Accuracy issues with my Makino...

[Joe788]

I think I have found the culprit. My spindle chiller unit isn't working correctly... This evening I ran a tool at 12k rpm for 20 minutes. Before running the tool I measured the standard and it measured at 5.00001". After running 20 minutes the standard measured 4.99919", or SHORTER by .00082". This is in only 20 minutes. After running surfacing toolpaths for several hours it is probably much worse. The chiller unit cools the spindle AND ball screws and if it is running cold things will shrink and parts will be cut small, which is exactly what I have been experiencing. If I measured a tool at the beginning of a program and did extended cutting with indexing and rotary moves and tool shrinkage would be amplified resulting in things being off by a mile, like .0032" I will be calling Makino service first thing in the morning.

 

Bob, what was the machine doing prior to the 12k rpm, 20 minute run? I'd try it again, and continue it running for another 20 minutes. Most likely, it'll still be the same -.00082.

 

.0008 is a little excessive, but if you were dead cold beforehand, I could see that happening. That machine probably has a 30ish minute warm up cycle, no?

[Bob W.]

The machine had been setting for about an hour and prior to that it had been running all day. I plan to gather more information and I have added a probing program that outputs macro data (M296) to the production loop. By the end of the day I will be able to graph the measurements and get a better idea of what is going on. I also have granite standards on the way (square and parallel) along with a spindle test bar so I can start at the bottom as Sticky recommended.

 

The thing that concerns me is that the tool got shorter. I have done similar tests in the past and the tool would grow by .0002-.0003".

[Guest]

The thing that concerns me is that the tool got shorter.

How is your drawbar pressure. Perhaps increasing? You're not testing BigPlus tools right?

[Bob W.]

How is your drawbar pressure. Perhaps increasing? You're not testing BigPlus tools right?

 

Not sure, where would I check? I don't have BigPlus but if the machine sets for 20 minutes after the test the tool length returns to normal.

[Sticky]

You need to have a drawbar force gage to check. I don't think that is your issue, if your retention system uses belleville washers its actually impossible anyways.

 

What is your cooler displaying for temperature? That is lame they use the same chiller for the screws and spindle.

 

Newbee touched on something quite important, most chillers can be configured to either work towards a target temperature, OR a within a percentage range of ambient. I would recommend using the percentage of ambient.

 

Does your chiller have a standalone control or is it tied into the NC? Either way, in most cases you can view the temperature the chiller is at on the chiller itself. You should maybe monitor it today while you are running parts and take note of temperature fluctuation compared to ambient changes and spindle rpm.

[Joe788]

Not sure, where would I check? I don't have BigPlus but if the machine sets for 20 minutes after the test the tool length returns to normal.

 

You'd need a drawbar force tester.

 

You never want to wing a spindle right back up to high speed, even if it's only been sitting for an hour. You need to run the warmup program again. I can believe a .0008 jump going from idle for an hour to 12krpm for 20 minutes. There are a handful of different temperature sensors in the spindle, and they (should be) are calibrated for the machine at continuous operation. You wing it up from a cold start, and you get a big jump in bearing temp, with no change in cartridge temp (or something along those lines) and the comp is gonna go stupid.

 

After the machine has been running at high speed for an hour, observe the spindle load while running at 14,000rpm. Now let it sit for an hour, and spin it back up to 14,000rpm. The load will be 3-4x higher. That's your bearings being tortured.

[Bob W.]

Does your chiller have a standalone control or is it tied into the NC? Either way, in most cases you can view the temperature the chiller is at on the chiller itself. You should maybe monitor it today while you are running parts and take note of temperature fluctuation compared to ambient changes and spindle rpm.

 

I spoke to Makino and my chiller uses a thermistor in the casting and sets the temperature based on that. After the current cycle I plan to power down the machine and measure the resistance of the thermistor and check the value against resistance/ temperature tables in my maintenance manual. There is not temperature display on my chiller, just a light that shows it is on and functioning.

[Sticky]

I spoke to Makino and my chiller uses a thermistor in the casting and sets the temperature based on that. After the current cycle I plan to power down the machine and measure the resistance of the thermistor and check the value against resistance/ temperature tables in my maintenance manual. There is not temperature display on my chiller, just a light that shows it is on and functioning.

 

My machine does the same but I can view and modify the parameters on the chiller controller.

 

 

Looks like Makino is making this one easy for you

[Bob W.]

After much testing I think I have it figured out. I ran some longer tests consisting of a surfacing routine that runs the tool at 12k rpm and lasts 30 minutes. I scheduled this in my DNC queue four times with probing routines between each instance. The probe measures the position and Z-depth of my ring gage and outputs the result to a macro file that can be opened in Excell. The first thing this test does is probe the ring gage and establish a baseline. I ran the typical warm up program and then kicked this off. Between the start of the program and the end of the first surfacing routine there was .0005" of movement in Z only (~30 minutes elapsed). After that there was less than .00005" movement in any axis with most at .00002" :-).

 

So my final verdict is that I was running an insufficient warm up program and the laser has a significant variance (+/-.0005") which has had me somewhat chasing my tail, and my B and C axis are offset by .00035" which could be improved. So here is my plan of attack going forward:

 

1. I plan to get a tool setter, set all of my tools together and load them into the machine at the same time so they will be accurate with respect to each other.

2. I then run a good long warm up program (~20-30 minutes) that really gets the tool up to speed toward the end

3. Immediately after the warm up cycle I run the program which will have a probing routine that re-sets the offset in X, Y, and Z.

 

With this approach 99% of the movement will be taken up by the time the program is running and the probing routine will account for this by resetting the offsets. This, along with grouping critical operations together and programming common sense should allow me to hold better than .0005" on these molds. I still plan to verify a few other things such as the axis squareness, etc... But I am confident this will yield good results. Thanks for all of the help and insight. Squeezing out that last .0005" really increases the challenge exponentially, even with a nice machine.

[Zoffen]

Squeezing out that last .0005" really increases the challenge exponentially, even with a nice machine.

 

3D tool comp?

[Guest]

Squeezing out that last .0005" really increases the challenge exponentially, even with a nice machine.

For every zero to the right of the decimal the price doubles, then for every .0001" beyond .0009" the price increases 30% for exactly that reason.

 

Bob, you want to increase the accuracy of your machine? Do yourself a favor and run in Metric. Your whole drive, ball screw and control system is native metric. Do it and don't look back. It'll take a bit to get used to but for those times when you need to split .0001", you actually can.

 

JM2CFWIW

[Newbeeee™]

^^^Foggy,

I never realised this is true (the metric thing)? I often wondered if it made a difference because with the amount of divisions on encoders, I thought it wouldn't make a difference but???

[Guest]

It ABSOLUTELY makes a difference. The smallest increment on an Inch mode machine (without special options) is .0001". On a typical Japanese machine the smallest increment is .001mm or .00003" so should you need to split .0001" running in inch mode, you can't do it. Running in metric, you can split it 2~3 ways. The more important fact is everything in the system is metric. The control, the ball screws, everything. So by running a "Metric" machine in inch mode, you're forcing the machine to do a bunch of math just to get from point A to point B, unnecessarily.

[Newbeeee™]

[Joe788]

It ABSOLUTELY makes a difference. The smallest increment on an Inch mode machine (without special options) is .0001". On a typical Japanese machine the smallest increment is .001mm or .00003" so should you need to split .0001" running in inch mode, you can't do it. Running in metric, you can split it 2~3 ways. The more important fact is everything in the system is metric. The control, the ball screws, everything. So by running a "Metric" machine in inch mode, you're forcing the machine to do a bunch of math just to get from point A to point B, unnecessarily.

 

I think nowadays just about all of the Japanese builders are doing .00001 as the smallest increment in inch mode, aren't they?

[Guest]

You mean .0001" right? I've only seen one Japanese builder go .00001" out of the box and that's Yasda.

[Joe788]

You mean .0001" right? I've only seen one Japanese builder go .00001" out of the box and that's Yasda.

 

All of my Mazak Matrix controls are .00001. I haven't looked at new machines from anybody else in awhile so I just assumed everybody else was doing it by now too. It really is nice having that 5th decimal, especially when a tool is going to work on both sides of a part, effectively doubling the comp value.

[mike93]

we have mazacks as well ,if you post in .00001 mode will you get a better finish ,will the look ahead handle it,it wood nice to yous point to point withe out using GO2 GO3 to smooth out the master cam out put ??????

[Redfire427]

You mean .0001" right? I've only seen one Japanese builder go .00001" out of the box and that's Yasda.

Most of our Makino's have .00001" resolution. Only the S56 has .0001"

Our Mikron's are .00001" also.

 

Carmen

[GoetzInd]

Our A51 is .00001 but I believe it was an option.

 

Mike