Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
#317 SMW Zero point system
314 Plays2 years ago
Topics:
- MTDCNC visited Grimsmo
- Net 30 customers
- SMW Zero point system
- Willemin 2d wrap toolpaths
- 140 hours a week of Kern run time
- Fastems
- Tool life management
- Bambu 3d printer
Recommended
Transcript
Introduction to Manufacturing Journeys
00:00:00
Speaker
Good morning and welcome to the business of machining episode 317. My name is John Grimsmough. And my name is Sean Saunders. And this is the weekly podcast where John and John talk about their manufacturing journeys and businesses and what's going on in the shop and hiring people and the opposite of that and all kinds of manufacturing stuff. Is there more to unpack? Nothing. Okay, okay. Got it. What's new? What's been going
AeroX Miscollectors Visit
00:00:28
Speaker
on?
00:00:28
Speaker
Yesterday, we had a fun day. We had Tony Gunn from MTDCNC visit the shop for the whole day and his crew. That was really fun. He's a super cool guy. Yeah. Awesome. What were they doing up there? They're here with AeroX miscollectors, touring around a couple of different shops that use AeroX products. We're one of the poster child in Ontario for AeroX because we have eight of them. I don't even know.
00:00:55
Speaker
Um, and they're fantastic. So it was mostly centered around that like arrow X brought them up, um, and to, to do that. So, but it was cool to be able to, you know, show the shop, talk about things. I haven't done much YouTube lately, so it was kind of good to get back into it and like nerd out about cool stuff. Certainly got the bug back in my, in my pants to, uh, film more, you know? Yeah, that's fun. Yep.
00:01:24
Speaker
When that happens, do you find that you just allocate the day toward hosting or do you actually- Pretty much. Yeah. Even our whole manufacturing team, they got stuff done but it was a different kind of day, slower day production wise because they were around to help and ask questions and be part of the thing.
00:01:48
Speaker
move things out of our way if we needed them, things like that. We all had lunch together and it was good, but at the end of the day, everybody's like, man, I am fried. Yeah. It does take a lot out of your pipeline. Yeah, it's cool.
Customer Payment Challenges
00:02:05
Speaker
We, all right, so I did something, which we will see how it goes. There's a balance of doing something with conviction, but also recognizing I may regret this, but we have a customer. Sounds like the title of this podcast may regret this. Look, it's not a huge deal. I'm sure some folks listening will absolutely, you know,
00:02:31
Speaker
cheer this on. We have a pretty typical customer in that their initial purchase was in the thousands of dollars for the fixture plate in a kit, and then they continue to buy some accessories over the time. They are a pretty legitimate defense contractor, not a household name per se, but pretty legit.
00:02:54
Speaker
And their paperwork has fancy language about defense sector like priorities, blah, blah, blah. But these POs are to be blunt. For POs, they're annoyingly small, hundreds of dollars. Usually, we would
00:03:11
Speaker
Often we try to either push them in size higher or just force them to do a web order because it takes time and effort to process them as paperwork. Then in this case, chase down the payment. This is their third or fourth order.
00:03:27
Speaker
They repeatedly basically wait. So it's net 30 terms. They agreed to net 30. It's net 30 on the paperwork. And I wholeheartedly recognize some folks listening might tell you, you're kidding me. A lot of big defense contractors won't do better than net 60 or net 90. And I hear you, but for whatever better or worse, right or wrong, we have net 30 with them. And so after I usually wait five or 10 days after net 30, you know, in case the checks in the mail or, you know, obviously,
00:03:56
Speaker
So when I've emailed them now in the past two orders after about 40 days, their response after a day or two is usually like, oh, we have it over in XYZ department and it's going to be waiting on internal approval and then we'll get to it.
00:04:13
Speaker
And so again, this isn't the end of the world, but it rubs you the wrong way because it's kind of a dual standard of number one. These are the terms. And number two, they expect us to ship quickly. They've asked that before. So I wrote back to them and I said, on future orders, are you able to
00:04:33
Speaker
hold
Zero-Point System Launch
00:04:34
Speaker
to net 30, or would you prefer to be moved to prepay? And that was part of that question was in the update request in which they said, oh, by the way, he's over it. They used a fancy word, awaiting approval. And I'm kind of like, it is a $300 invoice. I don't, whatever. So they didn't even acknowledge that. So I just wrote back and said, appreciate the update. We'll move your account to prepay. They responded.
00:05:00
Speaker
Now look, what I find from folks I look up to is you can talk a big game. John Saunders here, I could talk a big game and I moved into pre-pay, but the reality is when they send in a big juicy order or even a $500 order and they are still on net 30, am I going to actually
00:05:17
Speaker
I'm actually going to walk away from the PO unless it's prepaid. We'll see. But you also don't get what you don't ask for. And we're small potatoes compared to them. But we're also doing OK. And I think they're buying from you. You're in charge here.
00:05:33
Speaker
I don't know. They want your product. Yeah. I mean, customer wins too, right? I don't know. To me, it's just like, look, it's not fair. Everybody else usually pays. And look, we get it. It's not a perfect world. And many net 30 customers will wait until the 31st day to write the check, which means you don't get it for 36 days. That's fine. This is more just about what's the process, especially as I step away or try to step away from some of this
00:05:58
Speaker
stuff like, look, how do we handle it? And what's what's fair to the customers at our pan time. So I'll leave it at that. But keep
Production Orders and Customer Demands
00:06:05
Speaker
you posted if they I bet you they are going to laugh at this and just completely disregard my little man's comment on it. But okay. Yeah, you got to try. Oh, that's fascinating. And it's like, on a tiny hand. I'm like, I'm sorry, you're going through this. But also it's like,
00:06:24
Speaker
not the end of the world. It's just annoying, it sounds like. Totally. Right? But it is the reality of business, right? So the way we do this now is we keep a manila folder. I'm still the one in charge of this. I won't be forever. And these are all the outstanding
00:06:43
Speaker
invoices, and then accounting system flags them when they're past due. And then we have a on our PO process sheet, which is the on the staple to the front of every purchase order. It's an internal SMW purchase order process sheet has all the good information, special instructions, where are we in the stage, stuff like that. We have a little section about what we're doing post 30 days default, who's emailing to the customer, what's the response?
00:07:09
Speaker
If somebody told me, oh my gosh, we're late or we're on vacation, or I'm sorry, I'll get to it. We're fine. But the whole like, wait till 45 days, then to give you some BS excuse and then say like, okay, well, I mean, whatever we stick to, if you want next 60, I can have that conversation. But you're going to stick to what you do. That's my soapbox.
00:07:31
Speaker
Yeah, fun times. For the most part, it sounds like the majority of your customers are like website, prepay, online orders. Not a big deal.
00:07:41
Speaker
That's a good question. By order volume, or by number of orders, for sure, website. But by dollar amount? I don't know. It's probably 30%, 40% purchase orders. Interesting. Just because a lot of the POs are, we're doing a pretty large custom plate set for a mix of Okuma and Toyota machines for a big company now. So that's the equivalent of your average, it's equivalent of 15 average web sale prices.
00:08:11
Speaker
Interesting. You could do, I guess you don't really have creditors, huh? No.
Competing with German Zero-Point Systems
00:08:20
Speaker
Everything's buy and then we ship. Back in the day, we did pre-orders, which I mean, we made it work, but I will not do it again. Take money upfront and then take too long to deliver. That's not a good place to be.
00:08:41
Speaker
Well, that's a good, interesting question. So we're, um, I've actually been working a lot on the zero point system, um, to come back to that. I know I've brought up a few times intermittently. Um, we had a design philosophy change end of last year, early this year, which was an excellent thing. I'm happy with where it's at. And then, um, honestly, we wanted to wait for the Wilhelmin to be up and running to help make some of the parts, which it is. It is amazing. And, uh,
00:09:10
Speaker
So we have a functional 3D printed prototype already. And we have a machined one that will probably be done today. And we'll end up using that internally for a couple of weeks probably. And then we'll start making, assuming there's some functionalities, OK. We've got a very short two or three user beta test
00:09:36
Speaker
or like alpha test really list of folks that we'll send them out to. But then we do have, and we'll put in the podcast description, we have a signup list with quite a few folks on it that are interested. So they asked the question is, I know you're different as like a kind of a consumer product, but
00:09:54
Speaker
Do you do a, would you consider doing a limited quantity discount kind of an early user? I guess it's a question of what are you doing? Are you just giving them a discount from being an early adapter? Are you looking for feedback? How that is a product? I haven't really gotten there yet to see what I want to do on that.
00:10:09
Speaker
Yeah, like the early adopter program or the first run, 10% off or something, with full disclosure that the design will change in three months of feedback kind of thing. It might not be perfectly fleshed out yet. Yeah. I think people would go for that. I mean, I know it'll work. I know you guys. It'll work fine. It can just be better with more experience.
00:10:38
Speaker
Yeah, and it's actually not even, cause my sort of opinion, especially on a first dev product is you always take care of the customer. Like, look, if you, you know, um, if you need to do the right thing, you do the right thing. I thought more is a, as a call to action, like, Hey, you know, we're going to have, we're going to have a hundred orders or a hundred pieces as the first mover advantage. If you want it on this, here's the opportunity to take advantage of it. Uh, I guess it's very Kickstarter thing. Otherwise you can go into the normal bucket.
00:11:07
Speaker
Which will be later, I guess. Do you want to make the hundred and then wait for feedback before you make any more? Yeah, that's actually, that is a good point. That's probably the best thing to do. It's annoying because you're like, sweet, we're in production now. Stop. Wait two months for feedback. Yeah, but I like that. You don't want to overproduce. For sure.
00:11:30
Speaker
It's been the first product that we've ever worked on truly as a team, so I am very much detached from being the lead on this. That's cool. Which is
Wilhelmin Machine and Saga Clips
00:11:40
Speaker
great because it gives me the chance to give feedback from a bit of an unknowledgeable perspective. I'm not intimately in the know on it. On the flip side, it feels very uncomfortable to not
00:11:51
Speaker
have been locked set through the process. Imagine if somebody in your department just brought you a flashlight or a new night. Yeah, it would be uncomfortable. Yeah, it is. But the reality is I need to get there.
00:12:08
Speaker
But yeah, it's good. I am legitimately excited, both for the product and legitimately excited because it really, I think, matures our product offering to where you have a true... Can you sell it to me? What's the elevator pitch here?
00:12:23
Speaker
Yeah, look, this is very much Red Ocean. There are quite a few players from the Unilock to Shunk to Aroa center that make a similar repeatable quick change, automatically actuated. So air style actuated mechanism. OK.
00:12:41
Speaker
Um, and so ours is going to be at a, you know, more, what we think is a very, uh, profitable to us, but also very much more competitive price point compared to your average say German made. Yeah. Cause my, my shunk Vero bases are like two grand a base. Yeah. Yep. And the pulse that itself's like 200 bucks. Mm-hmm.
00:13:06
Speaker
And so I think we've got an opportunity to deliver a product that has very acceptable or respectable tolerances, but also doesn't need to cost two grand. All right. Yep. That's cool. And is it made specifically to fit on your fixture plates?
00:13:26
Speaker
The initial version will. We are including some feature sets and holes that would allow it to not always have to be married to that. The initial plan would be. I think that's a good idea. You're making a product family. You're in the family now, so now you can use all of our cool stuff. You have, I would assume, hundreds if not thousands of fixture plates out there by now. This might fit on those.
00:13:55
Speaker
Oh, yeah, absolutely. Well, right. That's cool. Yeah, we will. What we aren't as sure about is the hobby version to whether we do a could be say just aluminum versus steel and slightly more forgiving tolerances for the version of
Implementing Zero-Point Systems
00:14:09
Speaker
the same style. Or we have a different design that was kind of our first one last year. I still really like it that we might do for the hobby version. Again, it would not now that we have the Willam and oh my god, these parts here. Look at that.
00:14:25
Speaker
Oh, that's got some features in it. Yep. Oh, maybe tapered five axis on the tip angled features, five axis deeper. Yes. Such a good element for it. Yep. So it's it's it's running. That's sick. It's cool. It's also a good segue. And then I'll stop hogging the mic here. But um,
00:14:51
Speaker
Like last night, I had a meeting, I came back at 3.30. I wanted to manually run some parts on the horizontal to help out because we're behind on one product. Got those made. And then, so basically when Garrett left it too, I did not have him start the night run because I wanted the machine to be idle when I got in at 3.30.
00:15:07
Speaker
So I did my work for 10 minutes and then I switched it into the overnight mode and hit run. And it was awesome to pop into the screen. You just do a double check of what's scheduled to run. And all six tombstones were already set up and scheduled to run. Nice.
00:15:24
Speaker
We don't have every face of every tombstone used. And some of the faces that we have set up for tombstone faces are not by all any means the most efficient. But we are at that point where we're sort of we can start to be we're start to be we're start to hit the limit of that machine. But
00:15:40
Speaker
10 of the, let's see if there's six faces, excuse me, four faces, six tombstones, 24 faces. I'd say eight or 10 of them are not full-time products, but it's just nice to have them set up. So those, we're going to start switching over into the zero point system. So where we can
00:15:57
Speaker
Use Fusion, we can push the offsets from Fusion, so we can do both coordinate system offset push to the Akuma, but also use probing and some macros we wrote to do a validation to make sure you're not running the wrong program on the wrong tomb center on the wrong base. And then when we want to run a particular product that's not super high volume, we don't have to do any more setup, you just put it on there and hit go.
00:16:21
Speaker
That's the greatest thing with a palletized four or five axis machine that has a million tools in it because your setups are just like everything's locked for for full production. But then you have a few pallets, few vices, if you whatever for just screw it around and all your setup tools are there and all your roughing tools and like outside of production. And it's so easy just to make a thing.
00:16:46
Speaker
And then there is like the way I even do it now is if I'm doing a test program, I'll run it as a scheduled program. So that when that's done, it just goes right into night production. Say that again. Like I can either run a program just by itself, like John's test vice, whatever. Um, and then it'll stop and it'll like wait for me to do another command. But if I scheduled it in our palletized run, then it will finish that and then go right into production afterwards.
00:17:14
Speaker
Yeah, that's a huge gripe that I don't have on the Akuma that I think it's borderline unacceptable. Couldn't, I don't know, could you just schedule your testy program? You, if you're running the machine, of course, manually. You have to actually like turn the key or something. Yep.
00:17:36
Speaker
Interesting. Yeah. In mine, it's just the difference between which master program I'm opening, whether I'm opening the, the end file, like, you know, make a park.h or palletmanagement.h is what I use. Hmm. Yep. Not cool. Couldn't you keep it in automatic mode and schedule your test program first?
00:18:01
Speaker
Uh, honestly, no, not even because let's say you're working, let's say tombstone five is in the machine and that's going to be your, Hey, I just want to quickly test out, uh, file one thousand one dot NC. You can keep operating that over in that file after you. So let's say you post that file from fusion and then
Programming Challenges
00:18:20
Speaker
you schedule palette five, which.
00:18:23
Speaker
would go ahead and allow you to hit cycle start from the APC side. It would start running that program. As soon as that program is done, you cannot rerun that program without taking the tombstone, lazy-suzing it out of the machine, and sometimes even to the low station, and then retelling the machine that it's fake been reloaded, and then running again.
00:18:44
Speaker
Oh, so if you just want to like, even if you want to air cut something or whatever. Basically, it thinks if you run Tombstone 5 in the PPC mode, it thinks it's ineligible to be run again until you pull it back to the operator load station to quote unquote reload the material. I guess I get the logic there, but yeah, kind of annoying if you want to play and produce at the same time. Anyway. You work around it.
00:19:17
Speaker
Um, speaking of Wilhelmin, so I've been running mine quite a bit too. Um, I finished these bushings that I was making out of a two steel. Um, so we're going to send those off to heat treat this week. And then I started working on the saga clip, which is the whole reason I bought this Wilhelmin. Um, so I spent the weekend programming that getting all my tool paths ready. Um,
00:19:43
Speaker
On the other day, I stayed late, and I ran a couple tool paths. I ran the turning tool paths, turning the OD, grooving some stuff. That went great. A little bit of chatter on the turning, just because my part's sticking out kind of far, and I figured that out, whatever. I'm not concerned. And then I started to do the milling, and I did a 2D pocket with a wrap feature, because it's like a 300 degree pocket. It doesn't go all the way around.
00:20:11
Speaker
Fun fact, don't do 2D wrap on the Willyman yet because the post will crash the tool. And so it's no big deal, but I lost a high-feed end mill. Basically, imagine if you go down to touch the surface and then you type A
00:20:32
Speaker
You know, 200, right? You would think the turning spindle would just rotate, but it doesn't. It rotates eccentrically around some invisible pointed space that isn't oval. And so I'm watching this and I took a video and I sent it to CJ and I sent it to some of the Willowman guys and I'm like, what's going on here? Like is my machine, does it have the center of rotation programmed wrong? Is some parameter like
00:20:57
Speaker
We've messed with my machine pretty hard with the FANUC update and the Inch conversion and everything. I'm like, okay, something's not set right, whatever. We'll go through all the parameters. It looks good. It looks good. Everything's good. Our current theory between CJ and Jim and Marcus from Wilhelmin and some other stuff is that the machine's fine.
00:21:18
Speaker
Willeman just doesn't read 2D-wrap programs properly because it seems like we wrote a test program that was like, go down to Z this, move to X this, rotate A this, and go out and do that a couple of times. It's a simple, simple program. I watch it and it's moving all eccentrically and all weird. Then we try a different feature and it does it about half as bad.
00:21:42
Speaker
And then CJ actually tried it on his new Wilhelmin and he's like, same results. What? Oh, okay. So you and Grant, be careful with 2d wrap. Um, and CJ is working on a solution right now. Well, I'm sure we'll figure it out, but it's interesting how it could not work because I know it's something about, I don't understand this, but a tool center point, like TCP, something, something versus, uh,
00:22:12
Speaker
The way it interprets a five-axis move like a multiple axis around the center point, Wilhelmin apparently does it weird. But have you done 2D contour wraps features? Technically, I don't think I've done a 2D contour try yet. Although my test program should be that. Because that would be the same. The same it should be. Like 2D Pocket, 2D Contour, but wrapped should give you the same code, more or less.
00:22:42
Speaker
Well, this probably doesn't help whatsoever. But I can tell you that the new Master Fusion module works, which Mastercam uses module works, by the way. But the Deburr feature in Fusion that has three, four, five axis modes, we're using the five axis Deburr
00:23:01
Speaker
I saw your video of that. That looks tasty. There's a little bit of what we call it like staccato-y. Wiggle. Yeah, which I will get figured out. It's kind of funny because I've seen that on Haas machines before and people are always like, oh, Haas can't handle it. I'm like, no, it's the code. Here's a good point. We've got to get it figured out. I tried.
00:23:24
Speaker
tried to program a deburr feature over the weekend and I couldn't get it to reach. It did a couple little lines but I wanted it to go this way and I gave up on time and energy to mess with the settings to get it to work properly but it looks like a really cool feature, like a tool path.
00:23:44
Speaker
Well, and it lets us do something I think you otherwise can't do, which I think is also a little bit disappointing that Fusion doesn't have a solution. Let's say, take a can of pop, like a cylinder. If you had a hole on the side of it, that hole, the circumference of the hole is now along the three axis curve of the side. You can't,
00:24:10
Speaker
We talked about this on WhatsApp. You can't deburr that infusion with any sort of a chamfer tool set that has an offset tip. You can have it cut with the center tip of the tool, but it's not great to do it that way. And I get that the chamfer won't be perfect because it'll be a different amount of tool engagement as it rotates through that sort of three-axis angle. But what's annoying is that the Trace toolpath doesn't respect your
00:24:36
Speaker
some radial, negative radial, sorry, I'm getting super noisier. Yeah, no, I know exactly what you mean. Yeah, whereas the DBERT toolpath will take that tool and it'll turn it into an A-axis, AZ loop. Right. So it'll give you the correct chamfer because it's rotating the A, so your tool has the correct engagement with the chamfer angle as it goes. Interesting. Yeah, I know on our RAS candles, we do that kind of three-axis chamfer.
00:25:04
Speaker
But as the three axes move, not anything rotating. And I think we're getting really good results, but I don't think it's an absolutely perfectly even chamfer around. It's close, though. It's really close. Yeah. I don't care. And like I told you, it's either using the trace toolpath or the project toolpath. I don't remember which. Probably project is what I would guess. And then you just mess with the offsets, the stock to leave, and the
00:25:31
Speaker
chamfer size and all that to make it look like what you want it to look. Yeah, and for production, to solve my dilemma, you could offset three-dimensionally offset or patch, and you can absolutely drive the toolpath you want that way. It's just annoying that you have to go through those steps versus just offset the tip, please. Well, so much of the time, if you want something specific like you're drawing a sketch to make it do what you want it to do.
00:25:58
Speaker
Yeah, exactly. It's nice to have that ability and that power. I was just showing Garrett that we are running hobby mod vices on the horizontal now. Those used to be on a VF2. And the pattern of them, we hold the material in mod vices to make them in dual stations. So the material is pretty close to each other left to right. So when I'm
00:26:23
Speaker
decking part number two, I don't want it to risk hitting part number one. And this is actually also, I wouldn't call it a limitation of fusion, but it's a quirky scenario where I actually create dummy solid models of where the parts are on the setup. Can I only program one part and then we pattern it across four offsets or eight offsets?
00:26:47
Speaker
But I create dummy parts that show me where the other workpiece of material are so that I can simulate and look at if it's got it. And most of the facing operations are now sketch driven traces because we want to control exactly how, I know you do this all the time, like how the tool comes in, moves over, cleans up this area, doesn't clean up that area, and so forth. Yep. And those little tricks you just learned over time that I guess it's our job to share more.
00:27:15
Speaker
Cool.
Tool Load and Life Management
00:27:17
Speaker
So Willem is going good? Willem is going great. Last week I got that power turning off the rebooting problem. I think just tightening all the power cables, every little connector in the enclosure, I think that solved it. It's been fine. Hasn't rebooted since on a tool change. Good.
00:27:36
Speaker
So that's good. Yeah, I've been running it quite a bit. And I'm kind of stuck on this 2D-wrap toolpath, because it's the toolpath that I want to use several places on the saga. It's exactly what we do in the Nakamura. It's like, put tool down, rotate, A or C axis, pull tool up. Like, perfect. I don't know. Maybe there's some five-axis toolpaths that do the same thing that I would have to play with. But I really just want to get these to work.
00:28:05
Speaker
It's actually quite a dangerous thing, post-wise, if it's just do a 2D-wrap toolpath, crash the tool, or if it's worse, crash a holder, crash a spindle, crash. That's what gets me, I don't know enough about the kernel, why wouldn't be happy with that? We're figuring it out. My team is on it. Sorry, I'm yawning here.
00:28:32
Speaker
What else has been going on? Um, that's all I got for my notes so far, but current's going good. We pulled, uh, two weeks ago, I calculated 140 hours of spindle time. It's incredible. Yep. Out of 168 in the week. Uh, very happy with that. And the only thing we need to run till Monday is more fixturing material, more pallets, which just showed up today. Um, yeah.
00:29:01
Speaker
Like we have all the bases. We even have the rough tombstones. I just need, I needed more clamp material. I needed more top fixtures, more like stuff. So all that stuff just came in today. About $2,000 of raw material, like clampy material and random brass rods just for playing around on the lathes and stuff like our local metal store. I've had this shopping cart going for six months now and I finally just hit go. It's funny.
00:29:29
Speaker
So that's cool. The Fastoms, you know that company Fastoms that makes a row of automation sales or automation solutions. They were at that Okuma tech showcase thing and their shirt logos had a little like a red circle and the number 8760 in it. 87, okay. Do you know? 87. No, it doesn't ring a bell.
00:29:56
Speaker
Yeah, I didn't know either. That's like a metal grade, a grade of metal. No, way better. Yeah. Number of hours. Oh, that is cool. I like that a lot. Which is 168 times 52. Yeah, yeah. Whatever it is. Yeah. I think you're right. 24 times 365. Yeah, but you said it right, 168 times 52.
00:30:24
Speaker
Cool. No, it's off by one day. I guess there's... Leap year? I don't know. Yeah, I don't know. How's that off by one day? I don't know. Math don't lie. You're 24 short, 24 hour short, 52 weeks in a year. 168 hours in a week and 52 weeks in a year. Somebody out there is killing themselves thinking about this. 87.60.
00:30:50
Speaker
Yeah. Well here, think about it this way, 365 divided 52 is seven and seven days, 7.02 days. So there is a rounding anyway. Cool. Yeah.
00:31:06
Speaker
Okay, so Fastem's... You were talking about 140 hours a week in the current and I'm like, that's crazy to think you could run a machine 87, I mean, I'm sure there are machines out there that are too close to that because you just keep them loaded up like that. They just never stop. Yeah, exactly. And we've finally gotten, I mean, over the past two, three years, we've gotten our tool life management so dialed that, gosh, I don't know if we ever break in mills on the current, like one a month.
00:31:36
Speaker
That's incredible. Two a month, maybe. And it's just like, oh, crap, whatever, no big deal. But for the most part, it is dialed. Like I've said before, we're exporting the tool table to an Excel spreadsheet on a Raspberry Pi. We're crunching the data through some Python scripts to estimate our tool life looking forward based on the schedule that we've scheduled a machine at.
00:32:01
Speaker
So we know like, oh, we've got, you know, 10 pallets lined up. They all need this. So tool 100 has to be replaced in order to successfully finish this round. And it just tells us that.
00:32:12
Speaker
It's so fantastic. Just this morning, Angela was asking for a look ahead feature. Right now, it's just based on the actual schedule that we've put in the machine. He's like, what if I want to double that? In the tool table in the current, maybe add another row where it's like, I want this times two. I've scheduled it once, but I want to run it again. Can you estimate the tool life for the schedule plus whatever the adders are? I was like, I think I can do that so that
00:32:42
Speaker
so that let's say he wants to get the tools perfect and let Steven reload the whole machine in the morning and then Angela wouldn't have to show up to change tools. It's prepared for that much or it would tell us how many sister tools we should be adding to supply this because we don't really use sister tools right now because the system is so tight. I have so much respect for that, John. That's incredible. Cool. Thank you.
00:33:10
Speaker
We need to solve this and I actually was asking at that same Akuma event some help on tool light management and I actually think it's not going to work because you can set things like load limits or time limits but all it's going to do is stop the machine.
00:33:32
Speaker
100 tools get used every night. Statistically, it's going to stop the machine every single night for some reason. Do you have that happen? No, because to be blunt, we're not using tool limit. We use break detect on certain tools that we know are higher risk and that will also stop the machine.
00:33:54
Speaker
I know one solution we could do would be to combine load limits with sister tooling, but it's not that graceful of a solution because I can't afford nor do I have the space to do sister tools everywhere. Exactly. Where do you stop the line?
00:34:11
Speaker
And the way we want to solve the problem is to take tool 49. Here's what I want to do. And this is actually on my list to start just researching. But Yokuma displays the current tool load in the cut. There's got to be a way, and there might not, because I'm nervous about hitting roadblocks here. But I want to start dripping that data out to a database or a spreadsheet so that I can hopefully look at what's the current high limit
00:34:38
Speaker
by either, hopefully by, I don't think I can do by operation within a part, but at least by part like file. Yep. That's how I do it. So then we have this database and you can start looking at the trends because I don't know and I don't frankly care if tool 49 is averaging four to five.
00:34:56
Speaker
percent spindle load, then all I want to do is we
Optimizing Tool Usage
00:34:59
Speaker
could do the Python math or whatever you want to call it, script parsing this data to say, okay, as soon as that gets to eight or nine percent, we just want to flag it in the morning and you can just look through a printout that says, okay, these six tools are above their trending average load limit. We let them still run, but you should take a look at them or replace them.
00:35:19
Speaker
Yeah, you're gaining data and then you can visually look at the end mill under microscope and be like, oh yeah, it is kind of torn. So we do everything time-based, not load-based at all because like all of our tools are eighth-inch or smaller. You know, we have some quarter, some three-eighths but load limit doesn't work for me.
00:35:39
Speaker
Because an eighth inch end mill is not going to take more load if it's got a flute missing. Not really. Yeah, right. For the bigger roughing tools, anything quarter and above, that's actually hawking a bit of material, but I'm still pulling like 3% spindle load. Yeah. So it's not accurate for me. But I found time-based and then also
00:36:03
Speaker
knowing, okay, so running rask handles tool 100 runs for six minutes on that palette on that individual program. So
00:36:14
Speaker
It knows last time it ran six minutes. So next time it's going to run six minutes. So it uses that math to calculate like, oh, you've got six of those scheduled. So it's going to take 36 minutes. But your limits at 100 and your current times at 92 and you don't have 30 minutes of whatever. So it tells you while the machine's running, it tells you ahead of time. Like if you replace it now, you're good to go. No, I mean, that would be a better solution. But I don't.
00:36:43
Speaker
I'd rather be reacting to the machine intelligence of looking at the load. I don't think this will work, but would you have to figure something out because it's not good to not have anything in place, but again, just stopping the machine. It sucks. It sucks. I think with sister tools, you could put the worn out tool in the NG pockets. There's four NG, which I think is Japanese for no good. NFG is the American translation.
00:37:13
Speaker
Um, but, uh, that is, um, it's just not the same as, well, first of all, we'd run out of potentially run out of those pockets. And I just wait. So it, it actually, if it actually puts it in a set, a special separate pocket in the tool matrix, that's kind of neat. It is kind of neat. Whoa. Yeah. Is it a random matrix where it puts them anywhere at any time? It's like tool 100 goes in 100.
00:37:40
Speaker
pocket in the wine rack 100 always. Yeah. Okay. Or in the NFG pocket, if it's bad. Just NG, John. It's a family friendly.
00:37:52
Speaker
Which my kids now respond to that. I used to say jinx growing up. They say Coca-Cola sandwich. What? I've never heard that. I guess proof that we don't know our kids. Yeah, they've got their own thing going on. Super sus. Yep. That one I know. I just learned it. Oh, yeah.
00:38:17
Speaker
Yeah. Cool. Okay. So thinking about
Tool Load Tracking on Okuma Machines
00:38:20
Speaker
how you guys make your pallets, fixture plates, like you're drilling a big hole, you're tapping a big hole, you're face milling, like these take horsepower. Yes. Right. So I think tool load could work for many of your tools, but like any graving end mill, if it's gone, I mean, tool break to tech would get that.
00:38:39
Speaker
fixture plates run during the day with a person there. This is only horizontal stuff, which is mod vices, everything else that runs unattended. My question really is, do you use a lot of small tools that will benefit from load detection at all? Small tools won't, you mean? Right. Yeah.
00:39:01
Speaker
No, but we use plenty of quarter inch, three eights, drills and mills, high feed mills, where there's enough spindle load to know. And look, the other thing I've thought about doing is just setting up a GoPro and running it, the GoPro during a certain part and then just getting an idea to set the load limits manually. But then you're just left with this, like what happens then? And maybe I'm just not educated enough on Akuma options because if you could
00:39:30
Speaker
Let's say you want to have a 15% load limit on a drill. If it violates that load limit, but it allows the tool to keep running, it just flags it, emails you. The Akuma can't email you that I know. It just says, hey, by the way, last night, that tool went over that load limit. Just want to let you know. That would be freaking perfect. Right, right.
00:39:49
Speaker
And like you said, if you GoPro the control, at least for a data acquisition standpoint, a couple nights, it would tell you something. You might. I thought about upworking. The video feed should be able to get sent to somebody that's smarter than me to then have, not AI, except an overuse word these days, but like have a computer program watch the video and scrape the percentage values using OCR.
00:40:17
Speaker
Interesting. It says, give me the output. I don't sit there and scrub through it all. That sounds like a $70 solution that's waiting to happen. It could be a $270 solution. It'd still be freaking worth it. That'd be cool. My iPhone now can take a picture and then OCR all the text in the picture. I assume you're an Android guy, right? Yeah, I'm an Android guy. I haven't really done that.
00:40:44
Speaker
All I know is I can take the camera app and look at a QR code and it takes me right to the website. I'm like, yay. I'm curious. You must be able to do that because frankly, I thought. Just from the camera app or from some special. No, so I have my phone open right now and I need it as a Mari tool drill card and it gives you a little text pop up and see if I can do that. Yeah.
00:41:07
Speaker
from the camera. Look, yep, it just pulled that up. And that's now actually didn't recognize the T because it's that weird green. But it just lets you copy and paste like a serial number or a, you know, something it's super annoying, like a Mac address for a ethernet device. Right. You just copy and paste it into your phone.
00:41:25
Speaker
think mine can do that. It'll like, if I'm taking a picture of a document, a piece of paper, it'll like outline it and be like, you just want to outline just this document, not the background or anything. Oh, interesting. That's mildly helpful too, but. Cool. Fun stuff.
Fixture Production Planning
00:41:44
Speaker
That flew by. What are you up to today?
00:41:47
Speaker
Uh, I have several texts from the Wilhelmin people to see this, uh, thing. So I got to look at those really interested. I got Marcus suggested one thing to test, um, just to gather some data. So let's see if that works and then.
00:42:05
Speaker
start sorting in and figuring out what I'm going to do with all that steel that just came in for fixed material. I need to make a plan. What I really want to do is map out this chunk of material is going to make rask top clamp and here's the program and here's the fusion file and here's how to run it and here's how to set it up. Angelo, go ahead.
00:42:25
Speaker
It's called a process man. Yeah, exactly. Yeah, process man. Those things are clutch. Yeah. Yes. Cool. Yeah, I did watch your video on that. Exactly.
00:42:40
Speaker
Because we have a lot of parts to make, probably dozens of individual fixture parts to make, some multiples, but just a lot of different stuff. Just to duplicate more rask tombstones, which is a tombstone in three operations and then also 20 clamps. They're all different and they're all weird. But we've made them all before, so it's literally just rinse and repeat. Why not send out the clamps to be metal 3D printed? That's a really good question.
00:43:11
Speaker
I don't know. I hadn't really thought about it. I think, I mean, there's shapeways, 3D hubs, zometry are all viable options, but like, we have, oh, sorry, to be completely transparent, I have, ooh, we do that once. It's not like I'm an avid activist. I don't mean to be, you know, making stuff up here, but I know a buddy who did it, and it was surprisingly reasonable, and you know, depending on the stress properties and tolerance, I bet you it could be fine. Yeah, that's really fascinating.
00:43:40
Speaker
And now I'm in that quandary where it's like, but I already have all the material. I already have all the programs. You load it and go. I'll probably just do that. But for the future, that's a really neat idea. It's the 3D printing of the metal world. So on that note, we had some really weird shapes.
00:44:01
Speaker
for the zero point system. And OP-1 in those weird shapes was fine. Like one of them was in the EMC, one of them was in the lathe. So no big deal, because you just use the material itself to work the OP-0. OP-2, not so much fun. And I didn't want to make custom crazy soft jaws, so we bamboo'd them all.
3D Printing in Manufacturing
00:44:20
Speaker
And I did. Is that a word though? I bamboo'd them? Yep. It's like a, between a swishiest machine and a...
00:44:27
Speaker
And so it was great. Like one of them was a circle circular part, but it had some areas that need to be relieved. And then I 3d printed it with a little pin in the jaw to cloth the part. And then I just put the coordinate system offset in the soft jaw. So I don't even have to probe the top padded part. And it worked great. I mean, look, it's not perfect.
00:44:47
Speaker
It won't work for production for a variety of reasons, but to get one done, totally easy. And I'll tell you, that's one of the things I'm now loving. My perspective on fast 3D printing has completely changed. I didn't think it was important whatsoever before because we don't produce a high quantity of parts each week or month or whatever. But having that ability to just print something you haven't done in an hour, it lets you iterate faster and more people have time on the printer throughout the day.
00:45:16
Speaker
I will now never go back. It's now the only choice because I don't know how to describe it. It's just so much better. That's amazing. Yeah. I'm quite tempted to just order two of those and be done with it. Yeah. It's been great so far. Cool. Cool. All right. You up to anything cool today?
00:45:38
Speaker
Um, programming some of those zero point parts, which it's super fun. Yeah. It's super fun. Getting some stuff in a row ducks row for new person to start next week on logistics and operations coordinator. Sweet. And I got starts on Tuesday too. Ah, Thursday here. Cool. Um, and then I, I'm probably guilty of making a little bit of a too long of a to do list for stuff like Akuma to life value stuff. So I probably need to just sit down and go be a surgeon on that. Yeah.
00:46:08
Speaker
That's all. Cool. With the tool life, I think you need to gather data and then analyze and see what the next step is.
00:46:15
Speaker
Yeah, that's what I don't know is if the Okuma is displaying that value, you would think there's some way to capture it. I don't know if it's possible or I don't know if it's possible without being a Okuma software partner. Yeah. TMac from Caron Engineering basically does this sort of thing, but it's five-figure software. And it's not what I want right now. I've heard good things about it. Yeah.
00:46:46
Speaker
Yeah, anything that's like a live display, you can't just output a macro with a dprint or something because it's a live display. It's like constantly changing.
00:46:57
Speaker
Look, I would rather hire a, oh my God, this could be amazing, hire an upward person to do a like VGA or like HDMI split for output of the OSP into a inboard processing screen and start scraping the data real time based on the file name and the tool number and the mode value and build your own darn system that's real time adjusting and reacting to what the machine is showing. Yeah. What are you waiting for?
00:47:26
Speaker
Yeah, that's cool. Did you just come up with that or has that been rolling around your head? It's been chewing on it before because I kind of laugh when people say you can't do something. I'm like, let me tell you something. You don't know me very well.
00:47:41
Speaker
Well, it goes back to the Tormach days when I wanted the machine to send me a text message when it was done. It was in my New York basement. And you kind of couldn't do that with PathPilot. So I just put an LED, a photo sensor over the spindle light in the machine. And when the spindle light went out, it tripped a if this is an atom on Arduino. So it's kind of like this hacky way of interfacing with the machine. The data is there. You're just not allowed to access it unless you get real creative on how you access it. Right. So you get real creative.
00:48:13
Speaker
Scraping a screen section in real-time OCR shouldn't be that crazy these days. You're correct. I know nothing about it, but I know that programmers are very smart these days, and that's like child's play to people, you know? Yeah. Oh, that's cool. Cool. All right. I'll see you. Have a good one. Bye.