Clem's CNC Control Project - All Comments

RE: When you have to abandon a project? When is it worth pushing on? The sunken cost fallacy at work.

dougw — Sun, 13 Oct 2024 16:16:00 GMT

I like the mountain hiking analogy and your insights into electronics projects. To add to your points, there is always going to be some opportunity costs to working on a project and there will always be some aspects of it where it is just drudgery work where you aren't learning anything, but is needed to support the aspects that do provide benefits. Even completing some drudgery work can be a satisfying accomplishment.

There are some costs to stopping as well - dismantling and disposing, but the main cost is the stigma of failure. It is smart to rationalize this into being less important or even adjusting the goal to be one of learning rather than a working system. Indeed the completely working project is usually only one of many goals in a project.

I like e14 projects in that we can define or redefine the goals as needed. We can travel part of the way up the mountain on roads built by someone else, and we can forage over to another road in an effort to make progress, or we can park the project until some new inspiration brings it back to life.

I get enormous satisfaction from making a project work, and the harder it was to get working, the more satisfying it is when it works. If you never tackle things that are so hard that there is a high risk of failure, you will never achieve maximum satisfaction from completions. In this regard I view "failures" as a necessary aspect of obtaining maximum fun from projects.

I often set multiple goals for a project, where some have high probability of succeeding and some are risky stretch goals. The easy goals are enough to motivate the project but the stretch goals can turn into a a monumental success. Most projects end up having some setbacks, and often overcoming these provide the biggest satisfaction in the project.

RE: When you have to abandon a project? When is it worth pushing on? The sunken cost fallacy at work.

BigG — Sun, 25 Jun 2023 15:55:56 GMT

I prefer to park or shelve a project for awhile if it starts to become over complex or too tricky to resolve. It's amazing how different a parked projects looks when you come back to them. The success rate is so much higher on a 2nd attempt.

RE: When you have to abandon a project? When is it worth pushing on? The sunken cost fallacy at work.

shabaz — Sun, 25 Jun 2023 15:36:34 GMT

You got further than me! And learned a lot along the way.

I too had an idea to self-construct and purchased a small testbed with a couple of motors one Xmas, but got quickly disillusioned when I saw the poor state of open source elements that would be needed. I thought Machinekit would be a good software base, but looking deeper, the available code made no good use of real-time features in Linux and the underlying hardware (BeagleBone Black), although they offer paid services, perhaps that has better code, but I didn't want paid, closed source for that part of my project (I had reasons).

If the desire is easier prototyping, then there are so many low-cost services, personally I would rather it be their problem

Even making things in larger volumes, it's likely more productive to just pay for a CNC operator to set it all up and churn out dozens or hundreds of an item.

Your needs may be different, but it sounds like you absolutely made the right decision since you mentioned that you forgot the project purpose, i.e. it was not critical to your needs either.

RE: When you have to abandon a project? When is it worth pushing on? The sunken cost fallacy at work.

robogary — Sun, 25 Jun 2023 14:06:21 GMT

Prototypes, mockups, alpha versions builds are never cost effective, altho they can be cost minimized. The adventure of the journey and facing the challenges is the life blood of build.

RE: Cursed connectors, why is there no middle ground option?

shabaz — Wed, 24 May 2023 17:52:38 GMT

I found a decent manufacturer M12 and a cheap M12, and took some photos.

First off, here are photos of the cheap version. This is the mating end view:

The other end of the connectors. This particular model looks like it will accept cable up to 5mm through the gland, but maybe up to 6 mm with the gland more splayed out.

These are the screw terminals inside the connector body:

Here's a decent manufacturer one, from Farnell I think, but I can't find the product code. This wasn't OTT-expensive, it was probably about 5-8 GBP. I don't know which manufacturer this is. I've done a search in my order history e-mails but I can't find the code since the text M12 is so generic.

Wire end of the same decent connectors:

With the screw removed, the socket terminals can be pulled out, and the left one is from the cheap version. It is more poorly machined from the outside. However I'm actually surprised that it's pretty good overall. I was expecting it to be worse, perhaps without a split, but it seems to be manufactured with the correct cuts, just without whatever polishing operation the decent manufacturer one has. It's brass I think, and no plating. I don't know what that means for corrosion.

The fit is very snug. I think it's a good electrical connection, at least with a newish connector such as this one (this connector is at least a couple of years old).

In summary, the cheap connector is not bad, it's better than I originally thought. However obviously for critical stuff, a decent manufacturer version should be used.

RE: Cursed connectors, why is there no middle ground option?

dougw — Wed, 24 May 2023 15:59:33 GMT

I like AMP plastic CPC (circular power connectors) for cost.

[embed:ad4490ac-0ce8-4f2d-a04c-9285646d5689:2716de2e-7734-45fe-8b42-7d87e8a61f0c:text=&style=link&farnell=592882&newark=&cpc=&avnetemea=&avnetema=&avnetasia=&comoverride=&cmpoverride=&AreaInteracted=rte-content&bom=]

RE: Cursed connectors, why is there no middle ground option?

shabaz — Wed, 24 May 2023 14:35:31 GMT

As another option, avoiding drilling holes, could be to replace one side of the enclosure (or a portion of the enclosure) with a PCB.

This style of D-sub looks very good, it is PCB mount, and would allow you to eliminate all wiring inside the enclosure, since you could just route PCB traces to whatever internal connector you wish to use. D-sub is robust enough for industrial equipment, provided it's not in an area that will vibrate much, and with paralleled pins I believe it can handle the current you likely expect (I'm guessing you're controlling stepper motors?). Also, this will still fully seal the inside of the enclosure from dust, since only solderable holes are drilled.

If you go down a PCB panel route, then there are other options too, for instance other shaped panel mount connectors, or using aluminium PCB services for complex cut-outs.

However for low-cost circular connectors, even the cheap aliexpress M12 will be better than the other 'aviation' connectors. I might have one or two of the Aliexpress ones, I can take a close-up photo of them if you require.

RE: Cursed connectors, why is there no middle ground option?

dougw — Wed, 24 May 2023 14:11:02 GMT

For high current, I like pluggable terminal blocks, they are so much easier to wire up, rewire and maintain. You might also check automotive connectors for low cost, high current.

RE: Cursed connectors, why is there no middle ground option?

shabaz — Wed, 24 May 2023 13:28:09 GMT

Hi Clem,

The connectors in your photo remind me of the ones that used to be on consumer CB radio gear, they are not good quality as you say.

The decent connector is M12 (there's also a smaller version, but M12 is a good size), it's used in industrial equipment, and I've seen it in railway gear, so it's trusted, and it's not expensive considering what it does, but definitely won't be anywhere near the price of the consumer CB style connectors (which of course just have the aviation name but are not really used in aviation).

There are cheap M12 connectors on ebay/aliexpress as shown in the photo below, however they are not 100% perfect, they are a tighter fit if you try to mate the cheap clone with the decent known manufacturer versions. When both the plug and socket are from decent manufacturers, the fit is smooth. They are still miles better than the CB style connectors.

Even cheaper could be D-sub style connectors, they may support the current you require (perhaps with using 2 pins per connection), and they have a small amount of protection and have screw fitting to prevent disconnection. It could be protected from dust by having all the cables enter an area through a separate area, through a bend (for instance) and then plugged and screwed in. I think it's a great option personally, but required drilling of non-circular holes usually!

Cheaper than that too could be Molex Mini-Fit Jr, however that would need a crimp tool, and would need some dust prevention techniques such as the separated wire entry point as mentioned above.

RE: Embedded Systems and the future of CNC-like Machines

shabaz — Wed, 22 Mar 2023 00:09:02 GMT

There's an example of this called Machinekit, there is a version that runs on BeagleBone Black. However the devil is in the detail and in this case all is not rosy, because actually implementing it is full of hurdles for anyone who wants to DIY it. There is a small industry around it, with people selling hardware and selling Machinekit support as I understand.

I gave it a shot and was astonished at the mess that it is. There's no good documentation about how the BeagleBone Black should be interfaced, and some of the builds do not even use real-time capabilities in Linux, for anyone unwilling to pay for support I guess! Also, the real-time engine inside the BBB (known as PRU) is supposed to be used, but (you guessed it), that's not necessarily supported or configured and there's little documentation on it.

It is possible to use a Pi or otherwise, but personally I'd be sceptical unless I was sure the developers had gone to some effort to use real-time capabilities and it was reasonably stable, because the last thing one wants is for the software to hang for a little bit while motors are spinning. The hardware should be very stable too.

RE: Let me untangle my Spaghetti – Clem’s Numerical control

DAB — Thu, 16 Feb 2023 19:57:33 GMT

Hi Clem, have you looked for any cross talk on the wires?

The high current spikes on the stepper motors could introduce problems on signal lines.

RE: Let me untangle my Spaghetti – Clem’s Numerical control

shabaz — Wed, 15 Feb 2023 19:46:14 GMT

Hi Clem,

Regarding: "Matching ferules to the right cable size is not always trivial as they seem to be mostly made for AWG sizes and not metric cable sizes"

It's a topic there is little literature on! : ) I too came across this issue a while back, and wrote this guideline, I don't know if it will help, but it's a starting point, it specifies between what range of AWG values will fit inside ferrules that are then crimped into a square cross-section. So, for instance, if your wire is (say) 1.2mm2, then that is 17 AWG using online calculators, which means the red ferrule is (the table is just a guideline however!) best for that (the color coding is different for France and German standards, but it is easy to see which is being ordered usually). The H value in the table indicates the minimum diameter hole needed for the crimped ferrule to fit in, so you can measure your screw terminals with calipers to see if it will fit or not. (The assumption is that the crimper is making a square shape).

RE: Let me untangle my Spaghetti – Clem’s Numerical control

shabaz — Wed, 15 Feb 2023 17:45:15 GMT

I too thought about this a while back (I have a small CNC table that needs a controller), and have unfortunately made zero progress for years. At the time, I was considering using ready-made cables, to save having to do as much crimping.

The plan was to create PCB adapter boards (shown with red arrow) that would plug or screw onto the drivers (e.g. using terminal blocks, or terminal pins), and adapt out to a connector (e.g. perhaps short RJ45 cables [shown with grey arrow], since they come in a range of convenient lengths). For the power connections, I would do a similar thing (shown with yellow arrow), and either parallel up many pins of a RJ45 connector (I did check the PoE standards to see how much current is possible per pin, but I've forgotten what I concluded at the time), or crimp those, e.g. Molex Mini-Fit Jr or similar, again paralleled connections if higher current needed. Having two connectors for the power would mean I could reserve the possibility of daisy-chaining power like a bus, or just plug into one connector for star-type distribution, i.e. I'd worry about that problem when it came to plugging it all together, and see what worked out best.

RE: Let me untangle my Spaghetti – Clem’s Numerical control

dougw — Wed, 15 Feb 2023 16:03:59 GMT

This what I call a brave project - it takes intestinal fortitude to tackle something of this magnitude. Even the space requirements to work on it would be a logistical nightmare for me. Kudos for persevering...

RE: Let me untangle my Spaghetti – Clem’s Numerical control

dougw — Wed, 15 Feb 2023 15:58:14 GMT

Is it possible to use a larger bootlace terminal that will accommodate all the strands and then flatten it a bit in a vise so it fits in the screw terminal?