[pippin88]

I used to live 400m from the beach - steel rusted ("surface rust") if left uncovered overnight in my shed. When the seas are up / big waves, there is visible salt spray in the air.

Now I live about 4km from the beach. Now it takes months for the same sort of rust.

[GeckoSub]

In packing mode, heading out on assignment tomorrow, so not my wordy self right now:

@craig, sorry! Kiwi. Yeah, less scorching but salty nonetheless
@pete, I did think of running a dehumidifier in the workshop when I am not around. Way cheaper and better for the environment than aircon, I'd think. And:
@pippin, good point on distance to the ocean, I already turned down one free workshop space. It was 20m from the water, in a wooden house. Didn't wanna take that chance. The island I am on is about 7-8km wide, so the most I could ever get away would be 4km, as you. But will try my best to find a space towards the middle, then.

[GeckoSub]

How To Home A Rotary/4th Axis?

Hi,
(...)
I do rather suspect that I will use index homing for both the trunnion and fifth axis previously pictured. I'll want very accurate and repeatable way to zero the fourth and fifth axes.
I'm hoping that if I can zero them well and repeatably that I can still use the mill as three axis, with the vise still sitting atop the fifth axis. The majority of the parts I make a small and
mostly brass, aluminum and plastics. If the parts are small enough I would not have to remove the trunnion off the bed of the mill.....and that would be a major time saver.

Craig

This stirred my mind, thanks And a rotary does sound like a pretty near perfect candidate for zero pulse homing.

I had originally planned on having the rotary come way down the road, perhaps 6-9 months after the machine was up and running. But that might be changing. First of all, for the sake of getting the base right you guys urged me to really think the table layout and Z-height of the base through (thanks!). Also, I have found myself thinking about specialty tools, workholding, 2nd, 3rd ops, etc, at all sort of weird times a day already. I guess the bane of a (hobby) machinist in spe. Haha.
All this has firmed up the idea that the rotary will be incredibly helpful for the particular parts/product I want to make. So, now I am thinking the rotary might come as soon as I have an OK grip on the machine.
But how to home it?

If I stick to UCCNC, which I think I will, I see 2-3 options.
1). Indicate the part/rotary vise each time which I'd rather not do.

2). Use a switch, but I wonder how that works because I guess it would trip once per revolution, at least. Is it easy enough to tell enable it for homing/indexing and then disable it again right after? Or is that the sort of things macros are for? I have a feeling, just setting up UCCNC and wiring the machine will take plenty out of me and if this needs a macro, not sure how good I would be at that.

3). I think it may be possible to have the Delta A series do a zero pulse homing routine in the drive. I may be wrong, but I think that's one of the differences between the B and A-series. So, let's say the A series can do that, how would you trigger that from the software side. Is that another "dreaded" macro...? (I can get Delta A2 used too. A bit less than double what the B2 are, but still not too bad).

I may have left something obvious out

Also, I mentioned early in this thread that I had a 2 week, self-imposed deadline to send the base drawings off to the factory. The beauty of self imposed deadlines is that the guy who set it, wont hold it against me when they are moved....
And there's a cashflow issue, too. Haha. But as mentioned things are looking brighter on that front with the assignment I am about to fly out on. So, now the deadline for getting the base started and parts shopping spree is when the payment has come in + some days to finalize the base design if not done already by that time. In other words, no immidiate hurry. Maybe something like 3-4 weeks.

[peteeng]

Hi All - This brings us to the 5 axis programming and how to zero the machine to the CAM. Are people using kinematic zero for this or some offset method? I think fusion uses kinematic zero ? Peter

[joeavaerage]

Hi,
homing is a realtime operation, it has to be. The machine needs to respond to a home switch event within micro seconds. But UCCNC is running
on a Windows platform which is NOT realtime, ergo a home switch event being reported to UCCNC could take many milliseconds, then the PC/UCCNC has to make a decision and that
gets propagated to the motion buffer which could easily be several hundreds of milliseconds long. The around the loop delay could be several hundred milliseconds, way WAY WAY too slow
for Homing. No, UCCNC is not realtime and so cannot do homing. It is the hardware motion controller that does homing. Sure it does it with data from UCCNC but the motion controller
has to home autonomously. Thus there is no secret macro that will make this happen as the macro is within UCCNC and cannot enable realtime behaviour.

If you want index homing then you have to use a motion controller that supports it. By-in-large motion controller behaviour is not user programmable....so if it does not have a feature you can't put it there.
You could alternately use LinuxCNC which is realtime, and thus you could program index homing if its not already there

You maybe correct about the A2 series servos being able to home themselves.

The B2 series has digital inputs which can be assigned certain properties. There is one called CCLR (page 7-96 of the ASDA B2 manual) that clears the pulse counter. Presumably you could trigger this from the Z signal
of the encoder. I suspect with some care you and clever programming of UCCNC you could make a B2 series servo home itself to an Index pulse.

To be honest I think you are making a real challenge for yourself. If you require Index Homing get a CNC software/motion control system that provides that feature. Mach4 has it for example.

Craig

[joeavaerage]

Hi,

Are people using kinematic zero for this or some offset method? I think fusion uses kinematic zero ? Peter

Fusion can do either.

My intention is to compose the Gcode (FusionCAM) where the work coordinate centre is the rotational center of the machine. But you could just as easily compose the code around a location
on the part. The post drafts Gcode that allows the machine controller to apply the required kinematics which in effect translates the work coordinates to the rotational center.

Most of the high end controllers use this kind of kinematic approach, but many hobby level controllers do not have it. If you have the budget you can have anything you want....

Craig

[GeckoSub]

Hi,
homing is a realtime operation, it has to be. The machine needs to respond to a home switch event within micro seconds. But UCCNC is running
on a Windows platform which is NOT realtime, ergo a home switch event being reported to UCCNC could take many milliseconds, then the PC/UCCNC has to make a decision and that
gets propagated to the motion buffer which could easily be several hundreds of milliseconds long. The around the loop delay could be several hundred milliseconds, way WAY WAY too slow
for Homing. No, UCCNC is not realtime and so cannot do homing. It is the hardware motion controller that does homing. Sure it does it with data from UCCNC but the motion controller
has to home autonomously. Thus there is no secret macro that will make this happen as the macro is within UCCNC and cannot enable realtime behaviour.

If you want index homing then you have to use a motion controller that supports it. By-in-large motion controller behaviour is not user programmable....so if it does not have a feature you can't put it there.
You could alternately use LinuxCNC which is realtime, and thus you could program index homing if its not already there

You maybe correct about the A2 series servos being able to home themselves.

The B2 series has digital inputs which can be assigned certain properties. There is one called CCLR (page 7-96 of the ASDA B2 manual) that clears the pulse counter. Presumably you could trigger this from the Z signal
of the encoder. I suspect with some care you and clever programming of UCCNC you could make a B2 series servo home itself to an Index pulse.

To be honest I think you are making a real challenge for yourself. If you require Index Homing get a CNC software/motion control system that provides that feature. Mach4 has it for example.

Craig

Sorry, def still making rookie mistakes. Yes, the controller, not the software, runs the homing!
I do like the ProbeBasic screenset for LinuxCNC, which basically looks "heavily inspired" by Pathpilot but maybe there's a bit of poetic justice in there as Tormach runs on Linux. Anyhow, not sure of the ins and outs, history and morality/legality of all this. I just liked that it looked "modern". But then I started reading and watching tutorials and LinuxCNC is a bit more than I am willing to dig into right now.
The controller for UCCNC does Mach(4?), too so it's not completely out of the question. I think Engineers Anonymous made a screenset that looks nice (there's a video on YT with a lot of crap talking). And yes, looks shouldn't matter, but if I have the choice, I vastly prefer a pleasant looking UI most other things being equal.

Anyways, back to homing. I don't mind at all a simple switch. Just wondering if it's easy enough to set up? I mean it would trigger once every revolution, right? So, is it easy to enable it for homing and then disable it again for the machining?

[joeavaerage]

Hi,
yes UC motion controllers (made by CNCDrive, the manufacturer of UCCNC) will run Mach4, however if UCCNC (with a UC motion control) cannot do Index Homing then Mach4 (with a UC motion control)
won't be able to do it either. Index Homing is a feature of the motion control, and you can be bound that if a UC motion control could do Index Homing using Mach4 as software
then CNCDrive would make bloody sure that it would work with UCCNC.

UCCNC and Mach4 compete head on, if one has a feature its highly probable that the other will get it eventually. I must say I am surprised that UC controllers cannot do Index Homing, I assumed that
they would, certainly my Ethernet SmoothStepper does, and as a direct competitor thought UC controllers would also.

One thing about Mach4 which I believe has advantage over UCCNC is that you can make your own screenset very easily indeed. All the screen editing functions are built in. NFS, the manufacturer of Mach4 did this very
deliberately so OEM manufacturers could customize Mach4 to their liking and make it a distinct offering to customers. For this reason the screen editing tools are pretty strong. LinuxCNC is highly customisable but is
a much harder proposition than Mach4 unless you are a C/C++ expert.

A simple Home switch would work just fine, but you might find a variance a a degree or so. That's where you Metrol switches may come into their own. I have yet to make a decision, I'll certainly start with either
a switch or an opto-sensor and see if that is good enough. If it is I'll leave it alone, if however I need closer angular repeatability then Index Homing is on the cards.

So, is it easy to enable it for homing and then disable it again for the machining?

Can really only speak about Mach4/ESS and who cares?. The Home switch is monitored and acted on ONLY while the machine is being homed. Once it is the switch is ignored. Limit switches are different,
but Home switches are only significant when the machine is homing. With Mach4 you could certainly write a macro to disable the Home input, but I rather think its redundant.

Path Pilot is a variant of LinuxCNC. The current Hass control solution is a variant of LinuxCNC. LinuxCNC is open source, so you could take it a make your own special version and sell it.
Officially any LinuxCNC code that is included in your own version is supposed to remain open source....although I don't know that anyone gets too worked up about it.

Craig

[GeckoSub]

yes UC motion controllers (made by CNCDrive, the manufacturer of UCCNC) will run Mach4, however if UCCNC (with a UC motion control) cannot do Index Homing then Mach4 (with a UC motion control)
won't be able to do it either.

Ah, yes, that does make total sense. So, back to likely sticking to UCCNC.

One thing about Mach4 which I believe has advantage over UCCNC is that you can make your own screenset very easily indeed. All the screen editing functions are built in. NFS, the manufacturer of Mach4 did this very
deliberately so OEM manufacturers could customize Mach4 to their liking and make it a distinct offering to customers. For this reason the screen editing tools are pretty strong.

I believe you are right, I think I have seen people allude to this on the UCCNC forum.

LinuxCNC is highly customisable but is
a much harder proposition than Mach4 unless you are a C/C++ expert.

Yeah, I had one stab at Arduino years back and even that was not easy for me. So, probably wise not to go the Linux route for me
Plenty of other stuff to learn.

The Home switch is monitored and acted on ONLY while the machine is being homed. Once it is the switch is ignored. Limit switches are different,
but Home switches are only significant when the machine is homing

Perfect. This was what I was still pondering but I am yet to operate a CNC, so some concepts I get, others I still can’t picture. But I think I get it now. It’s as simple as homing only kicking in when you actually ask it to and then once it’s done that, even it the homing switch gets turned on, the controller just ignores it. So, yes it could be switched on every time the rotary gets around to it, but it wont make the machine set a new coordinate or anything like that, it’ll just be ignored. Cool, then a nice switch should suffice.

A simple Home switch would work just fine, but you might find a variance a a degree or so. That's where you Metrol switches may come into their own. I have yet to make a decision, I'll certainly start with either
a switch or an opto-sensor and see if that is good enough. If it is I'll leave it alone, if however I need closer angular repeatability then Index Homing is on the cards.

Yes, Metrol does have a few options for “sideways” action and not just a straight plunger.

[GeckoSub]

Hi All - This brings us to the 5 axis programming and how to zero the machine to the CAM. Are people using kinematic zero for this or some offset method? I think fusion uses kinematic zero ? Peter

Brings you! Or you and Craig, rather, hahah. My head would exploded if I had to think 5 axis at this point. Wouldn't be surprised if some years down the line, it's a rabbit hole waiting to be crawled into. Now, no, no, no...

[pippin88]

LinuxCNC does not require C/C++ knowledge.

There are setup wizards.
The Hal and ini code files are pretty simple.

[GeckoSub]

LinuxCNC does not require C/C++ knowledge.

There are setup wizards.
The Hal and ini code files are pretty simple.

Yeah, I watched a few tutorials. But I also realized what's simple for some may not be for me. I read enough posts that said it wasn't too bad, that most could do it, but then when I watched the tutorials I still felt it wasn't. I mean, I might be able to just sink my teeth into it and get it done.
But I also have enough other stuff I have to learn so I'm thinking, start with something a bit easier for now. Most of you guys are serial-builders and I am "afraid" that gene is in me as well. I always wanted a tiny, stiff CNC-lathe and I could use a long, narrow foam/wood blank router - so there's ample opportunity to get into LinuxCNC later. Just for now, I am afraid Linux would end up taking too much of my time...
That said, ProbeBasic looks sweet!

[joeavaerage]

Hi,
I use Mach4 ($200USD) and an Ethernet SmoothStepper ($225USD), and have been using for eight, nearly nine years.
My machine has cost about $20,00USD to date, and more if I count the fourth and fifth axis. So, the CNC software solution and the associated motion control is
2.1% of the total. Another way of looking at it is that Mach4/ESS cost $425 and one servo is $438....and I have five servos!!!

Use whatever solution you are happy with and be done with it. Everyone has an opinion that this is better than that and so on.......but guess what most of them
don't even follow their own advice and actually use something else altogether. Cost differences are immaterial in the larger scheme of things.

I've been using a $425 software/motion control solution for 81/2 years or $50/year. I use Destiny Tools Raptor endmills on tough steels at $20 each, and buy ten or a dozen every year.
So I pay more in just this one type of tool than I do for the CNC software and motion controller! When I think about it, I pay $780NZD + $2200NZD per year for Fusion 360 and Fusion Machining
Extensions....so the CAD/CAM software is way more than the CNC software. So, choose whatever software you like, and then live within it's limitations.

Craig

[GeckoSub]

Hi,
I use Mach4 ($200USD) and an Ethernet SmoothStepper ($225USD), and have been using for eight, nearly nine years.
My machine has cost about $20,00USD to date, and more if I count the fourth and fifth axis. So, the CNC software solution and the associated motion control is
2.1% of the total. Another way of looking at it is that Mach4/ESS cost $425 and one servo is $438....and I have five servos!!!

Use whatever solution you are happy with and be done with it. Everyone has an opinion that this is better than that and so on.......but guess what most of them
don't even follow their own advice and actually use something else altogether. Cost differences are immaterial in the larger scheme of things.

I've been using a $425 software/motion control solution for 81/2 years or $50/year. I use Destiny Tools Raptor endmills on tough steels at $20 each, and buy ten or a dozen every year.
So I pay more in just this one type of tool than I do for the CNC software and motion controller! When I think about it, I pay $780NZD + $2200NZD per year for Fusion 360 and Fusion Machining
Extensions....so the CAD/CAM software is way more than the CNC software. So, choose whatever software you like, and then live within it's limitations.

Craig

That's good advice. Pretty much on many things in life
Even if the initial cost of a machine can sometimes be kept low'ish over time, upgrades, add-ons and like you say, just the tools and CAD end up far outweighing what the controller hardware cost. (For "hobby" machines, of course).

It's funny, cuz my real job is photography and in that, I rarely ever looked at prices. I knew what I needed, I had long since picked my "eco system". A new powerful laptop every three years or so, new camera body every 2-3 years, lenses last longer and not much point or need in mulling stuff over too much. Also, of course, made easier by the fact that they were my necessary tools which would pay for themselves. I use past tense as only right now am I beginning to get assignments again but still not to the point where I can run out and buy stuff and not think about the price, haha.

[GeckoSub]

LinuxCNC does not require C/C++ knowledge.

There are setup wizards.
The Hal and ini code files are pretty simple.

I know I just said, I wouldn't do Linux but... I really, really do like ProbeBasic. And at the risk of sounding shallow, I do enjoy good UIs. On some small scale, it just makes me want to use a gadget/system more if it's pleasant to the eye and mind.
And since this build is going to take a while, even after all the parts and base arrive, perhaps if I need to stick my head into Linux for a week or two, that's alright, too.
I can postpone that decision a bit though as once the base/frame design has been finalized and drawings sent off - which has now been delayed by at least 2-3 weeks - it'll still take about 6-8 weeks to arrive from China. Same for the other parts that'll need to order.

[pippin88]

Not hard to download the LinuxCNC live installer and have a play.
There are simulations built in, so you don't need CNC hardware to have a look.

A standard machine is easy to set up with the wizards.

I would never go for software step generation. A hardware step generator is much better.
Mesa and pokeys can be used with LinuxCNC
Or you can go with ethercat but that is harder to get going, and less common.

[GeckoSub]

Not hard to download the LinuxCNC live installer and have a play.
There are simulations built in, so you don't need CNC hardware to have a look.

A standard machine is easy to set up with the wizards.

I would never go for software step generation. A hardware step generator is much better.
Mesa and pokeys can be used with LinuxCNC
Or you can go with ethercat but that is harder to get going, and less common.

Yeah, I looked into it enough to figure out that sticking to a Mesa card and Ethernet was the fairly trouble free most often taken path. Hmmm, ok, it's back on the shortlist then. I think I may even have signed up on their forum. This whole endeavor has been ongoing for a handful of years, the pandemic just put a halt to all of it.
As Craig said, it's actually not that big a deal. You get used to what you have, you make it work, and figure out its features when you need them.

[GeckoSub]

Servo Drive Control Mode?

Hey again guys,
Sorry, it took a while. Had some paid work in Indonesia and stayed on for a few weeks of holiday and then had to wait to get paid.
But now, I have a bit of an urgent question as there's deal on some used Delta B3 servos and drives I'd like to pounce on.

Basic question but what control mode do we need for basic CNC?
PT (pulse train) Mode?
So, the L version of the B3 drives?

Or do all of them do pulse+direction and perhaps, the control modes are some sort of higher level wizardry which I wont ever need?

[joeavaerage]

Hi,
you require the 'L' version.

Only the L version has Step/Direction type input.

If you had a controller like a Galil or LinuxCNC then you could use Analog Voltage control, then all models would work, or naturally if you have
an Ethercat controller then the Ethercat model is appropriate.

The L model is the cheapest and is what you require.

Craig

[joeavaerage]

Hi,
the 'PR' mode is a set of internal programmable registers (up to 100 with the B3), with each register corresponding to a position.

You would program 20 registers if you had a 20 tool ATC carousel for example. Each register would be programmed for the numeric position for a given tool location.

For instance tool 14 may correspond to a position 567876 in encoder counts. Tool position 15 might be 727876 encoder counts or a difference of 160,000 counts or 360 degrees of rotation
of the servo shaft, which through a 20:1 reduction means 1/20th of a rotation. Convenient.

PPS: whoops just realised that my example is based on the encoder of the B2 series servo that I use, and it has a 160,000 count/rev encoder. Same principle applies to the B3,
just the numbers may differ.

A controller or PLC would command the drive 'go to register 15' say and the driver would energise the servo under PID control to achieve that position.

Very useful no doubt, but its not what we require, we require that the drive be able to take the servo to any arbitrary position, and therefore Step/Direction (or analogous method) is required,
ie the 'L' version only.

PS: That last statement is not technically correct, you CAN achieve arbitrary position performance by using Analog Voltage Control AND a controller like Galil or LincuxCNC or Masso
or similar. All of these solutions use the controller to close the position loop. This method was the only way many years ago (say 80's), and is still viable....but its significantly more complex.
If this build is your first foray into servos I would be extremely reluctant to recommend the extra complication this would involve, not because it cant work or even work extremely well,
but rather the extra complication is likely to daunt you to the extent you may NOT get a usable result. The lower risk option is the 'L' model....take it.

Craig