What's new
What's new

Servo bar accuracy?

GiroDyno

Hot Rolled
Joined
Apr 19, 2021
Location
PNW
Does anybody have experience with the Servo Bar bar feeder and can you attest to its accuracy/repeatability?

I have been having issues with the feeder performing double pushes, under feeding and even loading a whole new extra bar while there is an existing 2-3' long bar loaded. All these issues different issues come up running the same program and it seems like a different variation of one of these problems happens each time a new bar is loaded.

I found the counterbalanced sheet metal end of bar switch design to be pretty unreliable and easy to get stuck down so I added some weight and shimmed it to ensure it would function correctly, but I am still getting issues.

Yesterday I measured a a bar before I loaded it and compared it to the measured value, it was off by 3". I then ran G105 Q5 per the manual to reset the end of bar switch position and was able to get the correct measurement, but after running a whole bar I found the measurement on the next bar was off again.
I reloaded/remeasured that same bar several times and found the length changed each time by +/- 1/2", with rapid override set to 5% the error dropped to about +/- 0.02" but that is unbearably slow; I'd rather go back to running one bar at a time!

Any ideas what the issue could be?

This is a new piece of equipment to me so it could absolutely be my error, but the intermittent issues while running a single program make me think there could also be something up with the machine.
 
We have the same problems on ours. About all we have been able to do is make sure the material is clean and chamfer the end. If the material is dirty it transfers to the liner and the material drags. Also look at your clamp/unclamp time and the barfeeder push speed both of those can affect it.
 
Thanks for the confirmation!
I ran several bars this morning without intervention. All the bars measured incorrectly right from the start but were all within +/- 0.25" but by the end of the bar some were off by as much as 3.0".
All of the bars ended up short on the initial push, so I scrapped the first part off each bar but that's better than pushing long and crashing so I'll call it a win.
If I know it will never be able to keep track I will accommodate that by painting the ends of bars to easily spot a short part in the parts bin, at this point I mostly just want to avoid crashes from overfeeding.
 
We usually load bars at 25% rapid. I would estimate they are within 1/8" everytime.

If there are settings you are questioning, I can look and see what mine are.

What year/model lathe and bar feeder?
 
We usually load bars at 25% rapid. I would estimate they are within 1/8" everytime.

If there are settings you are questioning, I can look and see what mine are.

What year/model lathe and bar feeder?
It is a 2014 ServoBar 300

IIRC, I set parameter 316 (FEED INCHES PER MINUTE) to 12500 instead of the default 25000 to try and slow down the measuring feed
I will also try to have all bars are cut to the same lengths in the future, we had a bunch of different length bars cut +/- 3.0" which I'm sure didn't help

How much stock are you adding for the initial push?
 
Mine is a 2015. Par 316 is set to 25000. Having the bars to same length shouldn't matter.

My way of setting up a new part is...
Enough initial push to fall into the parts catcher
Incremental is part length + parting width + 0.01
Load bar and run G105
Touch part off in Z and subtract 0.03

I have run into issues pushing the bar if the material is a couple thou over nominal and binds a bit in the collet, but not more than 10 thou.

It sounds to me like it could be a problem with the flapper used to measure the bar.
 

Here is a video showing how to set your reference point if the chuck on your machine is different than what previous owner ran.

I believe is set my min retract to 15" I see in the video his is at 12. This is the location where the push bar sits while the lathe is running.
 
In current commands I zero out all the push variables, and hit origin to reset everything
I run G104 Q5 with a reference bar of known length to calibrate the bar feeder
Then G104 Q4 to load a bar (I measure and record the length of this bar and all other bars before they are loaded into the hopper) and jog the bar flush to the front of the jaws, then +.02"
G105 Q2 to set the length of the bar, it does an initial push of 0 because I cleared all those values when I started my setup
I check in current commands what the length of the bar is vs the measured length I took before loading it, this will usually be pretty close for the first bar

In my program it reads a user defined variable (bar diameter) to determine the initial push which is 2.5xD
It then reads a user defined variable to set the length of the part
More macros determine how many will fit into a bar sticking out 2.5xD - 0.5" clearance to the chuck
Using the number that will fit into a 2.5xD length, the length of each part + part off + facing it calculates the incremental push
It writes the initial and incremental push values to the bar feed variables and calls up the first G105, which pushes the bar 2.5xD
Starts a loop cutting parts
When the bar gets down to 3" it spits it out and loads a new one

This new measured bar length is not as close as the first and the more bars it runs the further off it gets, after a few bars the length is off by maybe 3"
If I'm lucky it reads one long, then one short, then one long, short, long, short, etc... and it stays pretty close, but if I'm unlucky it keeps reading then short, short, short, short and eventually I get a weird push where there's 18" hanging out of the chuck!
 
are you using a dead-length collet chuck? if not, that will introduce some length differences, which change based on how long your bar currently is.
i'm pretty sure there's a way you can use the turret to push against, so that way you always have a known length.
my other guess is maybe there's something funny in your macro program. why not use the built-in page on the control to handle push lengths and behavior?
 
are you using a dead-length collet chuck? if not, that will introduce some length differences, which change based on how long your bar currently is.
i'm pretty sure there's a way you can use the turret to push against, so that way you always have a known length.
my other guess is maybe there's something funny in your macro program. why not use the built-in page on the control to handle push lengths and behavior?
I would agree, try running without macros and see how it behaves.
 
I agree we are probably introducing some error with our workholding; we are not using a dead length collet, rather some old pie jaws. Even so the feeder holds the bar pretty close to position while the chuck is closing so it shouldn't lose much. I take a .025" facing pass on each push and it always cleans up, so I'm not losing more than 0.025" from push to push.

Is it possible to have the feeder run up to a hard stop? I know you can have it hold a position and you can butt bars back up to it for loading parts into the chuck.
The pneumatic feeders I'm familiar with push to a stop, they just never had magazines, if I could get this to monitor the V load for a stop and keep loading new bars I'd be a happy camper.
 
are you using a dead-length collet chuck? if not, that will introduce some length differences, which change based on how long your bar currently is.
i'm pretty sure there's a way you can use the turret to push against, so that way you always have a known length.
my other guess is maybe there's something funny in your macro program. why not use the built-in page on the control to handle push lengths and behavior?
Not sure about the Servo 300, but our newer barfeeder will not push to a stop like the lns ones will.
 
I'll see if there's a variable for the V axis load, if there is you can advance the V axis incrementally while reading that value and stop it when it spikes. That is how I run our subspindle for transfers vs moving to specific B positions, which allows me to compensate for variations in part thickness.
If there are also M codes to control the bar loading then it could be driven independently from the bar feeder page.
 
I'll see if there's a variable for the V axis load, if there is you can advance the V axis incrementally while reading that value and stop it when it spikes. That is how I run our subspindle for transfers vs moving to specific B positions, which allows me to compensate for variations in part thickness.
If there are also M codes to control the bar loading then it could be driven independently from the bar feeder page.
https://www.haascnc.com/service/onl...s/lathe-operator-s-manual/lathe---macros.html
#1266 it looks like .
I'm interested if you can get it to work. I hadn't thought about using the load to do it. Of course mine are NGC controls unfortunately. I can't even use the q commands with mine.
 
When I was running a bar feeder a long time ago, I would program a cut off cycle before any other tools. Had it "cut off" in front of the zero by .025".
Had issues with heavy initial facing and a few times broken tools because of too long feeds. The cut off cycle took time but saved me a quite a few times.
 
When I was running a bar feeder a long time ago, I would program a cut off cycle before any other tools. Had it "cut off" in front of the zero by .025".
Had issues with heavy initial facing and a few times broken tools because of too long feeds. The cut off cycle took time but saved me a quite a few times.
Good idea.
I was already starting by rough facing the bar, part-off would work just as well and not really any slower.
Scary to think about an extra 24" slug parting off tho!
 
When I was running a bar feeder a long time ago, I would program a cut off cycle before any other tools. Had it "cut off" in front of the zero by .025".
Had issues with heavy initial facing and a few times broken tools because of too long feeds. The cut off cycle took time but saved me a quite a few times.
I used to have our machines doing the same thing until we started breaking cut off blades on the longer initial pushes. I went back to using a g72.
 








 
Back
Top