laser cut platework and several are getting parts printed that would have been castings previously. and soon this kind of thing may be cheaper than the raw materials at our level.
However can a robot assemble all these fiddly bits into a complete loco? There are complex consumer products coming off the production lines but those items have been specifically designed for machine assembly nothing like a faithful reproduction of an historic machine.

Now, where did I put those clear packaged OC71s I got sixty years ago for my model railway automation sensors.

 

[noel shelley]

On noel shelley Said:

….Do you know why the OC71 was painted black ?

Yes, but I won’t spoil the fun by letting the cat out of the bag just yet!

Julie

[noel shelley]

On noel shelley Said:

….Do you know why the OC71 was painted black ?

 

Yes, but I won’t spoil the fun by letting the cat out of the bag just yet!

Julie

Second, and last, attempt at replying.

[duncan webster 1Participant @duncanwebster1]

Never mind such modern stuff as OC71 transistors, I remember red spot and white spot. Probably all came off the same production line, but all got tested and the better ones got a blob of the appropriate paint.

[Swarf MakerParticipant @swarfmaker85383]

And those early ones were packaged in nice blue faux velvet lined boxes!

[Nigel Graham 2Participant @nigelgraham2]

Where have Julie’s contributions gone? Just two white spaces. Bloomin’ computers! ‘Ood ‘ave ’em?

 

Mike –

With respect I think you have missed a fundamental point by which model-engineering as such is not necessarily in danger, but is changing. Just as it has changed since the days of making small models very laboriously on little lathes that had to be very versatile to allow us to make many of the components.

To use CAD / CAM methods you need their far newer, specific skills; which for the latter does mean understanding materials, tools and processes even if the stepper-motors replace our hands for the dial-twiddling. If anything it adds a layer of skill.

If you go one step further, creating your own project rather than copying an existing design, you need know how to design it, whether you use CAD and perhaps NC machines, manual draughting or rough sketches (on the backs of the address-labels used for posting ME&W?).

I struggle to learn complicated software including CAD, but I do appreciate its advantages and it does not remove my need to make and assemble the real parts, and hope they work.

 

Noel –

It will be very sad if making is replaced entirely by buying, but it does allow owning something the buyer is genuinely unable to make.

In the end, whether model-engineering, knitting, painting or writing music, the value lies in the continuation of both the willingness and ability of people to use their own brains and hands to create things; and that of the wish of the viewer or listener to appreciate such creations. For this ideal, the computer is only a tool; not an end in itself.

 

Bazyle –

The early kits that gave such things a bad name were nowhere near Artificial Intelligence / Idiocy – AI at the time was something agricultural. Nor even of CNC machine-tools, if these things were involved. Their faults were by design flaws and perhaps low quality-control; not inherent characteristics of manufacturing technique.

I helped diagnose why one of these kit examples, a steam-wagon resembling a Foden ‘C’-type to about 2″ scale, would run but not stop. The regulator would open but not close properly: inherent design error compounded by poor part manufacture.

That’s not a matter of any software that may or may not have been harmed in its production; but of the people who designed it and made the parts.

 

I wonder how many of these early kits lie forlornly dusty and rusty, their disillusioned owners unable to succeed with them.

…

The OC71 Transistor (“Valve, Electronic, OC71” according to the MoD Stores catalogues)… Ubiquity in action!

 

+++++

A little light relief –

Just spotted this side-bar ad. for a production metal-cutting circular-saw. It’s not clear if abrasive-disc or toothed blade.:

“tube mill machine.

CONTOR Cold Flying Saw Cutting Online In High Accuracy Without Burrs and Low Noise”.

I operated sawing machines professionally for some eight years and never thought them “mills”, nor especially aerodynamic. “Online”: means? “High Accuracy”: I hope so.

Does the last clause mean lacking “Low Noise” as well as burrs?

Sorry – I’m the sort whom advertising copy-writers must hate!

 

[David SeniorParticipant @davidsenior29320]

I had spotted that youtube video on Claude/Freecad integration a few days ago and decided it was worth a closer look. I always want to make models of ‘things’ but I never get round to it. I have a large library of old technical books, and I love those old ‘how to do technical drawing’ books that have pretty much complete designs in them. My problem has been that I struggle with doing the CAD modelling, and I don’t want to make models without the correct details. Nor do I really want to just build commercial kits – I would prefer something more unusual.

Combining Claude and Freecad has been a complete revelation for me. I scanned several pages out of a book with a fairly unclear design of a steam engine, and all the text, and fed it to Claude. That was at 3.30 this afternoon. I’ve had to do a lot of typing to clarify and correct the models, but it is basically finished. Now someone skilled and fast in 3D modelling could probably do it faster, but I can’t.

Hopefully in the next few days I will be looking up 3D printing prices for the ‘castings’, and getting on with the bits made from bar stock. So far I think it is just fantastic!

Dave

[Andy StopfordParticipant @andystopford50521]

That’s interesting, David, any chance of some images of what Claude produced?

A thing to be aware of if you’re having parts 3D printed is that there are limitations to the process, particularly with regard to overhangs and internal passages, and especially if you’re printing metal (plastic is easier).

I would imagine it would be possible to tell Claude to allow for this, but it might force quite major changes to the design.

[David SeniorParticipant @davidsenior29320]

The engine was out of Machine Drawing by Jones.

 

This is where I am at the moment. Most of the engine is done. I still need to correct the eccentric and valve mechanism. The pump casing is still very crude and I’m not sure whether it will ever be great. That casing is the first attempt, and I haven’t tried to improve it yet.

I think it is fair to say that if you can’t do the 3D modelling yourself it is fantastic tool. If you are proficient with your CAD package then less so. I am working my way through all the details at the moment telling Claude what is wrong, and they are being corrected, but the communication is the issue. If my son (who is a skilled CAD jockey) was sat at the computer and I could explain to him what I wanted changing it would be a lot quicker.

Once I am happy with everything I will decide on a scale, get Claude to shrink to suit, and then ask it to change certain dimensions to suit, for instance, unmachined metal bar sizes and threads. Where the component needs machining the dimensions are less important.

I’m surprised by your comment on 3D printing. My experience is that there are very few limitations on modern metal printing because it is all being done in a bed of metal powder, so support is not an issue. There may be surface quality issues with certain orientations and it is important to ensure that the loose powder can be removed at the end.

Dave

[Andy StopfordParticipant @andystopford50521]

I’m going by all the recommendations I’ve found, e.g. https://www.protolabs.com/en-gb/resources/design-tips/designing-for-direct-metal-laser-sintering/

Of course they don’t really give real-world examples – what have you found possible to print where supports are impossible? I’ve been working on a cylinder block design which requires jumping through a lot of hoops to avoid <35 degree slopes (the minimum that Craftcloud recommends).

[David SeniorParticipant @davidsenior29320]

I haven’t been involved directly with the detail of metal printing but my company does. I guess I’ll find out when I try and get them printed!

[Andy StopfordParticipant @andystopford50521]

Oh right – well when you find out, do tell (if it isn’t proprietary knowledge)!

[David SeniorParticipant @davidsenior29320]

The Claude/FreeCAD combination did appear to make a reasonable model of the centrifugal pump – though it was hard work explaining to Claude what it should look like!

 

Dave

[Author]

Posts