UMC-500 flooding, not ready for production

[empower]

I ran a Mikron HSK63 machine (similar to this haas) at 18k rpm for an hour. Re measure the tool in the Blum laser.
.0002"

BTW, not THAT much more expensive. To me, a much better value, hands down.

you're funny comparing mikron to haas :P

 

[empower]

We've noticed this as well.

Up to 0.003" variation per side at its worst, but the error is doubled when hitting two opposing sides of a part with a facemill, so more like 0.005"-0.006".

Probing Z for every part and then using DWO reduces the error to a more reasonable amount.

at a certain point, its worth it to spend more money to not have to fuck around with silly shit like this.

 

[empower]

Hello ,

We received a brand new UMC 500 SS 10 days ago.
First look very good , very powerful machine B and C axis had probably 0,001 deg backlash. I didn't expected such accurate rotaries, probably because of the scales.
After working on the machine a few days I noticed unrepeatable tool offsets. And I mean by 0.05mm up and down during the day. For me this is huge number.
I spoke a few days with the technicians to try to figure out the problem but all the variable we could imagine were not the cause of this problem.
I did a special warm up cycle routine, my factory is climate controlled so the temperature variance is not the problem.
Between a cold and warmed up spindle the difference is about 0,01mm which for this kind of machine I think is acceptable.
MRZP calibration was done several times - major differences in Z , very small in x-y.

We made a 50cm concrete foundation especially for the machine about 2 months before arrival. ( plenty time to cure ).
The machine is not anchored.

After a week of headbanging I understood that another local customer who received the same machine in the same shipment has the same issues ....
Now we wait a response between HAAS and our HFO.

The only thing we could think about is a mechanical or software issue... Backlash in Z is bellow 5 microns. X-Y have a little bit bigger backlash but bushing bores I interpolated were spot ON.
So mechanics seems to be fine.

Any info on your problem ?

you're not alone. we got rid of ours due to very similar issues, on top of other unrelated issues. worst purchase our shop ever made. returned it after 8 months.

 

[GeorgiMirov]

you're not alone. we got rid of ours due to very similar issues, on top of other unrelated issues. worst purchase our shop ever made. returned it after 8 months.

How did the return procedure went ? Please give some more info as I am in the same situation right now - do I return it and get something better or keep on headbanging. This machine was delayed about 2 months and the projects I got it for are still not in the works and this is huge problem.

Did they gave you some money back ?

 

[Orange Vise]

You get this much variation between every part? Or total variation throughout the whole day?

That's about the same growth I can see on my 2017 750SS, up to .003"

Over the course of a week-ish.

 

[coffeetek]

Over the course of a week-ish.

Our UMC-500 is triggering tool breakage detection set to 0.0078" because of the growth. I had to set it to 0.5mm ~0.02" just for detection to be able to pass on tools that are not broken.

 

[cameraman]

We've noticed this as well.

Up to 0.003" variation per side at its worst, but the error is doubled when hitting two opposing sides of a part with a facemill, so more like 0.005"-0.006".

Probing Z for every part and then using DWO reduces the error to a more reasonable amount.

The Pierson workholding guy has reported similar numbers / values for the UMC 500.

 

[cameraman]

might be worth measuring the 'Droop" in the Y axis head fully extended + torsion and deflection in-cut in the Y axis.

If the XY plane of the spindle is not coplanar to itself i.e. is a progressive non-linear curve of increasing "Droop" in the YZ plane then that could throw off the MRZP / kinematic calibration routines.

I have this "Vibe" that they (HAAS) are using the wrong "Math" for this... [These are intrinsically non-linear systems (mathematically).].

I.e. the simple algorithms they use assume that commanded moves of the spindle head are orthogonal - which clearly they are not.


+ Haas have Z spindle growth routines to back away from the part (not violate the geometry).

Depends maybe on threshold temperature values for that.

Worth checking the true squareness of the machine against granite angles and test bars etc.


Unbalanced tools on spindle bearings that heat up too much could be another (minor) factor. + no specific spindle chiller.

+ whatever final distortions to the machine once bolter to it's foundation + correction thereof physically.

 

[D Nelson]

might be worth measuring the 'Droop" in the Y axis head fully extended.

If the XY plane of the spindle is not coplanar to itself i.e. is a processive non linear curve of increasing "Droop" in the YZ plane then that could throw off the MRZP / kinematic calibration routines.

I have this "Vibe" they are using the wrong "Math" for this... [These are intrinsically non-linear systems.].

I.e. the simple algorithms they use assume that commanded moves of the spindle head are orthogonal - which clearly they are not.


+ Haas have Z spindle growth routines to back away from the part (not violate the geometry).

Depends maybe on threshold values for that.

Worth checking the true squareness of the machine against granite angles etc.


Unbalanced tools on spindle bearings that heat up too much could be another (minor) factor. + no specific spindle chiller.

It’s not the spindle doing it we ran cycles with the spindle off just moving the axes for about 20 minutes then sopped it on a indicator you could actually see the needle going back to zero in about 30 minutes of sitting still
Don


Sent from my iPhone using Tapatalk Pro

 

[cameraman]

It’s not the spindle doing it we ran cycles with the spindle off just moving the axes for about 20 minutes then sopped it on a indicator you could actually see the needle going back to zero in about 30 minutes of sitting still
Don


Sent from my iPhone using Tapatalk Pro

Wow,

That's a little disturbing,

Was the indicator measuring expansion or contraction of the ball screw on the X axis ? [As it cools / contracts ] ???

If you don't mind me asking. Not sure where you have your indicator (all good).

Just spit balling , the HAAS UMC 500 is much smaller machine but on the 750's the extended "Droop" of the Y axis is known or various work arounds need to be made (for really flat parts that span the whole breadth and depth of the table). Theoretically the smaller machine could be "Tighter" / more rigid - smaller force loop, but seems mechanically vulnerable with the head fully extended - which if you have multiple parts on a tombstone tilted 90 degree would be most of the time (given that the trunion is still quite long in relation to the rest of the machine.).

The HAAS 1000 looks very big - lotta iron and volume. BIG Palate system looks pretty amazing.

Pierson workholding guy was citing lack of lateral rigidity of the spindle assembly fully extended- in cut . i.e. he surmised the cutting forces were moving the spindle as the trucks and linear slides were not rigid enough or that the back end of the riding spindle assembly needed to be longer; i.e. the BRIDGE that this all rides on needs to be deeper (longer in the Y dimension) and maybe the ram assembly needs to be wider to compensate.

Might dig about...

 

[D Nelson]

No just the z axis would shorten up as it warmed up then just let it sit there cooling down resting on the indicator it would come right back Down to zero where it started out from when cold. The older 1000 doesn’t do it near as bad but it does it some. We notice it most when doing 2 sides of a part and ours is like between .0025 and .003 so about 1/2 that would be the amount of variation but it’s definitely doing it. Our machines bore nice holes also. I like the machines overall. A few things piss me off like the spacing between the pallets while in the pool are not symmetrical who would of thought that shit up we made fixtures that were at maximum width put them in and one station they hit each other. The other is you can’t probe with the b axis off zero. We wrote a macro so it can be done just stuff like that that gets you mad.
Don



Sent from my iPhone using Tapatalk Pro

 

[empower]

How did the return procedure went ? Please give some more info as I am in the same situation right now - do I return it and get something better or keep on headbanging. This machine was delayed about 2 months and the projects I got it for are still not in the works and this is huge problem.

Did they gave you some money back ?

documentation and lawyers. yes, they refunded us.

 

[mkd]

might be worth measuring the 'Droop" in the Y axis head fully extended + torsion and deflection in-cut in the Y axis.

If the XY plane of the spindle is not coplanar to itself i.e. is a progressive non-linear curve of increasing "Droop" in the YZ plane then that could throw off the MRZP / kinematic calibration routines.

I have this "Vibe" that they (HAAS) are using the wrong "Math" for this... [These are intrinsically non-linear systems (mathematically).].

I.e. the simple algorithms they use assume that commanded moves of the spindle head are orthogonal - which clearly they are not.
.
.
.

.

Comp'ing a single linear axis is simply table driven, At this point, comp this much.
On machines like these, there are two main variables. Spindle growth, and trunnion growth.The OEM should at least have something for the spindle on the software /sensor side. Then use kinematic to comp the table periodically.

 

[mkd]

you're funny comparing mikron to haas :P

Yes it is funny.
Funny that people buy UMC's at all.


When you can get 100% more machine for 30% more coin which ends up saving the business a significant chuck of labor EVERY SINGLE WEEK, the correct choice is clear, when you can think past the first six months of payments.
IMHO, of course

 

[coffeetek]

Yes it is funny.
Funny that people buy UMC's at all.


When you can get 100% more machine for 30% more coin

Not really. You can buy 2-3 UMC-500's for 1 Mikron. Mikron MILL E 500 U is from $247,320 to $345,060 depending on options... And its 3+2 machine.

 

[mkd]

Not really. You can buy 2-3 UMC-500's for 1 Mikron. Mikron MILL E 500 U is from $247,320 to $345,060 depending on options... And its 3+2 machine.

That's very incorrect. A Haas 500 optioned up comes to $303k and is still far less machine. Build a quote yourself. This is what is comes to.
The E500U Micron with a world class Heidenhain control will destroy the Haas in a Full 5axis Impeller test. Ask me how I know.

 

[coffeetek]

Not going to call you a liar, but you are very very wrong. A Haas 500 optioned up comes to $303 and is still far less machine. Build a quote yourself. This is what is comes to.
The Micron with a world class Heidenhain control will destroy the Haas in a Full 5axis Impeller test. Ask me how I know.

You must be high and you are not sharing what you are smoking with us, shame on you

Loaded up UMC-500 with almost all options available is about ~$138,506

 

[CarbideBob]

A big step is finding ways to make a lot of chips fast.
An then comes how to get rid of them and what kind of changes to make here?
Extra pumps or pressure?
An added end of cycle that does not cut anything but clears and moves chips like a wash down? Most would not consider that part of the cnc program but helps sometimes.
Think of it a robot with a hose in hand. Not cutting metal but supporting cutting metal.
Bob

 

[mkd]

You must be high and you are not sharing what you are smoking with us, shame on you

Loaded up UMC-500 with almost all options available is about ~$138,506

read it and weep
$303,545
any Micron will wipe the floor with a Haas

 

[mkd]

View attachment 320754
read it and weep
$303,545
any Micron will wipe the floor with a Haas

Correction; The 500SS model comes to US$324,745
My bad.
Still doesn't have 48hp, 20,000rpm, highly durable direct drive rotaries and proven Heinenhain Kinematics adjustment.