feed rate mismatch between planes

[planckscale]

Hi!
I am looking forward to build my first home made 3 axis CNC router although for now still digging into basic concepts.
Maybe it is a silly question, but I can't get my head around this:
Suppose I come to a setup where the XY plane and Z plane drive have different feed rate limitations, for instance a 100IPM bound over XY but only 10IPM on the Z axis. what would happen then if I send a downward diagonal cut ? Will the controller somehow restrict the XY feed rate in order to perform the correct path ? If it is the case, is the XY plane feed rate always bounded by the Z limit or only when Z moves are involved?

Just to give some context to my question, I was thinking about using HDT3M timing belts and pulleys for horizontal movements and a 5mm pitch ball screw assembly over Z (in direct motor coupling mode). Based on my quick calculations this would give me a stepper frequency offset of about 400 RPM between planes. In this kind of setup would it be better to maybe use a lighter stepper on Z, to catch up on the torque curve? I know that one could also stretch the ball screw pitch, but I really want to keep my depth accuracy to a maximum.

Any help with this would be greatly appreciated,
Thanks in advance.
Cheers,

Similar Threads:

• Need Help!- Steppers stall/oscillate at low feed rate. High feed is OK.
• feed rate vs override feed rate
• Australia feed rate and plunge rate help required please
• Okuma mill feed rate jumps to rapid feed
• Need Help!- Feed rate Ovverride also Increases rapid rate.

[ger21]

In your scenario, the X and Y will slow down if required due to the slower Z axis. If the Z is not moving, then the X and Y will run at their commanded speeds.
Belts will stretch and flex, and are not the best choice for motion.

[peteeng]

Hi Planck - As Gerry explained the motion will be limited by whatever you set the max velocities to, by the motion controller. The controller will move the machine at the maximum allowable speed within the config settings. In regard to belts HDT belts are more for power transmission then positioning. Their round teeth are designed to slid slightly in the pulley so as to cope with high power transmission and power surges. Look at AT belts they have a tooth section designed to snug into a pulley and maintain the position. AT is also much stiffer then A section. Nearly twice as stiff for the same size. Depending on your design you can get "extra" accuracy by using 400step per rev motors. They double your mechanical accuracy. I've used them on my latest machine and I'm impressed vs more microsteps or gearing. I'm a fan of belts as well. In Australia we can't seem to get reasonably priced racks...Cheers Peter

[planckscale]

Hey, thanks for the great advices. I will take a look at AT belts asap.
I should have specified, my intention is to use two face to face belts, as in the Everman belt drive system (see attachment). I feel this really is a brilliant setup, since it cancel's out almost all the longitudinal stretch while keeping the benefits of belts (reduced cost, scalable axis, speed, ...). Yet it seems to be an uncommon setup, it makes me wonder if I miss some practical drawback's.

[ger21]

Regardless of the type of belt, when used in long lengths, they flex and stretch.
The "Servobelt" system eliminates those issues, though.

I was going to use it, but eventually determined that it was easier, and cheaper (or similar cost) to just go with helical rack and pinion.

400 step steppers are not readily available, and are likely a lot more expensive.
If using steppers with that system, you really need a belt reduction or gearbox, between 3:1 or 5:1.
Steppers work best when you get 1/2" to 1" of travel per stepper revolution. This gives you smoother motion, and more power.

[peteeng]

Hi All -The issue with belts is length agreed. But Plank hasn't stated the size or use of his machine yet. There is an issue with the servobelt or everman design. The teeth of a belt are designed to change shape as they conform to the pulley. If you get two pieces of belt and engage them "flat" you will be able to slide them back and forth slightly. This negates the benefit in my view. If you can get R&P at the right price then thats a clear winner. But on a small economy machine a belt can be the answer. But you have to size the belt right. Most small machines have undersize belts. So use the biggest belt you can put on the motor you select or if you can do the calcs do them right... Gerry is also correct in that once you use double the belt length in the everman design the cost does add up. In Oz my belt systems run at about $100/m but a R&P system is $400/m so I'm up against a wall there.... Gerry if you have a helical rack supplier would you share this info?

re everman design you usually glue the bottom belt down? this means as the active belt stretches over time it will mismatch the static belt. Even if you don't glue the static belt down I think getting the two belts balanced in tension so their length agrees is a tough call over time. Plus the complication of the tooth mismatch gets worse. Better to use a big stiff belt and tune it occasionally if you use a belt.

400 step motors are around in N17 frame and they are the same cost as 200 steps (see stepperonline). So depends on what motor size Planck is looking for, and yes a reduction gets the motor working in a better operation zone, but then you can use a bigger motor, usually cheaper. Balancing the speed to force to cost trade-off can get tedious....Peter

[planckscale]

My project is not exactly small, I am targeting a 3 by 1.3 meters milling area and 300mm clearance (for professional wood working). Since I unfortunately have a budget limitation using timing belts is appealing. I have the feeling that a 3 meters screwball drive is anyway a bad idea because of rotational torsion and lengthwise bend. I understand that racks and pinion would technically be a better choice, but the price difference I have seen is huge. Apparently there are Chinese retailers selling 30mm width reinforced polyurethane belts for as few as $6/meters. At this price I am ready to change them every 100 work.hours if necessary. It would still be about 3 times cheaper than $400/m racks wearing after 1000 hours. Ger21 I am surprised that you have found similar cost.

For the everman belt drive system you are indeed supposed to glue the passive belt. Peteeng I hear your concerns about teeth mismatch, maybe I should just do a proper benchmarking test and re-evaluate from there.

My motor size will probably be 5 amps NEMA34 at 8.4 Nm (1200 oz.in) holding torque. I will look if I can find a 400 steps version for it, throwing in gear boxes surely adds some mechanical complications but once done I imagine that it is a great plus for machine tuning.

[ger21]

There is an issue with the servobelt or everman design. The teeth of a belt are designed to change shape as they conform to the pulley. If you get two pieces of belt and engage them "flat" you will be able to slide them back and forth slightly.

No, the belt is "preloaded" by the drive pinion, so that all the play is taken out under the bottom rollers.

re everman design you usually glue the bottom belt down? this means as the active belt stretches over time it will mismatch the static belt. Even if you don't glue the static belt down I think getting the two belts balanced in tension so their length agrees is a tough call over time. Plus the complication of the tooth mismatch gets worse. Better to use a big stiff belt and tune it occasionally if you use a be

These are non issues, for the same reason I mentioned above. It's really a pretty nice system.

Gerry if you have a helical rack supplier would you share this info?

$40/meter, plus $33 per pinion.
My machine is 1.7m x 3m, and the rack and pinion will be about $500 delivered. I'm having a friend bore the pinions to 14mm for me.

I'm using AC servos, and picked up Chinese gearboxes for about $150 each.

So, about $1000 for the rack and pinion plus three 10:1 gearboxes. And I don't have to fabricate the belt drive system that the servobelt required, just bolt the gearbox to a plate.

25mm belt is about $3/ft for 25mm T5, which is a bit cheaper than the racks. I bought the gearboxes after deciding to go with the helical rack, but they would also work with the servobelt, greatly simplifying it.

[peteeng]

Hi Planck - By professional woodworking do you mean cabinet work, joinery, timber sign engraving or something else? If its a high volume application you should look at buying a second hand machine or a chinese machine and refit it or use it until it needs rework. Building a machine to a "professional" level is a big task and will chew up lots of time. There are lots of hiccups along the way. If by professional you mean accurate for guitar building or something precise but lower volume that then this is possible. You need to find a commercial machine to benchmark against that does what you want. Then you need to copy this as close as possible and cost down as much as possible without compromising the specs. eg a commercial machine does not use belts for obvious reasons. Home built machines tend to go over budget and over time. Not that this is wrong just your at the beginning of a big learning curve and usually your first CNC is not your last. So in some ways its better to make a half sheet machine even in plywood and get through the learning curve (hardware. software, electronics, tooling and suppliers) as fast as possible then build the bigger machine. Or look at kits...Took me 2 years to build the first one then 3 months to build the second.... Cheers Peter

If you want to hold better then 0.1mm a belt can do this, if you want to hold better than 0.01mm then belts are out. Once you are clear on your precision or accuracy then you can move onto the subject of cutting speed and rapids. Which gets back to your original question. Seems you want to do fast Z axis work so you need to figure out what cutting/plunging speeds you need to achieve your economic or cycle time goals on a typical object. Once you have this the answers nearly become self evident... then it comes down to $$$ and time

Re belts - if you use them mount them vertical, they do collect sawdust and if they are horizontal they will misbehave as they fill up especially if they are below or near the bed level... Peter

[ger21]

Hey, thanks for the great advices. I will take a look at AT belts asap.
I should have specified, my intention is to use two face to face belts, as in the Everman belt drive system

Mike Everman has said that you need to use T belts, and not AT. It's in the long Servobelt thread here. I believe that the clearance in the T series belt is advantageous in this application.

If you want to hold better then 0.1mm a belt can do this, if you want to hold better than 0.01mm then belts are out.

The actual ServoBelt system from Bell-Everman claims ± .01mm repeatability. (under ideal conditions).

Re belts - if you use them mount them vertical, they do collect sawdust and if they are horizontal they will misbehave as they fill up

I don't think the ServoBelt will work vertically, as the top belt is completely unsupported.
Dust and chips are a serious problem, though, and is a big reason I 'm moving toward rack and pinion.

[peteeng]

Hi All - That's part of the conundrum with the Everman system for me. If you use a less stiff belt with clearance (and you have to use 2 of them) and you have to use twice the belt length and depend on the idlers compressing the belts together to remove the clearance then glue the static side down (what if you glue it down too short or too long?) it all seems to be complex. If you use a big stiff belt then it's easy. If the belt is sized right and the electronics are correct then under "ideal" conditions a single belt run should hold 0.01mm. The pretension on the belt is very important for this and I'm working on a simple auto preload mechanism to achieve this. But it's not easy. As Gerry has found its not a great $$$ step forward if you can find an R&P at the right price.

To expand on the belt thing a bit from my experience. If the application is a single belt run then 0.1mm is easy and 0.01mm is doable. EG across the gantry. If you use a belt for the long axis so you have two belts in parallel then you have two elastic systems to keep in tune and this is hard. Combining the stepper motor cogging and the springiness of the belts means the gantry will walk/stagger a bit as it goes down the rails. Plus the system will have hysteresis. eg if you go to a point from one direction it will stop somewhere, but if you come at it from the other direction it will stop slightly different somewhere. And I feel the Everman will have the same issue over time. So for the long axis if you are trying to hold 0.01mm over time with a long run then R&P is the solution.

If "wood working" is happy with >0.1mm then I'm sure using a good size belt (25mm plus for the 3m run) will do the job, be cheaper then R&P and you just have to keep it calibrated every month until it settles down then it seems to be Ok for every 6 month check or until you notice somethings not right. But having said all of that if you spend the $$$ in the right place R&P is clearly the target for anything production or commercial thats needs to be right, minimum fuss over time.. People seem to spend a lot of time and $$$ on super benches and then skimp on the motion parts. Motion parts are $$$ because they are important!!

Better to proportion your spending on the bits that really matter. I highly recommend writing down your machine objective, its required specs and design the machine in full including costs before you start. Once you start building you create a train of events that sometimes move away from those costs and objectives if you start too early. But then this maybe part of the fun or frustration... . Peter

[ger21]

For a regular belt drive, I think you need much wider than 25mm, or there'll be too much flex, and potentially poor cut quality.

And i think the Servobelt is better than you think, but can be difficult to DIY properly.

[peteeng]

Hi - I think the better belt idea is where the active belt is just a small continuous loop like a tractor tread that bears onto the static belt. The metres of extra active belt are not doing much. All systems are limited... Peter

[planckscale]

Hi,
Sorry if I am not clear enough, it puts me a lot of effort to write in English, especially when it comes to technical language.

peteeng: Hi Planck - By professional woodworking do you mean cabinet work, joinery, timber sign engraving or something else

By woodworking I mostly meant custom furniture and layout design that I want to develop on top of my usual activities (which is joinery probably? this goes beyond my English abilities ).

peteeng: your at the beginning of a big learning curve and usually your first CNC is not your last.

Well this alone motivates me a lot. To be fair taking the hard path is something I usually do, I enjoy building from scratch. Plus I don't expect to have large volume applications, at least in the near future. And also the beast has to be somehow foldable because of work space necessities. Thus buying a second hand router is not really an option for me. On the subject of accuracy, reliable XY motion at 0.2mm is good enough for my needs. I have to be a bit more accurate on the Z axis, 0.05mm would be great. Speed is not really my priority, but rather to maintain good cutting conditions without losing steps. This means reaching at least a stable 120 IPM feed rate (horizontal) , 300 IPM would be fantastic. For now I don't bother much about the rapids, I will see where I get with the friction and all. My initial question was more about not "clamping" the machine because of a slower feed rate over one particular axis. I am glad to know it will not.

On the belt design subject, I clearly don't have yet enough experience to add any valuable comment. Hopefully in a few month I will be able to come back to this better informed. Many variables are at play, my goal is mainly to get a well balanced setup. This said I like the Servobelt option for it's ingenuity.

ger21: Mike Everman has said that you need to use T belts, and not AT. It's in the long Servobelt thread here.

Thanks for the link. I started to look into it and it is an interesting read I must say. By the way you seem to have yourself done some serious investigations of the system. Did you use this in working conditions? I saw that you made a neat geared drive out of it. Very inspiring.

ger21: And i think the Servobelt is better than you think, but can be difficult to DIY properly.

Could you elaborate on this last remark ? What specifically is difficult to DIY ?

Oh well, it looks like I am getting caught in a serious adventure. And I have plenty to dig in from your messages.

Cheers,

[ger21]

No, I never tested the sample I made. I was going to redesign it, and make it stronger, but decided to go in a different direction.

Someone else built one similar to mine, and is just getting his machine running now.
https://www.cnczone.com/forums/ardui...ml#post2293096

The difficulty lies in the precision required. This really depends on the equipment you have available, and your abilities.

You should be able to get far more than 300ipm. You might want to even go with slightly smaller 960 oz motors, with a 6 or 7 amp rating. These should actually give you higher speeds.

[peteeng]

Adventure Away!! You won't loss steps...The 0.2mm means a belt is fine and I recommend you use a ballscrew for the Z which means the 0.05mm is fine as well. 300ipm is 7000mm/min which is easy for belts. I do finishing at 15m/min can run at 20m/min using 16mm belts. Most heavy cutting is in the 1000-2000mm/min range. So get more detail going and the transport/folding issue will need some serious thinking. A light bench is a challenge for such a big machine. But if you can afford the R+P thats the go for a 3m machine. Cheers Peter S

Boiling down the Everman idea it's the glued down "rail" thats the key element. If you use a looped belt for the active side you save a lot of belt cost. It also simplifies the active side a bit I think. Attached is a sketch of the tractor drive. I've seen it somewhere. I think this is a step forward from the Everman drive. The active side drive needs more wrap as drawn. I'd try to do it with 2 idlers not 3. If the two idlers are close together this will provide a good wrap on the high drive pulley. Plus you will need a spring system to push the "drive" onto the "rail" just like an R&P system... I'd run the "rack" vertical so it didn't fill up with dust...good luck

The belting is polyurethane and is very smooth and inert. You will need to use a scotchpad or something to take the gloss off the belt. Plus I'd use a contact cement to hold it on... or an industrial strength double sided tape would work, easier to apply. Peter

[peteeng]

Hi All - Here is a photo of an AT belt engaged on the flat. The gaps are serious. I can't see how this can be compensated for. A "T" belt is supposed to have more clearance again?? or does the T have less clearance? Here in Oz we have a saying... Please explain??? It can be rocked back and forth quite easily... Peter

Edit - I dug around and found a piece of T belt. Looks same as the AT! The T has a 40deg trap while the AT is 50deg. The T has 1mm "cover" while the AT has a 1.5mm cover in the 5mm size. This means the is more steel wire in the AT then the T so its stiffer. But the clearance is a worry no way to fix that....

[ger21]

You really need to read the ServoBelt thread.
https://www.cnczone.com/forums/linea...570-forum.html

The pinion pulls the top belt tight, removing all backlash.

[peteeng]

Hi Gerry - I have read the thread prior and to stretch the belt to compensate that much clearance is a fundamental issue I think. It is not adequately explained. So we stretch the belt to create a mismatch to make up the gaps, we sort of longitudinally compress or displace the active belt onto the static belt with the idler geometry. Then we can't fully load the teeth as there is no bottom clearance to allow the tapers to engage. Happy to accept the system will hold >0.1mm over time (so will a std belt system) but not >0.01mm over time in a router application. Too many niggles & fiddles for me. Has someone out there done this on a hard working router? Peter

Edit - I have sent a request to Optibelt asking if there is a belt profile that matches when used flat. I'll keep you all informed.

[wizard]

Plankscale - it is going to be hard to see eye to eye with you! ???????????? (Plano scale so tiny)

My project is not exactly small, I am targeting a 3 by 1.3 meters milling area and 300mm clearance (for professional wood working). Since I unfortunately have a budget limitation using timing belts is appealing.

Still I wouldn’t use timing belts on anything but the largest axis. As noted economics is an issue. On the other hand well designed belt drives are not exactly cheap.

As for the Everman system I’ve never seen one in real life. The first thing that came to mind seeing the drawing is jam ups and glitches due to material getting caught in the belt drive. You can have similar issues on rack and pinion if the orientation isn’t protected but I can see the belts closing and effectively grabbing crap.

While I don’t have a formal name for the design a similar approach is to drop the glued down belt. Such arrangements are used extensively in industry and provide significant damping of belt oscillations which can make belt drives with the belts in free space problematic. Just realize that the precision machining required for a well designed belt drive is a factor in costs. It is easy to look at the cost of a belt and smile and then have that smile turn to a frown when you add in all the extras. If you don’t have a machine shop to DIY much of the structure your costs may be much higher than anticipated. For example the yokes that the idlers and steppers mount to will be expensive to have built.

I have the feeling that a 3 meters screwball drive is anyway a bad idea because of rotational torsion and lengthwise bend.

It has been done. Is it a good idea? That is harder to say. At some point racks make sense.

I understand that racks and pinion would technically be a better choice, but the price difference I have seen is huge. Apparently there are Chinese retailers selling 30mm width reinforced polyurethane belts for as few as $6/meters.

I don’t have experience with the Chinese belts so can’t comment on them. The good thing here is that you are at least considering wider belts. Going too narrow is certain failure. All that said even 30mm can be considered too narrow. However I’m imagining you are driving from both ends of the gantry. In any event I just need to restate that the cost of the system is likely to inflate your build costs significantly.

I’m not a big fan of the AT series of belts. There are many tooth profiles to choose from so your best bet is to glean from the net all the engineering info you can find. While I won’t suggest a tooth profile to use, I will state that the HTD/GT series of belts perform very well. On the other hand I’ve grown to hate AT and similar belts.

At this price I am ready to change them every 100 work.hours if necessary. It would still be about 3 times cheaper than $400/m racks wearing after 1000 hours. Ger21 I am surprised that you have found similar cost.

Timing belts in all their various forms can be very durable in normal operation in a well designed system. I’ve seen a lot of failures over the years from premature damage. That is something gets caught in the drive and snaps a few cords (or all of them). Some timing belts have better wear characteristics than other so you need to consider that.

For the everman belt drive system you are indeed supposed to glue the passive belt. Peteeng I hear your concerns about teeth mismatch, maybe I should just do a proper benchmarking test and re-evaluate from there.

Or just design out the one belt.

By the way it would help if you could define usage a bit better. I can see belt drives being perfectly useful for some woodworking applications. In other instances it might be a terrible choice.

My motor size will probably be 5 amps NEMA34 at 8.4 Nm (1200 oz.in) holding torque. I will look if I can find a 400 steps version for it, throwing in gear boxes surely adds some mechanical complications but once done I imagine that it is a great plus for machine tuning.

In such systems gear boxes are often required to get the positional accuracy required with steppers. A 1.8 degree step can be pretty fours on a 50 or 60 mm diameter pulley.

Here are some suggestion related to the use of timing belt drives in industry. In this case the experience is with respect to large machines such as injection mold machine or as small as pick and place machines handling items a few microns thick.

1. Nothing in this world is perfect. Sometimes you have to compromise and that is often budget related
2. No matter how good the engineering the drives must be installed correctly. It doesn’t matter if it is a belt, leadscrew or rack, proper installation is a must.
3. Good quality timing belts really don’t stretch much. A retensioning might be required after initial run in. After that you are likely to have belt breakage, specifically the steel cords.
4. The cords will eventually break due to the higher installation tension. Timing belts used in motion control often require far more tension than one would apply to a belt doing power transmission. In any event loose belts do motion systems no good.
5. Once a belt is damaged it is toast! You can’t take a die grinder to it, weld it or otherwise try to repair it. Often you can’t even manage a field expedient to get the machine running.
6. Gates and others have engineering resources on line that can greatly help with drive design. However stay away from the minimal requirements. Idler pulley diameters can impact belt life for example.
7. Belts need protection! As noted above it is easy to damage a belt well before its expected lifetime. One issue is foreign object damage and the other is E-Stop braking. Foreign object damage can very destructive taking a belt out instantly. So you need to consider the expense of guarding for the belt or design protection into the machine.
8.. I don’t like to see one drive system as being superior to another. Belt drives can be very capable but how well they will work for you is unknown. It comes down to what you are doing and your expectations.