Advanced Search

TLDR: get the largest harmonic reducer you can afford and build a 4th axis around it.

The two (basic) issues with rotary axes are backlash and compliance.  Eliminating both gets very expensive.

A planetary reducer - especially a low-cost one - will have backlash, and so will a worm drive.  Both mechanisms are generally suitable for uni-directional indexing, but bi-directional movement will result in artifacts or dimensional errors.  Features machined from multiple orientations will not line up.

Worse still, if off-centerline cutting forces are applied, the part can rotate even with no commanded movement.  Imagine drilling straight down along the centerline - there's no rotational force on the rotary axis so all goes fine.  However, if you try to drill holes offset from the rotational axis the cutting forces will cause the rotary axis to move as the force takes up the mechanical backlash.  Same thing happens when side-milling above the centerline.

This problem can be reduced or eliminated by adding a brake to the rotary axis directly on the output shaft (platter).  A motor brake (stepper or servo) will stop back-driving, but not eliminate cutting forces taking up the backlash.

The above may not be an issue for light machining close to the centerline... but even very low-load engraving will result in artifacts due to backlash.  My mill came with a worm drive 4th axis and I quickly abandoned it due to the backlash issue.

Harmonic reducers eliminate backlash (more or less), and have come down in price dramatically in the past 4-5 years.  The downside with harmonic reducers is compliance.  Due to the nature of the flexible strain wave gear/cup, machining which imposes a high rotational force can cause the flex gear to twist under load.  Once the force is removed the reducer 'springs back' in to the desired position, but this flex can cause tool chatter or dimensional errors.  Generally speaking, a bigger reducer will flex less than a smaller one under the same rotational force since the larger flex cup/gear is stiffer.

As with planetary or worm drives, a suitable brake will stop chatter or flex in a harmonic reducer.  But a brake cant work during continuous rotational machining.

For light(ish) machining that isn't really far off the centerline, harmonic reducers are an easy solution to the backlash problem.  I now have a 32mm (through-bore) harmonic drive in a 6" diameter rotary axis and it works quite well.  But if I had wanted a 8" or larger mechanism - or I had a more powerful mill using larger tools - I would probably have gone up to a 40mm (or larger) reducer to increase the off-centerline stiffness.

At the upper end of rotary axis mechanisms are cam-roller drives and direct drives (torque motors).  Cam-roller mechanisms, especially preloaded versions, have zero backlash and are extremely stiff, fast, and accurate.  Direct drive motors are also excellent, but may not be as fast as cam-roller drives unless wound for higher speed (at the potential loss of holding torque).

Both are eye-wateringly expensive and are not as common on the surplus market as harmonic reducers.

Hey Scotta,I saw that you recommended flashing the ST-Link with CMSIS-DAP firmware. Just out of curiosity, I tried flashing it using the ST-Link, and it worked! In fact, it was smoother than the CMSIS-DAP probe I bought. The reason is, the CMSIS-DAP probe needed some extra configuration to work on Windows, while the ST-Link just worked out of the box. So, for anyone thinking of flashing the firmware, I’d suggest using the ST-Link V2 with PyOCD directly.

The 7I85 has built in 120 Ohm termination when used is RS-422 (line driver) mode.
no additional termination is needed.

1. Should be OK, just get a good quality one like MeanWell or Delta
2. TTL encoders should work as is with 7i85, so should glass scales rated at 5V, incremental types that is, and both differential and single ended also work on most Mesa cards.

Hmm, possibly damaged 7I92T  or 7I77?

If you unplug the 7I92 and 7I77 (with all power off), re-power the 7I92T, what does:

mesaflash --device 7i92t --addr [ip_addr] --rpo 0x1000

report?

(ip_addr = 192.168.1.121 or 10.10.10.10 depending on 7I92T jumpering)

Ok thank you.

I think I am gonna look into precise switch because I know nothing about encoder...

I don't even know if my stepper have an encoder in them, I think it is optional (?).

I run closed loop 48v stepper with CL57T drivers on a Mesa 7ixx.

Thank you for the info.

I am trying a piece ATM with skewkins (linked above), it look to work fine too.

Less skewing options than matrixkins (I think) but I only needed X/Y skew.

ok update...today i got a working 3 and a working 4 axis plot to my machine configuration...the 3 Axis working plot is just fine all buttons working correctly.but in the 4th axis plot the Zero B button did not work. DRO works correctly to my b axis.and when i do single ref for all axis (not refall button) i cant jog my machine and become a error like this from the attachment.

is this an error from porbe_basic or from linux_cnc? iám running linux cnc 2.9.0 pre ...

edit: I solved the Zero B error... G10 L2 P{ch[0]} B0.0 needs to be G10 L20 P{ch[0]} B0.0

Hi hmnijp,

Thanks for your ngc files.






 

Hi,
I'm working on building a 3 axis CNC inspection system. There will be 2 Y axes with one as the slave axis. I plan to use linear encoders on all X, Y and Z axes. For this, I have got the 7i96S and 7i85 Mesa boards. I had two queries regarding the electrical circuits.

• The power supply rating is 5 V. Is a 5V 2A PSU suitable?
• The datasheet for 7i85 says that the RS422 encoders should be terminated with a 120 Ohm resistor and a 26LS32 RS-422 differential receiver. What would the circuit look like in this case? For reference, this is the encoder I plan to use, in the Line Driver model. m.alibaba.com/x/16C29T?ck=pdp

I'm very much new to this, so I apologize if these are very basic queries. I'd really appreciate any insights.

I have used Matrixkins and it is very useful, but you need to compile the comp file yourself

Glad you got it working.
There is another sticky to fix this
forum.linuxcnc.org/ethercat/53785-instal...thercat-repositories

I'm also seeing issues with the GPG-keys in 2.9.3. I just did a fresh install from the iso found on linuxcnc.org/downloads/ and the first command i ran was:...not sure why it isn't working...

I'm but no means a Linux guru, so my attempt to fix this was to try to jumble together a solution from this thread. I ended up fetching a new key:...but that didn't work, so i ended up deleting the old keys and copying the new one into their place:
...i'm pretty sure this isn't the correct way to fix it, but now i can run apt update without errors. Maybe it can help someone else. I believe this is a bug in 2.9.3, since it was a completely fresh install(with available internet connection) and it was literaly the first command i ran and it failed... 

That is really cool, once I have my existing machine fully tested I intend to start playing around with all the extra stuff you've implemented. When you say several devices per bus, is that now as per the standard i2c protocol, so 128 devices ? I understand that lots of i2c devices only have a couple of address options so that doesn't help.