Mazak Impulse - Oak retrofit build log

All things related to Centroid Oak, Allin1DC, MPU11 and Legacy products

Moderator: cnckeith

ashesman
Posts: 395
Joined: Thu Dec 03, 2020 4:54 am
Acorn CNC Controller: No
Allin1DC CNC Controller: No
Oak CNC controller: Yes
CNC Control System Serial Number: none
DC3IOB: No
CNC12: Yes
CNC11: No
CPU10 or CPU7: No

Re: Mazak Impulse - Oak retrofit build log

Post by ashesman »

So, it been a long couple of months. I have managed to scrape up a few hours a week plus the remaining parts to complete my control board. I finally have a set of motor drives that appear they are going to work after quite some trial and error.

My goal has always been to try do my retrofit non invasively and finish up with a machine that performs as if not better than it did on the old Mazak controller. To be honest it is going to be hard to beat that old controllers motor drive capabilities. But there are plenty of other advantages of the upgrade that make it worth while!

Anyways, below are some pictures of the control panel. Today I fitted it to the machine and connected up most of the wires. Depending on how bad the storm is tonight I should be able to get it running tomorrow.

The control panel is on a door behind the machine so this presents some challenges in terms of clearance as the door opens and strain relief of the wires.

I started by mocking things up on a piece of wood. Note the first dinky little spindle VFD! Sorry about the sideways pictures, the forum seems to put them all in landscape. If you click them they are the right way up.
20210222_203436.jpg
Then on a sheet of 6mm aluminium I constructed the necessary cable entry brackets. A 6mm sheet was used so I had enough thickness to drill and tap into it as it needs to be screwed hard up against the door. For all the main components I fitted countersunk screws from behind so they made little studs.
20210331_211227.jpg
After spending way more time than I am willing to admit, I eventually managed to crunch everything onto the panel. Parts are three deep in places!
Everything was a compromise. Things crammed closer together than I would have liked. Not sure yet if interference is going to rear its ugly head! I fitted the OEM heatsinks to the backside of the panel and behind the DMM drives. This was mainly so it looked like it used to. It may also help with a bit of cooling! Note the final VFD size, even this one doesn't have a high enough peak current rating but was all I could fit.
20210527_211905.jpg
20210527_212623.jpg
I eventually decided to mount the PC and its power supply under the OAK as I don't want it up on the control panel.
20210515_185415.jpg
I manhandled the door back onto the machine. It is much heavier than it used to be!!! First thing I did was close it to make sure everything cleared. No problems.
20210527_213828.jpg
Then a full day today joining up wires. I am slow at wires! All that is left now is the coolant pump which is currently too short and the main power supply which I couldn't work out how to do nicely yet!
20210529_175349.jpg
slodat
Posts: 793
Joined: Thu Apr 12, 2018 11:16 pm
Acorn CNC Controller: Yes
Allin1DC CNC Controller: No
Oak CNC controller: No
CNC Control System Serial Number: none
DC3IOB: No
CNC11: No
CPU10 or CPU7: No

Re: Mazak Impulse - Oak retrofit build log

Post by slodat »

How did your spindle stuff end up? Keeping the original motor?
ashesman
Posts: 395
Joined: Thu Dec 03, 2020 4:54 am
Acorn CNC Controller: No
Allin1DC CNC Controller: No
Oak CNC controller: Yes
CNC Control System Serial Number: none
DC3IOB: No
CNC12: Yes
CNC11: No
CPU10 or CPU7: No

Re: Mazak Impulse - Oak retrofit build log

Post by ashesman »

Yup. Kept all original motors. I have literally done zero mechanical disassembly.

The Hitachi SJ-P1 can drive it nicely. It cant quite reach the 0 to 6000 rpm in 0.34 seconds that the original controller did or slow it quite as fast but I can live with that. I just hope I can get it to do spindle orientation nicely. The drive has an orient function built in but doesnt work on PMM motors.

For any one reading in the future, to achieve full acceleration and deceleration the 1.5kW spindle motor needs about 65A peak and around 20 ohm braking resistor. You will need a motor drive of around 5.5kW capacity to achieve this. I used a 3.5kW drive as that was all I have space for.
martyscncgarage
Posts: 9912
Joined: Tue Mar 28, 2017 12:01 pm
Acorn CNC Controller: Yes
Allin1DC CNC Controller: Yes
Oak CNC controller: No
CNC Control System Serial Number: none
DC3IOB: No
CNC12: Yes
CNC11: Yes
CPU10 or CPU7: Yes
Location: Mesa, AZ

Re: Mazak Impulse - Oak retrofit build log

Post by martyscncgarage »

Very nice clean wiring on your control panel ashesman!
Marty
Reminder, for support please follow this post: viewtopic.php?f=20&t=383
We can't "SEE" what you see...
Mesa, AZ
ashesman
Posts: 395
Joined: Thu Dec 03, 2020 4:54 am
Acorn CNC Controller: No
Allin1DC CNC Controller: No
Oak CNC controller: Yes
CNC Control System Serial Number: none
DC3IOB: No
CNC12: Yes
CNC11: No
CPU10 or CPU7: No

Re: Mazak Impulse - Oak retrofit build log

Post by ashesman »

It was a long weekend and rained for most of it so I spent the best part of three full days working on my machine. I had some highs and some lows along the way. But anyways, here is a progress update.

I spent a lot of time doing the finishing touches on PLC code and PC configuration. A big part of the job was configuring the jog panel custom buttons.
20210607_152240.jpg
Here is a summary of button functions:
  • The Tool L and Tool R buttons extend and retract each tool rack. They are interlocked and each button lights while the rack is out.
  • Manual Tool will run a macro to allow performing a manual tool change. Not complete yet.
  • Aux 1 and 2 have no use yet.
  • Home sends the machine to home position if you hold it down for a few seconds and lights up when at the home position.
  • Set X0, Y0 and Z0 use Aux functions to allow quickly setting part origin. Aux 3 does the final data input. I didn't know it needed an extra key to finalise input so didn't get one added! Ideally each button will light when at current WCS home but I haven't worked out how to do that yet.
  • WCS/MCS switches between machine and work coordinates by sending an Alt+D key press to CNC12.
  • F8 sends an F8 key press to CNC12. Allows switching between G code and graph when running a program.
  • The blank blue key was spare but now brings CNC12 to the front when pressed. Super handy when another window is covering CNC12.
Capture111.PNG
In case you are quick enough to work out that the last three button functions cant be done in PLC code, I achieved this by writing a PC application that will run as a service in the background. It uses the Centroid Skinning DLL to read memory bits from the PLC and Windows API calls to focus and send key presses to CNC12. Took a few hours and works mint.

The next mission was a full day of battling with motor drives and tuning. Also diagnosing why the DYN4 drives would not reset faults. I ended up at half the autotuned gains to get the drives to not fault on high speed moves. I got to a point where motion seemed reliable but I am not at all happy with it. The gains need to be low causing the drives to not be as rigid at stationary as they should be. Any higher and they fault on full speed moves.

I was able to get to a point where I could do all the setup finishing jobs. The self test failed as index pulse is too close. I need to move the home dogs anyway so didn't bother rotating the motors on the lead screws yet.
20210607_152150[1].jpg
I spent some time checking the backlash. I could have done with a finer dial gauge I think but it is pretty close. 0.01mm (~3.5 tenths) backlash on X and X and effectively 0 on Z as is always loaded. Backlash seems consistent over most of the travel. Not bad for an old machine.
20210607_152240[1].jpg
I had been intending on using a 17" touch screen mounted vertically to run CNC12. I wasn't sure how this would work out but after using it for the last few days I am sold on it. CNC12 runs in the upper half of the screen so there is room to edit G code below, display a setup sheet or even run Fusion 360! I bought a grunty enough PC so can run CAM or CAD software on it if need be but I don't think I will. The next major job is to build a user console to mount the display and jog panel on the front of the machine. There is not a lot of room and I want to keep it in the old style of the machine so will be a sheet metal challenge for me!

Overall I am finally making progress now but to be completely honest I am pretty disappointed with it all. I tried really hard to end up with a better machine than I had but it just isn't happening! The Centroid part is OK. Besides CNC12 crashing a few times :? the rest of it has been pretty good. The CNC12 user interface leaves a lot to be desired and offers little functionality over my 1995 controller! But I knew that before I started.

The big gripe I have is with motor drives. The motion control on this machine was flawless before. Rigid, fast and reliably precise to within 0.001mm every time. You said move 0.001 mm and it did and would sit exactly on that number. The DYN4 drives just don't seem to be up to the job. They constantly flutter back and forth 0.001 to 0.003mm. If you jog in 0.001 steps they move in jerks of 0.003. You can push the machine head and see it move on the DRO before the DYN4s catch up and correct it. And if you have high enough gain to get a more solid control, they graunch while moving at full rapid speed and then fault. Then cant be reset!!!

The Hitachi SJ-P1 spindle drive also turned out to be a fail. I have it setup and tuned now so can accelerate 0-6000RPM in 0.5s and back to zero in 0.35s. It will go full speed to reverse in fractions of a second (~0.25s at 2000 RPM). However, the motor drive has something weird in its control that causes it to over current fault if you try and start or reverse the motor at below 1500 RPM. I have tried fiddling every setting! The big cockup here was I assumed it could do closed loop control and orientation with an encoder on PMM motors as it said so in the manual. It turns out that it also says in the fine print that it cant use an encoder on PMM motors!!!

So, I am kind of f#$ked at the moment as I cant reliably get the spindle to start at slow speed and looks like it will never be able to orient correctly which means tool changing is out of the question! I will get there though. I haven't given up yet...

Thanks for reading if you made it this far!
martyscncgarage
Posts: 9912
Joined: Tue Mar 28, 2017 12:01 pm
Acorn CNC Controller: Yes
Allin1DC CNC Controller: Yes
Oak CNC controller: No
CNC Control System Serial Number: none
DC3IOB: No
CNC12: Yes
CNC11: Yes
CPU10 or CPU7: Yes
Location: Mesa, AZ

Re: Mazak Impulse - Oak retrofit build log

Post by martyscncgarage »

When you get an index pulse too close to switch, I simply remove the servo, turn it a half turn and put it back in, OR move the home switch tripper.
You say "CNC12 Crashing a few times" please clarify that statement and post a fresh report.
When the drive faults, what does DMMDRV report? Have you had a conversation with DMM about this particular issue?
All axis faulting or just one particular axis?

DMM DYN4 drives are OK for basic machines. I had good luck with them on a Fadal TRM. But I only pushed it to 100IPM
What rapids did you have before and what is the best the DMM drives are doing now? What size motors did you have before and what did you buy from DMM? Did you consult with them on sizing?
When do they fault upon acceleration or reversing?
You might post a video of what it is doing to YouTube and share a link here and with DMM.
ALSO post a FRESH REPORT.
Reminder, for support please follow this post: viewtopic.php?f=20&t=383
We can't "SEE" what you see...
Mesa, AZ
ashesman
Posts: 395
Joined: Thu Dec 03, 2020 4:54 am
Acorn CNC Controller: No
Allin1DC CNC Controller: No
Oak CNC controller: Yes
CNC Control System Serial Number: none
DC3IOB: No
CNC12: Yes
CNC11: No
CPU10 or CPU7: No

Re: Mazak Impulse - Oak retrofit build log

Post by ashesman »

martyscncgarage wrote: Mon Jun 07, 2021 5:50 pm When you get an index pulse too close to switch, I simply remove the servo, turn it a half turn and put it back in, OR move the home switch tripper.
You say "CNC12 Crashing a few times" please clarify that statement and post a fresh report.
When the drive faults, what does DMMDRV report? Have you had a conversation with DMM about this particular issue?
All axis faulting or just one particular axis?

DMM DYN4 drives are OK for basic machines. I had good luck with them on a Fadal TRM. But I only pushed it to 100IPM
What rapids did you have before and what is the best the DMM drives are doing now? What size motors did you have before and what did you buy from DMM? Did you consult with them on sizing?
When do they fault upon acceleration or reversing?
You might post a video of what it is doing to YouTube and share a link here and with DMM.
ALSO post a FRESH REPORT.
I didn't rotate the motors on the shafts yet as need to move the home switches to the limit switch positions as there is an area past them where the machine could stop then fail to home as already past the switches.

I took a photo of CNC12 crashing last time. It was an unhandled exception error. I later deleted the photo as I have support cases open with Centroid, DMM and Hitachi so didn't need to be complaining about that as well!

The drives fault on LOST_PHASE. This seems to happen when they reach a constant speed rather than during acceleration. They never fail any other time. Reducing the gains has almost eliminated the problem but now control is weak! All axis have the fault. I have asked DMM for help but not sure if they can offer much as the drives have only three settings so what can I change to fix it!

DMM were given all the information on the machine, the motors and the feed rates I wanted to achieve. They are 750W Panasonic. DMM seemed to think it was not an issue. The motors are capable of pushing the axis easily at 40m/min (1575 IPM). The old controller was limited at 24 m/min (945 IPM = 2400 RPM) and I am trying to run at that. Currently much over 15 m/min (590 IPM) is unreliable.

Report is attached.
Attachments
report_0008DC111213-1214201066_2021-06-08_19-23-11.zip
(994.12 KiB) Downloaded 137 times
ashesman
Posts: 395
Joined: Thu Dec 03, 2020 4:54 am
Acorn CNC Controller: No
Allin1DC CNC Controller: No
Oak CNC controller: Yes
CNC Control System Serial Number: none
DC3IOB: No
CNC12: Yes
CNC11: No
CPU10 or CPU7: No

Re: Mazak Impulse - Oak retrofit build log

Post by ashesman »

After taking a day off to sulk about the state I was at in my last update and send a few tech support emails, I got back into it again. Another at least 10 hours have been spent on the Hitachi SJ-P1 spindle drive. I had completely written it off and had my ABB sales rep friend over this morning giving me heaps for not just using an ACS 880 (which in hindsight I should have!).

BUT, I had a win tonight and managed to get it working beautifully. Well, so far so good. I still have some work to do yet but it is finally looking like the SJ-P1 might stay. Along with a ton of other fine tuning details, the major key was getting the IMPE (initial motor pole estimation) function dialed and increasing the no load current. Without closed loop from encoder which is completely not an option for a PMM motor on the SJ-P1, some sort of starting function must be used. The drive offers a few options. I chose IMPE, it lets out a little squeak and delays starting a little but beggars cant be choosers! Here is a video of it starting, stopping and reversing up to 6000 RPM. I would say this is 70% as good as the original drive.



I had a quick play with rigid tapping to see what was going to happen. Looks like it is going to work. The rigid tapping function seems to momentarily disable the spindle at the bottom which is a bit unnecessary but I couldn't find a parameter for that! Here is a video showing tapping at 1000 RPM. 1.25mm thread pitch. The squeak is the motor drive doing its position estimation. It sounds worse in the video than in real life! The feed rate doesn't look fast enough to me in the video, probably OK though!



Tonight's mission is to write a PID controller to do spindle orientation. I hope the drive will handle it!

I am still having grief with the stupid DYN4s though. I can tune each one to nicely do rapid traversal of each axis over and over without any faults. Put them together in a program and always faults with lost phase and one time over current. Faulting only seems to happen at full rapid speed. My suspicion is that these drives are too pussy!
tblough
Posts: 3072
Joined: Tue Mar 22, 2016 10:03 am
Acorn CNC Controller: Yes
Allin1DC CNC Controller: Yes
Oak CNC controller: Yes
CNC Control System Serial Number: 100505
100327
102696
103432
7804732B977B-0624192192
DC3IOB: No
CNC12: Yes
CNC11: No
CPU10 or CPU7: No
Location: Boston, MA
Contact:

Re: Mazak Impulse - Oak retrofit build log

Post by tblough »

My guess about the delay at the bottom is you used Intercon to program the rigid tap cycle snd didn't set the pause at the bottom of the hole to 0.
Cheers,

Tom
Confidence is the feeling you have before you fully understand the situation.
I have CDO. It's like OCD, but the letters are where they should be.
ashesman
Posts: 395
Joined: Thu Dec 03, 2020 4:54 am
Acorn CNC Controller: No
Allin1DC CNC Controller: No
Oak CNC controller: Yes
CNC Control System Serial Number: none
DC3IOB: No
CNC12: Yes
CNC11: No
CPU10 or CPU7: No

Re: Mazak Impulse - Oak retrofit build log

Post by ashesman »

tblough wrote: Sat Jun 12, 2021 8:01 am My guess about the delay at the bottom is you used Intercon to program the rigid tap cycle snd didn't set the pause at the bottom of the hole to 0.
I definitely set the intercon delay to zero. And reduced the parameters that control rigid tapping, but I guess not by enough! I will check parameter 69 and 82. Also I found that parameter 36, bit 5 can be set to zero to not turn off the spindle when at the bottom of the hole. I will play with those settings at some stage but for now I need to get spindle orient and tool change working!
Capture8.PNG
Post Reply