Bob La Londe
2024-05-20 00:41:14 UTC
As I might have mentioned sometime back the Hurco KMB1 has developed
backlash and become untunable in the Y axis, and mid job went from
tolerable to catastrophic backlash on the X-axis.
I had decided to scrap it out, and if it had been less work to do so it
would probably already be gone.
The thing is its got the second biggest work envelope of all the mills
in the shop, and the biggest of the CNC mills. Until I bought the South
Bend 10x54 knee mill the Hurco was the biggest work envelope mill in the
shop. Its still the heaviest mill at about 4000lbs.
I am probably going to do something stupid, but I'm doing it anyway.
I'm eliminating the servos and putting closed loop steppers on it.
Yeah, yeah, yeah... "Servos rule and steppers drool." I already have 4
absolutely massive Nema 34 closed loop steppers, and one that's even
bigger. They were for another project which I may never get to, and
well I can buy more or use the servos on it instead.
I'm eliminating the belt drive on the Y axis. I think it may have wound
up compounding the backlash on that axis. It was also sucking up some
of the usable power. It will get a direct drive. Either a flex plate
coupler or a massive helical coupler. I thought about making it direct
drive before, but the servo sticking straight out the front makes for a
big doghouse to bump into right about crotch height (depending on the
knee height). The extra heavy stepper is shorter than the servo, and if
I write the end program macro and tool change macros to move the saddle
all the way out from the column it will be mostly under the table top
enclosure anyway. I also no longer need to stand in front of the machine
doing manual work with the MPG since I have a half decent manual machine
now.
Both the X & Y axis will be getting new ball screws and new fixed end
bearings. I found a vendor with some decent quality (claimed) metric
ball screws about the right size and better backlash than Hurco claimed
for that machine in the manual. I'll have to pick a micro step count
that comes out to a nice even steps per inch, but I've done that with
metric ball screws before.
** Hurco claimed .003 or better in the manual I have. When I first got
it running I was getting much better than .001 not counting flex. For
aluminum there really was no significant flex. In steel it would run
better than .003 including flex all day, and of course I could program
multiple cleanup passes with an "MO" stop to measure and hit the stop
button or hit resume. I have to add that was total flex. I didn't
realize back then how much of that was the tool itself.
The X axis will be configured they same as the Y, but with one of the
regularly massive NEMA 34 steppers. I may need to make a new doghouse,
or I may be able to get away with just cutting that one shorter.
I don't know about the Z. Currently the backlash is good, and most of
that is taken up by the weight of the quill anyway. I am thinking if I
eliminate the belt drive on it as well I might be able to use that
recovered power lost in the belt for a little more acceleration. If I
go direct drive I will have to make a new head cover for the top since
the motor would be sticking up.
Ideally I would love to see reliable 30iss (or better) and 250-300ipm.
Yeah I ran that much with the original GeckoDrives and they kept frying
even though Gecko swore they could take it. I kept turning down the
current, rapids, and acceleration, and the Geckos still kept frying.
Later I ran the servos with an ammeter connected while adjusting the
gibbs, and they never should have been dying. The Geckos just suck.
The Dugong 160-35s have been pretty good. Acceleration initially was
insane as was rapid speed, but once I started tuning them, and reducing
following error they were little better than the Geckos... except they
didn't fry if you looked at them cross eyed like the Gecko drives.
I really like the number of error outputs on the drives. I can setup a
diagnostic bank to show me what error "exactly" is happening instead of
having to guess or go plugin the laptop with servo tuning software and
hope I can duplicate it.
Probably I'll go back to my old breakout board too. All the inputs are
settable as high or low trigger. Not so with the current BOB. I had to
use low current trigger relays for some of the senors to swap the high
and low. Blech. I'll keep the ethernet motion controller instead of
going back to the USB one. The BOB plugins for them are the same on
both boards. The current BOB also didn't tune as cleanly for 0-10 RPM
output, and forward reverse triggering. The old one was dead nuts at
every speed I ever checked it from 96RPM (3hz) to 3600 (120hz) after the
initial setup.
It's been a while. I'm probably going to have to relearn a lot of the
setup.
I'll probably be getting rid of the old Mitsubishi VFD for spindle motor
control. It tops out at 120hz. I've got a little turret lathe that
might go on. I have since learned that Hurco used to do an upgrade to
these machines to bump them to 6000 rpm (200hz) without changing the
spindles or upgrading the bearings. I put new bearings in this spindle
a while back. Not that many hours of run time though. After break-in
they run quiet without getting hot, so I guess I must have done it
right. I also have another set of new old stock original bearings in
the control cabinet. If I bump this spindle to 6000 rpm any program I
already have for the Tormach (5120 rpm) will run on this machine. I
already have a slightly larger spindle on hand I used briefly for a
companion spindle that will run up to 400hz. I've programmed these VFDs
before, so setting it to top at 200hz should be pretty easy.
It seems Mach 3 does not support rigid tapping. I had hoped to pick a
good RPM and use the Hurco for tension compression tapping (pseudo rigid
tapping). An upgrade to Mach 4 or a switch to LinuxCNC may be in order
down the road. LinuxCNC doesn't run my motion controller so that would
mean another BOB and motion control swap, but LinuxCNC (core of Tormach
Pathpilot) is really clean and has a great trajectory planner. I have
grown to really like the PathPilot software running my Tormach. The
thing is I have never successfully setup LinuxCNC before. I tried a
couple times, but went back to Mach 3 because I just knew how to set it
up. Oh. I can't run any Tormach Code file after all. I use tension
compression tapping on it all the time. Maybe copy the Tormach style
library to a Hurco library and change all the TC styles to
thread mill styles.
One thing at a time. Today I am getting the servos off and the ball
screws out for the X&Y axis.
backlash and become untunable in the Y axis, and mid job went from
tolerable to catastrophic backlash on the X-axis.
I had decided to scrap it out, and if it had been less work to do so it
would probably already be gone.
The thing is its got the second biggest work envelope of all the mills
in the shop, and the biggest of the CNC mills. Until I bought the South
Bend 10x54 knee mill the Hurco was the biggest work envelope mill in the
shop. Its still the heaviest mill at about 4000lbs.
I am probably going to do something stupid, but I'm doing it anyway.
I'm eliminating the servos and putting closed loop steppers on it.
Yeah, yeah, yeah... "Servos rule and steppers drool." I already have 4
absolutely massive Nema 34 closed loop steppers, and one that's even
bigger. They were for another project which I may never get to, and
well I can buy more or use the servos on it instead.
I'm eliminating the belt drive on the Y axis. I think it may have wound
up compounding the backlash on that axis. It was also sucking up some
of the usable power. It will get a direct drive. Either a flex plate
coupler or a massive helical coupler. I thought about making it direct
drive before, but the servo sticking straight out the front makes for a
big doghouse to bump into right about crotch height (depending on the
knee height). The extra heavy stepper is shorter than the servo, and if
I write the end program macro and tool change macros to move the saddle
all the way out from the column it will be mostly under the table top
enclosure anyway. I also no longer need to stand in front of the machine
doing manual work with the MPG since I have a half decent manual machine
now.
Both the X & Y axis will be getting new ball screws and new fixed end
bearings. I found a vendor with some decent quality (claimed) metric
ball screws about the right size and better backlash than Hurco claimed
for that machine in the manual. I'll have to pick a micro step count
that comes out to a nice even steps per inch, but I've done that with
metric ball screws before.
** Hurco claimed .003 or better in the manual I have. When I first got
it running I was getting much better than .001 not counting flex. For
aluminum there really was no significant flex. In steel it would run
better than .003 including flex all day, and of course I could program
multiple cleanup passes with an "MO" stop to measure and hit the stop
button or hit resume. I have to add that was total flex. I didn't
realize back then how much of that was the tool itself.
The X axis will be configured they same as the Y, but with one of the
regularly massive NEMA 34 steppers. I may need to make a new doghouse,
or I may be able to get away with just cutting that one shorter.
I don't know about the Z. Currently the backlash is good, and most of
that is taken up by the weight of the quill anyway. I am thinking if I
eliminate the belt drive on it as well I might be able to use that
recovered power lost in the belt for a little more acceleration. If I
go direct drive I will have to make a new head cover for the top since
the motor would be sticking up.
Ideally I would love to see reliable 30iss (or better) and 250-300ipm.
Yeah I ran that much with the original GeckoDrives and they kept frying
even though Gecko swore they could take it. I kept turning down the
current, rapids, and acceleration, and the Geckos still kept frying.
Later I ran the servos with an ammeter connected while adjusting the
gibbs, and they never should have been dying. The Geckos just suck.
The Dugong 160-35s have been pretty good. Acceleration initially was
insane as was rapid speed, but once I started tuning them, and reducing
following error they were little better than the Geckos... except they
didn't fry if you looked at them cross eyed like the Gecko drives.
I really like the number of error outputs on the drives. I can setup a
diagnostic bank to show me what error "exactly" is happening instead of
having to guess or go plugin the laptop with servo tuning software and
hope I can duplicate it.
Probably I'll go back to my old breakout board too. All the inputs are
settable as high or low trigger. Not so with the current BOB. I had to
use low current trigger relays for some of the senors to swap the high
and low. Blech. I'll keep the ethernet motion controller instead of
going back to the USB one. The BOB plugins for them are the same on
both boards. The current BOB also didn't tune as cleanly for 0-10 RPM
output, and forward reverse triggering. The old one was dead nuts at
every speed I ever checked it from 96RPM (3hz) to 3600 (120hz) after the
initial setup.
It's been a while. I'm probably going to have to relearn a lot of the
setup.
I'll probably be getting rid of the old Mitsubishi VFD for spindle motor
control. It tops out at 120hz. I've got a little turret lathe that
might go on. I have since learned that Hurco used to do an upgrade to
these machines to bump them to 6000 rpm (200hz) without changing the
spindles or upgrading the bearings. I put new bearings in this spindle
a while back. Not that many hours of run time though. After break-in
they run quiet without getting hot, so I guess I must have done it
right. I also have another set of new old stock original bearings in
the control cabinet. If I bump this spindle to 6000 rpm any program I
already have for the Tormach (5120 rpm) will run on this machine. I
already have a slightly larger spindle on hand I used briefly for a
companion spindle that will run up to 400hz. I've programmed these VFDs
before, so setting it to top at 200hz should be pretty easy.
It seems Mach 3 does not support rigid tapping. I had hoped to pick a
good RPM and use the Hurco for tension compression tapping (pseudo rigid
tapping). An upgrade to Mach 4 or a switch to LinuxCNC may be in order
down the road. LinuxCNC doesn't run my motion controller so that would
mean another BOB and motion control swap, but LinuxCNC (core of Tormach
Pathpilot) is really clean and has a great trajectory planner. I have
grown to really like the PathPilot software running my Tormach. The
thing is I have never successfully setup LinuxCNC before. I tried a
couple times, but went back to Mach 3 because I just knew how to set it
up. Oh. I can't run any Tormach Code file after all. I use tension
compression tapping on it all the time. Maybe copy the Tormach style
library to a Hurco library and change all the TC styles to
thread mill styles.
One thing at a time. Today I am getting the servos off and the ball
screws out for the X&Y axis.
--
Bob La Londe
CNC Molds N Stuff
--
This email has been checked for viruses by AVG antivirus software.
www.avg.com
Bob La Londe
CNC Molds N Stuff
--
This email has been checked for viruses by AVG antivirus software.
www.avg.com