A Dreadful Likelihood of Dual Chain (or belt) Drives

[Beefy]

Hi Everyone,

I've built my 2.4 x 1.5m plasma table using dual chain drive on the X-axis. Did my very first test cut and then had a haunting thought about the accuracy implications of dual chain (or belt) drives which are driven from a single cross shaft as mine are.

I squared the gantry to the table at one end then realised that due to the high likelyhood of the chains not being equal lengths over a 2.4m distance, the gantry could be ran to the opposite end and would be out of square. And there's nothing that can be done about this. Even if the total length of chain was exactly the same length from new (for a given number of links of course) it would be unlikely that wear and stretch in the chain would occur equally and so you could only square the gantry to the table at one end but when at the opposite end would be out of square by the length difference of the chain.

So if anyone is considering a single shaft drive with only one drive motor and accuracy is a concern then maybe it would be better to spend the extra cash and drive each chain individually so the steps per mm / inch could be calibrated for each side.

After the above thoughts I think rack and pinion is the very best way for a plasma table. No stretch, more dirt resistant if they are mounted with the spur gear underneath, no tensioning issues.

Just a thought for anyone who's thinking of doing the same design as me. It's my first table and like a lot of other people I've been doing the money saving thing and have done a lot of living and learning as a result.

Anybody had any such issues with accuracy ?

All the best for the New Year.

Keith.

[ger21]

So if anyone is considering a single shaft drive with only one drive motor and accuracy is a concern then maybe it would be better to spend the extra cash and drive each chain individually so the steps per mm / inch could be calibrated for each side.

If you use mach3, and slave two motors, they must be set exactly the same, so you can't calibrate them independently.

[Beefy]

Now that is something I didn't know. Thanks very much for that, it changes things quite a bit. I have been through the Mach manual several times yet I'm learning I can easily miss things. Now that you've told me that I think I can recall the manual saying the slaved axis recieves exactly the same pulses.

Seems then that the ONLY way to get the accuracy in question is to use dual gear racks where the manufacture has been accurate enough to give them "identical" lengths.

I'm actually still in a bit of shock over this new information. When money permits if didn't go the gear rack route, I would have invested in a 2nd motor and reduction pulleys, + another Gecko drive, for the second chain, yet have been in exactly the same position as I am now.

Well it's great to know where I stand now. Thanks again, your reply has educated me and saved me some future dummy spitting.

Keith

[ger21]

Just to be sure I'm giving you 100% correct info, I'll do some testing later today, with the steps per inch of the slave set to half that of the master, to see what it does. If it even lets me.

[jimcolt]

The industrial machines (read expensive) typically use a servo drive on each side of the gantry (usually inertia matched, ac brushless) coupled to a zero to low backlash planetary gear box and helical rack and pinion drive. Each side is treated as its own axis, and most machines leave the factory after being calibrated with a laser interferometer to compensate for gear rack tolerances. The Industrial CNC controls can map tolerances along the length of the motion, store them in memory, and compensate for them continuously. This technology may not be necessary on smaller footprint machines....but many industrial cnc machines have very long travel lengths.....I have seen 350' long machines in shipyards!

On the lower end of the spectrum, there are machines with a single gantry drive motor with a cross shaft and standard rack gears on both sides. These maintain accuracy and squareness pretty well. I have never seen a "chain drive" commercial cnc plasma....but there are a few that use long "toothed belt" drives....typically these are lower accuracy machines use for sheet metal HVAC (ductwork) applications.

Jim

[Beefy]

Gerry this is interesting. The Mach manual does indeed say that the slaved axis will recieve the same pulses as the master axis (during normal use it adds !), YET I can go into motor tuning and set different values for steps per mm/in. However I don't have dual drive axis to test this out. The very fact that the manual states "during normal use" seems to imply that this is not set in stone. If you manage to do those tests that will be very interesting. If not I'll do a separate post on this subject.

Jim, you've answered some questions that have been in my mind several times in the past. I sometimes suffer from thinking deeper than is necessary and one of those things was where a drive system could be worn at a point along its length, say where lots of cutting was done. The designers of those big machines obviously leave no stone unturned when it comes to setting up and maintaining accuracy.
I haven't noticed many people talking about how they set their machines up accurately, especially getting the gantry perpendicular to the X-axis. Apart from having a giant and accurate set square, or the method I used where I attach a fine tip pen and use Pythagoras to measure and see if I've got 90 degrees, have you heard of any other tricks (I guess the laser setup gear may be expensive to hire).

Cheers,

Keith.

[ger21]

I just went out and checked it. My X axis is 6400 steps/inch. I set the A axis to 3200 steps/inch. I jogged it back and forth, and it was no different
Mach3 appears to ignore the A axis setting, and the machine works the same as it does with both set the same.

[Beefy]

Thanks for that Gerry, guess I'm not getting off the hook that easily. The only way around this that I can think of at the moment would be to not have the A-axis slaved, and modify the post in my cam program so whenever a command is given to the x-axis, and identical command is automatically given to the A-axis. However, I have absolutely no idea how to do that.

Keith.

[Torchhead]

If you use mach3, and slave two motors, they must be set exactly the same, so you can't calibrate them independently.

Ger: You sure? My first servo plasma table when I calibrated each side was a little different on the steps per unit side to side. I was able to set the slave slightly different and it corrected the issue. I tested it over several feet of run and multiple times. Maybe I was dreaming......It's been several years now and that was a ballscrew table (bad mistake for plasma).

TOM C

[ger21]

Read my post #7 above.

[start here]

Hi Everyone,

I've built my 2.4 x 1.5m plasma table using dual chain drive on the X-axis. Did my very first test cut and then had a haunting thought about the accuracy implications of dual chain (or belt) drives which are driven from a single cross shaft as mine are.

I squared the gantry to the table at one end then realised that due to the high likelyhood of the chains not being equal lengths over a 2.4m distance, the gantry could be ran to the opposite end and would be out of square. And there's nothing that can be done about this. Even if the total length of chain was exactly the same length from new (for a given number of links of course) it would be unlikely that wear and stretch in the chain would occur equally and so you could only square the gantry to the table at one end but when at the opposite end would be out of square by the length difference of the chain.

So if anyone is considering a single shaft drive with only one drive motor and accuracy is a concern then maybe it would be better to spend the extra cash and drive each chain individually so the steps per mm / inch could be calibrated for each side.

After the above thoughts I think rack and pinion is the very best way for a plasma table. No stretch, more dirt resistant if they are mounted with the spur gear underneath, no tensioning issues.

Just a thought for anyone who's thinking of doing the same design as me. It's my first table and like a lot of other people I've been doing the money saving thing and have done a lot of living and learning as a result.

Anybody had any such issues with accuracy ?

All the best for the New Year.

Keith.

I too am building a table using dual chain drive on the X-axis (but mine uses one motor per side) so I read your post with interest.

After a little thought of what you described, IMHO, I doubt it will be a problem. I think the chains or belts will need to operate under extreme tension for a long time before any significant stretch is measurable.
And if (or when) you find that it is starting to affect the accuracy of the cutting you should be able to simply tighten one side....
I would think a simple test can quickly determine if you have a problem. By temporarily disconnecting one side, get the table to move the gantry a known distance - say 2 meters, & mark that point. Now reconnect that side, disconnect the other side, and repeat the same 2 meter move. If the marks don't line up, you can see which one is further along and how much to tighten the chain on the opposite side.

[Beefy]

Hi Start Here,

Thanks for your input. Terry Parker has been using chain for years and he says it's never been an issue for him, but he told me something very important about chain drive and which way to do it, so I thought I should pass this onto you seeing as you are building one.

My table has the drive sprocket mounted at each side of the table, and idler sprockets at the opposite ends for tensioning. So the sprocket is stationary and the chain moves. Problem with this is you get a lot more chain "bouncing". Terrys system has stationary chain stretched the length of the table and his drive sprockets are mounted each side of the gantry. So in effect the chain is a rack and the sprockets are the pinions just like a rack and pinion setup. Even with my chains quite tight you still get the bounce especially on sharp turns and this fast bouncing shows up in the cuts. In my cam program I'll use "loop sharp corners" to try and prevent this, plus I'll use the smallest tips I can to keep the speed down. Not a great machine design.

The one thing that has interested me with a lot of the replies on this topic is that so many are saying they don't think it will be a problem. Nobody seems to have actually tested / measured their systems to verify it definitely is not a problem. I will be checking it out after all the other tasks I have on my plate but I personally could not just hope for the best and think it won't be a problem. Let me give you an example.

I did all the calculations to enter steps per mm in Mach. Synchronous pulley ratios, chain pitch and number of sprocket teeth per revolution. If the chain is manufactured so precisely then my steps per mm should have been dead on. I used Machs calibration feature to measure the distance travelled over the 2.4m run and it was not exactly as calculated. Tolerances are a very standard thing in any manufactured item so I didn't expect my steps per mm to match exactly. With my chain pitch there's 260 links over 2.4 metres. Each chain links allowable tolerance can add or subtract to a length when you've got 260 of them. If they do end up precisely the same length I would feel very lucky, and likewise I would be very surprised to find them stretching by exactly the same length. Like I say I will be doing initial tests on this and checking these lengths every so many hours of use.

Another thing is google some chain manufacturers and see what they have to say about initial wear-in stretch. Mine was predicted to be 2.4 mm longer after wearing in. Definitely not the type of thing I will think / hope will be OK.

At the end of the day it probably won't be an issue unless I'm cutting a large item where the sides need to be accurately square. Plasma cutting itself is not that accurate due to the bevel on the sides, and getting the kerf offset very accurate. The biggest issue I'm finding now is the flop in the chains affecting the cut quality, mainly after a sharp turn. When money and time allow I'll be swapping to rack and pinion with spring loaded pinions to give zero backlash. No stretch, no bounce.

Cheers,

Keith.