Machsupport Forum
Third party software and hardware support forums. => CS-Lab => Topic started by: Hood on May 17, 2013, 09:49:44 AM
-
CS-Lab have just implemented rigid tapping in the IP-A and IP-S controllers when used in conjunction with the Encoder module.
Here is a video of the first tests, it is mild steel and I am tapping at a conservative 300rpm and reversing at 600rpm, all seems to be working perfectly :)
http://youtu.be/GNL2XdsJz2E
Hood
-
Doing what they said could not be done.......... again! This is great Hood! And, you can reverse at a different speed than you go in.......... this sounds very good. How do you choose the forward and reversing speeds and/or the difference between them? Is that set in gcode or the plugin or where? Tell us more.
Brett
-
As rigid tapping is not supported in Mach3, its only a pseudo G84, the guys at CS-Lab have used a macro.
In your code all you need is M84 instead of G84 and the parameters are
P - pitch
Q - depth
S - Forward RPM
R - Reverse RPM
If tapping a left hand thread the pitch is written as a negative.
There is some sample code included with the macro, it is as follows.
%
G18 G21 G40 G50 G80 G90
G0 X0
Z0
M84 P2 Q-30 S50 R100
G0 Z5
X10
M5
M30
%
Hood
-
HIYA Hood is that with a servo spindle or a VFD spindle ? The trick is to get the VFD spindle version working. As we have had rigid tapping with a servo for years now.
Just a thought, (;-)
-
I have a servo but it will work with any spindle as long as it has an encoder fitted.
Would probably even work with a spindle that the speed is controlled manually, as long as Mach controls the Fwd and Reverse.
Hood
-
This controller just keeps getting better all the time. All the things we ever wanted with Mach3 are becoming possible with it. Also saw in the change log that they did some work on the threading code, wonder if the pullout got worked on as well.
The tapping looks great. Thanks for the video, Hood. And the macro is nice too.
Dan
-
Oh... also meant to ask - is the Z actually electronically geared to the spindle so that manually rotating the spindle would cause the Z to move? Guess you'd need to enable the M84 macro for this?
Dan
-
Nice work
I'm impressed.
-
Terry
Just run it again for you :)
Slowed the accel of the spindle, was previously 0.8 and now set to 10 to simulate the slow response from a VFD.
Here is a vid showing it tapping the same as before.
http://youtu.be/2DE4sAgLPX8
Hood
-
Dan
Yes its everything I have wanted in a controller, in fact its actually more than I had ever thought possible and its only getting better as time goes on.
Regarding the threading with the lathe, no they can not do anything about that with Mach3 as it is a Mach3 fault.
HOWEVER!!!!!!!! I eventually got some time with Brian yesterday and did some testing of altered code, as you will see in the video it is almost perfect, in simulation at least. The only problem now seems to be when the thread has finished Mach just sits and will not continue with the rest of the code. We ran out of time but I am booked in to do some more testing with Brian on Tuesday so fingers crossed.
I had hoped to do more tests today on the lathe but the spindle drive is brain dead and I dont have a spare of the same kind. I will fit a DSD tomorrow but it means I have a lot of soldering to do as the plugs are different on them compared to the DDM drives.
http://www.youtube.com/watch?v=BHlTshnxiac
-
Thanks Robert :)
Hood
-
The threading in the simulation looks perfect! Engages incredibly quickly and retracts fast as well. Fingers crossed for your next session with Brian :)
Dan
-
Yes it does, was wanting to actually cut some threads today and to slow and speed the spindle with the override but the spindle drive decided not to allow that ;D
Regarding your other question, yes the Z is synchronised with the spindle encoder but not sure if it would work by rotating manually, may well do but no way for me to test on the Chiron.
Hood
-
Thanks, Hood.
Dan
-
Hood in slow accel mode how well does it control Z depth for blind hole tapping? THAT will be the true test.
GOOD job so far (;-) TP
-
Terry,
I will be able to tell you tomorrow with 100% certainty but I am sure it doesnt go any deeper. I can easily check tomorrow as the axis DROs are updated from the axis encoders so I will be able to see if they do go to the depth specified in the code.
I believe even with a VFD spindle control is closed loop (assuming you have an encoder on it) I have never tested it as I dont have a VFD spindle on any of my machines, but there is PID tuning for the 0-10v analogue out for VFD spindles.
Hood
Hood
-
Hood,
That's really impressive. Somewhere very deep down my to-do list (read wish-list) is to try rigid tapping with Arduino. My machine is already equipped with VFD and a differential encoder on the spindle motor. After seeing your posts it has moved up many notches on the to-do list. Let's see when I gather enough courage to give it a try :).
Zafar
-
Be sure to keep us updated Zafar :)
Hood
-
Hi Terry,
couple of vids for you, first is showing an exceptionally slow accel of spindle (20 seconds to get to full rpm, normally 0.8 set)
It shows that the depth is observed exactly when rigid tapping.
http://youtu.be/A29bYyWzrlo
Second vid is same accel but this time I have done a peck tap, 5mm then 10mm then 15 then finally 20mm. Just did it in Alu that is 1/4" thick but I am sure you will agree if it wasnt pecking exactly right the threads would be stripped :)
http://youtu.be/UbNizRlRT1g
I have to say I think the IP-A is the mutts nuts :)
Hood
-
Thanks HOOD, You answered all my questions (;-).
I agree the unit appears to handle Rigid tapping very well.
(;-) TP
-
Wow, that's amazing Hood. Need to show that to those snooty LinuxCNC guys. ;)
-
Well thinking more about a VFD tapping. A servo will still KNOW what the accel/deaccel values are not mater how fast or slow you run it. So going to a set depth is not a problem
With a VFD you will NOT know what the values will be(different for each speed you run it) so the unit MAY or may not work well for blind tapping with a VFD.
More testing needed using a VFD.
Just a thought, (;-) TP
-
I also ran the peck thread again into the previously peck tapped hole but this time at full accel of the spindle (0.8 seconds to full RPM) and again it worked flawlessly as can be seen in the video.
http://youtu.be/d87kv3GSqvY
Hood
-
Good tests and videos, Hood. Thanks.
Dan
-
Wow, that's amazing Hood. Need to show that to those snooty LinuxCNC guys. ;)
LOL one of the loudest exponents of LinuxCNC and Mach bashing has seen it but failed to comment by the looks of it ;D
Hood
-
Well thinking more about a VFD tapping. A servo will still KNOW what the accel/deaccel values are not mater how fast or slow you run it. So going to a set depth is not a problem
With a VFD you will NOT know what the values will be(different for each speed you run it) so the unit MAY or may not work well for blind tapping with a VFD.
More testing needed using a VFD.
Just a thought, (;-) TP
Replied to you on the Yahoo however I will post my reply here as well just so people dont think I am ignoring you ;D
Terry, my spindle (as well as axes) is controlled via +/-10v analogue signal and
the loop is closed back to the CSMIO and it is tuned with the PID in the CSMIO.
If you have a VFD you will be using the 0-10v analogue signal from the CSMIO,
again the encoder goes back to the CSMIO and again it is tuned in the CSMIO.
So in summary I can not see how there would be a difference, both analogue
voltage, both tuned in CSMIO, both using encoder feedback.
Hood
-
No probs Dan, I am just so delighted with it I like to show off what it is capable of ;D
Hood
-
Bravo......... Bravo!!!!!! ;D It just keeps getting better. Very impressive Hood! Even the coolant is over the top! ;D
Brett
-
Nah its just fine, did have two locline hoses on the other week but that was OTT so went back to 1 ;D
Hood
-
Is it me? I saw the video - very nice! It is always cool to see a converted machine that runs as good or better than the original control. Nice work.
Mach has been able to rigid tap for a while with external hardware that supports it.. Kflop and now cs-labs (there might be more). ( http://www.cnczone.com/forums/benchtop_machines/125999-grizzly_x3_rigid_tapping_1000rpm_video.html ). 'mach' doesn't rigid tap. The external motion card does. (as seen by the unconventional gcode to make it work...) That is the main argument. Linuxcnc with its realtime control does all the motion and syncing (plus integrated ladder logic). This makes it so any external hardware (even the printer port) can rigid tap (or do other fancy cool things like gear hobbing...). While with mach - you have to make sure the motion hardware you buy supports the things you want to do. (plus you can configure everything in one place. - hood had to tune and setup cs-labs, mach, and his plc. With linuxcnc - this could all be done in one place.) And no - you don't have to be a programmer to do it.
Knowing the work that goes into a conversion of that scale - again - very nice work!
sam
Wow, that's amazing Hood. Need to show that to those snooty LinuxCNC guys. ;)
LOL one of the loudest exponents of LinuxCNC and Mach bashing has seen it but failed to comment by the looks of it ;D
Hood
-
It has been possible to do rigid tapping for a lot longer than that Sam, as long as you had a servo or stepper on the spindle, no other hardware required. Was it a good way, well not really, you had to use swapaxis() and to be honest I just used the floating holder as it was easier and did a great job.
Regarding tuning in one place, well its all done in Mach, the plugin is in Mach ;)
I presume you have to tune your spindle loop in LinuxCNC?
Now did I need to use the PLC, nope, CSMIO has a load of I/O and lots of expansion capability if required. So why did I use a PLC, well two reasons,
1. I had one sitting doing nothing.
2 I know my way around the direct logic PLCs so its easy. Doing it in Brains and/or VB would take me longer, but it could be done and has been by others.
The ladder in my PLC runs to, I think, 17 pages of A4 when printed out at 60%, lots of I/O on this changer and I wanted lots of interlocks, you can even see one working in one of the vids. The toolchange takes a lot longer to happen simply because I had such a slow accel/decel set for the test and the spindle took so long to drop to 200rpm which is the tool change rpm and the toolchange will not happen until that rpm is seen.
Even if in the future if I decide to go with LinuxCNC (about as likely as you using Mach, but never say never ;) ) then I would still use the PLC, for the simple reason I would not have to learn the ladder logic in Linux. That is assuming LinuxCNC supports Modbus?
One thing I really like about the CSMIO is the capability of using high res encoders, that means very tight control can be kept.
Hood
-
Sure - stepper/servo spindle as an axis - sure. That has been possible on both platforms for a long time. (I don't think it is a good way either...)
I did not know the plug-in within mach allowed for pid tuning and such - neat. I don't know what you mean by tuning the spindle in linuxcnc. Yes - if the spindle has its own pid loop for speed control - you would tune it within linuxcnc. Some do that. The vfd on our hmc does good enough that I never setup closed loop speed control. It isn't required for rigid tapping as the axis is slaved to the spindle. The axis just stays in sync with whatever the spindle does. (don't know if that is what you where asking..)
I don't know how standard the plc language is in linuxcnc. I didn't have any trouble wrapping my head around it - having only some exposure in college... I am about 50/50.. I did the tool change/chain/pallets logic in ladder and the spindle gear shift logic in 'comp' (a C like language for making realtime components in linuxcnc) I could wrap my had around the 16 gear spindle programmatically vs ladder. Again - vs the brains/vb - that isn't realtime. it happens when it happens. In linuxcnc I know that the ladder is going to scan every 1ms and the realtime logic is going to happen every X ms or us..
There is modbus support - but I cannot give you any details as I have never used it. I mostly see people using it to interface with their vfd's
I hear you with switching.. I don't think I would either. I like the control that have with linuxcnc.
with mesa hardware (5i25 with server daughter board - 6 axis + 48i/o + expandable $239) you can count to 2mhz.
Again - awesome conversion.
sam
It has been possible to do rigid tapping for a lot longer than that Sam, as long as you had a servo or stepper on the spindle, no other hardware required. Was it a good way, well not really, you had to use swapaxis() and to be honest I just used the floating holder as it was easier and did a great job.
Regarding tuning in one place, well its all done in Mach, the plugin is in Mach ;)
I presume you have to tune your spindle loop in LinuxCNC?
Now did I need to use the PLC, nope, CSMIO has a load of I/O and lots of expansion capability if required. So why did I use a PLC, well two reasons,
1. I had one sitting doing nothing.
2 I know my way around the direct logic PLCs so its easy. Doing it in Brains and/or VB would take me longer, but it could be done and has been by others.
The ladder in my PLC runs to, I think, 17 pages of A4 when printed out at 60%, lots of I/O on this changer and I wanted lots of interlocks, you can even see one working in one of the vids. The toolchange takes a lot longer to happen simply because I had such a slow accel/decel set for the test and the spindle took so long to drop to 200rpm which is the tool change rpm and the toolchange will not happen until that rpm is seen.
Even if in the future if I decide to go with LinuxCNC (about as likely as you using Mach, but never say never ;) ) then I would still use the PLC, for the simple reason I would not have to learn the ladder logic in Linux. That is assuming LinuxCNC supports Modbus?
One thing I really like about the CSMIO is the capability of using high res encoders, that means very tight control can be kept.
Hood
-
I did not know the plug-in within mach allowed for pid tuning and such - neat.
Yes often the case when trying to score points ;) should really research before you do, that is why I dont criticise EMC as I know very little about it ;)
I don't know what you mean by tuning the spindle in linuxcnc. Yes - if the spindle has its own pid loop for speed control - you would tune it within linuxcnc. Some do that. The vfd on our hmc does good enough that I never setup closed loop speed control. It isn't required for rigid tapping as the axis is slaved to the spindle. The axis just stays in sync with whatever the spindle does. (don't know if that is what you where asking..)
Well it was really just because you were saying I had to tune my spindle and it sounded like EMC magically could mind read your VFD so tuning was not required ;D But I see now the main thrust of your argument was tuning in different place from other set-up but that is now known by you to be wrong ;D
Yes tuning is not important for rigid tapping, however precise depth control then becomes trial and error, something I am sure you would soon fall into with use on your particular machine at particular spindle speeds in a particular material etc, ie just try, measure then increase/decrease in the code to suit.
I don't know how standard the plc language is in linuxcnc. I didn't have any trouble wrapping my head around it - having only some exposure in college... I am about 50/50.. I did the tool change/chain/pallets logic in ladder and the spindle gear shift logic in 'comp' (a C like language for making realtime components in linuxcnc) I could wrap my had around the 16 gear spindle programmatically vs ladder. Again - vs the brains/vb - that isn't realtime. it happens when it happens. In linuxcnc I know that the ladder is going to scan every 1ms and the realtime logic is going to happen every X ms or us..
Here you go again, real time, is it needed for a tool change? I think not. Brains are extremely fast but constrained by Mach3's update of 10Hz for non-critical I/O but if 10Hz is not fast enough for your toolchanger then I would be very surprised. And as far as "it happens when it happens" well no, it happens within the expected time frame.
Mach4's update will be much faster but Brains wont be in Mach4 ;)
with mesa hardware (5i25 with server daughter board - 6 axis + 48i/o + expandable $239) you can count to 2mhz.
Ahh come on now Sam, thats only half your war cry, tell me again, I seem to have forgotten, how much does LinuxCNC cost? ;)
2MHz is quite good, especially for hardware at that pricepoint.
Now I do need to say one thing Sam, I do respect what you have done with your machine but more importantly I respect your integrity. I recall some guy listening to John S babbling on how Mach4 wont have threading with the PP etc, so he went to Yahoo and asked Art directly, again trying to get the info he wanted to hear so he could score points.
Art replied but it was not the info he wanted to hear. The guy never posted the info from Arts reply on the homeshop forum as it would not be possible to score points.
You however reported that info and my respect for you went up considerably :) You did still manage to get a wee dig in though ;) but that is why I enjoy this banter.
Hood
-
Yes - hitting the bottom of a hole is defiantly a pain. Our spindle mass is huge so acc/decel are long. (that is one thing A axis servo spindle tapping has over this :) ) Commercial machines that actually do servo spindle tapping call it 'sync tapping' iirc
I don't have blinders on (too much) I try to be fair as much as I can be ;) (I still think linuxcnc is the bestest of course. )
You have been 'one' with mach and have made it do thing I never thought it could. (with help from external motion devices <- I am a bitch - I know...) ;) I see you always going above and beyond to help people with their setups. 'send me your xml and I will take a look' We need more experts on both sides to keep pushing the envelope.
Now back to your rigid tapping thread. You did a great job on a very very technical conversion. Keep up the great work!
sam
I did not know the plug-in within mach allowed for pid tuning and such - neat.
Yes often the case when trying to score points ;) should really research before you do, that is why I dont criticise EMC as I know very little about it ;)
I don't know what you mean by tuning the spindle in linuxcnc. Yes - if the spindle has its own pid loop for speed control - you would tune it within linuxcnc. Some do that. The vfd on our hmc does good enough that I never setup closed loop speed control. It isn't required for rigid tapping as the axis is slaved to the spindle. The axis just stays in sync with whatever the spindle does. (don't know if that is what you where asking..)
Well it was really just because you were saying I had to tune my spindle and it sounded like EMC magically could mind read your VFD so tuning was not required ;D But I see now the main thrust of your argument was tuning in different place from other set-up but that is now known by you to be wrong ;D
Yes tuning is not important for rigid tapping, however precise depth control then becomes trial and error, something I am sure you would soon fall into with use on your particular machine at particular spindle speeds in a particular material etc, ie just try, measure then increase/decrease in the code to suit.
I don't know how standard the plc language is in linuxcnc. I didn't have any trouble wrapping my head around it - having only some exposure in college... I am about 50/50.. I did the tool change/chain/pallets logic in ladder and the spindle gear shift logic in 'comp' (a C like language for making realtime components in linuxcnc) I could wrap my had around the 16 gear spindle programmatically vs ladder. Again - vs the brains/vb - that isn't realtime. it happens when it happens. In linuxcnc I know that the ladder is going to scan every 1ms and the realtime logic is going to happen every X ms or us..
Here you go again, real time, is it needed for a tool change? I think not. Brains are extremely fast but constrained by Mach3's update of 10Hz for non-critical I/O but if 10Hz is not fast enough for your toolchanger then I would be very surprised. And as far as "it happens when it happens" well no, it happens within the expected time frame.
Mach4's update will be much faster but Brains wont be in Mach4 ;)
with mesa hardware (5i25 with server daughter board - 6 axis + 48i/o + expandable $239) you can count to 2mhz.
Ahh come on now Sam, thats only half your war cry, tell me again, I seem to have forgotten, how much does LinuxCNC cost? ;)
2MHz is quite good, especially for hardware at that pricepoint.
Now I do need to say one thing Sam, I do respect what you have done with your machine but more importantly I respect your integrity. I recall some guy listening to John S babbling on how Mach4 wont have threading with the PP etc, so he went to Yahoo and asked Art directly, again trying to get the info he wanted to hear so he could score points.
Art replied but it was not the info he wanted to hear. The guy never posted the info from Arts reply on the homeshop forum as it would not be possible to score points.
You however reported that info and my respect for you went up considerably :) You did still manage to get a wee dig in though ;) but that is why I enjoy this banter.
Hood
-
Sam can you not tune your VFD in Linux? I am sure you can but maybe just not bothered?
With the CSMIO I could use a VFD and still tune it so that precise depth is possible but never had to as dont use VFDs.
Regarding the Chiron, not really that hard, just have to sit down and think about things and think of ways to do what you need.
Your Kearney would just have been the same I imagine, in fact any machine is that way, just a matter of scale of problems to solve..
Hood
-
...those snooty LinuxCNC guys. ;)
You guys know that was a joke right? I have the utmost respect and admiration for everyone from both the Linux & Mach camps.
Moreso on the Mach side 'cuz they patiently & kindly train me on an almost daily basis.
After perusing the Linux forum, I'm sure I'd get great support there too if this ol' dog wanted to learn new tricks.
Did I get that right Sam...just like you told me?;)
-
Great thread, Guys. Nice machine too, Hood.
I recently received a Novakon Torus. It can do rigid tapping too and I have been working with John at Novakon on it.
Sheetcam outputs nice rigid tapping gcode, but when I try it in standard Mach, once at the bottom of the tap, the spindle starts a few thousandths of a second before the Z starts lifting. That is a crying shame.
This machine has a servo spindle and it has a switch on the BOB.
You don't have to flip the switch each time, but I keep it switched to tapping.
It still does all other milling and drilling well too.
That switch makes it like the spindle is the A axis I think and then A and Z are synced with some code snippets that are adjustable.
I have it all working on my first test tapping run now except that I cannot get it to stop in between the drilling cycle and the tapping cycle. Mach stops at tool change fine earlier in.
Is there a code snippet I can use to force Mach to stop and stay at the tool change position?
I'll post some video when I get it doing right.
Thanks in advance.
-
Here is a code snippet, BTW.
I have different tool numbers, so Mach should stop. I have it set that way in the general settings.
N1980 (Operation: Drill, SPRING_HOLE, T12: Drill, 0.0938 in diameter, 0.28 in Deep)
N1990 S500 G00 Z0.1000
N2000 M09 (Coolant off)
N2010 M05
N2020 Z3.5000
N2030 (Drill, 0.0938 in diameter)
N2040 T12 M06 G43 H12
N2050 M08 (Flood coolant on)
N2060 S500 M03
N2070 G00 X-0.4675 Y-0.4661 Z0.1000
N2080 Z0.0197
N2090 G01 Z-0.070 F3.0
N2100 G00 Z0.1000
N2110 Z-0.0503
N2120 G01 Z-0.140
N2130 G00 Z0.0700
N2140 Z-0.1203
N2150 G01 Z-0.210
N2160 G00 Z0.0000
N2170 Z-0.1903
N2180 G01 Z-0.280
N2190 G00 Z0.1000
N2510 (Operation: Tap, TAP_1, T11: Rigid tap, 0.167 in x 32 TPI, 0.5 in Deep)
N2520 S150 G00 Z0.1000
N2530 M09 (Coolant off)
N2540 M05
N2545 Z3.5000
N2550 (Rigid tap, 0.167 in x 32 TPI)
N2610 G00 X-0.2027 Y-0.7467 Z0.1000
N2615 Z0.0394
M98 P10000
Z.5
This is the code at the end of the program.
O10000 ( S/R Tap 32TPI .5" Deep )
M49
F 200 ( Arbitrary Feed Rate USE F200 )
G91 ( Switch to Incremental Mode )
G1 Z-0.5 A16 ( Plunge 1/2", Rotate Spindle 16 Times CW )
G1 Z 0.5 A-16 ( Retract 1/2", Rotate Spindle 16 Times CCW )
G0 Z 0.1 ( Raise Spindle 0.1" )
G90
M48
M99 ( End Subroutine )
%
-
It sounds like you may have something similar to Simpsons external SwapAxis controller, in fact it may even be it as I know he did have some companies interested in it.
Why it will not stop at the toolchange I am unsure, it could be the version of Mach you are using or it could be because of the macro or plugin required for your external swapaxis controller.
I would really need to see documentation on it to see exactly what it is and what it does.
Hood
-
I'm not sure what it is.
The manuals are not ready for these machines yet. I don't recall what the labeling on the BOB said if anything.
I do have a picture I think, but not that great.
The switch is the little black slider in the lower left.
-
Looks to be some propitiatory board and without info no way to know what it is doing I am afraid.
Hood
-
I have found some documentation on some of the control parts like Drives on the Novakon site, but haven't even searched for the BOB yet.
I did manage to figure out how to get it to stop for tool change though. A simple M01 command after the move to tool change position is all that was needed. That is provided that is checked in the General config screen. :)
So all that is working and I am just filling in my offsets now, so I should be able to run a few parts fully on the machine later today.
I have been manually drilling and tapping these parts for years now. It will be nice to finally let a machine do it.
Novakon is coming out with a PDB and an ATC that I will need to save some pennies for. That would be a very nice upgrade.
Thanks, Hood.
I'll post a video when I have something.
-
The M01 wasn't what was missing.
It was the actual M06 with the tool number in front of it.
The tool number in the parentheses is what was throwing me. That don't count. ;) At least not as Gcode. ::)
-
Here is the full tool change code where the tool number is 11. Most of you guys know this.
T11 M06 G43 H11
Here is the video I promised as well. All setting are very conservative.
http://www.youtube.com/watch?v=JQJhLSyJkYM&list=HL1369656394
-
Looks to be working well for you :)
Hood
-
This looks really good, I wish I purchased the threading module when I ordered my IP-A now.
On thing I'm not clear about though, will I need to add and encoder, a tacho or both to my spindle?
Thanks
Jim
-
Encoder would be needed. Double check with CS-Lab that you can tap with a VFD but from what I understand you can.
Hood
-
thaanks hood, I'll check with cs-lab before splashing the cash