Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
#438 Pissed: Motion sensors when moving machines
80 Plays5 minutes ago
Topics:
- Within Tolerance podcast shoutout
- Pissed: GPS sensors when moving machines
- First part on SMW's Brother
- Speedio optimization at Grimsmo
- SMW Dec 18th hosting Toolpath webinar
- Grimsmo vision system deep dive
- Ai tip - feeding it the right imformation
- Shopify integration
Recommended
Transcript
Introduction and Podcast Goals
00:00:01
John S
Good morning. Welcome to the business of machining episode 438. name is John Saunders.
00:00:07
johngrimsmo
And my name is John Grimsmo.
00:00:09
John S
And we hope that everyone listening is a little bit better off in, on their passion and awareness, uh, to be manufacturing entrepreneurs. And you can learn through our successes, but also through our mistakes.
00:00:21
johngrimsmo
And this is always trying to be an open conversation between two friends who are, you know, trudging through the mud in the same time.
00:00:21
John S
Yes.
00:00:27
johngrimsmo
Sometimes, you know, sometimes one of us is doing better than the other.
00:00:29
John S
Yes.
00:00:30
johngrimsmo
And here we talk about it, private conversation.
00:00:33
John S
Uh, private
Successes and Challenges in Podcasting
00:00:34
John S
conversation. Well, I'm going to open with two things. One is probably one of the coolest things that's happened to me. And the other one, I am very, to put it politely, very hot about and pis and pissed.
00:00:45
johngrimsmo
Polar opposites. Okay.
00:00:47
John S
um The compliment, i don't I'm sure I didn't mention this last week. I think it just i think I got the news right after um Dylan from With Intolerance and Protea Machining DMed me that the episode that he hosted with me as the guest was his number one listened to episode of 2025.
00:01:04
johngrimsmo
Nice.
00:01:05
John S
And- I don't mean to, I just, it was the best compliment because it can't be fabricated. Like it just was, he did a great job with the interview and the questions and I hope it.
00:01:12
johngrimsmo
Yeah. he did I listened to that one. Yeah.
00:01:15
John S
Yeah. And so, I mean, thank you, Dylan. Like I, I, it was a very flattering, it's a fun little flex, but it's also, did I mention that? feel like I'm saying this. mentioned it last week. Am I repeating myself?
00:01:24
johngrimsmo
Nope. I think you you posted it online real quick after the fact.
00:01:25
John S
Okay.
00:01:27
John S
Yeah.
00:01:28
johngrimsmo
You had, yeah, no, that's wonderful, man.
00:01:29
John S
Yeah.
00:01:30
johngrimsmo
Good for you.
00:01:31
John S
And no disservice. He's had so many talented people on. He does a great job with his episodes.
00:01:34
johngrimsmo
Absolutely. Yeah.
00:01:35
John S
Um, but all the more reason why i'm like, that's pretty fun. Like feels good to have that little validating. And I i posted on a little bit of a share slash rant afterward about like what I think that means to be in this day and age of things.
00:01:47
John S
And I hope that that story, but what we do rings, rings true.
00:01:50
johngrimsmo
Yeah, and I was listening to that when it came out, and I learned some things about you that I never knew. And we talk every week, and it's like, yeah, it's nice.
00:01:55
John S
Ooh. Isn't that funny? Yeah. Well, he did. You ever watch Hot Ones on the the wings?
00:02:01
johngrimsmo
A couple times, yeah, yeah.
00:02:03
John S
That Hoshan Evans does a great job or his team does because they'll be like, in 12th grade, you did not get the lead role in the high school play, Annie. And the person is like, how did you know that?
00:02:13
John S
Like, I don't know if Dylan did his homework or what, but I was, so I mean, he clearly knew what questions to ask.
00:02:13
johngrimsmo
Yeah, yeah.
00:02:19
johngrimsmo
Exactly. You know, that's great.
Controversial Haas Policy Discussion
00:02:21
John S
um So that's my high. Here's my low. I'm i'm i'm pissed. But i'm gonna I want to keep this diplomatic because I i want to be a good custodian to the industry and and so forth. I also think it's a major PSA. um We purchased a new Haas UMC 400 SS. It was delivered. The service tech is here installing it. We don't know exactly where we want the machine to be positioned long term because we have a robot coming for it and we're going to actually, that's robots going to come quite soon. And then we also, when we hopefully sell the horizontal, we'll end moving the shop around even more. So, you know, we have rigging skates, moving this machine like that couldn't be less of a big deal. You know, it takes 20, 30 minutes to get up on skates. We move it around, we're done.
00:03:04
John S
The service tech goes, oh, you can't do that anymore. said, what? And he said, yes, if you move that machine one inch, a Haas factory tech has to do a service call and come out to your facility.
00:03:15
John S
And I am upset because our sales guy never told us this.
00:03:15
johngrimsmo
What?
00:03:18
John S
And I think that's ah So it's just a complete, there's just, you know, there's sales, there's service techs that will, or excuse me, there's sales folks that are honest and candid with you and and work with you. And then there's people that hide the monkey. And I have zero tolerance for the latter. And the excuse of like, well, we didn't know you didn' know. No, you are aware this is a major change. And hold on, it's, there part of this gets better, but not really. um Um, and this, ah all this came about for me, I found out yesterday and that I've done some homework this morning.
00:03:47
John S
So it's all very fresh news. I want to continue to digest this, but I think it's a real PSA to anyone who's considered considering buying a Haas machine of which we have bought.
00:03:49
johngrimsmo
Okay. Mm-hmm.
00:03:55
John S
I don't even know at this point, 12, 15, like we love them. We had great experience with them. We, we like, um, To be very fair fair, when we first started this, Haas offered us a consignment thing like they do with many other influencers. Nope, we have bought the machines. Like I wanted to, kind of goes back to my, with intolerance, being wholesome, like I just, I like our story and not just being a shill for somebody.
00:04:16
John S
Anyway. So I assumed incorrectly that this would be on the UMC five axis machines for reasons that lots of people have heard, but I very rarely meet people who can site cite actual laws around things like ITAR and things like the actual paperwork.
00:04:31
John S
It's still a whole lot of talk about, oh, this machine could be used for certain defense or or energy applications.
00:04:36
johngrimsmo
Yeah.
00:04:37
John S
um Different builders handle different ways. um So I'd like to learn more of a little bit about that. And frankly, if anyone listening wants to share some specific, I'd be happy to listen, but here's what I know so far. The the bad news is that, no, it's not on the Haas UMC machines.
00:04:53
John S
It's on every Haas machine.
00:04:56
johngrimsmo
Since.
00:04:57
John S
I don't know. the so the ah thing I saw this morning on the Haas website referred to a February 2025, which would be 10 months
00:05:05
johngrimsmo
Some recent, yeah.
00:05:07
John S
And my opinion, which doesn't necessarily matter, is I have no problem with reasonable measures that that speak to a machine tool not being able to sold outside of the United States, certainly to certain countries, and maybe even some hurdle of a machine being moved physical locations. um I don't love that, but I understand that if I need to move my machine to a different facility and within Zanesville, great, um that that I cannot move it three feet and that that then requires an on-site visit, um, is I mean, unacceptable.
00:05:42
John S
I told I'm yeah, it's unacceptable. Good news is, um, number one, apparently you can call and you can get a five day grace period. So, um, I guess if we, i don't know if we didn't know that and we moved the machine we're like Oh my God, the machine's bricked.
00:05:56
John S
We can call, they give us a code. It gives us five days for them to then get out to us to like per,
00:06:01
johngrimsmo
They still have to come out to you.
00:06:02
John S
They do sell to come out to you.
00:06:03
johngrimsmo
Yeah, yeah.
00:06:03
John S
I also am am told, but I want to kind of try to verify this further, that that can only occur once in the lifetime of the machine. So you, you've materially changed the value proposition of a Haas machine at this point, because now if I buy, if I go to buy a machine, a Haas machine, i use my guests, I'm going to ask, Hey, have you already used up that code?
00:06:10
johngrimsmo
Hmm.
00:06:13
johngrimsmo
Yeah.
00:06:20
John S
Um, and, um,
00:06:23
johngrimsmo
Well, doesn't matter because the text still has to come out.
00:06:27
John S
No, no, no. What I mean is like that code has being able to actually use that code once has value. So if I buy a Haas machine from somebody, I'm going to want to know if that code is still around. If I want to use it down the road.
00:06:40
John S
It makes sense.
00:06:40
johngrimsmo
But if you use the code and then a text still has to come out, how's that much different than a tech having to come out without the code?
00:06:47
John S
Well, in this case, because it took me a week to get a a person here just tot install the darn machine.
00:06:51
johngrimsmo
Yeah, OK.
00:06:51
John S
um So I understand you have to have a tech come out if you move the machine sell to somebody in Buffalo, New York. But like, anyway. um So, and the fact that it's an onsite service call seems ah ridiculous.
00:07:06
John S
Now I understand some of the security natures around that, but i mean I'm not a, i don't know. There's other ways builders handle this with paperwork, with affidavits, with notaries, with documentation, GPS sensors that track the machine's location, not just
00:07:16
johngrimsmo
Yeah.
00:07:19
John S
ah a one foot move. So I'm pretty pissed. I'm pretty hot about it. I don't i think it's uncool. I'm disappointed to see that this how they're handling it. I also want to give them a chance. I reached out to few people. I'm hoping to hear more. At this time, so far as I know, um other builders are not doing this.
00:07:36
John S
Yeah, I'm aware that some machines have GPS location sensors.
00:07:40
johngrimsmo
yeah
00:07:40
John S
I'm not aware of any that have this level of them. um I asked some friends at competing builders right now for things like, hey, if I went and bought a U500 or if I bought a Doosan or a Herco, they didn' do not those even the five axis don't have those sensors on them.
00:07:50
johngrimsmo
Mm-hmm.
00:07:57
johngrimsmo
Yeah, because it's always touted about like, this machine's too accurate. So therefore, it can't go to these countries or whatever.
00:08:04
John S
Yes.
00:08:04
johngrimsmo
So when when I bought my Mori in 2015, I heard that it has some sort of sensor like this on it. I don't know i don't know too many details, but I know when we got it delivered, i mean, the tech was there already and he kind of mentioned it to us.
00:08:11
John S
Okay.
00:08:17
johngrimsmo
And then when we moved it across the street, um the riggers came, they slapped it on the truck, they brought it back here, it turned right on. And...
00:08:24
John S
You never worried about it.
00:08:25
johngrimsmo
never worried.
00:08:25
John S
Yeah, sure.
00:08:25
johngrimsmo
i was ready to but like it it didn't need it.
00:08:27
John S
Yeah. Yeah.
00:08:28
johngrimsmo
And that was, I don't know, like a quarter of a mile, not even.
00:08:33
John S
Yeah.
00:08:33
johngrimsmo
So I, I see value, like if you put a radius, if it moves more than five miles, then that's a different story.
00:08:41
John S
Yeah.
00:08:42
johngrimsmo
But within your own facility? That's annoying. That's that's.
00:08:46
John S
there There's an ownership question around, wait minute here.
00:08:48
johngrimsmo
Yeah.
00:08:48
John S
This is a, if this, I've heard rumors that the early sensors um were so sensitive that they were getting disrupted during the cut. So i want to be careful to not play off into the unreasonable realm. If they've fixed that, then they fixed it. But this idea that you can brick a device um inadvertently, um,
00:09:08
John S
And I now have to incur, you know, a service call is probably a thousand bucks for someone to come out, the drive time, be here for an hour or two and go back. um i um I have an issue with that.
00:09:20
John S
So I need to learn more. I'll keep you updated. I just, I'm going to, the PSA I wanted to do right now. Oh, by the way, I'm aware of this whole ah Russia debacle with Haas, whatever, a year ago.
00:09:30
John S
I don't care about it. I'm happy to be an American that understands national security and whoever did what to get anything somewhere else, I don't care about it. And I understand, again, if there's some reasonable things that have to be done to make sure whether it's a CNC machine or whether it's a Glock 19 doesn't get exported fine, but this approach and not being told about it and this way they're handling it, I don't
Exploring New Machinery and Automation
00:09:51
John S
agree with.
00:09:51
John S
Yeah.
00:09:51
johngrimsmo
Yeah, yeah. Bummer. sorry Sorry to hear that.
00:09:55
John S
Yeah, that's my rant.
00:09:56
johngrimsmo
Yeah.
00:09:57
John S
Yeah. um Yeah. Well, um ah my list, and then I'm going to let you run with this. We made our first brother part. um We are learning some really good stuff about LLMs. Oh, I have some an event, two events I'm doing, participating, I wanted to mention, but I'll do it later.
00:10:19
John S
And I want to ask you about your time tracking update stuff. But what do what do you got?
00:10:21
johngrimsmo
Oh, yeah. Well, let's let's close out the Speedio conversations. You can do your side and then I've got Speedio optimization on my list.
00:10:29
John S
Okay.
00:10:29
johngrimsmo
So let's let's do that. You first.
00:10:34
John S
It's installed. The first day the apps guy was here. Justin from Toolpath was actually down for fun, both because of the Brother, but just in general to hang out. He helped us a ton, so thank you. And there's been a ton of people in that Brother community have done some really good things with macro improvements for the Bloom probing.
00:10:48
John S
Ed right now is kind of running that machine the way we've, I won't say informally, but it's not super rigid. Ed is kind of learning how to run the machine. I'm going to handle the cam of the new parts we're going to run in production. And then Grant and Courtney are going handle the automation that lights out, your robot side of it.
00:11:06
John S
That sounds very structured and siloed. It wasn't the point.
00:11:09
johngrimsmo
sounds great
00:11:09
John S
It was more just, yeah, we you needed somebody to be there with the apps guy and then Ed can teach.
00:11:13
johngrimsmo
yeah divide and conquer yep yep
00:11:15
John S
Bingo. Um, so we we don't we don't have coolant in the machine cause we're waiting for the Jorgensen conveyor to show up So we dry cut some aluminum went fine and then we just dry cut some steel touching off tools.
00:11:25
John S
Um, nothing more exciting than that other than we're, we're getting ready to to rock and roll with it.
00:11:30
johngrimsmo
Nice. What probes tool setters did you end up with?
00:11:35
John S
Bloom Z nano table probe and the bloom spindle probe. i don't know if there's a model option on it, but.
00:11:38
johngrimsmo
Okay. do Do you remember if it's infrared or wireless?
00:11:43
John S
It has LEDs. Um, I, so I do, I believe it's infrared and not radio.
00:11:48
johngrimsmo
think they all have LEDs. I think it's infrared.
00:11:50
John S
Oh, I, I'm, ah I'm fairly against radio, but not the hill I want to die on.
00:11:50
johngrimsmo
Okay. Right.
00:11:57
johngrimsmo
Yeah, I think I can't remember. I feel dumb. I can't remember. I think I changed mine from infrared to radio for reasons and drawing a blank right now.
00:12:03
John S
I think you did. Yeah. Yeah. I remember that whole thing with the signal issue and all that.
00:12:08
johngrimsmo
Yeah. Yeah.
00:12:11
John S
Um,
00:12:13
johngrimsmo
Well, cool. Super cool.
00:12:14
John S
Yes. What's your optimization going on here?
00:12:17
johngrimsmo
So I've spent the past two days. um We're running the Speedio a lot, which is awesome.
00:12:23
John S
Okay.
00:12:24
johngrimsmo
And we're starting to run it's starting to become a bottleneck um in Fjall production specifically.
00:12:27
John S
Interesting.
00:12:30
johngrimsmo
And it's great, except we're not utilizing the Aroha Palette Changer for production. We're using it to ah you know grab the next...
00:12:41
johngrimsmo
grab the RASC pallet, and then we manually load a bunch of RASC blades on this pallet.
00:12:43
John S
ha
00:12:45
johngrimsmo
So its you know it could be a 10-minute cycle, could be a 40-minute cycle, or maybe even an hour cycle, depending on what we're running. And then the Aeroa grabs that pallet, gets the Fjall pallet, does that, and then we manually load Fjall parts on that part.
00:12:58
johngrimsmo
So the operator has to come, like set a timer on his watch or on his phone or whatever, and every so much to keep the machine running.
00:13:05
John S
Yeah.
00:13:06
johngrimsmo
And it's because it's had some pallet-changing issues. where it brings in the new pallet and there's these air purge blasties underneath the the feet of the pallet that measures back pressure to be like, is the pallet installed?
00:13:20
johngrimsmo
Is there a chip in the way?
00:13:20
John S
De Benedetti,
00:13:21
johngrimsmo
Is there is is the air feedback installed? higher or lower because there might be a chip in the way and and it's tuned to like know that. And it's been out of tune for quite a while. So i I spent two hours digging into it and I got it retuned. And luckily I had notes from when I did this on the Kern in 2021.
00:13:42
johngrimsmo
So I was able to do it, which was pretty sweet.
00:13:45
John S
Yes.
00:13:46
johngrimsmo
And so now I got the, it's kind of like a threshold limit of like, cause we have small pallets and then we have bigger pallets and they have different air feedback values. Um, so you need to kind of just really play with it to get it to do what it's supposed to do, but I got that all tuned up and that's perfect.
00:14:02
johngrimsmo
So now the actual act of changing pallets should be totally reliable. I will test that either today or tomorrow by literally calling 22 pallets in a row with, with an empty program in it.
00:14:12
John S
Yeah, right, right.
00:14:15
johngrimsmo
Um, So that's awesome. And then the other thing is we are manually tweaking some macros on every blade based on the results of the last blade for bevel side to side accuracy, et cetera.
00:14:28
John S
Yeah.
00:14:29
johngrimsmo
Over the years, I've had various ways of probe the bevel, log the result, comp the wheel, do it again and try to do it all automatically, especially on the current.
00:14:40
johngrimsmo
And that worked okay-ish. um But the... system we're using right now with the guys doing it is a fairly manual blade per blade tweak.
00:14:51
johngrimsmo
And it's getting us results. And then like when you automate any system where you're relying on tight tolerances, you know you make three in a row, 10 in a row. And before you were tweaking every blade, and now you're just hoping they're all gonna work.
00:15:05
johngrimsmo
um There is a level of ah trust or risk that you have to put into it to be like, and I'm about to tell the guys, I'm like, run three blades, just tell me what happens.
00:15:15
John S
Sure.
00:15:16
johngrimsmo
And then i wonder if we are chasing our tail and over comping by doing it manually after the fact.
00:15:21
John S
Yeah.
00:15:23
johngrimsmo
And more importantly, try a picture in your head. i don't know if you've seen enough pictures of how we mount the fuel or brass blades to the side of the fixture.
00:15:31
John S
Oh, I remember that quite well.
00:15:32
johngrimsmo
Yeah. And then there's these two little clamps that hold the flat part of the blade, exposing the tapered bevel. Um, and they have to be small enough and clear enough that the grinding wheel can come down and not hit into them.
00:15:38
John S
Yep, yep.
00:15:43
John S
Mm-hmm.
00:15:44
johngrimsmo
Um, What I've noticed after we do op one, where the other side of the blade is still totally flat against the fixture, we grind them we machine and then grind op one.
00:15:52
John S
yeah
00:15:55
johngrimsmo
And then I can walk up to the machine with my finger and I can push on the bevel and I can flex the blade enough to see coolant come out.
00:16:02
John S
Yes. i remember you mentioned that.
00:16:03
johngrimsmo
Yeah. And so a while ago I put an indicator on it and I bought a force gauge.
00:16:08
John S
Cool.
00:16:09
johngrimsmo
that measures pounds when you push on it. And I was able to notice after eight pounds, the blade is now flush and it moved, I forget the answer, three thou or something like that.
00:16:11
John S
and Okay.
00:16:19
johngrimsmo
So if I choose to go down this rabbit hole, which I'm just about to make that call in the next day or so, ah I think I might've talked to you a couple months ago about this, is putting switchable magnets behind that part of the bevel. And i i designed it a couple months ago. I looked all into it, learned all about magnets, and then I shelved it. And now it's like coming back and I'm like, you know, it might be time to like thoroughly, thoroughly test that out. Cause I bet you we're seeing more variation from the deviation of one blade's moving three thou, one blade's moving four thou, the next blade's moving two thou. And we're chasing our tail comping things
00:16:56
johngrimsmo
due to a mechanical variance in the part loading or the stresses in the metal itself or whatever it is.
00:16:58
John S
Yes.
00:17:04
johngrimsmo
And it was cool using the force gauge to know eight pounds because I could do the math and be like, can I get eight pounds of magnet pull from my internal magnets?
00:17:14
John S
Yes.
00:17:14
johngrimsmo
And I can, I think I can get 10ish or something like that.
00:17:15
John S
Okay. Yeah.
00:17:18
johngrimsmo
So, you know, that's kind of where I'm at right now. And I'm really looking at the whole system again. I haven't looked at the speedio in a long time, which is nice because I'm just, it's just runs, but we have some seriously inefficient systems going on and, uh, segwaying into my, my business note for the day of hard work versus optimization.
00:17:39
johngrimsmo
Um, You know, Angela's talking about staggering some shifts so we can keep that machine running longer or having a guy come in on Saturday before Christmas or something to like, you know, get more blades out the door before the end of the year, which is great.
00:17:49
John S
Yeah.
00:17:53
johngrimsmo
But that's the hard work side of it. And it's like that's it's a fairly clear path of like, yeah, just throw resources at it, throw throw man time, man hours. to make more products or on my end to optimize and to figure out the bottlenecks and use this a row of pallet changer that we put so much effort into doing it.
00:18:10
johngrimsmo
And we have multiple pallets already. It's just the consistency of result that we're not sure about. So yeah, so I've spent the past two days, you know, quite a bit of time closing in on 10 hours of, uh, of working on that.
00:18:23
johngrimsmo
And I'm making some like notable progress right now.
00:18:26
John S
Good.
00:18:26
johngrimsmo
Yeah. So
00:18:28
John S
Can I rewind you to understand this at a simple level?
00:18:32
johngrimsmo
Yes.
00:18:33
John S
The Aroa Compact 80, I know it's shared between the Brother and the Kern, and that works.
00:18:38
johngrimsmo
Yep.
00:18:38
John S
Like you can call up a Brother. From the Brother, call up a pallet. So that's loading a Aroa-style pallet into Am I crazy? It goes into a shunk base? Or am I confusing...
00:18:46
johngrimsmo
It's an MX knockoff Eroa chuck. But it looks just like an Eroa chuck.
00:18:52
John S
So you have a Shunk system separately on that machine as well?
00:18:54
johngrimsmo
Yeah.
00:18:55
John S
Okay, that's i'm confusing.
00:18:55
johngrimsmo
I technically have five pallet positions on that machine. Four chunks and one Eroa.
00:18:58
John S
Okay. Thank you. Okay, so it's the ROA that's... Okay, so that was that answers one of my questions, which is that that ROA chuck, even if it's a knockoff, that actually has an error feedback to the ROA Compact 80.
00:19:08
johngrimsmo
Yes, and the Auroa controls that.
00:19:11
John S
Interesting. And you're not, of course, it's it would be a bad idea for long-term reliability to to defeat that by like ah ignoring in the the air check of the vacuum pressure, et cetera.
00:19:22
johngrimsmo
um Correct. yeah I don't know, I never thought about that, of like disabling it or not not caring or whatever.
00:19:28
John S
Yeah.
00:19:30
johngrimsmo
It's nice and when it works great.
00:19:33
John S
Yeah, and i I'm very hesitant to endorse this idea because I think but I was just talking to Justin a lot about this because we're really're really on the precipice of starting to do robot part loading.
00:19:42
johngrimsmo
Yep, yep.
00:19:43
John S
I respect chip control chip management and validation and checking.
00:19:45
johngrimsmo
Yeah.
00:19:47
John S
So I would say keep I'm just kind of curious why that was a flag. um And then, and you, so you'd have to make 10 more identical pallets or you already have them versus what you, okay.
00:19:58
johngrimsmo
Right. Yep, I already have a bunch.
00:20:01
John S
But you already have a bunch, okay.
00:20:02
johngrimsmo
Yep.
00:20:03
John S
So you don't even have to do,
00:20:03
johngrimsmo
And then assuming palette to palette is the same, um you know We can check them, we can validate, we can measure, but they're all tapered surfaces. So you'd like have to CMM them to know if palette to palette is identical.
00:20:17
John S
Okay, yeah, because that matters here, huh?
00:20:19
johngrimsmo
It does, yeah.
00:20:20
John S
And you don't want to have, you could, yeah it's such a good conversation about when probing is the right move and when it's ah trying to compensate for a poor design. not No, forgive me, I'm not throwing shade, but yeah.
00:20:30
johngrimsmo
Yep, yep. Yep. And and what what I love about the Speedio and the current, obviously, but the Speedio with our high accuracy package, which gives us five decimals inch, um it can hit a 10th or less of tolerance, but everything else comes into play.
00:20:34
John S
Yeah.
00:20:48
johngrimsmo
Thermal tool wear, probe deviation, probe, the you know, the probe centering itself.
00:20:48
John S
Yeah.
00:20:53
John S
Yeah.
00:20:54
johngrimsmo
not Not reading properly.
00:20:54
John S
Right.
00:20:55
johngrimsmo
I just calibrated it right before the podcast and it was off by 1.3 thou. Measuring a whole diameter, 1.3 thou bigger.
00:21:00
John S
Yeah. Yeah. yeah Yeah.
00:21:04
johngrimsmo
and And it just walks. I don't know when it's the last time, but a couple of months. So I'm contemplating staring at the machine, mounting a master ring gauge in a corner of the machine and calibrating it every morning before every critical feature using it as a sanity check, because if ever I, you could, but then it's another palette change.
00:21:21
John S
Put in your role.
00:21:24
johngrimsmo
If I can just have it in cycle in process, um, it might be ah an option.
00:21:25
John S
So.
00:21:28
John S
Okay.
00:21:30
johngrimsmo
Um, But then if any chips will get on it, then it's going to read weird.
00:21:33
John S
I know. Right. we whatever we're Right.
00:21:34
johngrimsmo
And like the coolant I don't care about, but chips I do care about.
00:21:34
John S
Right.
00:21:37
johngrimsmo
So you could wash down before, but how complicated does it get? you know I could put a sphere in there, which is interesting and I'm curious about, because there is a calibration cycle for probing on a sphere.
00:21:41
John S
Yeah.
00:21:51
John S
the ball mm-hmm
00:21:52
johngrimsmo
But imagine a sphere, it's like round on top, and there is one center point at the very, very top. If you're off center, it's going to Z wrong.
00:22:00
John S
who
00:22:00
johngrimsmo
And then down the sides, there is one band halfway down the ball that is the biggest diameter. If your Z height is off by tenths, your to your diameter is going to be wrong.
00:22:10
John S
So we've been through this exercise or Alex has for some of the hard gauges that we've either ordered or made for buck chucks.
00:22:16
johngrimsmo
Okay.
00:22:16
John S
And you're not, you're correct, except it's not linear. If you're off in the Z height a little bit on a diameter of a sphere, it's not as consequential.
00:22:20
johngrimsmo
Of course.
00:22:24
John S
And you would just, I assume just do a loop. You would do it twice. You would probe it and it would get quite close and you would retouch off Z, retouch off XY on that.
00:22:31
johngrimsmo
Yeah. Yeah.
00:22:32
John S
And you'll be good. I mean, I think you'd be plenty good.
00:22:33
johngrimsmo
yeah
00:22:36
johngrimsmo
Yeah. And I think like, I know when we, when we use the sphere in our CMM, it like scans around the whole thing and it spirals down and it finds the fattest point and it goes down and up and down.
00:22:43
John S
home
00:22:45
johngrimsmo
And I don't know what the speedio bloom cycle does, but I want to run it and find out.
00:22:52
John S
Yeah. So you already, in a weird way, the fact that you already have 10 fixtures made is nice because it means you don't have to have the anxiety of deciing deciding how to design the fixture, let alone making them.
00:23:02
johngrimsmo
Yeah, exactly.
00:23:05
John S
the The magnet would then become part of each fixture or it would live
00:23:05
johngrimsmo
Yeah.
00:23:08
johngrimsmo
It would would go into new newly made fixtures, but yeah, it would be I'd bore it out from the long end pretty much and install it inside. And it would basically invalidate all the old fixtures I have to make new ones, I think, unless I can modify them.
00:23:22
John S
Oh, would, okay. Is it an electromagnet or
00:23:26
johngrimsmo
No, it's it's a physical, like how a Noga base works. We talked about this a while ago where it changes the induction flux path or whatever it's called. And it basically cancels the magnetism by creating an internal short of magnetism versus if you rotate it the opposite direction, it's like two opposing magnetic fields.
00:23:44
John S
Yes.
00:23:46
johngrimsmo
Then the magnet wants to go outside the field and into your part.
00:23:50
John S
So how would you actually the magnet?
00:23:52
johngrimsmo
you'd ah have a like a push rod that would rotate one of them 90 degrees.
00:23:57
John S
But you would do that via a machine?
00:23:59
johngrimsmo
You would do that during during loading manually.
00:23:59
John S
on it
00:24:03
John S
Ah, okay.
00:24:03
johngrimsmo
Right?
00:24:03
John S
I'm thinking that you're gonna use the spindle to smell to do this.
00:24:03
johngrimsmo
So you put a blade on there, no, no.
00:24:05
John S
Okay, okay.
00:24:06
johngrimsmo
Load a blade on there, activate it manually, and then it sucks down.
00:24:06
John S
No, that's easy. Okay.
00:24:09
johngrimsmo
And then when you take the blade off, unload it, and it should work, yeah.
00:24:11
John S
Yeah, that's easy to or easier. I'm thinking electromagnets or more complicated.
00:24:15
johngrimsmo
Yeah, it gets complicated, right?
00:24:16
John S
Sure, sure.
00:24:17
johngrimsmo
Which could be cool too, but but yeah.
00:24:17
John S
Okay. It'd be interesting. I would almost, I would almost posit that yeah, you needed eight pounds of, of the fishing gauge scale, weight scale measurement.
00:24:25
johngrimsmo
Yeah,
00:24:27
John S
I bet you a magnet that does even less than that might actually solve your stability issue.
00:24:30
johngrimsmo
yeah.
00:24:31
John S
You know what i mean?
00:24:31
johngrimsmo
Yeah, right?
00:24:32
John S
Did you watch the Colin Furze on the bike?
00:24:35
johngrimsmo
I did the magnet bike.
00:24:37
John S
Dude, I they actually didn't, I didn't finish it in fairness.
00:24:37
johngrimsmo
I did watch that. That was really cool.
00:24:40
John S
I'm probably halfway through it, but like, holy cow.
00:24:42
johngrimsmo
Yeah, yeah.
00:24:43
John S
Hilarious.
00:24:44
johngrimsmo
Yeah, I don't watch a lot of his videos, but that came up and I was like, yeah, I'm going to watch that before bed.
00:24:47
John S
Right. Well, when Justin was here from Toolbat, he's a big bike rider, Grant's a big bike rider. So they were, they went on a big ride together. And then we were talking about like building the bike and, and you magnets are great, like awesome.
00:24:55
johngrimsmo
Nice.
00:25:00
John S
And yeah, it's funny. Yeah.
00:25:02
johngrimsmo
Yeah.
00:25:04
John S
So what's your next step on this?
00:25:07
johngrimsmo
On the speedio.
00:25:08
John S
and the In the magnet, I think.
00:25:10
johngrimsmo
Oh, the magnet thing. um Well, I want to wrap up the active speedio things. I calibrated the probe just now. um Pallet changing works. I want to run 20 pallets just to make sure sure it works.
00:25:23
johngrimsmo
And then... I want the magnet thing to be like the last hurdle.
00:25:27
John S
Okay.
00:25:28
johngrimsmo
Like right now, if we run three pallets in a row with no changes in between and see what the results are, um that will tell me if the magnets are worth doing or not.
00:25:37
John S
Say that again?
00:25:39
johngrimsmo
So if if we actually make three Fjellblades in a row on three different pallets without without comping between them,
00:25:44
John S
Oh, oh, yes. Yeah.
00:25:47
johngrimsmo
and and track the results. ah do do that Do I need the magnets? Are the results all over the place? Or are we or were we chasing our tail over comping, under comping, over comping, under comping?
00:25:55
John S
Yeah. All right.
00:26:01
John S
Can you, are you able to leave, ah but I'll offend you with large numbers, 5,000 stock to leave when you do this test that way, if that they're all wrong, you can still fairly easily, like he really easily rework them?
00:26:13
johngrimsmo
You could.
00:26:17
johngrimsmo
Yeah, interesting. Yeah, it would get you the answer. You would just, you have to rework them in that case.
00:26:23
John S
You'd have to re-work either way, but you're at least not.
00:26:26
johngrimsmo
Well, it might just work.
00:26:29
John S
Uh, yo, you don't care about not all you just care about parallel or
00:26:33
johngrimsmo
um I care about with within a range. It needs to be nominal.
00:26:36
John S
interesting. Okay.
00:26:37
johngrimsmo
Yeah.
00:26:37
John S
Fair enough. Yeah. Yeah. I'd be, I'll be very curious see you get along. Like I want that to work.
00:26:43
johngrimsmo
Yeah.
00:26:43
John S
Magnus is just such a weird, like the elegance of like, okay, how do you stop chips from small chips from adhering? how do you clean them? If you can disable it, like, do you, um,
00:26:52
johngrimsmo
he Well, that's the thing. If you can turn it off, like how cool is that?
00:26:57
John S
Yes. Well, or like, I'm talking out loud here, but if you could even like um cast or mold in a really durable, almost a glass-like thin layer that creates a, that the magnet lives behind and that way you can wipe it off.
00:27:12
johngrimsmo
Yeah, pretty much well. Yep.
00:27:14
John S
Okay.
00:27:14
johngrimsmo
Yep.
00:27:14
John S
Something like that. Yeah.
00:27:16
johngrimsmo
Yeah, between ah the housing that I'm going to make and the potting compound that I'll buy and use to kind of seal it like an epoxy.
00:27:25
John S
Or maybe you just leave the feature 5,000 thick on the steel aluminum fixture wall. That way it's one surface there anyway.
00:27:32
johngrimsmo
Yeah.
00:27:33
John S
Sorry, I'm talking about.
00:27:34
johngrimsmo
Yeah, pretty much. Yep. You want to keep the magnet juice as close as possible to the part.
00:27:39
John S
Totally.
00:27:39
johngrimsmo
But yeah, yeah. No, it's interesting. it's ah I definitely want to proof of concept it, you know, like get everything else stable so that the guys are at least more productive, like easily.
00:27:45
John S
Yeah.
00:27:49
John S
Uh-huh.
00:27:49
johngrimsmo
um Because we're so close. It's like put so much time and effort into getting this Aeroa to work and we're not using it because we don't we don't trust the process yet.
00:27:52
John S
Yeah.
00:27:58
johngrimsmo
But you guys will see that very soon. It's like any automated process. You're like, okay,
Upcoming Events and Community Engagement
00:28:03
johngrimsmo
two worked. Are 20 going to work?
00:28:05
John S
I know, right? i had this, I didn't do it, but when we had when we had the Tormach robot, on I was almost just to have it pick up a piece of a top jaw stock and move it to another table, pick it back and move it.
00:28:15
John S
I was like, do you think that could run for a month straight? Like doing that every four seconds for month?
00:28:17
johngrimsmo
Right.
00:28:20
John S
I mean, it's a colossal waste of power, I guess, or whatever. But like, i don't know, something, it's like, how would it fail?
00:28:24
johngrimsmo
Yep.
00:28:26
John S
And then I'm like, but how would it, how would it actually never have some weird glitch? I don't know, you know, huh?
00:28:31
johngrimsmo
Yeah. Yeah. And it's those failure recoveries that, uh, make automation tricky. Um, Like i restarted the speedio this morning and it turned the AROA off or it East stops the AROA, which if the current was using the AROA, that would be bad.
00:28:45
John S
Oh, that stinks.
00:28:48
johngrimsmo
And it's, I think I remember there's a way I can wire it differently, like the East up circuit to separate them.
00:28:48
John S
Yeah.
00:28:55
John S
Yeah. Yeah.
00:28:57
johngrimsmo
But yeah, it's kind of weird.
00:28:58
John S
yeah Wouldn't you stop the current though, at least.
00:29:01
johngrimsmo
No, it doesn't. So if the the the current was off or like not running at the time, but the current would keep running.
00:29:01
John S
Yeah, that's good.
00:29:06
johngrimsmo
But if it was mid pallet change to the current, then you'd be stuck in a, you know, mid pallet change. And then you got to recover from that, which is fun.
00:29:13
John S
Yes. Yeah. absolutely Yeah, by fun, you mean not fun.
00:29:18
johngrimsmo
Yeah, we're we're good at it now, but it's, yeah. Yeah.
00:29:22
John S
Right, right. oh Okay, i want to mention two things. Number one, next, ah the 18th, is that a Friday? It's Friday. No, thursday excuse me, Thursday, the December 18th, I am hosting a one of Toolpath's webinars. don't exactly what they're calling them. um And they've done, they just did one with Chris Eppicini.
00:29:44
John S
They've done one with other folks in the industry. And they asked me if I wanted to do on this topic in particular. And I've always loved the AMA style format. I mean, frankly, kind of inspired by the Dylan type, like, hey, have a conversation.
00:29:54
John S
Like, you know, because I don't want to, I don't want to come with an agenda.
00:29:54
johngrimsmo
Yeah.
00:29:57
John S
But if you want to talk about our kuma horizontal or williman or you want to talk about strike marking quitting my day job i'm sure let's talk you know how can i be helpful so um i'm sure social media stuff will go out for the sign up just a free zoom webinar thing but we'd love to see folks on the um i was little on that ama um i don't know if they're goingnna ask make you sub submit questions ahead of time or it'll be real time ama um either way Then the other one, which is long-term bigger picture, which I think you may have just chimed in on is in April, there's a event called the summit on the summit.
00:30:04
johngrimsmo
Cool.
00:30:16
johngrimsmo
Sweet.
00:30:27
johngrimsmo
Yeah. Yeah, exactly.
00:30:30
John S
It's a three or four day. Everyone in the machining world seems to be rallying behind this event at in California at mammoth mountain. Um, tons of awesome people are going to be there from heavy hitters, you know, Carl Bass and Mark Terry very all the way to little guys like me and hopefully you will like, um,
00:30:43
johngrimsmo
Yeah.
00:30:47
johngrimsmo
exactly yeah
00:30:49
John S
But like, seriously, tons people are already going to be there as panelists, hosts, and talking to Al about you setting up the event. I was like, i don't want sales pitches. I don't want showing. I want like a people community to get together to talk about manufacturing webinars, not webinars, seminars, roundtables, fireside chats, et cetera, and scheme.
00:31:06
johngrimsmo
Yeah. Yeah. I'll reach out to me and he said, you should, you should come to this. And I looked at it and I thought about it and I saw all the people that were even just speaking, the let alone the people that are attending.
00:31:14
John S
Yeah.
00:31:15
johngrimsmo
And i was like, Oh, I really got to make this work.
00:31:16
John S
Yeah.
00:31:19
John S
Awesome.
00:31:19
johngrimsmo
So yeah. So I'll be there.
00:31:21
John S
Yeah. Looking forward to anyone who can, can make it out there. if you don't ski there, I think there's plenty of stuff to do, but obviously if you do ski, um, should be awesome. And I think they're even going to figure out a way to get the mountain open, uh, like half an hour early for the attendees. It'll be shoulder season. It's not going to be crowded anyways, but it should be awesome.
00:31:38
johngrimsmo
Sweet.
00:31:40
John S
Um, what's next on your list?
Advancements in Vision Systems
00:31:43
johngrimsmo
um You want to talk about telecentric lenses?
00:31:48
John S
yo Dude, no, no, because I can't handle it. I watched the train video and I'm like upset that I didn't know about this.
00:31:56
johngrimsmo
Right. Okay. So to give context to that train video, if you guys haven't seen it, Edmunds Optics on YouTube, you'll find it. They have, to be fair, that is a massive telecentric lens. It's like a foot or more, two feet in diameter.
00:32:07
John S
Okay.
00:32:09
johngrimsmo
and it probably costs $20,000 or something stupid like that.
00:32:11
John S
Okay.
00:32:13
johngrimsmo
And it's it's it actually... led me down the wrong understanding of tele watching that video and some of the other ones, because they have this um depth of field that's like a foot long or something where the the thing is actually still in focus.
00:32:19
John S
Interesting.
00:32:30
John S
Yes.
00:32:30
johngrimsmo
And so in this model, they have they have a train on rails that goes back and forth and back and forth and back and forth. And there's like people on benches and it's a tiny little model, right? And trees and stuff. And a telecentric g lens basically looks at the entire picture without any parallax.
00:32:44
johngrimsmo
And it's so hard to explain it and just look straight forward.
00:32:46
John S
No, it's not. You just lay down, put drinking straws inside of a toilet paper roll and look through that.
00:32:50
johngrimsmo
Yeah.
00:32:53
John S
You cannot see anything but straight lines out out of it.
00:32:55
johngrimsmo
exactly from the entire field of view of the game and with this massive lens they could see a whole like model scene um whereas the i bought two lenses um they both came in super exciting i got one from ebay toronto and one from ebay in china but it's an american lens um really really nice and the field of view is about like half an inch
00:32:58
John S
Right.
00:33:12
John S
Mm-hmm.
00:33:19
johngrimsmo
in diameter. So when your microscope or whatever is looking through it, you can only see a half inch object.
00:33:21
John S
Okay. Sorry.
00:33:24
johngrimsmo
So you couldn't even see like an entire quarter.
00:33:27
John S
Interesting. Okay. I'm with you.
00:33:29
johngrimsmo
And also these lenses have a um depth of field, a focal distance of like a millimeter. So like your your your stage, your microscope has to be the perfect Z height from your part um to be in focus.
00:33:40
John S
Oh, wow.
00:33:49
John S
Okay.
00:33:49
johngrimsmo
So I had a gauge block on its side, like a small one. And I could i could lift it up like by hand, lift this lens up with the camera on top. And i could have the top in focus, but the bottom would be all blurry. And then I could lift it down.
00:34:03
John S
Can I ask you to explain?
00:34:05
johngrimsmo
Yeah, for sure. i know i'm
00:34:06
John S
So, well, what's the difference between focal length and depth of field?
00:34:11
johngrimsmo
I don't know. I think they basically mean the same thing. um
00:34:15
John S
Well, there's one measurement for um how far away an object needs to be.
00:34:21
johngrimsmo
So there's working distance, which, yeah, and and I'm constantly wrapping my head around all these terms and a bunch of other terms too.
00:34:23
John S
Maybe that's I'm thinking of. The depth of field, sorry, go ahead.
00:34:33
johngrimsmo
But basically... how much of your viewable object is in focus in let's call it Z height.
00:34:39
John S
Mm-hmm.
00:34:40
johngrimsmo
And watching the train video, you're like, there's like a foot of working distance or like depth of field.
00:34:40
John S
Yeah.
00:34:45
johngrimsmo
Like I can have the object close.
00:34:46
John S
It's crazy. Yeah.
00:34:47
johngrimsmo
I thought I could literally film like myself with this lens and this camera. um The other caveat to these lenses is because they're so long, no light gets in them.
00:34:58
johngrimsmo
like you need very powerful, very direct lighting for, for any image to show up whatsoever.
00:34:59
John S
Got it.
00:35:05
johngrimsmo
So it, it worked exactly.
00:35:05
John S
This is going to lead. Yeah. You're going to end owning like a phantom too now.
00:35:08
johngrimsmo
It works for a vision system because you have a white backlight and all that light comes straight up.
00:35:12
John S
Got it.
00:35:13
johngrimsmo
And with the white light interferometer, you have beam splitters and lenses and LEDs, and everything should bounce right back up to the light, uh, to the camera perfectly. So All that to say, i thought I would have a lot more um area of focus, let's call it, in Z.
00:35:30
johngrimsmo
Like I thought I'd be able to see more of a part in focus, but I can only see like a sliver.
00:35:33
John S
Yeah.
00:35:35
johngrimsmo
So if I want to scan, say, a SagaPen tube standing upright,
00:35:40
John S
Oh.
00:35:41
johngrimsmo
Because the telecentric lens, I don't have the blowout of parallax, but I can only focus on a small sliver as I scroll down.
00:35:44
John S
Yeah.
00:35:49
johngrimsmo
So for my manual vision system, I had to design in a Z hand crank.
00:35:49
John S
Interesting.
00:35:57
johngrimsmo
And I'm like, well, I want it to be like really accurate. So I found a super high pitch um metric M4 0.3 pitch.
00:36:05
John S
ah wow.
00:36:06
johngrimsmo
like super fine, like a hundred TPI or something like that. Uh, but long, like six inches long.
00:36:11
John S
Interesting.
00:36:12
johngrimsmo
So I'm building that into a 3d printed hand crank and little ball detents so that every, like i can rotate and that's one thou of the movement. And I can go,
00:36:21
John S
Okay. Oh, nice.
00:36:23
johngrimsmo
So that's going be cool. And then building that into the system. So otherwise, i'm i'm last thing I have to do is finalize the design and the 3D print of mounting all these things together.
00:36:35
johngrimsmo
But otherwise, the vision system is relatively functional and like super exciting.
00:36:39
John S
Dude, awesome.
00:36:40
johngrimsmo
The software is, I can't wait to show you. It's going so cool. And it works great.
00:36:44
John S
The software is this the open source DTF or whatever.
00:36:47
johngrimsmo
OpenCV is the package.
00:36:49
John S
Oh, okay. Nevermind. That's, that's the Python package. I'm thinking of the white light and from our software that Cyrus uses.
00:36:56
johngrimsmo
Which I'm not using, but yes, sort of.
00:36:58
John S
Okay.
00:37:00
johngrimsmo
um
00:37:00
John S
So you're using open CV with some secret sauce programming, all that to make measurements.
00:37:03
johngrimsmo
Yeah, like like all programmed.
00:37:03
John S
And so yeah, right, right, right.
00:37:05
johngrimsmo
The OpenCV is like a back-end ability. it's not You can't do anything with it unless you program you're your front-end.
00:37:12
John S
Sure. Yes.
00:37:14
johngrimsmo
um So I did that, spent about a week or so. making the coolest you know Python-based user interface with video feed and and clickable draw lines, draw circles, auto-capture light to dark regions.
00:37:22
John S
Sweet.
00:37:29
John S
Yes.
00:37:30
johngrimsmo
So I can click three points on a circle even roughly, and it automatically finds the circle.
00:37:34
John S
Finds it. Oh, my God.
00:37:36
johngrimsmo
And i with my Thor Labs calibration disk that has 10 millimeter cube, five millimeter cube, all the way down to 10 microns, not only can I see how close I can resolve,
00:37:36
John S
I love this, John.
00:37:43
John S
Mm-hmm.
00:37:49
johngrimsmo
can I resolve a 10 micron cube or, ah but I can use that to calibrate my line to line distances.
00:37:55
John S
Yeah, dude.
00:37:55
johngrimsmo
And it's like, it's so cool.
00:37:56
John S
I love this. You, I mean.
00:37:58
johngrimsmo
I've been doing it. I've been doing it all at home, like at my desk. And then when the kids walk by, was like, check it out, check it You got to see this. You got to try the software. Let me, let me show you.
00:38:09
John S
Remember the mentor thing I mentioned to you at some point about, you know, mentioning something to somebody in a mall parking lot or whatever, or whatever. It's like,
00:38:18
johngrimsmo
i don't know.
00:38:19
John S
um actually i think i'm probably completing two stories one is real quickly one is uh this guy would watch people buy his competitors product this sounds super sus now but in the early ninety s this wasn't as sus and walk up to the person in the mall parking they're like hey i saw you bought this why this product just curious why and then he would say ah very quickly like
00:38:30
johngrimsmo
yeah
00:38:37
John S
explain that he was actually with the competitor, but he just want to understand it. But you've ultimately got somebody to answer a question where they're not going to give you a biased response as you so often do when you know that they're approaching it with a certain perspective of
00:38:49
johngrimsmo
yeah
00:38:50
John S
what they want they want to hear. The other thing is kind of like looking for product feedback when you aren't asking for it. So it's kind of like the whole, you know, if you walked into a machine shop and just said, oh yeah, I have a new type of aluminum that's magnetic or whatever. And everybody would be like, oh, want that. Like you didn't have to ask them. They be like, I want that. I'm telling you as your friend right now, not to curry favor with you or to or to to boost your ego of all the things you and I've talked about over the last 10 years of one thing I've thought about that you should consider turning in a product, this is it.
00:39:20
johngrimsmo
Yeah.
00:39:23
johngrimsmo
That's wild. yeah I don't know. we'll We'll see how it turns out.
00:39:27
John S
OK.
00:39:27
johngrimsmo
It's like I wanted to build the interferometer first. That's all I wanted. I wasn't interested in a vision system at all. I've seen them at shows.
00:39:33
John S
Yeah.
00:39:34
johngrimsmo
I've seen the stare at the key ends, the whatever. And they're neat, right?
00:39:36
John S
Yeah.
00:39:37
johngrimsmo
They're neat. they're But they're only good for 2D features, through whole 2D features.
00:39:42
John S
Sure.
00:39:42
johngrimsmo
And a lot of our stuff is three d However, with the OpenCV plugin, you can turn on different lighting, like upper side lighting, like a ring light or whatever, and you can take ah an accurate photo of the top of your part.
00:39:51
John S
It's
00:39:55
johngrimsmo
And OpenCV can recognize your engraving line width.
00:39:57
John S
just so cool.
00:39:58
johngrimsmo
And you can be like, oh, that engraving is supposed to be 3,000 wide, but it's only 2,000 wide, flag red.
00:40:03
John S
Yeah.
00:40:03
johngrimsmo
And I'm like, oh, baby.
00:40:05
John S
Yes. Dude, I love this.
00:40:07
johngrimsmo
oh So, yeah.
00:40:08
John S
And it's so you. sorry like it's so you.
00:40:09
johngrimsmo
and And I've been filming all of this at home, kind of the progress I've made so far. so it's going to really cool video that I'm super excited about. And it's, I haven't been this excited about some like big project in in a ah long time.
00:40:23
johngrimsmo
Like this is, it actually feels like, like dirty.
00:40:24
John S
Good.
00:40:26
johngrimsmo
Like I shouldn't be doing it sometimes. I have more important things to do, but there is enough substance and enough like value to this that it's just so exciting.
00:40:36
John S
Awesome.
00:40:37
johngrimsmo
Yep.
00:40:39
John S
but i want to keep i'm go make it I want to keep hearing about this. um
00:40:43
johngrimsmo
Yeah, and it's hard to like wrap it all into a conversation and actually explain it properly. So I'm hoping the YouTube video has enough context to show what's going on.
00:40:54
johngrimsmo
But the second lens that I got is an avatar and it's adjustable. So the zoom range goes from 0.7, which is zoomed out to 4.5, which is zoomed in a lot.
00:41:01
John S
Interesting. Yeah, yeah.
00:41:06
johngrimsmo
And I got it like two days ago and I've been playing with it. It's like, yes,
00:41:10
John S
A telecentric lens that has zoom? That is mind-blowing, John.
00:41:14
johngrimsmo
And there are lens add-on attachments that add or remove further magnetism. So it all comes down to how much working area do you want to see?
00:41:24
johngrimsmo
Like, do you want to see a half-inch circle? you want see a quarter-inch circle? Do you want to see a 10,000 circle at multiple super magnification?
00:41:28
John S
Mm-mm.
00:41:32
johngrimsmo
And then depending on your camera, the unit is microns per pixel.
00:41:38
John S
Okay.
00:41:38
johngrimsmo
So a camera has 1028 whatever pixels.
00:41:38
John S
Yeah.
00:41:44
johngrimsmo
how how big of a micron feature can each pixel resolve does it make sense so the my big telecentric lens for the vision system gets me 10 microns per pixel which is fine for like as i'm using it i'm like that's great i can see all kinds of features um
00:41:49
John S
Yes. Yeah.
00:42:00
John S
Yeah.
00:42:06
John S
10 microns is four tenths though. So it's a number that's not like, it's kind of big in the sense of like, that makes sense. Like
00:42:14
johngrimsmo
Yeah, but so if you were measuring a hole that was 4 tenths, it would show up as one pixel on your your camera.
00:42:22
John S
yeah. Which is yeah. But one pixel on a screen is like a pixels are tiny, right?
00:42:26
johngrimsmo
but But when you zoom in on the on the photograph taken from this this system, you're zooming in on one pixel.
00:42:29
John S
You can zoom in on the pixel. Okay. Yeah. Yeah.
00:42:34
johngrimsmo
So assuming everything's square and perfect and your hole was exactly 4 thou or 10 micron, 4 tenths, or 10 micron, it would show up as one pixel.
00:42:39
John S
Four tenths.
00:42:44
John S
Yeah, that's legit.
00:42:44
johngrimsmo
But you can't measure a diameter of one pixel. It just shows up as a pixel.
00:42:47
John S
Yes.
00:42:48
johngrimsmo
So, but if your hole is say a hundred thou or, you know, one, eight, seven, five or something like that, that's so many pixels wide and and it averages around the outside. So even though you only have four times per pixel, like your hold is still insanely accurate.
00:43:05
johngrimsmo
And so as I want to be able to image one 10th hole differences accurately and,
00:43:10
John S
he Yeah. Well, the relative of that should be really interesting.
00:43:14
johngrimsmo
Exactly.
00:43:15
John S
that's cool.
00:43:15
johngrimsmo
So then the second camera that I got fully zoomed in, ah gives me one micron per pixel, not 10 microns of the other lines, one micron per pixel.
00:43:22
John S
Yeah, John, that's nuts. 40 million per pixel.
00:43:25
johngrimsmo
Yeah, exactly.
00:43:25
John S
This is nuts, John. And that's because of the glass you bought, the telescope.
00:43:27
johngrimsmo
And yeah, yeah, exactly.
00:43:29
John S
Okay. Yeah.
00:43:30
johngrimsmo
Yeah. And the camera you put on it, all cameras have different microns per pixel natively, et cetera, et cetera.
00:43:35
John S
Okay.
00:43:35
johngrimsmo
There's a lot to it. but And if I'm understanding this right, because a wavelength of light is roughly 500 nanometers, half a micron, if you zoom in more than that, it won't work, I think.
00:43:50
John S
We just hit the end of the the road.
00:43:51
johngrimsmo
I think. So I got a bit more sussing out to do there.
00:43:53
John S
yeah.
00:43:55
johngrimsmo
But ah so for the interferometer, which is going to measure basically 3D surface topology, I'll be able to zoom in. I want to see grain structure of metal.
00:44:07
johngrimsmo
And I think I will.
00:44:07
John S
Dude. Dude.
00:44:09
johngrimsmo
I think I'll be able to do that.
00:44:12
John S
And I know you're buying on eBay and it's used, so this is a big caveat emptor, but like hundreds of dollars or thousands of dollars for these?
00:44:18
johngrimsmo
About 300 bucks each.
00:44:18
John S
Yeah. Yeah, that's great.
00:44:20
johngrimsmo
For these used lenses, new, like if you were building a product to like stare at and Kian's make new lenses are probably thousand or $2,000 each.
00:44:21
John S
Yeah. Mm-hmm.
00:44:28
John S
Okay.
00:44:28
johngrimsmo
Um,
00:44:29
John S
But not like not a deal breaker at all for what you're talking about. The potential. No, lots of software and work to do.
00:44:33
johngrimsmo
yeah, exactly.
00:44:34
John S
it yeah, it's different if you're like, I'm going to go experiment with a $13,000 piece of glass.
00:44:38
johngrimsmo
Exactly. Yeah. to Experiment.
00:44:39
John S
Right, right, right.
00:44:40
johngrimsmo
No, but, but even still, like I came into this, like not wanting to spend a lot of money because, you know, sometimes things are tight right now.
00:44:41
John S
Yeah.
00:44:47
johngrimsmo
And I'm like, yeah, it's, it's part like super valuable in part.
00:44:47
John S
Sure.
00:44:52
johngrimsmo
I'm just obsessed and want to this project. um
00:44:55
John S
Yeah.
00:44:55
johngrimsmo
But it's going to be insanely valuable for what we're doing. So I'm just super jazzed and I'm glad, I'm i'm glad to hear your excitement for it too. Cause it's good.
00:45:05
John S
Dude. Yes. Awesome.
Leveraging AI and Tech for Efficiency
00:45:09
John S
I'll wrap up my side with an AI tip and then ah the quote-unquote business tip. AI tip is pretty dumb, but I got to say it. I was naively too often using chat and Claude, etc. as just being like, hey, write a script.
00:45:25
John S
you know For example, Python, maybe it's okay to do this in Python because Python is so ubiquitous and it's learned off of so many Python data sets, etc. But... I was making the mistake with certain things like machine manuals or G-code examples or post processors or the brother and probing and the Fusion API.
00:45:43
John S
What you really need to do is find ah the specific example I was working on was the Fusion API. it was It was failing to create Python scripts for Fusion to both do stuff in CAD and then in CAM.
00:45:56
John S
And part reason is that it would chat.gpt will never respond. You'd be like, ah I'd rather you give me some pointers. It's just going to like do it. You know, it's just going to make it up.
00:46:04
johngrimsmo
indeed
00:46:04
John S
And it was failing. And as soon as I fed it, i was like, hey, here is the fusion API PDF. That actually doesn't exist, which is one of the complicated factors of how I needed to get this fed into chat.
00:46:16
John S
But if you feed it a PDF of like, hey, here's the manual for the API. Use this to write code to do this. This success rate is...
00:46:23
johngrimsmo
No way. Yeah.
00:46:24
John S
like not 100 but like way it makes totally makes sense but um same thing like we asked we we were stumped on that we actually we still are stumped on the brother ftp server and when you just ask chat like hey help me with the brother dz user control ftp server you're going get an answer but it's kind of like ah if you upload my manual which we did that to we've been doing a lot of that to notebook lm and be like hey in this manual you'll tell me the pdf much better answer concern
00:46:28
johngrimsmo
Yep. Yep.
00:46:48
johngrimsmo
Yeah. Yeah. I remember struggling with the FTP server like a lot to get the speedio doing what I wanted it to. But now it works great. And it didn't work this morning.
00:46:58
johngrimsmo
I restarted the speedio and now it works great.
00:47:01
John S
Yeah, that guy did great.
00:47:01
johngrimsmo
But it's like we're deprinting a lot of information after every blade or every probe result or tool life or tool diameter, or things like that, because I want to be able to track what's what's actually happening.
00:47:05
John S
Oh,
00:47:11
John S
Yeah, we'll get it figured out.
00:47:12
johngrimsmo
So.
00:47:14
John S
i'm not about to i don't not I'm not putting out the bat sign yet to get help, but um business tip, I think I've probably mentioned this in the past, but it's it's a good enough point.
00:47:14
johngrimsmo
Yeah. No, of course. But.
00:47:22
John S
um we Everything we do at Saunders runs through Shopify. So even POs and other stuff goes through Shopify. That means we have one system, tracks all orders.
00:47:30
johngrimsmo
Yeah, same.
00:47:33
John S
And the real reason I do that is that even if somebody calls in and has a problem that results us reshipping a product, that is also an order. It creates a standardized process that means the customer gets tracking. We know the followup, the orders can tie into each other. And then we flag those right now as four different categories, I believe. QC, shipping error, shipping damage.
00:47:55
John S
There's a fourth, I don't remember. But we only started doing that, and this is my like biggest punch, takeaway that I hope to um impose on people or or teach is that we only create data for reports when we actually want to use it.
00:48:08
John S
It's so tempting to be like, oh, that would be good to know. Let me add that to ah a data dump or a report.
00:48:11
johngrimsmo
Yeah.
00:48:13
John S
And like, nope, all I want to know is over the course of, say, a quarter or a full year, how much money did we tie up with with quality control issues, shipping errors when we make a shipping mistake, shipping damages when the product gets dropped or whatever.
00:48:26
johngrimsmo
Yeah.
00:48:27
John S
And that system has been great. It helps quantify it. It's also not creating unnecessary BS reporting shenanigans that everyone's like, i don't know, we just do this because we were told to do it. Yeah, that's my lesson.
00:48:37
johngrimsmo
Nice. Yeah, we do that via spreadsheet manually, but we are categorizing all customer return issues, all things like that so that we're not doing it, but you could look at it and sort by category and be like, like give me a report manually on all all the times that, you know, how many returns did we get this year?
00:48:49
John S
Yeah. Yeah.
00:48:53
John S
Yeah.
00:48:57
johngrimsmo
How many issues did a customer have? How many whatever?
00:49:01
John S
Actually, if I may say so, where that's fun is that like we realize, okay, shipping a hobby plate, we ship a lot of hobby plates to Europe now. And that's, in some respects, the worst outcome because it's a lower ticket item, lower margin, but shipping is expensive. And so if a $200 fixture plate aluminum goes to Norway and gets damaged, that's really a bad outcome for everybody. And so that's the situation where you can then use whatever Python scripts or the the way the world we live in to be like, okay, every time a plate that meets a certain definition and is international gets shipped, our brother printer who ships out the order then prints out a ah separate sheet of paper behind it that says the order that just printed out XYZ requires special packaging.
00:49:45
johngrimsmo
oh
00:49:46
John S
because I don't want to ship every hobby plate with this additional stuff. Cause this kind of s sinks to say it, but the reality is if I ship a hobby plates in Nebraska and it gets damaged, we'll just replace it. Shipping there's 15 bucks versus $180.
00:49:56
johngrimsmo
yeah smart yeah automate the system so that the person doesn't need to be like oh that's an international order or oh that's one of these countries that you know yeah cool man
00:49:56
John S
Right. Right.
00:50:15
johngrimsmo
That's all I got.
00:50:16
John S
Did you have a business tip?
00:50:17
johngrimsmo
I squeezed it in, hard work versus optimization.
00:50:21
John S
Remind me of the...
00:50:23
johngrimsmo
About like running the speedio versus automating it.
00:50:25
John S
Oh, 100%. Yeah, sorry sorry.
00:50:26
johngrimsmo
Yeah.
00:50:27
John S
Totally get that. Okay. Yeah, yeah,
00:50:30
John S
Good. i love it. um See you next week.
00:50:33
johngrimsmo
Yeah, i'm just going to look at the calendar real quick.
00:50:37
John S
We'll probably be... I'm good next week, and I'm assuming...
00:50:38
johngrimsmo
17th is next week. 24th is the week after, I think. I think we're working on the 24th. don't know. We'll figure it out. But okay.
00:50:47
John S
Okay.
00:50:49
John S
Shiba, take care.
00:50:50
johngrimsmo
I'll see you next week.