I'm gradually becoming seduced by the idea of CNC on a mill. The
question is whether to convert my Tom Senior M1 to CNC, or whether to
buy a KX1 or a KX3 - to be honest I like the idea of a new toy!
- Has anyone here got any opinions on the KX1 or KX3 compared to an M1?
- Has anyone in the South got a KX1 or KX3 running that I could have a
look at (within striking-range of Winchester)?
- How easy/hard is it to convert an M1 to CNC? (I'll ask in the TS
owner's group too)
- Never having used CNC before, if I bought a KX3 would I want to keep
the M1 for manual work or is it reasonable to plan to do everything with
First off, YMMV, however...
I decided to go down the "convert a decent old fashioned mill" route
for several reasons, I know I'm starting with a quality machine, I
know I can spec EVERY single component to my own desires, and perhaps
most of all because I think that having done a conversion I will be
more able to understand and therefore use the thing properly...
I also think it works out a lot cheaper, cost per quality wise.
Motion Control Products for your steppers and drivers
The rest of it you can make easy enough, but if you'd rather buy then
Reliance Precision Mechatronics for things like couplings and belts
and pulleys and ballscrews....
I tell you one HUGE benefit, you already know by sound and feel and
all that good stuff how all your tooling and your mill sounds when
things are optimal and when they are not, you get all that for free
when doing a conversion.
Any questions ask away, or if you're in the southwest you can always
drop by and nose around.
Cost wise I can convert both the mill and lathe to CNC for less than
400 quid all in, and that is all quality branded industrial products
made to spec, no fleabay stuff, no chinese stuff, none of that.
When I was initially researching this I asked arc for a graph of
stepper torque vs rpm, they couldn't help, I rang MCP and spoke to the
bod there and he just rattled it all off, plus they had a far greater
range, plus I found that they were actually cheaper than arc or anyone
else, most of all my decision was based on the fact that MCP don't do
anything but steppers and servos, YMMV
Just checked and on the smaller drive MCP don't have the same as Arc,
theirs is 2.5 amp at =A324.10 + vat =3D =A328.31
Arc's is 3 amp at =A328.95 including
On the larger drive, again both Leadshine drives from China
MCP is =A370.66 + vat =3D 83.03
Arc is =A367.00 inclusive.
I only know this as I have just had a machine and some bits delivered
for a conversion and the customer was going to go to MCP but went to
arc instead and saved himself =A364 just on the drivers.
I'll let you into a small secret, regardless of what's printed on the
data sheets BOTH these drivers carry the same Leadshine part number
and come from the same factory.
The MCP ones state 68 volts typical, which is where the Arc ones will
be at 80 volts with some headroom.
Signal power is also roughly the same 10 ma compared to 12 ma because
they are the same driver.
Who's worried about 2 ma signal power anyway? a decent breakout board
will take care of that.
I also do not want to be putting words in anyones mouths or
denigrating one supplier with respect to another here but just point
out that both products ARE the same.
It's then up to the individual to deal with whoever they want.
Couplings or HTD belt pulleys can be bought cheaper believe it or not
from RS Components, they can beat anyone else hands down, I know they
can be expensive on some items but on these that are good.
Cheap ballscrews can be bought from Moore.
Watch the accuracy though as to where you buy.
This 0.003" per foot for some companies means it WILL be out 0.003"
per foot, some like NSK or the other big makes will list this as a
So buying a class 5 from a cheap supplier will mean it's a class 5,
from THK or NSK, chances are you will get close to a class 3.
How did you solve the problems of backlash and leadscrew wear? I've got
new nuts(!) but there's always going to be some play/wear.
I've read that ball leadscrews aren't a good idea for CNC/manual mills
because when used manually there isn't enough resistance - but I'm not
convinced that wouldn't be insurmountable by either programming
resistance via the steppers or adding a friction device - - any thoughts
Backlash and leadscrew (wear) mapping is built into Mach3.
well, you have to really lock down the axes when manual milling.
Yeah, ballscrews are bloody expensive by the time you've bought decent
ballnuts and bearings, and the fact is a ballscrew will not suddenly
make your machine more accurate, they reduce stiction and friction to
minimal amounts but that doesn't actually affect accuracy and
You can always retrofit ballscrews later, if you think it is worth
I've just scanned through the manual and found that backlash can be
programmed so that it "knows" how far to back-up, but I think this
disables corners and a couple of other functions - so maybe ballscrews
will be needed.
I couldn't find anything about leadscrew mapping, where is this documented?
Whilst scanning I noticed that glass DROs can be connected - a nice feature!
...and unless you go for antibacklash ballnuts, they won't remove
The big "plus" with ballscrews over and above the increased efficiency
is that they wear much more slowly than conventional threads, so the
accuracy, repeatability, etc. changes very slowly over time compared
with conventional threads.
To be honest I'm still learning the software and configuring the
system, so I can't point you at official documentation for each
feature that I have found and played with...
Agreed about the DRO input, in my case it makes my playing an
iterative game of pinning the tail on the donkey.
Oh come on, ballscrews DO wear and suffer other forms of damage, and
my leadscrews are at present 30 years old and yet by "coming back" on
backlash I could approach a thou manually, with the DRO considerably
Sure, ballscrews wear LESS than trapezoidal by definition because
there is less friction and stiction, but "less" is a relative term,
and PLEASE do not let us fall into the trap of thinking that just
because our CNC software allows us to "work" to an apparent accuracy
of a thousandth of a micron our actual physical machine tool and
tooling suddenly got orders of magnitude more accurate.
My X axis "glass" scale is about 750 mm long and has a coefficient of
linear expansion about a tenth of aluminium or a fifth of the cast
iron machine table, but even so, 15 degrees celcius ambient range
between seasons means worrying about a theoretical trapezoidal
leadscrew wearing at maybe a tenth of a thou (if you do not maintain
it) a year is pretty pointless.
My initial tests show that the discrepancy between what the
uncalibrated mach3 software things the positions are and what the DRO
reports they actually are for any meaningful sized work that will fit
in my mill is of the order of 1.1 thou on the Y and 2.2 thou on the X,
already that is better than the deflection I'll get running a 3 mill
endmill on a cut with a full chip load per tooth... if I can get a
practical repeatable accuracy out of this thing under CNC control of
approaching a thou I will be over the moon.
Practical repeatable accuracy means measuring the part machined
compared to the dimensions it is supposed to be, actually measuring
the finished part, not reading the numbers off the idiot box.
For 99% of scenarios a practical repeatable dimensional accuracy of 5
thou will make me happy and consider the investment of time and effort
in the CNC conversion to be throughly worthwhile.
Perhaps Mr Stephenson will have some comments about how rigid 16mm end
mills are when told to cut a perfectly perpendicular slot in something
nice and soft like alu, and how perpendicular the CMM says the
finished product is after a single pass.
I asked a similar question on the Mach3 forum, here's the reply (for info):
"This reply is contentious but IMO there is no sensible way to
compensate for backlash on a machine your size because the cutting
forces will drag the table around. The control system will be
hopelessly non-linear as the motor has to turn quite a few degrees
befor the table moves at all then it suddenly starts moving at the
commanded speed. Little Taigs etc. have enough friction that some users
get away with Mach's backlash compensation. But it can never be better
that a bodge.
Fit ballscrews or, probably better, sell the Senior and buy a machine
that already has them to retrofit with Mach."
This implies that stiction may be a problem too, have you done much
metal cutting yet to see how well it works in practice?