another 4th axis...... and a 5th

Home Model Engine Machinist Forum

Help Support Home Model Engine Machinist Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

blighty

Well-Known Member
HMEM Supporting Member
Joined
May 3, 2011
Messages
410
Reaction score
67
i have some jobs coming up that could really do with a 4th axis. my cnc controller box is already geared up for a 4th and 5th axis, and with the advent of fusion 360 making its self free to all hobbyist (thanks Autodesk) it is now possible to run a 5 axis mill.

so i have been hitting the numbers to find out gearing between the stepper and the table. i know that normal rotary tables are 90:1 that would be great, but they do suffer from backlash. this can be made to a minimum but you will still have some. next option would be timing belt. problem with that is your now down to very low gear ratios e.g 6:1 other wise pulleys tend to get big. next is antibacklash worm drive. this will now put us back up to the 90:1 plus. and the last one is the harmonic drive. no backlash, high ratios....... not cheap.

it seems that the easiest option would be belts and pulleys. that then brings up the gear ratio. if you take the step of the motor, plus microstepping divide it all by 360 etc etc. and make the gearing 3.6:1 (20t:72t pulley) for each step you will get 0'3'0 deg. this works out well. but if your part has a 4"rad and you machined at 4" you would get a flat that is 0.12mm wide at each step. on an 8 inch part would this really be a problem? depending on space i could go to a 144 tooth pulley or up the micro step count.

as this one day will sit on top of a 5th axis both 4th and 5th should really have the same gearing and all sit in a compact design.

I have all the gear i need to build it now, as in steppers and controllers so it should be a quick build :D:D:D:D

thoughts??
 
Hi

I fretted over the same problems when fitting a 4th axis to my CNC router. I finally just dug very deep into my pocket and bought a zero backlash gearbox that fit directly to my stepper motor. Works great but expensive if you can not find a used one like I did.

Mark T
 
Having fitted a stepper to a 90:1 rotary table, I found that you can't get much more than 4 rpm out of the average stepper motor. I think a 5:1 timing belt drive with a good microstepping drive would do the trick. My DM542A stepper drive does 25x microstepping on my plasma table under LinuxCNC and it only cost about $30. Thats 5000 steps per rev so the resolution would be about 52 seconds when I eventually add a rotary axis. Surely that is enough?
 
Having fitted a stepper to a 90:1 rotary table, I found that you can't get much more than 4 rpm out of the average stepper motor. I think a 5:1 timing belt drive with a good microstepping drive would do the trick. My DM542A stepper drive does 25x microstepping on my plasma table under LinuxCNC and it only cost about $30. Thats 5000 steps per rev so the resolution would be about 52 seconds when I eventually add a rotary axis. Surely that is enough?

think your right about 52 seconds being enough, but it would all depend on the diameter of what your cutting. im thinking of making the table 8" so i can work thinks out from there. was also thinking about the step count. as in what does it do when there is a measurement thats in between your step count. e.g you can only get 1deg per step what happens when one day you need 1/2deg step.
 
had a play with some numbers.

i was a bit out from what i said before. if you have a stepper stepping at 10miro steps with a gearing of 3.6:1 you will get a step of 0.05deg (0'3'0)
that would mean if you took a cut then did one step took another cut, over a 4" rad you would get a flat of 0.088mm (0.0034)
had a look at the micro stepping values my driver can do and one of them is 50. so doing the same cut you now get a 0.02mm (0.00078')flat and 0.01deg (0'1'12)
would be nice to get rid of the '12. considering you probably get 0.1mm and more in head deflection, dont think 0.02mm over 8 inches is nothing to worry about.

20170312_201337.jpg
 
This is interesting to me as I've fabricated a stepper motor adapter to my 6" rotary table but I have yet to use it...(insert excuse here), though I did generate some G-code from Fusion 360 to try it out.

I have the typical 90:1 reduction - 1 deg = 50 full steps (1/4 turn on stepper / input).
But the driver I currently have is typical of the cheap ones($15), micro stepping by the power of 2 (1 2 4 8 16 or 32 micro steps). I'm going to try the 32 micro steps and see how it does, as a smaller resolution may be of more benefit than speed - time will tell.

You are probably right that the absolute distance of each step at the 4" rad isn't critically important though small is nice.
I think that - depending on your controller - having a small "exact" number of degrees per step would be the desirable configuration, so that there is no rounding error when the controller is calculating the moves, though not being a mathematician or programmer I could be wrong and there are other ways of calculating angles (rads, etc) and cumulative moves that may be in use behind the scenes to accommodate error stack up.

I use LinuxCNC and the rotary axis motion is defined by degrees (default / my choice).

I think it also has to do with the CAM software and the post its using for the controller and the rotary axes, though I would think most will output degrees.

Do keep us informed of your progress.

Mike
 
This is interesting to me as I've fabricated a stepper motor adapter to my 6" rotary table but I have yet to use it...(insert excuse here), though I did generate some G-code from Fusion 360 to try it out.

I have the typical 90:1 reduction - 1 deg = 50 full steps (1/4 turn on stepper / input).
But the driver I currently have is typical of the cheap ones($15), micro stepping by the power of 2 (1 2 4 8 16 or 32 micro steps). I'm going to try the 32 micro steps and see how it does, as a smaller resolution may be of more benefit than speed - time will tell.

You are probably right that the absolute distance of each step at the 4" rad isn't critically important though small is nice.
I think that - depending on your controller - having a small "exact" number of degrees per step would be the desirable configuration, so that there is no rounding error when the controller is calculating the moves, though not being a mathematician or programmer I could be wrong and there are other ways of calculating angles (rads, etc) and cumulative moves that may be in use behind the scenes to accommodate error stack up.

I use LinuxCNC and the rotary axis motion is defined by degrees (default / my choice).

I think it also has to do with the CAM software and the post its using for the controller and the rotary axes, though I would think most will output degrees.

Do keep us informed of your progress.

Mike

using Mach3 at the mo, and with mach like LinuxCNC you can ether run the 4th in constant mode or step and dir. 1 unit is taken as 1 deg. have found a thread on how to change the post possessor to get mach to work with Fusion 360 and 4th axis. not found one with 5th yet.

i'll get the sizes of the pulleys i need and draw it up. would be nice if i could lay the stepper on its side but that would mean having a 90deg twist in the belt. at the mo it will be about 5.5inch think. that's from the bottom to the top of the 4th axis table.

as for bearing to support the shaft, was thinking angular roller near the work end and a plane bearing with thrust race the other, but may go for roller both ends, or i could............:wall: its never easy is it :wall:
 
Blighty, I will be very surprised if you can get up to 50x microstepping. There are a lot of other timing issues that will impact and limit speed in the real world.

In my instance, I have dedicated hardware step generators on a commercial quality external ethernet control card (Mesa 7i76e) able to run at 10 Mhz and the best I could get was about 90 kHz with 40x stepping. I did shorten the step time to see if I could squeeze a few more pulses in a second but never tried to see If I could push it past 40x. Also be aware that there is a limit to how much accuracy microstepping can add so you can't really assume that you have microstep resolution. Thats why Gecko does not do more than 10x microstepping with their stepper drives. What you do gain (with good quality microstepping controllers) is a much smoother motion.
 
Blighty, I will be very surprised if you can get up to 50x microstepping. There are a lot of other timing issues that will impact and limit speed in the real world.

In my instance, I have dedicated hardware step generators on a commercial quality external ethernet control card (Mesa 7i76e) able to run at 10 Mhz and the best I could get was about 90 kHz with 40x stepping. I did shorten the step time to see if I could squeeze a few more pulses in a second but never tried to see If I could push it past 40x. Also be aware that there is a limit to how much accuracy microstepping can add so you can't really assume that you have microstep resolution. Thats why Gecko does not do more than 10x microstepping with their stepper drives. What you do gain (with good quality microstepping controllers) is a much smoother motion.

as i said.........:wall:its never easy:wall:

in that case ill have a play with a stepper at 50 micro and see what happens. i remember increasing the step count on me lathe. i didn't run a job, was just seeing what it would do. from what i remember it got smother, quieter and a lot slower the higher you went. would start skipping steps if you went to fast. if you kept the speed down it was fine, but saying that and as i didn't run a job i have no idea how accurate it was.

i've just put in order in for the bearings. was about to order the belts a pulleys. then i saw your post. i'll hold off till i have a play. but again, saying that if the stepper is just sitting there i wouldn't know if its keeping position. i could rig something up on a surface table with some sort of flap on the stepper. clock up the flap, spin the stepper at various micro steps, speeds and see what happens.
 
The one I pointed to, with postal charges and 20% import duty from China should come to just over £400 to the UK.

I got a 5" 6 jaw chuck from AliExpress for £140, and only had to pay £4 import duty, so the above price could be cheaper.

There are hundreds of this type available from AE, some with chucks, some without, all about the same price.


John
 
blogwitch....

sorry i should learn to read stuff... well all of it anyway. i just saw the US bit amused it was from the US. dont tend to get stuff from the US as they cant seem to make up the mind how much to charge. i spent 50 odd quid on some bearings from the US and got stung 48 quid. someone sent me a pulse jet to fix 140 to send it 60 odd quid get it.
 
Blighty, When I did my tests I used a cheap $10 tacho off ebay to measure RPM of an unloaded stepper. You can calculate the RPM and compare with what you measure to see if you are missing steps. Once I hit the limits, I backed off a bit.
 
looking at the table blogwitch posted. one axis has 8:1 the other 6:1. i have seen a lot of 4th with 6:1. may be there is something about 6:1 that gives you the best of both worlds. or it could be a case of "well that guy did it, so will i" kind of thing.

if you go with 6:1 you get 0.03deg per step. opposed to 0.05deg @3.6:1 going by that i would need a 20t and a 120t. HPC gears stop at 84t at £69 (http://www.hpcgears.com/pdf_c33/12.72-12.77.pdf). i could make my own pulleys.... hang on don't you need a 4th axis for that:wall:
 

Latest posts

Back
Top