[TeeTomTerrific]

I have finally decided to take the CNC plunge after a couple years of lurking, 3D printing and because I really want one for my woodworking hobby. I’ve decided to go the DIY route and more adventurously design my own machine. Below is an almost finished mechanical plan for the design I have come up with. The 3D rendering is just missing drag chains, proximity limit switches and a few angle braces here and there that would hardly show in this view. I’m not an engineer but I sometimes play one at work since the word engineer is in my official title and I have stayed at a Holiday Inn Express so I’m just barely qualified to design my own machine.


Even though I’ve learned a lot over the last couple of years, there are still some things I’m still fuzzy on, and since I don’t actually have the math and physics background that comes with a mechanical engineering degree, or much real world CNC experience, I have a few questions for the experts here to help me make some informed decisions.

First of all here are some of the specs of this design:
• Ballscrews are 20mm diameter
• Linear Rails are 20mm wide
• Motors are NEMA 34 and what is shown are Chinese 8.0nm Closed Loop
• Table area is 750mm x 1300mm
• X axis will travel a full 750mm so the spindle center point can go edge to edge
• Y axis will travel 1310mm so the spindle center point can go from font edge to 10mm beyond the back edge
• Z axis travel distance is 140mm from 20mm off the flat extrusions to leave room for spoil board strips. (That should be enough for flat work and low relief carving. Much more and the moment arm force could be an issue)
• Spindle is Chinese 3.5KW Air Cooled with ceramic ball bearings

My first question is about what ballscrew lead to use. 5mm vs. 10mm, (SFU2005 or SFU2010) and why choose one over the other?

My second question is around correctly sizing the Motors for the designed dimensions listed above. I am showing this design with 8.0nm motors, but is that totally over-powered or over-torqued for this design? The cost difference between a 4.0nm and an 8.0nm Closed Loop Motor isn’t that much in the grand scheme of the overall cost of this project so if being overly large is not a problem then I’d probably go that way but I’m a bit out of my element on this one. I need some solid technical advice on this decision. (My tendency for Tim Allen grunting about “more power” is not a sufficient justification... AAARRGHHROOO!)

Aside from that I’ll listen to any other advice any of you experts want to give. (I don’t guarantee I’ll follow it, but I will consider what’s offered objectively.) Thanks in advance for your feedback and I hope this isn’t TLDR.

Tom

[routalot]

I suggest you calculate the force applied by those steppers on your leadscrews.Then visualise the likely effect on the position of the gantry beam.After which you may decide to add some triangulation to hold things in place.The gantry beam also seems very high for a machine with a 140mm Z travel and keeping it low will do a lot for the rigidity of the machine.It may also be useful to model on trays for drag chains as they are easier to integrate at an early stage.Good luck with the build.

[awerby]

I'd vote for the fast (10mm) screws, since those NEMA 34 motors will likely have fairly high inductance, and while they've got lots of torque, won't want to go too fast.

[dazp1976]

From experience with a medium mill
.
Had a 1600oz (12nm ish) on Z axis in beginning.
Literally BINNED IT! Would not run more than 1000mm/min, noisy, over travelled with it's own inertia.
Swapped it for a 1080oz. (7.7nm ish) A little better but maxed at 2200mm/min. Still not perfect.
Upped voltage from 60v to 80v (110vdc driver) and hits 3000mm/min with ease.

X,Y axis both had 1080oz on those, similar issues with velocity limit and inertia over travel. Swapped these out for nema24 566oz (4nm) keeping them on 60v. Tested to 3000mm/min and do it easy.

Bigger is def NOT always better.
In my view. A mill has a little friction due to dovetails, a router has rails and is looser. You don't need 8nm. I would use 4nm BUT, I would use higher voltage drivers if they're nema 24 size, not the 50v 'kitted' ones.
These will allow the table size to grow WHEN you want it too and still have plenty of velocity.

[ger21]

My first question is about what ballscrew lead to use. 5mm vs. 10mm, (SFU2005 or SFU2010) and why choose one over the other?

What you really want is 20mm pitch, if available.

My second question is around correctly sizing the Motors for the designed dimensions listed above.

Personally, I'd go open loop 4.5Nm steppers. Low inductance, 6-7 amp, running at 72V. Should be a bit cheaper.


As someone else mentioned, lower the gantry beam. If your Z travel is 140mm, then there's no reason the gantry clearance needs to be much more than that. Lowering it will make it much better.

[dazp1976]

What you really want is 20mm pitch, if available.




Personally, I'd go open loop 4.5Nm steppers. Low inductance, 6-7 amp, running at 72V. Should be a bit cheaper.


As someone else mentioned, lower the gantry beam. If your Z travel is 140mm, then there's no reason the gantry clearance needs to be much more than that. Lowering it will make it much better.

For nema24 3.5nm
These
https://www.omc-stepperonline.com/digital-stepper-driver-2-4-7-2a-18-80vac-or-36-110vdc-for-nema-34-motor-dm860t
These
https://www.omc-stepperonline.com/nema-24-stepper-motor/p-series-nema-24-bipolar-1-8deg-3-5-nm-495-74oz-in-5-0a-60x60x97-8mm-4-wires-24hp39-5004s

This pairing on 70+vdc flies!

For nema34 4.5nm
These
https://www.omc-stepperonline.com/nema-34-stepper-motor/p-series-nema-34-bipolar-1-8deg-4-5-nm-637-38oz-in-6-0a-86x86x74mm-4-wires-34hp29-6004s

Same DM860T drivers as above but run them on 80-90vdc instead.
I've done it and my 34's now kick ass.

[TeeTomTerrific]

Thanks for the feedback folks. This gives me some food for thought. 2010 sounds good and I'm not so sure about a 2020. In case anyone is interested I have added pictures hereof the specs on the motors I have been considering. iI this data changes opinions already expressed here please let me know.

[TeeTomTerrific]

As someone else mentioned, lower the gantry beam. If your Z travel is 140mm, then there's no reason the gantry clearance needs to be much more than that. Lowering it will make it much better.

Gerry,

The bottom edge of my X gantry beam is 215mm from the Table top not any added spoil board width. The X gantry beam is 160 mm wide so the top of the X gantry beam is 365mm from the table surface. The Z axis Gantry beam bottom edge is at the bottom limit of the X mount plate surface so If I lower the X beam it will lower my Z Axis travel distance. Are you suggesting I should have less than 140mm of Z axis travel? In order to have more Z axis travel with my design I would have to increase the length of the z axis the same amount as I lower the X beam and that seems counter intuitive to me because it will re duce the total working height. With the current design I have a full 140mm of working height. The spindle body mounting rails are 190 mm in length and I could conceivably raise it by 90mm while lowering the X Beam but that will leave me with only 50mm of working height or less than 2 inches and I need more than that for the things I want to do.

If I'm interpreting what you mean incorrectly then please give me some specific measurements for what you are suggesting because I played with the gantry height before I settled with what I have and that is over 100mm less now than where I started.

[peteeng]

Hi TTT - Here is a spreadsheet for calculating the torque required for your machine. Big N34s are not the go. N23 or N24 with high voltage will do better. I would also put a web or 2 on the outside of the columns...You won't need twin rails on the sides. I have modelled these many times and as the square cars support a high torque and are quite stiff, the second rail does virtually nothing just adds lots of $$$....I'd also use a motor with a brake on the Z axis so when the power goes down the Z axis does not fall...especially if you go 20mm pitch. Peter

[routalot]

One further suggestion,you ought to go through the mental exercise of actually picking up each part and assembling the machine.How will you thread the cables,can you reach any nuts inside box sections,how will the spindle get attached to the Z axis and at what stage?Its one thing to create a model of the machine and all too easy to overlook the practical difficulty of making it happen.

[dazp1976]

Hi TTT - Here is a spreadsheet for calculating the torque required for your machine. Big N34s are not the go. N23 or N24 with high voltage will do better.

Tried to explain that earlier. Took me a lot of money wasted to realise this when I started out.
Seems adamant on 34's for some daft reason. Machine isn't exactly big and calling for it tbf.
Closed loop steppers aren't what people believe them to be and are a bit pointless.
For 'proper' closed loop use it's either servos or nothing.

[dazp1976]

Thanks for the feedback folks. This gives me some food for thought. 2010 sounds good and I'm not so sure about a 2020. In case anyone is interested I have added pictures hereof the specs on the motors I have been considering. iI this data changes opinions already expressed here please let me know.
If

Write the 12nm off IMMEDIATELY!!!!!.
That chart is a bit BS. That's with them on a bench test.
You cannot get much rpm and the resonance is horrendous on lightweight aluminium machines.
I was that p****d when I got duped into trying one as a novice, I smashed it to bits then chucked it.
If serious about closed loop then look at AC servos.
Closed loop steppers are a bit pointless.

[TeeTomTerrific]

Hi TTT - Here is a spreadsheet for calculating the torque required for your machine. Big N34s are not the go. N23 or N24 with high voltage will do better. I would also put a web or 2 on the outside of the columns...You won't need twin rails on the sides. I have modelled these many times and as the square cars support a high torque and are quite stiff, the second rail does virtually nothing just adds lots of $$$....I'd also use a motor with a brake on the Z axis so when the power goes down the Z axis does not fall...especially if you go 20mm pitch. Peter

Thanks for the spreadsheet peeteng! I'll plug things in and see what happens and compare to other motors. Thanks for the tip on the Z axis brake. I'm not exactly sure what you mean by put a web or 2 on the outside of the columns and would appreciate it if you would elaborate. I hear what you are saying on the extra Y axis rail and will try an alternate drawing with it. My original reason for using the additional rail was to reduce the side force on the Y beam aluminum extrusions and with a wider extrusion stiffen the table base. I'm not opposed to reducing the overall cost, but alternately that cost for the extra side rail isn't huge. I guess what I really want to know is if the extra side rail is detrimental if I were to leave it in? I guess I'm not done modeling just yet...

[TeeTomTerrific]

One further suggestion,you ought to go through the mental exercise of actually picking up each part and assembling the machine.How will you thread the cables,can you reach any nuts inside box sections,how will the spindle get attached to the Z axis and at what stage?Its one thing to create a model of the machine and all too easy to overlook the practical difficulty of making it happen.

This is actually something I have been doing throughout the design and modeling phase and I totally agree with the concept. However I will go back and do it again as a mental exercise. I still need to add drag chains and cable routing and so your suggestion is timely.

[TeeTomTerrific]

Tried to explain that earlier. Took me a lot of money wasted to realise this when I started out.
Seems adamant on 34's for some daft reason. Machine isn't exactly big and calling for it tbf.
Closed loop steppers aren't what people believe them to be and are a bit pointless.
For 'proper' closed loop use it's either servos or nothing.

dazp1976 I hear you loud and clear. I'll take a closer look at the motor recommendations you made in your earlier post. Thanks!

[ger21]

Are you suggesting I should have less than 140mm of Z axis travel? In order to have more Z axis travel with my design I would have to increase the length of the z axis the same amount as I lower the X beam and that seems counter intuitive to me because it will re duce the total working height.

No. If you were to lower the gantry beam, you'd move the Z axis bearing blocks on the back of the spindle plate up by the same amount. So your travel stays the same, but the Z axis position relative to the beam moves up. This reduces the amount the Z axis hangs down below the beam, making the Z axis more rigid.

Any gantry height above what you can use just reduced rigidity.

[ger21]

Seems adamant on 34's for some daft reason.

While I'm with you, my preference is for small Nema 34's vs large Nema 23/24's. They can handle higher voltages/higher heat better, and the larger shafts are better with bigger ballscrews. And they can still spin very fast at high voltages. I'm talking about something in the 460oz-in range.

Here's the thing with those closed loop steppers. So you have an 8Nm stepper. At 1000 rpm, it has maybe 1.5Nm of torque. Acceleration is linear, so it takes the same torque to accelerate to 100 rpm as it does to accelerate to 1000 rpm, at a given acceleration rate.
So if you want to run your motors at 1000 rpm, you can only accelerate at a rate that the 1.5Nm will allow. And if you lose steps at 1000 rpm, you don't have any additional torque to make them up, until the machine slows down to a speed where it has more power. Bat this point, your parts will likely be ruined already.
It's nothing like a servo, that has full torque throughout the entire RPM range, and up to 3x more peak torque in reserve.

[TeeTomTerrific]

While I'm with you, my preference is for small Nema 34's vs large Nema 23/24's. They can handle higher voltages/higher heat better, and the larger shafts are better with bigger ballscrews. And they can still spin very fast at high voltages. I'm talking about something in the 460oz-in range.

Here's the thing with those closed loop steppers. So you have an 8Nm stepper. At 1000 rpm, it has maybe 1.5Nm of torque. Acceleration is linear, so it takes the same torque to accelerate to 100 rpm as it does to accelerate to 1000 rpm, at a given acceleration rate.
So if you want to run your motors at 1000 rpm, you can only accelerate at a rate that the 1.5Nm will allow. And if you lose steps at 1000 rpm, you don't have any additional torque to make them up, until the machine slows down to a speed where it has more power. Bat this point, your parts will likely be ruined already.
It's nothing like a servo, that has full torque throughout the entire RPM range, and up to 3x more peak torque in reserve.

So if I read this correctly you are with me on the NEMA 34 for the 20mm ball screw at ~4Nm, but opposed to the closed loop route? Are you advocating for a NEMA 34 servo instead of a stepper motor or are you fine with the stepper just not a closed loop variety?

Killing two posts with one reply here. I think I now understand how to lower the X beam per your explanation and thinking about it some more and I will be remodeling a lot of parts over the weekend removing the lower rail on the Y axis and lowering the X beam 50mm to 60mm. I'm probably going to lengthen the Y axis 50mm and move the spindle forward by 50mm to reduce the cant on the side supports when lowering the X beam.

[ger21]

So if I read this correctly you are with me on the NEMA 34 for the 20mm ball screw at ~4Nm, but opposed to the closed loop route?

Yes, but they need to be low inductance, 6 or 7 amp motors, and be paired with good drives, and use 60V-72V.

[dazp1976]

So if you want to run your motors at 1000 rpm, you can only accelerate at a rate that the 1.5Nm will allow. And if you lose steps at 1000 rpm, you don't have any additional torque to make them up, until the machine slows down to a speed where it has more power. Bat this point, your parts will likely be ruined already.

They don't 'catch up' on the lost position, apparently they just tell you they faulted.