I've been replacing the controller and upgrading the Novamill to LinuxCNC - making good progress and has been fairly straight forward. I've got two other machines to do the same thing to, so will use this thread as much to document it but also show my approach too.
Onwards!
LinuxCNC and MESA Upgrade
Moderators: Martin, Steve, Mr Magoo
-
madebymccoy
- Posts: 27
- Joined: Tue 13 Dec , 2016 23:36 pm
- Hardware/Software: Novamill
- in process of upgrading controller to Mesa I/O and LinuxCNC
- retrofitting head with power drawbar
- and next year, raise column by 70mm
Microrouter V3
Micromill 2000
-
madebymccoy
- Posts: 27
- Joined: Tue 13 Dec , 2016 23:36 pm
- Hardware/Software: Novamill
- in process of upgrading controller to Mesa I/O and LinuxCNC
- retrofitting head with power drawbar
- and next year, raise column by 70mm
Microrouter V3
Micromill 2000
Re: LinuxCNC and MESA Upgrade
New cable loom to plug into the Eurocard 96-way plug.
TE connector housing with TE crimped connectors. Printed heatshrink (yup, have the machine for that) and will be using mesh wrap once the loom is complete.
The limit switches need rewiring as the MESA is sinking not sourcing, but that's an easy job of a few DIN rail terminals.
TE connector housing with TE crimped connectors. Printed heatshrink (yup, have the machine for that) and will be using mesh wrap once the loom is complete.
The limit switches need rewiring as the MESA is sinking not sourcing, but that's an easy job of a few DIN rail terminals.
- IMG_7700.PNG (1.47 MiB) Viewed 7121 times
- IMG_7699.PNG (1.8 MiB) Viewed 7121 times
- IMG_7698.PNG (1.9 MiB) Viewed 7121 times
-
madebymccoy
- Posts: 27
- Joined: Tue 13 Dec , 2016 23:36 pm
- Hardware/Software: Novamill
- in process of upgrading controller to Mesa I/O and LinuxCNC
- retrofitting head with power drawbar
- and next year, raise column by 70mm
Microrouter V3
Micromill 2000
Re: LinuxCNC and MESA Upgrade
So, wiring loom fitted...
One snag, I found the 96-way plug on the Baldor had badly soldered pins so the connector wouldn't go on; some pins higher than others and some at angles. I ended up removing the board and heat sink and removing the board from the heat sink to straighten and get even pressure behind the PCB to avoid damaging the solder joints or PCB itself. Connector went on easily enough...
All hooked up, with the MESA 7i76 stepper I/O board fitted to DIN rail mounted in the position the original spindle isolator board was located. The 7i76 features and opto-isolated 10V analogue output, so no intermediary board is needed; making space and reducing complexity.
Best step is to fit some DIN rail terminals to modify the limit switches and E-stop. Then final task is the spindle 10v analogue signal cable plus the guard interlock and spindle "go" signal.
One snag, I found the 96-way plug on the Baldor had badly soldered pins so the connector wouldn't go on; some pins higher than others and some at angles. I ended up removing the board and heat sink and removing the board from the heat sink to straighten and get even pressure behind the PCB to avoid damaging the solder joints or PCB itself. Connector went on easily enough...
All hooked up, with the MESA 7i76 stepper I/O board fitted to DIN rail mounted in the position the original spindle isolator board was located. The 7i76 features and opto-isolated 10V analogue output, so no intermediary board is needed; making space and reducing complexity.
Best step is to fit some DIN rail terminals to modify the limit switches and E-stop. Then final task is the spindle 10v analogue signal cable plus the guard interlock and spindle "go" signal.
- IMG_7708.PNG (2.15 MiB) Viewed 7109 times
- IMG_7707.PNG (2.14 MiB) Viewed 7109 times
-
madebymccoy
- Posts: 27
- Joined: Tue 13 Dec , 2016 23:36 pm
- Hardware/Software: Novamill
- in process of upgrading controller to Mesa I/O and LinuxCNC
- retrofitting head with power drawbar
- and next year, raise column by 70mm
Microrouter V3
Micromill 2000
Re: LinuxCNC and MESA Upgrade
Some design/redesign thoughts...
The existing stepper drivers are okay as is the spindle driver for initial work. Long term I'll put some better drivers in and likely even replace the steppers - but better to get the machine running and decide if it's worth the time and money!
By keeping the old stepper drivers, it means interfacing with the Baldor's 96 pin eurocard connector. Easy enough thanks to info about - though I will post my pin outs shortly. I originally planned to use the same connector for the limit and e-stop, but found that the Baldor is for a sourcing interface and the MESA is sinking, so it'll be faster to just hookup the negative (58) to 12vdc from the MESA. I'll keep the ceramic capacitors, as they're there for a good reason! EM noise can play havoc with the limits, so best to keep the ceramic caps in place.
Likewise, similar story with the spindle driver. It's not great (its single direction), but it's a starting point. The manufacturer does make a reversing board for this model, but I suspect a VFD with closed loop speed control may be better if I ever want to machine threads etc. No rush for now.
The MESA features a wonderful 0-10V opto-ioslatoed analogue interface that acts like a potentiometer. There's a "min", "max" and a "wiper" pin, so this totally does away with the need to use a 0-10V interface board, and works with the existing spindle driver without any effort. The MESA has fwd/rev pins, so I could use the reversing board, plus an enable pin as is also required - all in all, a very good choice of IO board.
Whilst it's doable to use a parallel port and MACH3, they're just kludgy and less than an elegant solution. LinuxCNC offers much more flexibility, plus the MESA board protects the electronics and the host computer, which is kind of handy! I won't go into the LinuxCNC vs MACH3 debate, but suffice to say, having evaluated my needs, wish lists and plans, I feel LinuxCNC with a MESA IO board fits perfectly.
I've already had the machine running and roughly dialed in, using a temporary cable loom and a breadboard, so I'm happy with the wiring and method of connectivity. That's something I'd advise - do a rough test, and document it well! Then make up cable looms accordingly, otherwise you'll just have a rats nest and it'll make it really difficult to diagnose problems.
Something I have found, and suspect why the machine was "disposed of" to start with, is an issue with the X limit switch. Somewhere, it's going to ground it seems - the original control board just came up with an "L" warning (limit switches open), and when I hooked up to LinuxCNC, it always shows the X limit as open. Not a massive issue, it's easily repairable, but it justifies spending time and effort labelling, wiring neatly, using terminals and staying far, far away from "choc blocks" and solder splices. The grey plastic trunking in a control panel is to route wires, it's not to hide unmarked inline joints badly wrapped in PVC tape!
The existing stepper drivers are okay as is the spindle driver for initial work. Long term I'll put some better drivers in and likely even replace the steppers - but better to get the machine running and decide if it's worth the time and money!
By keeping the old stepper drivers, it means interfacing with the Baldor's 96 pin eurocard connector. Easy enough thanks to info about - though I will post my pin outs shortly. I originally planned to use the same connector for the limit and e-stop, but found that the Baldor is for a sourcing interface and the MESA is sinking, so it'll be faster to just hookup the negative (58) to 12vdc from the MESA. I'll keep the ceramic capacitors, as they're there for a good reason! EM noise can play havoc with the limits, so best to keep the ceramic caps in place.
Likewise, similar story with the spindle driver. It's not great (its single direction), but it's a starting point. The manufacturer does make a reversing board for this model, but I suspect a VFD with closed loop speed control may be better if I ever want to machine threads etc. No rush for now.
The MESA features a wonderful 0-10V opto-ioslatoed analogue interface that acts like a potentiometer. There's a "min", "max" and a "wiper" pin, so this totally does away with the need to use a 0-10V interface board, and works with the existing spindle driver without any effort. The MESA has fwd/rev pins, so I could use the reversing board, plus an enable pin as is also required - all in all, a very good choice of IO board.
Whilst it's doable to use a parallel port and MACH3, they're just kludgy and less than an elegant solution. LinuxCNC offers much more flexibility, plus the MESA board protects the electronics and the host computer, which is kind of handy! I won't go into the LinuxCNC vs MACH3 debate, but suffice to say, having evaluated my needs, wish lists and plans, I feel LinuxCNC with a MESA IO board fits perfectly.
I've already had the machine running and roughly dialed in, using a temporary cable loom and a breadboard, so I'm happy with the wiring and method of connectivity. That's something I'd advise - do a rough test, and document it well! Then make up cable looms accordingly, otherwise you'll just have a rats nest and it'll make it really difficult to diagnose problems.
Something I have found, and suspect why the machine was "disposed of" to start with, is an issue with the X limit switch. Somewhere, it's going to ground it seems - the original control board just came up with an "L" warning (limit switches open), and when I hooked up to LinuxCNC, it always shows the X limit as open. Not a massive issue, it's easily repairable, but it justifies spending time and effort labelling, wiring neatly, using terminals and staying far, far away from "choc blocks" and solder splices. The grey plastic trunking in a control panel is to route wires, it's not to hide unmarked inline joints badly wrapped in PVC tape!
-
madebymccoy
- Posts: 27
- Joined: Tue 13 Dec , 2016 23:36 pm
- Hardware/Software: Novamill
- in process of upgrading controller to Mesa I/O and LinuxCNC
- retrofitting head with power drawbar
- and next year, raise column by 70mm
Microrouter V3
Micromill 2000
Re: LinuxCNC and MESA Upgrade
New block of terminals to terminate the home limits, e-stop and guard loops. All seems well, but this X axis limit issue is persistent.
I traced the length of the X home limit wiring and tested continuity at every stage, without any issues. It seems to only flake out when the panel is powered, indicating maybe severe noise or damage somewhere. The switch is open all the time; when powered off, continuity tests show the switch and wiring to work. Bizzare.
I've got to run a 3-core shielded cable for the analogue spindle control, and then I'm done in the "direct" conversion.
Playing with some controls, I'm not entirely happy with the quality of feedback from the interlocks. At present there's no feedback for the spindle, so if the guard is opened whilst there's feedback for that, LinuxCNC won't actually know if the spindle is rotating or not. Since opening the guard kills power to the spindle, LinuxCNC will need that logic to automatically pause the operation, which I don't entirely like, as the lag may cause issues. Would much rather the guard be soft interlocked, so LinuxCNC stops the motor, vs power being cut. Handy so chips etc can be blown away without having to worry about pausing the program (the guard will initiate pause which in turn stops the motor vs the guard killing the motor and LinuxCNC only knowing "because" the guard is open). Or maybe that's dumb. Either way, current logic will prevail. The MESA has lots and lots of IO, so I'll be hooking up a touch probe and a tool setting probe, tool release valve (for power drawbar conversion), maybe even a multicolour beacon for visual status of the machine. Also want mist coolant (I think flood is pushing it - how can I drain it without coolant getting under the machine or through the chip covers and have it flowing out the rear pulley cover?).
Anyway, all working sans X axis. May have to run a fresh X home shielded cable if it's damage or noise.
I traced the length of the X home limit wiring and tested continuity at every stage, without any issues. It seems to only flake out when the panel is powered, indicating maybe severe noise or damage somewhere. The switch is open all the time; when powered off, continuity tests show the switch and wiring to work. Bizzare.
I've got to run a 3-core shielded cable for the analogue spindle control, and then I'm done in the "direct" conversion.
Playing with some controls, I'm not entirely happy with the quality of feedback from the interlocks. At present there's no feedback for the spindle, so if the guard is opened whilst there's feedback for that, LinuxCNC won't actually know if the spindle is rotating or not. Since opening the guard kills power to the spindle, LinuxCNC will need that logic to automatically pause the operation, which I don't entirely like, as the lag may cause issues. Would much rather the guard be soft interlocked, so LinuxCNC stops the motor, vs power being cut. Handy so chips etc can be blown away without having to worry about pausing the program (the guard will initiate pause which in turn stops the motor vs the guard killing the motor and LinuxCNC only knowing "because" the guard is open). Or maybe that's dumb. Either way, current logic will prevail. The MESA has lots and lots of IO, so I'll be hooking up a touch probe and a tool setting probe, tool release valve (for power drawbar conversion), maybe even a multicolour beacon for visual status of the machine. Also want mist coolant (I think flood is pushing it - how can I drain it without coolant getting under the machine or through the chip covers and have it flowing out the rear pulley cover?).
Anyway, all working sans X axis. May have to run a fresh X home shielded cable if it's damage or noise.
- IMG_7774.PNG (1.85 MiB) Viewed 7094 times
- IMG_7773.PNG (2.05 MiB) Viewed 7094 times