Is Haas axis-zero on power-up-restart a precise operation?

On an axis-zero operation (power-up-restart), the Haas moves the axis until it strikes the stop switch and thus sets the zero.

I don't see how the stop switch can precisely measure the axis home -- I say this by looking at the design of the stop switch.

It seems, in practice, that if I re-home the axis's, then they do not return to the exact previous offset. It goes out by more than 0.0005" (½ thou).

Is this error unavoidable?

Is the Haas mill never meant to be turned off?

Is one's morning startup procedure supposed to include resetting your G58 (or whatever) axis offsets to match one's previous day's work?

What am I missing?

BTW I machined a 3" round pocket in mild steel and the inside diameter was accurate to 0.00015" (0.15 thou). So this machine is in good condition.

Thanks!

Paul (1997 Haas VF-Oe)

Well, a 1997 machine is a bit long in the tooth. I wouldn't look a gift horse in the mouth though.

The 2015 machine here repeats pretty well on startup, but there is a little thermal variation as the machine warms up. You could be seeing some of that. My understanding is that the servo's encoders tell the machine the exact position, the limit switch tells the machine what revolution of the encoder is zero vs .200" or somesutch. One of the first machines I worked on was an early 90's Fadal, and every once in a while it would start up exactly .200" off.

mhajicek said:

...servo's encoders tell the machine the exact position, the limit switch tells the machine what revolution of the encoder is zero....

Click to expand...

thanks, that makes sense

½ thou seems a lot though

abbeyroad1124 said:

thanks, that makes sense

½ thou seems a lot though

Click to expand...

For a 24 year old machine, half a thou is great! Unless it's been recently rebuilt, in which case you could get .0002".

0.0005” variation on startup is a lot?
That sounds pretty good to me without scales or probing cycle.

I would ask “whats changed” and if the temp over the day has shifted and the machine is 4c cooler in the morning 0.0005” is great!
The ball screw assembly and main casting will grow/shrink and so will your concrete, even that could change your z height/machine tram if your floor is moving.

Asking for better than 0.0005” in a non temperature controlled environment is asking for the lotto numbers weekly. Just not gonna happen IMO.
I could be wrong?

Stirling said:

...

Click to expand...

i have the AC set constant. so there would be no thermal variation.

back working on this today....

i just re-measured, and seems i had to add 0.001" to the Y to bring it back to the center of the work-piece.

i will just keep working and see what happens over time

One should expect plus or minus one count out of a machine tool
As worst two encoder counts if your controller is slow.
Long before Hass even dreamed of building machine tools the axis home routine was standardized.
On a cnc you do not home off of the home dog switch. Yes it seems so much underneath.
You go to the home dog and than go slow one direction or the other to the next encoder index line and that is zero is where zero is put.
This is why if you disconnect a servo from a ballscrew one has the reset the off to zero or grid shift counts.

Two things here go wrong. The dog trigger point is inline with the index, This rare but you get zero returns off one ballscrew rev every once in a while.
This one is easy as the error is huge or nothing but can be a frustrating SOB.
Second if the search speed to index is two fast the control can not grab the position fast enough.
Since this is time dependent normally if even five or ten counts lag it is normally decent repeatable but we are looking at cpu/encoder cycle rate in the digital time slice what it sees vs the speed things are changing.
Everyone wants fast zero returns on startup .

Outside machine tools stepper motor machines do not have a index pulse. Here you do home off a switch.

Unsure about this 5/10's number which seems so way out of line. Verified shift to the vise or part or something on the screen during zero return?
Bob

I do not know if a '97 differs from a '96 but the manual says the encoder Z channel signal must occur between 1/8 and 7/8 revolution from the HOME switch release or a 165/166/167/168 error occurs.
Homing moves the axis until the switch closes then reverses until the switch opens.
My '91 has a slightly sticky X HOME switch so I have to HOME all axis then single axis HOME the X again to get a consistent absolute position. Otherwise relying on the G54 position as the part zero will be off.

abbeyroad1124 said:

i have the AC set constant. so there would be no thermal variation.

Click to expand...

AC is constant here too. The machine will warm up over the first hour or two it's in use. The spindle motor lives at around 10-20 degrees above ambient, depending on how hard it's being used. Servos, ball screws, and frame will also warm up a bit. If you're curious, go over your machine with an infrared thermometer first thing in the morning before turning the machine on, write down the numbers in various spots, then go over it again after you've been running a while and compare the numbers.

When I'm running something with really tight tolerances, after doing the spindle warm up in the morning, I like to run a cycle or two with no part before putting the stock in; that gets everything at the temp it will be when running that particular program. This is for small medical device parts, 10 - 30 minute cycles.

abbeyroad1124 said:

On an axis-zero operation (power-up-restart), the Haas moves the axis until it strikes the stop switch and thus sets the zero.

I don't see how the stop switch can precisely measure the axis home -- I say this by looking at the design of the stop switch.

Click to expand...

If you read Bob's and Magno's response, you will understand the concept of the homing procedure.
But to put it in an even simpler term, how' bout this:
You give directions for someone to find your house.
You tell them that it is the first red door on the right from the end of the blind alley, but you must see at least 10, but no more than 50 white doors on the left, else you're on the wrong street.

So, the individual blindly walks down the alley in a relatively slow pace until he bumps into a wall, signaling the end of the line.
He then opens his eyes and starts to walk backwards, meanwhile he's looking at the doors on the right and counting the doors on the left.
As soon as he sees the first red door on the right, he stops and evaluates this:
Did I see 10 doors on the left?
If yes, then the next question is:
Did I count less than 50 doors on the left?
If the answer is yes to that one also, then all is well, I am at your door, and ONLY at your door.
If the answer is no to either of those questions, then I am in the wrong alley.

That very last scenario ( wrong alley ) in the control is then evaluated further to determine if this is a homing switch, encoder, axis motor or some other possible cause.

Can this sequence be off by as much as .0005? Not likely in today's controllers unless it's a piss poor design.
Your deviation is more likely caused by a simple temperature change that the control is not compensating for.

( well, at lease that is how I am trying to make a simple sense of what's up in the background. )