What's new
What's new

New Machine Day - Robot Invasion!

I watched the last video you uploaded: As nice as it worked it seems you need either more pallets, a huge drawer system or looong cycle times to run lights out. With the short cycle time of those parts the operator need to be there pretty often to reload ?
 
I watched the last video you uploaded: As nice as it worked it seems you need either more pallets, a huge drawer system or looong cycle times to run lights out. With the short cycle time of those parts the operator need to be there pretty often to reload ?
Where did you get to see the video that has been uploaded?
Thanks
 
That video was not supposed to be "public" yet, (my mistake in not making it such) and is no longer available right now.

The previous one Panza saw in Post #63 was not shot in UHD so I took it down.

The new one is shot in 4k and will be made public very soon once I get a chance to write and post pics to go along with it.

To be continued..............
 
Time for an update!

We have been making some steady progress with the robot and machining 1st op's on a variety of parts. No, we still have not made the leap to stacking parts on the material table, but with things progressing in the manner in which they are, it will not be soon before doing so.

Right now Andrew has been able to set up robot fairly quickly. When I say "fairly", it can be perceived as a very subjective term because let's face it, who knows how long it "should" take, right?

Well, according to the applications engineer who did the training, he insinuated it could be the better of a day to get it set up and making parts. Ok, wait....what? That is not what I signed up for, so let's break this down.

When learning to set up the robot for the "first few times", it was taking the better of a day to set up, yes.....absolutely!

Andrew though seems to have the knack for doing this however in very short order!

Once you do a few similar jobs, the gripper selection and array set up on the material staging table gets much easier to understand and adopt for the new job.

So, by comparison, to set up some of the parts he's doing with Wall-E, it might take 45 minutes in one of our VF-2SS machines, whereas with the robot cell, it is taking about an hour to an hour and a half. I can live with that, considering once it's set up, it's running until the staging table is empty and the finished parts table is full. The longest cycle time we've managed from table full to empty has been only 36 minutes, but, it's running through break, it's running through lunch, and more importantly, it's one out of five machines running that is completely unattended.

The table change over time from removing finished parts to setting up blanks is taking him about 60 seconds.

We have experienced a couple of crashes due to our own inexperience. When I say crash, I simply mean a broken end mill that was the result of an improperly seated part in the vise. Fortunately, someone close by heard that awful "SNAP" and the cycle was stopped before any more tools were lost. The part slipped in the gripper due to some gouges in the extruded side of the aluminum material and was placed in the vise off in the "X" axis, resulting in the .500" carb 3flt em trying to slot full diameter .755" deep at 12k rpm and 250ipm. It actually made it 1.5" into the material before it broke, so not too bad.....I guess.

The other blunder was due to the array on the pick up table being off and same as above, way off in the "X" so when profiling it just ripped that part right out of the vise and tossed into the back of the machine. Very glad it did not take out the wireless tool setter on the table.

After those two mishaps we've learned to double check the array dimension in the control, and also, that one set of grippers is not necessarily good for every part.

Here's a close up the grippers we're using on some knobs we're machining.
20220922_130526.jpg


Another safety feature we've implemented on every job now is to perform a break detect after the rough profiling tool. It adds about 10 seconds to the cycle time, but ensures that if the first tool breaks and no one is around to hear it and stop the machine, all the subsequent tools will not get broken as well as the machine will just stop and alarm out if it detects a deviation on the tool being checked by a tolerance we set in the control.



Time for some more pics!

Below is some 1" x 3.5" cut 3.250" 6061 material waiting to go into the robot cell to be machined into some knobs we make. With a qty of 200pcs on this order, Wall-E is the perfect choice for the 1st op on these parts.

20220922_130357.jpg


Andrew setting up the blanks on the material staging table. Yes, it looks cumbersome, but before investing in a larger table, we're keeping things simple for now. Once we get good at the simple set ups such as these, the larger ones will be much easier.

For now we are just using some 2" wide strips of aluminum to set the spacing between blanks.

20220922_130508.jpg


With the spacing set, he simply pulls out the spacer while holding the material.


20220922_130514.jpg



I did take some video of this part from start to finish with Wall-E putting them in and taking them out of the machine, but like a dope........it was not shot in HD and looked like total crap, so all I have are before and after pictures.

Here are the after pictures of the parts after coming out of the machine, thanks to Wall-E!

20220922_130412.jpg

20220922_130316.jpg

Here's a close up view of the same pieces.


20220922_130326.jpg


This is the finished product after being red anodized.

20220922_135353.jpg


That's all for tonight, been a long day!


To be continued...............
 
Thoughts that riff off this:

1. You may have forgotten how long it took you to make your first part, or your first batch of parts. You have to evaluate this thing after some time to get smooth with it. Evaluate it after 3mo or 6mo.

2. At IMTS Hermle (and perhaps others) was showing a Robot loading parts onto pallets, which then got fed into a pallet changing machine. And one of their folks admitted "That's a pretty complicated first step, we try to get customers to think about just a large enough pallet pool to start with". (Or close to that, not a literal quote.)

The point? If budget, floorspace, etc., allow, a pallet changer may be an easier scheme to absorb. But the robot arm systems seem to be net cheaper (don't have to duplicate all the workholding) and generally take up less space, at least at reasonable prices.

(I'm sure Russ and co would love a Kern, so long as somebody else paid for it...)
 
Anyone else notice the UR20 from Universal Robots at IMTS?

The thing is massive. We seriously considered a UR16 (16kg payload) before settling on the UR10e (10kg). The UR16 didn't give us enough reach.

The UR20 is, IMHO, a very big deal for machine tending. It has huge reach and sufficient payload capacity (20kg) to change pallets in a 5-axis machine, and it can also swap parts. You don't have to choose one over the other. Base price was low 60s.
 
Anyone else notice the UR20 from Universal Robots at IMTS?

The thing is massive. We seriously considered a UR16 (16kg payload) before settling on the UR10e (10kg). The UR16 didn't give us enough reach.

The UR20 is, IMHO, a very big deal for machine tending. It has huge reach and sufficient payload capacity (20kg) to change pallets in a 5-axis machine, and it can also swap parts. You don't have to choose one over the other. Base price was low 60s.

Massive is kinda relative. I had a Fanuc 420if for awhile. Iirc, it was rated for 420KG at a 10' radius at full speed. I decided against using it for machine tending. Figured it was big enough to take out a wall or tear a machine apart if something went wrong. I thought it'd be real neat to set several machines around it so it could tend all of them.
 
Anyone else notice the UR20 from Universal Robots at IMTS?

The thing is massive. We seriously considered a UR16 (16kg payload) before settling on the UR10e (10kg). The UR16 didn't give us enough reach.

The UR20 is, IMHO, a very big deal for machine tending. It has huge reach and sufficient payload capacity (20kg) to change pallets in a 5-axis machine, and it can also swap parts. You don't have to choose one over the other. Base price was low 60s.
I heard the UR brand cobots joints don't last very long in machine tending (wet and/or dirty) environments, has this been addressed in the newer models? Know of a couple shops that have run through multiple URs doing what seems like light duty work, but that was a few years ago...
IIRC Fanuc has a green robot that can handle 35kg, assuming the green ones use more or less the same hardware in the joints as the yellow ones, and those seem to last a good while.
 
Picture and Video update!!

Final I've had some time to post up some 4k video of Wall-E in action! Be sure to have your YouTube video settings set to 4k for watching!

Was able to have Andrew lock me inside the cage for some close up camera work!

But first, here are some pics of the parts we were machining in the video.

Enjoy!!


Here's some of the material blanks on the cart outside the machine waiting to go on the staging table for Wall-E. These are for some handles we make. Blanks are "1" x 2" cut 7.062" lg. 6061 aluminum.

20220928_125152.jpg


Due to their size, we could only get 5 blanks per cycle on the table. Again, we are just taking things slowly to work at getting good at the smaller staging area before moving on to a larger table. We're just using 1" square stock as a spacer between blanks.

It soooo much easier to figure out the smaller array at this stage with our limited experience. Besides, the job runs unattended with the exception of the minute it takes Andrew to change to out completed parts and load up new blanks. So for 15 minutes or so, it's running unattended.

20220928_124323.jpg




With the spacers removed, Wall-E is in position and ready to rock!

20220928_124357.jpg



While the part inside the machine is being machined, Wall-E waits patiently outside with the next blank ready to go in!

20220928_125047.jpg




This is the plunger that pushes the part down into the vise jaws after Wall-E drops it off in the vise. You will see this procedure in the upcoming video below.

The robot places the part in the vise, the jaws close, Wall-E lets go of the part and retreats back outside the doors. The auto doors close and the plunger rapids down above the part and then feeds down to press on the top of the part. The jaws open, the parts gets compressed down onto the locating surface of the jaws, the vise clamps, the plunger rapids to home and machining begins.

20220928_124525.jpg




And here it is! The first piece is done and back on the table!

20220928_125031.jpg



Andrew watches from outside to be sure everything is running as planned! The remote touch screen E-stop right by his side, just in case!!

20220928_125222.jpg




With the novelty now worn off somewhat, the rest of the guys are focused on the keeping the other machines making chips!!

20220928_125114.jpg



Here you can see some of the 1st op finished parts.

20220928_123721.jpg
20220928_123750.jpg


Below is the link to the video! Remember to set your YouTube video player to 4k setting for best viewing!!

Later,
Russ

 
I think that sort'a negates the porpoise of the cage, eh?


-------------------

Think Snow Eh!
Ox

Yeah, as we all know, there are ways around everything. As long as Andrew was around to unlock the cage and let me out, it was all good.

Besides, getting to old to go and try scaling the fence to get out on my own! :D

Later,
Russ
 
Hi Mark,

Thank you!

Long time since we've talked, hope you're doing well!

Let me know if you ever need anything!

Later,
Russ
Russ, If you recall, I retired back in April. So I am no longer leading up projects and buying machined parts. These days I fool around in my home-shop and ride my bicycles & motocycles when it's nice out. And when it's not nice out, you'll find me in the gym, in my shop, or maybe working on my house. :D

A former aquaintence has started an LLC. And he's trying to coax me out of retirement. But I am resisting. But if I get drawn back in, I'll keep you in mind.

I'll be watching your thread. Good stuff! After the videos, now I know what you look like. Your Boston accent on youtube is just like over the phone. It was always a pleasure to work w/ you and your team. :)

-Mark Long
 
It's hard not to contrast this with Wheelieking's robot integrated Brother experience. Seemed like Brother has this stuff down. Brand new machine setup and spitting out parts in a couple days iirc.

Haas sure doesn't seem to be winning it with any of the new things they're trying in recent years.

Yamazen did a LOT of work on Wheelie's set up before it was delivered, it was a turn-key to make a specific part vs new machine and robot. Still, considering how big the parts can be it seems that Haas is quite wasteful for that application. An S300 would use a lot less space and energy and be plenty big enough for that vise.

Exactly as David stated. I went with a "turn-key" package to produce one specific part. In doing so I worked closely with the (at that time) Yamazen apps. guy on the process. I had already been making these parts for years, so I had the process and tooling dialed as fast as it reliably could be. We just had to adapt the part to the robot cell. At the end of the day, the robot cycle time per part was actually slower than making them in my haas pallet-change machine. BUT! The robot would run lights out, did not take breaks, did not draw on the payroll, did not get vacation pay, did not require workman's comp, and throughput actually increased. You get the picture. I would hire a bot over a laborer every freaking time! What I had running on my floor was freaking AWESOME!

The problem for me was, because I was very much an absolute moron when it came to the Nachi system. (god I hated that damn pendant!) And seemed to have a severe learning block to the whole thing (I was struggling BAD). The training was a joke. The trainer was from the other side of the country. Only here for one day.
They basically gave me a 3-4hr rush course on how to run my part. My system was a one-trick-pony (for me). No way in hell I would have been able to efficiently run many different part#s in that set-up had I needed to switch gears. I am sure most of the blame for this rides my shoulders. But, it is what it is. When that job shut down, and the awesome apps guy had quit Yamazen just weeks after I signed off, so I basically had nobody local in my corner? I couldn't sell that rig fast enough! Yea, Yamazen can fly help in. Sorry, screw that! Not interested.
 
Last edited:
A bitter lesson from the world of large software projects applies here. Usually stated something like this: "If a system works, but is not understood, it is doomed to break and fail. If a system does not work, but is well understood, it can be fixed and thrive." Keep in mind that large systems (think "windows" or "google") can be way harder to understand than you might imagine.

It seems that Russ/Andrew are coming up the learning curve, and will be able over time to adjust the robot arm to do their bidding. Fast enough/well enough to be economically useful to them? I don't know. But it seems they can work with it.

Wheelie - you were stuck with something you didn't really deeply understand, and so when the world changed, it "broke" for you. It doesn't matter WHY you didn't grok it, that was just your reality.

Often with these sorts of systems, there is lots of tacit knowlege (e.g. "you have to counter rotate the end of arm tool so you don't overrun the rotation before the end of the op"), which can be difficult to absorb quickly.
 
Exactly as David stated. I went with a "turn-key" package to produce one specific part. In doing so I worked closely with the (at that time) Yamazen apps. guy on the process. I had already been making these parts for years, so I had the process and tooling dialed as fast as it reliably could be. We just had to adapt the part to the robot cell. At the end of the day, the robot cycle time per part was actually slower than making them in my haas pallet-change machine. BUT! The robot would run lights out, did not take breaks, did not draw on the payroll, did not get vacation pay, did not require workman's comp, and throughput actually increased. You get the picture. I would hire a bot over a laborer every freaking time! What I had running on my floor was freaking AWESOME!

The problem for me was, because I was very much an absolute moron when it came to the Nachi system. (god I hated that damn pendant!) And seemed to have a severe learning block to the whole thing (I was struggling BAD). The training was a joke. The trainer was from the other side of the country. Only here for one day.
They basically gave me a 3-4hr rush course on how to run my part. My system was a one-trick-pony (for me). No way in hell I would have been able to efficiently run many different part#s in that set-up had I needed to switch gears. I am sure most of the blame for this rides my shoulders. But, it is what it is. When that job shut down, and the awesome apps guy had quit Yamazen just weeks after I signed off, so I basically had nobody local in my corner? I couldn't sell that rig fast enough! Yea, Yamazen can fly help in. Sorry, screw that! Not interested.

I think if you would have gotten #1 a pallet load solution or #2 a robot load with intuitive easy software you wouldn't have sold that cell. You would have sold something else instead.
A pallet load system is more of an investment but its way more flexible .https://www.erowa.com/en/solutions
 
At IMTS I had a chat with a vendor who was showing a robot loading pallets that then got swapped into the machine. And one of their staff opined that "yes, this is very powerful, but maybe too complicated for many shops to absorb, we suggest they start with just pallets". Of course they sell pallet pools for their machines so there may be some bias, but the point is well made.

(In other words I agree with dstryr.)

On the other hand, the workflow needs to support the investment - both the cost of the hardware, AND the learning curve. Russ bought something that is supposed to be reasonably well integrated and famaliar (they clearly know haas controls) - so maybe that's the investment that makes the most sense for their work flow.

There will be a prize for shops that can get automation to work for lowest total investment (money, time, floor space, headaches) to make it economic for relatively low value parts. Good luck.
 
On the other hand, the workflow needs to support the investment - both the cost of the hardware, AND the learning curve. Russ bought something that is supposed to be reasonably well integrated and famaliar (they clearly know haas controls) - so maybe that's the investment that makes the most sense for their work flow.

There will be a prize for shops that can get automation to work for lowest total investment (money, time, floor space, headaches) to make it economic for relatively low value parts. Good luck.


Nailed it!

We do repeat production work. All of our machines our Haas machines, two 2018 VF-2SS machines, one 2018 ST-20 lathe, one 2020 ST-10Y live tooled lathe, and of course the newest, a 2022 Fanuc robot loader on a VF-2SS.

Any of the three machinists out there on the floor can walk up to any machine and almost blindly reach for the buttons on the controls. I can take any job off any machine (mills or lathes) and run it in any of the others using the same program.

Both lathes have the same exact tools in the same exact spots in each machine, same for all my VMC's.

Our set up times for any job that runs in any machine are typically around 30 minutes tops and that usually is due to material type or alloy change over. On the majority of our first op set ups, parts require the tools that live in all the machines all the time. Face mills, end mills .125", .187", .250", .375" and .500", and a .500" finisher are always loaded and always ready to go. .125" and .250" chamfer mills as well.

The most common drills and form taps are also in there and never come out, #6-32, #8-32, #10-32, and 1/4-20, plus a .750" x 82* c'sink, and #3 ctr drill.

We also have .625" and a .750" inserted drills always loaded in as well.

To finish it off, they all have either a Lang, or Haas chip clearing fan as well.

Tools #1-5 are always left open for tools that are job specific.

It is entirely possible for a new job to hit the floor and the machinist to open the doors, pop a piece in each vise, or 4 pieces in the 2 vises, probe the 3 axis' and run the graphics screen to check for errors and push cycle start. 10 minute set ups are very common.


So along comes our new robot loader in a new VF-2SS we already have thousands of program for.

In the beginning, it used to take Andrew a few hours to set it up, but that's ONLY the robot. The parts being run in it are the same exact parts we've been making for over 15 years.

The curved handles you saw in a previous post is a prime example of a 10 minute set up. The only tool needing to be set up was a .125" carb crem. Put it in the holder, put it in the machine, set it off the tool setter, drop a piece in the vise, probe the part, push cycle start.

Setting up the robot was just over an hour for Andrew to do and that was only because he got the part from the table to vise and back out only to realize he never undid his clockwise motions when exchanging parts in the machine so when it came out to drop it off, it hit the rotational travel limit, so he had to go back and redo those steps.

Once though it was set up, it ran all day long with the exception of him spending 60 seconds to remove finished parts with new blanks on the table. We very quickly had many jobs with the 1st ops completed and stacked up waiting to go into the other two VF-2SS' to have the remaining ops finished.


That is exactly how it will be for that machine with the robot. It will handle nearly all of our 1st op pieces for every job without having to spend any extra time doing anything to the machine, the vise, or the tooling. I am quite confident it will be very soon before he has robot set ups down to a science.


Once those struggles are overcome, the time invested in the learning curve now, will be money in the bank later on.

Later,
Russ
 
Last edited:








 
Back
Top