Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
#435 Kern float sensors image

#435 Kern float sensors

Business of Machining
Avatar
1.6k Plays14 hours ago

Topics:

  • Audio issues on podcast?
  • Waterjet mod vises
  • Kern float sensors
  • The power of vision systems and AI


Recommended
Transcript

Introduction and Overview

00:00:00
johngrimsmo
Good morning and welcome to the Business of Machining episode 435. My name is John Grimsmo.
00:00:09
John S
My name is John Saunders.
00:00:10
johngrimsmo
And this is the podcast where we talk about ah our manufacturing our businesses. I've got a couple of things on the list today. ah Echo, coolant current sensor, or current coolant sensor, water jet question for you, and nesting software.
00:00:26
John S
Ooh, interesting. Okay. mine My list is also somewhat short, although I've never not worried, but... Brother, machine tool accessories, Python, script and code, home assistant, bomb, audio issues, um drill grades for our steel plate production. that's it.
00:00:44
John S
Yeah.
00:00:45
johngrimsmo
All right, well, jump in.

Audio Issues and Listener Feedback

00:00:47
johngrimsmo
have a question for the audience. Does anybody hear an echo on my end? Because I can hear myself a little bit sometimes, but it seems to go away after like 10 minutes and it's driving me crazy. So if anybody else is like hearing it, just shoot us an email um or DM on Instagram or whatever. And just just let us know, because I'm curious.
00:01:06
johngrimsmo
We're thinking it's like a Windows input software something. I don't know, but...
00:01:12
John S
We got some emails since the last week's episode. More more than, every in while we'll get an email from somebody that's like saying something about the audio, which I want to be kind about. But like, we we're kind of like hands off on this podcast.
00:01:20
johngrimsmo
sure
00:01:23
John S
But this was a lot of emails specifically. And what's weird is I got a new computer. So I was like, okay, very much could be my fault. i have a new computer. um But I can't figure it out. You hear your own echo sometimes.
00:01:34
John S
And so we just want people to know, I don't hear your echo at all.
00:01:34
johngrimsmo
Right. But you don't.
00:01:38
John S
You sound cool.
00:01:38
johngrimsmo
Yeah.
00:01:39
John S
You you sound great, John.
00:01:40
johngrimsmo
Well, thank you very much, sir.
00:01:42
John S
So we appreciate everybody's patience. We're on it, but, um, and we, we've had some people offer to help fix it like post process. I'm like, no, no, no, no we're not. We need to fix the cause, not the symptom.
00:01:54
John S
Um,
00:01:55
johngrimsmo
Yeah, this podcast needs to be as little effort as possible. Just that's how it is.

Exploring Wazer Water Jet Enhancements

00:01:59
John S
bingo.
00:02:03
John S
right. Uh, go dive, go water jet.
00:02:06
johngrimsmo
um what's okay water jet so we eric's been using our wazer water jet more and more which is awesome um and he was asking me yesterday he's like i wish i had some vice solution for the water jet and i was like what about a saunders mod vice like i don't own any mod vices um but i was wondering about putting one on the wazer and if that would work Or, I don't know.
00:02:32
johngrimsmo
you like Imagine if you 3D printed a subplate with holes in whatever, and you mod vice to that, and you just cut through the center between the two vices.
00:02:39
John S
ah yes
00:02:44
John S
Yeah. The problem is that water jets usually going to cut through, the, the fixture plate, if you will.
00:02:48
johngrimsmo
The center.
00:02:51
johngrimsmo
Yeah.
00:02:51
John S
Um, and that's not good.
00:02:54
johngrimsmo
But if you had a hole in the middle where you're water jetting,
00:02:57
John S
Yeah, I know that would work. I mean, for sure it could work.
00:03:02
johngrimsmo
yeah I don't know what exactly he wants to hold, but um yeah, he was thinking about it.
00:03:03
John S
Um,
00:03:09
John S
Well, look, I mean, obviously DM me, we can get you some mod vices. yeah Also don't, don't repeat this online. Cause I don't want anybody to think it's a great idea. You could just 3d, we share the file. You get 3d print a mod device. And honestly, a 3d printed mod device might actually be a better fit for a water jet.
00:03:21
johngrimsmo
Better for this, yeah.
00:03:23
John S
You don't want, you don't need, and you don't want exorbitant clamping pressures.
00:03:28
johngrimsmo
Right, and yours are steel. um I think they would rust and real fast.
00:03:33
John S
We make the aluminum hobby ones. There's still still steel screws on them.
00:03:35
johngrimsmo
Okay.
00:03:37
John S
But yeah, I don't know.
00:03:37
johngrimsmo
That's fixable.
00:03:38
John S
you know, when we had a laser, it was like screwing down in, um like into the, I want to say there was like wood screws would work into the honeycomb.
00:03:46
johngrimsmo
Yep.
00:03:47
John S
And now if you're doing production type stuff, that, i sorry, that actually could make a ton of sense, John. You're right.
00:03:54
johngrimsmo
I've i've heard heard of people making production fixtures for the water jet and Wazer actually wants us to do that for ourselves to figure it out and make a video on it.
00:03:54
John S
that's what you're thinking. Yeah. Okay.
00:04:01
John S
Okay.
00:04:02
johngrimsmo
Um, cause it could be kind of cool. So Eric's wrapping his head around it, trying to figure out what he really wants.
00:04:07
John S
Yeah, right, right.
00:04:08
johngrimsmo
Um,
00:04:08
John S
No, that I'm more on board with. You could have a fixture plate that has a hole, a cavity in it, so you're not water jetting through the fixture plate and it has the correct spacing.
00:04:15
johngrimsmo
yeah. Hmm.
00:04:16
John S
Because the mod vise, the beauty of the mod vise is, you know, infinitely adjustable, small to big, but that also means you need a fixture plate that is full surface, fixture plate, full coverage, and you don't want that here.
00:04:25
johngrimsmo
Yeah.
00:04:28
John S
Yeah. Interesting.
00:04:30
johngrimsmo
Yeah. Eric actually took our second bamboo home. So he's been playing with it a lot, which is good.
00:04:36
John S
Sweet.
00:04:36
johngrimsmo
um Making RC car parts and all kinds of home stuff. And yeah, he needs to start 3D printing stuff, fixtures for the thing.
00:04:46
johngrimsmo
So he uses the long wood screws into the honeycomb plastic bed or whatever.
00:04:49
John S
Okay.
00:04:50
johngrimsmo
And he realized you can only go straight down so many times before you run out of good spots. If you put them in at an angle, he's like, that works great.
00:04:58
John S
For a while, yeah.
00:04:58
johngrimsmo
Go at an angle until you cut through the angled screw because it goes under your cut path, but it works.
00:05:02
John S
Right. Oh, you were it right, right, right. But i it's been a while since I've used the Wazer, but you don't need very much clamping force, if I recall.
00:05:13
johngrimsmo
No, you just need like several screws to make sure it doesn't lift, pick up, kick, twist, whatever.
00:05:18
John S
Right, right. What are you guys water jetting on it?
00:05:22
johngrimsmo
Mostly damasteel blades.
00:05:24
John S
Okay. Whoa.
00:05:25
johngrimsmo
also various time ascus things and i actually made a plate out of 80 thou titanium i came in on sunday and i' machined it a double-sided tape or two tape and super glue kind of thing and i tracked my time it took ah the the part was already designed it took me two and a half hours to set up this video load some tools break three end mills
00:05:47
John S
Whoa.
00:05:48
johngrimsmo
end up delaminating the tape anyway and finishing the parts and they're beautiful. And then it wasn't until I was finished that I was like, you know, this would have taken 20 minutes on the Wazer.
00:05:59
johngrimsmo
like
00:06:00
John S
Yeah, right? Sure. Yep.
00:06:01
johngrimsmo
And I don't run the Wazer yet. I'm sure I can figure it out, but I haven't. I can't just jump on and like without figuring it out. um But I could have given that to Eric on Monday morning and it would have been done.
00:06:13
John S
yeah
00:06:13
johngrimsmo
um There. And all of a sudden my echo is gone. No, it's still there. Never mind. whatever
00:06:19
John S
Well, keep me posted on the Wazer. i But i I commend you and fully endorse and support this idea of like, whether it's you or your brother or anybody else, like you want that to just be, just work.
00:06:21
johngrimsmo
yeah
00:06:30
John S
Not like the inconsistency of like, well, is the wood screw going to go in?
00:06:30
johngrimsmo
Yep.
00:06:33
John S
Is it going to hold? Yeah.
00:06:35
johngrimsmo
Yep.
00:06:39
johngrimsmo
Okay, your turn. What's up?

Drilling Challenges and Solutions

00:06:43
John S
Small win, but a good PSA type of thing. We have um had very inconsistent drill tip life in steel fixture plates. And this this really matters to us because of how much um how many steel how we spend on steel drill tips in production here.
00:07:00
johngrimsmo
the replaceable tip, tip drill insert things.
00:07:00
John S
And yeah. yeah we use the 870 style from Sandvik and we learn, I think I've talked about this before, as a very quick recap, we've learned that it was poking through the bottom side of our 4140, which has a very thin mill scale section. So it's not nearly as thick or difficult to say the outside has that becomes has from the flame cutting of the of the plate, but nevertheless, you have a mill scale from the rolling of the of the product, which which really messes with the drills life in an inconsistent but often devastating way.
00:07:35
johngrimsmo
Hmm.
00:07:36
John S
And um simple change. We realized Sandvik makes the exact same drill that we use series grade or series diameter in a different grade that's tougher. And that change same price, that change alone, i don't, I should have researched this before I got on, but I think we were formerly using like 1130 and there's maybe like 40, 300,
00:07:45
johngrimsmo
Oh.
00:07:55
John S
gray And it's kind of those things like the Laurens of the world, like anybody who really lives and breathes will be rolling their eyes right now. But I do think we all fall in that guilt of not realizing, oh, wow, you know, this company makes two or three different coatings or um this insert comes in a different coating or a different grade or a different edge breaker.
00:08:01
johngrimsmo
Yeah. Yeah.
00:08:12
John S
And it's it's very much a if it ain't broke, don't fix it until you don't necessarily realize. In this case, it was broken. Like we needed a solution, um but it's been great.
00:08:21
johngrimsmo
I find that things like that only become visible when you're doing consistent enough production that you can actually taste the difference.
00:08:29
John S
yeah Yeah.
00:08:30
johngrimsmo
um And we've seen that on our lads too, because we can get, you know say, 1,000 parts out of one insert and 200 parts out of a different identical insert um with a different coating or different grade or hardness or something like that.
00:08:43
johngrimsmo
And it can make like a big, big difference. And it's those, like you said, you look in the insert catalog and you're like, there's four different grades and three different coatings of the same insert. I don't know what I want. I don't know what's going to work best.
00:08:54
John S
Right.
00:08:55
johngrimsmo
And usually, we as machinists, we pick one, we start. And then we just run with it and that's all we know. And it's kind of expensive to test and tune a bunch of different ones. And, you know, the vendor's not always willing to hold your hand and suggest the perfect one and all that.
00:09:05
John S
so
00:09:10
John S
Well, they just don't, they just don't know. Like, and this is since they, they, yeah.
00:09:12
johngrimsmo
They just don't know.
00:09:14
John S
Um, I did pull it up real quick. Um, we were formally using the, uh, PM edge breaker and the 43, 34 grade. And just to keep this as confusing as possible, we have since switched to the KM three, three, three, four, uh, grade against price.
00:09:27
johngrimsmo
Hmm.
00:09:31
John S
I think same price. Yeah. Same price.
00:09:32
johngrimsmo
Great.
00:09:33
John S
Um, Yeah. so and
00:09:37
johngrimsmo
Yeah, we noticed that on our Willimon doing turning um using a V insert. And I changed from like the one we've been using on our Swiss forever, which is a small, rigid kind of machine.
00:09:48
John S
just
00:09:49
johngrimsmo
But in the Willimon with the B axis where the the spindle is holding your turning tool still,
00:09:55
John S
Oh,
00:09:55
johngrimsmo
I wonder if there's some like vibration or something in that motor that's holding the turning tool still that a tougher grade really helped there.
00:10:00
John S
oh
00:10:04
johngrimsmo
Cause otherwise I was blowing inserts every few parts.
00:10:07
John S
yeah Exactly.
00:10:07
johngrimsmo
Um, whereas on the Swiss, we can get like 2000 parts out of this exact same insert doing the same depth of cut, same material, same, everything tougher grade helps on the Williman.
00:10:08
John S
Right.
00:10:18
John S
Are you buying the expensive monolithic or integral turning holders? Are you shoving stick tools in stick tool adapters?
00:10:24
johngrimsmo
Most, mostly the monolithic ones and I blew one up and it was like $900 or something.
00:10:28
John S
Okay. Oh,
00:10:31
johngrimsmo
i was like, yikes. But the stick tool way kind of works too.
00:10:34
John S
got it.
00:10:36
johngrimsmo
I think I have one of those.
00:10:38
John S
Yeah. They're also not cheap, but at least that you can replace the stick tool.
00:10:41
johngrimsmo
Yeah, it's a big deal. Yep.
00:10:47
John S
What's next on your list?
00:10:48
johngrimsmo
Oh, we're flying through.

Coolant Sensor Troubleshooting

00:10:50
johngrimsmo
so since the weekend my obsession has been getting the kern coolant it's got a transfer coolant pump so all the coolant goes into the chip conveyor like most machines except that is not your coolant tank all that coolant gets pumped out into a the filtration system and then filtered through the paper band and then stored as clean chilled coolant um and a lot of chips get pumped to the paper band filter in that chip conveyor coolant storage system, um the factory system had these like stainless steel ball floats that would
00:11:30
johngrimsmo
that would get jammed up with chips. And so I replaced those with these IFM sensors. and I've always had weird problems with them. Chips would wrap around them.
00:11:38
John S
What's IFM?
00:11:39
johngrimsmo
IFM is a brand of sensor like Festo, but I've tried two different kinds.
00:11:40
John S
Oh, what kind of sensor is it?
00:11:44
johngrimsmo
I've tried capacitive and this is a wave guided radar sensor. So there's like a shaft sticking down into the coolant. That's like, millimeter shaft that the waves from the not ultrasonic but imagine the ultrasonic sensor sends weight radio waves down the shaft bounces off your your liquid and then comes guides back up the shaft and hits the sensor and it tells it oh it's 4.2 inches away and there's four outputs so like when the cooling's at you know 4.6 inches it's full turn on the pump and turn it off at 1.8 inches
00:12:20
johngrimsmo
And that's two outputs. And then high limit e-stop and low limit e-stop are outputs three and four.
00:12:27
John S
Okay.
00:12:27
johngrimsmo
So all important stuff. But this sensor is also clogging with chips around the outside and reading improperly and e-stopping the machine randomly. um So the plate that I made on the Speedio, which I could have watered jet, is called the launching plate.
00:12:44
johngrimsmo
And it helps to launch all these signals like better. um
00:12:49
John S
Oh, you just want a better reflector.
00:12:51
johngrimsmo
it's basically yeah and that's what the documentation says um so i made that and now i've been playing with it but it's being weird and it's not doing exactly what i think and everything works now except the low level e-stop doesn't work and i kind of want that to work case you run out of coolant the machine will stop um so i'm just fine tuning that but that's it's been a lot of time getting this work
00:12:55
John S
Yeah.
00:13:07
John S
Yeah.
00:13:15
John S
Yeah, right. And the in the original balls, how do those work from Kern?
00:13:20
johngrimsmo
So they are stainless steel shaft with a stainless steel floaty ball that just clearance slides up and down the shaft.
00:13:26
John S
Oh, okay. Fuel gauge type thing. Okay.
00:13:29
johngrimsmo
Yeah, exactly. And the floaty ball is buoyant because it's hollow and pretty lightweight. But the clearance between the floaty ball and the shaft chips would get stuck in there because this chippy environment is like super chips. And we do a lot of small machining. So we're making small chips.
00:13:45
johngrimsmo
um And there were two of those. One was low high, and the other was e-stop low high.
00:13:51
John S
Okay.
00:13:52
johngrimsmo
And one or both of them would always get clogged up, so we'd have to clean them often and have machine e-stops and problems like mid-run when the five-axis is all twisted up.
00:13:59
John S
Yeah. Yeah.
00:14:02
johngrimsmo
um
00:14:04
John S
Yeah. You don't want a better, you don't want a better warning about if things are clogged, you want the things to not be clogged.
00:14:10
johngrimsmo
to not be clocked, exactly.
00:14:11
John S
see
00:14:12
johngrimsmo
So I'm cautiously optimistic with this launching plate and like doing the sensor properly kind of thing.
00:14:13
John S
Gosh.
00:14:18
johngrimsmo
um Might be better. Otherwise, i did quite a bit of research on different styles of sensors. There's ultrasonic sensors that literally, I don't know, they shoot an ultrasonic wave at the surface, totally non-contact, no probe, no nothing, except it doesn't really read well in foam.
00:14:32
John S
Mm-hmm.
00:14:37
johngrimsmo
So that's not ideal. There's what CJ came up with is a pressure sensor, which basically simple way is put a tube in the water and put a pressure sensor way up top outside of the business end.
00:14:51
johngrimsmo
And it measures the water column. though you know You got five inches of water. That's however many PSI of downward pressure that the water tank is pushing on that sensor. Other ones have submersible sensors where you can literally just drop it to the bottom of the tank.
00:15:05
johngrimsmo
and it measures how much water is above it. um
00:15:08
John S
Yeah. yep who Yeah.
00:15:10
johngrimsmo
Impervious to foam, impervious to a lot of things. It it kind of sounds like the best option, but to make to make it work for me, I would need to rig up a four output system to interpret the results.
00:15:24
johngrimsmo
um You can do it with an Arduino or a Pi or something like that, but I kind of convinced myself, I don't want to mess with all that right now. I just need it to work.
00:15:35
John S
Interesting.
00:15:37
johngrimsmo
And then the other one I like, which I'm pretty sure is on Amish's DMG, is a toilet bowl float.
00:15:37
John S
Yeah. i
00:15:44
johngrimsmo
Like big rubber floaty ball thing that sits on top of the coolant and your mechanism is outside. So the switch is is like way above an outside and I kind of really like that system.
00:15:53
John S
yeah
00:15:57
John S
John, for sure. Well, that's like what we built and then never used for the horizontal was the same idea of a float, uh, a float in the horizontal because I didn't want to look in the tank. So you just put a float in there with a yardstick and a color markings on it.
00:16:08
johngrimsmo
Yeah.
00:16:11
John S
And it just tells you the coolant tank level in a way that I is really impervious to chips. Like, um,
00:16:17
johngrimsmo
Yeah.
00:16:19
John S
It just is.
00:16:20
johngrimsmo
yeah
00:16:21
John S
And it's funny when people, but this is just like an overall commentary in the world. Like people will poo poo the toilet bowl thing as CJ gets grief for it. Oftentimes it's like, no, do you understand the genius? Like there are literally probably tens of billions of toilet bowl floats.
00:16:35
John S
Like they're good. Same thing with like the whole, like the Titanic sub use a PlayStation controller. That's so janky. I'm like, No, no It's like a very, very, very, very highly developed, well mass produced consumer electronic device that's way better than like a custom one off from some think tank.
00:16:48
johngrimsmo
It's a good point.
00:16:50
John S
or It just is.
00:16:51
johngrimsmo
That's a really good point. I never thought about it like that.
00:16:54
John S
Yes, percent. the.
00:16:56
johngrimsmo
ah for
00:16:58
John S
but so the the thing I can't, well, first off, this gets me, my mind going again on AI with Home Assistant, which I'll come back to in cameras.

Innovating Sensor Systems with AI

00:17:07
John S
It doesn't solve your problem because you don't want to know when it's clunked up.
00:17:09
John S
You just want to, a non-clunk.
00:17:11
johngrimsmo
It just needs to work. Yeah.
00:17:13
John S
But like, if you, if you used, uh, Humor me here. This isn't going to be your solution. But if you had a if you drilled a hole in side, of you're not gonna do this. If you drilled hole in the side of the tank, you put a really fine mesh filter over that hole with baffles before it. So the baffles will stop chips from getting through. The filter will, the mesh filter will screen will stop getting chips getting through.
00:17:32
John S
And then you have a aquarium pump agitator blowing at it to avoid, you know, the whole issue of like, Hey, the outflow from the dam gets trees built up against the grates. Like that's always what happens.
00:17:42
johngrimsmo
right
00:17:44
John S
Then you have an external, what is it? Bernoulli's principle, whatever the water levels will stay the same. Now you have an external cavity, an external tank that could be the size of a paint can that reflects your water level that will be pristine.
00:17:56
John S
And you can now use one of these commercial off the shelf sensors without having to worry about chip contamination,
00:17:56
johngrimsmo
he
00:18:02
John S
you're done. Now, again, I think a lot of that's probably a bad idea, but wouldn't that...
00:18:05
johngrimsmo
And that's like the obvious answer is to put mesh screens between the, where the cooling comes from and where the sensors are.
00:18:13
John S
since Yeah, that's what I'm wondering.
00:18:14
johngrimsmo
But I don't trust that that's not going to get clogged up with chips unreliably um way too fast. I think, yeah.
00:18:22
John S
That screen plus the aquarium blower.
00:18:22
johngrimsmo
Yeah.
00:18:24
johngrimsmo
that might help to keep it cleaner, but still like, Oh, that's a good point.
00:18:28
John S
and baffle And baffle, because chips can't go through baffles. That's the beauty of...
00:18:31
johngrimsmo
Yeah.
00:18:33
John S
3D prints, even just a four baffle. These were used all the time in aquariums, like protein skimmers and stuff.
00:18:39
johngrimsmo
Yeah.
00:18:40
John S
Then that's the simplicity of it. Now you're you're getting almost no chips to the screen and the screen's also being blown out. Actually, the screen could be blown outward with that pond pump thing or pulsated out.
00:18:51
johngrimsmo
With the baffles though, because the coolant level is constantly changing from two inches to five inches, how does it get over the baffles into the clean zone?
00:18:57
John S
Doesn't matter.
00:19:04
John S
You just have to have the baffle entrance, like a quarter inch off the bottom. And then you have it just come up. You come up enough. That's still like below your high level.
00:19:11
johngrimsmo
Oh, you go up, down, up, down.
00:19:13
John S
Bingo.
00:19:14
johngrimsmo
I still think the chips are, what's it called? Not buoyant, but they're like floating around enough that they would be everywhere. Maybe.
00:19:22
John S
Okay.
00:19:22
johngrimsmo
Yeah.
00:19:23
John S
Yeah.
00:19:24
johngrimsmo
But it's a good thought exercise. And it's been kind of fun thinking about this. Like, partly like, dude, stop wasting your time. Just solve the problem. But also like, no, it's actually really important for me in my future life to know how float systems work best.
00:19:38
johngrimsmo
Like, i want this information, you know?
00:19:42
John S
Has Kern changed this on the new machines?
00:19:44
johngrimsmo
They said they have, um but I can't tell you what exactly they're using nowadays. And I haven't bugged them too much about these. They actually sent me one of these IFM sensors that didn't work work as well for me.
00:19:55
John S
Oh, okay.
00:19:57
johngrimsmo
um So, what yeah.
00:19:58
John S
As I said, don't ask them because maybe they'll even hook you up or help you out. Yeah. yeah And CJ's thing is, pretty i i hope we're not like spoiling his IP here.
00:20:09
johngrimsmo
We can tease. Yeah.
00:20:10
John S
Yeah, CJ's thing looks pretty sweet.
00:20:12
johngrimsmo
yeah
00:20:15
johngrimsmo
um
00:20:18
johngrimsmo
yeah so that's It's been eating up a lot of my time since, I mean, it's Wednesday now, since Sunday.
00:20:19
John S
good.
00:20:22
johngrimsmo
but um That's
00:20:24
John S
Yeah. Is it worth creating a process where at 3.50 or, you know, 10 minutes before you leave or somebody leaves, they always go and physically physically clean out the chips just as solid band-aid?
00:20:37
johngrimsmo
what we have been doing for the past few months, but it gets forgotten about too easily and
00:20:38
John S
Yeah.
00:20:42
John S
So that's something you got to fix, John.
00:20:43
johngrimsmo
Well, it's not just that, but it's like the process was every week we clean it out.
00:20:43
John S
Between... you
00:20:48
johngrimsmo
And then the problem happens more than every week. And then, you know, but also the cleanup we were doing was literally the second you unplug it, the machine e stops and you have to like bring it over to the sink and clean it and put it back in.
00:20:54
John S
No, I know. I'm not saying that I have the best answers, but.
00:21:04
johngrimsmo
That's not good.
00:21:05
John S
No, that sucks for sure.
00:21:06
johngrimsmo
Yeah. So i'm I'm close. It's like I pretty much have a solution, but it's been interesting.
00:21:08
John S
Okay. Yeah. Yeah. No, it matters for sure. hear you.
00:21:17
johngrimsmo
It's the ultimate reliability on the current. Like you don't want to flood. You cannot have a flood and the machine needs to pump this coolant properly, reliably.
00:21:22
John S
Right.
00:21:25
johngrimsmo
ah It just needs to work. You don't want to e-stop either. It's just wasting time. And we've had so many random e-stops because of this over the past few years that i'm like, no, like what is, what is the right solution? Like what is the most reliable, bestest?
00:21:40
johngrimsmo
Yeah. Yeah.
00:21:41
John S
Yeah, I hear you.
00:21:45
John S
I like the float. The toilet bowl idea is pretty awesome.
00:21:47
johngrimsmo
it's pretty great. But, but for me to DIY that I need to, to turn that into four outputs, like through a software or whatever.
00:21:50
John S
And it's like...
00:21:56
John S
You don't have to have the you You don't want to flood the shop, but flooding the shop is an easy one to solve with water alarms, I think. Right?
00:22:06
johngrimsmo
Yeah.
00:22:06
John S
So
00:22:07
johngrimsmo
If they interview with the e-stop,
00:22:10
John S
you could, yeah. I guess I'm, I think I'd be willing to accept a solution that just had the toilet bowl high, low, like on off.
00:22:11
johngrimsmo
it's like you, you,
00:22:18
johngrimsmo
But overflow and no coolant should e-stop the machine as well.
00:22:23
John S
Explain that.
00:22:25
johngrimsmo
If there's too much coolant for whatever reason in the tank, I want the machine to e-stop. To stop pumping coolant.
00:22:35
John S
Okay, sorry, I'm slow. here I think I understand it, so hear me out. So you have a you have a bucket, let's pretend five gallon bucket. So when the bucket gets near the top, you want it to start pumping the bucket, the fluid out.
00:22:42
johngrimsmo
Yeah.
00:22:47
johngrimsmo
into the coolant tank yep
00:22:47
John S
But if it actually, yes, but if the bucket gets too high, I don't care if you're still pumping, you need to stop because something's wrong. So you, it's, yeah, that makes, need to stop.
00:22:56
johngrimsmo
yeah you need to stop pumping otherwise it's gonna flood which is an e-stop that's the best way yeah and then if if that bucket gets empty
00:22:59
John S
Well, ah yeah exactly, I see what you mean. Sure. That makes sense, John. Okay. Sorry.
00:23:08
johngrimsmo
ah it probably means there's something really wrong, like no coolant anywhere. Like you've already flooded all the coolant onto the ground. and
00:23:17
johngrimsmo
And now the machines not the machine's machining with no coolant, which is also really bad.
00:23:17
John S
Yeah.
00:23:19
John S
Sorry. Yes. we wouldn't it Wouldn't that cause a different alarm somewhere?
00:23:25
johngrimsmo
In the the secondary coolant tank, there are three float sensors there as well.
00:23:30
John S
Yeah.
00:23:30
johngrimsmo
So yeah, i would it would do that, but yeah.
00:23:30
John S
Yeah.
00:23:33
John S
Has it ever, that that first scenario situation where the five gallon bucket starts to overfill, has that happened?
00:23:39
johngrimsmo
Yes. Oh, yeah. Especially with the factory sensors, because the upper e-stop sensor would get plugged up with chips, and it would stay stuck in the middle, and it would just not not know.
00:23:46
John S
Okay. Got it. Didn't know I had to pump out.
00:23:50
johngrimsmo
Yeah.
00:23:51
John S
no Okay. Yeah, sorry.
00:23:53
johngrimsmo
So it's interesting. And shortly after my current, and I think they stopped using the stainless steel floaty balls and went to better sensor systems or something like that.
00:24:00
John S
Okay.
00:24:02
johngrimsmo
But...
00:24:04
John S
Interesting.

AI and Vision Systems in Manufacturing

00:24:05
johngrimsmo
yeah
00:24:07
John S
Well, that kind of segues to what I got to spend some time this weekend doing. not Not the answer here, I don't think, but I think it's worth keeping an open mind about the power of vision systems and AI.
00:24:18
John S
um you know You could have a camera trained at your ah sight glass on the side of a tank, which probably has a sight glass, and the site one camera could read that sight glass and make decisions about what to do.
00:24:30
John S
Now, I understand there's single points of failure or other flaws in this, but um I got the chance to spend some time with a buddy this weekend who is always an inspiration to hang out with. And he's building a, i mean, he has built a, um, raspberry pie with a webcam and he bought a hundred fake eyeballs. And now when I say fake eyeballs, these look as good as prosthesis. Like they look very, very good.
00:24:57
John S
He designed his own articulating eyeball socket. It's 3d printed with apparently bamboo cells, flesh colored, quote,
00:25:00
johngrimsmo
Look at it.
00:25:04
John S
seems quite like, don't know what, uh, gender profiling. This is what we've decided the flesh color is, but nevertheless, there's skin ish toned, PL, print. And so he has a, he has, doesn't have a hundred mounted yet, but he has 20 mounted on a, on a small wall.
00:25:21
John S
And when, when you walk into the view of the webcam, it, it finds you or multiple people. And then it decides whether all of the eyeballs are going to track you, or the eyeballs are going to split, whether they're going to bounce back and forth, whether they're going to go into creep you out mode. In fact, actually I'll show you, um I'll show you here if you don't mind real quick, one of these eyeballs working.
00:25:46
John S
Um, it is super weird.
00:25:48
johngrimsmo
That sounds weird.
00:25:51
John S
Um,
00:25:54
johngrimsmo
Okay, they're pretty big. Oh my god.
00:25:56
John S
Are they real? Yeah.
00:25:57
johngrimsmo
They're fast too.
00:25:58
John S
Super fast. And the eyelids open and close, which is part of what makes it what humanistic, anthrop whatever.
00:26:01
johngrimsmo
my gosh.
00:26:05
John S
Anthropomorphic, is that the word? um
00:26:06
johngrimsmo
Yeah.
00:26:08
John S
So the point of this wasn't just that it's really cool, you know technology, three d printing, making, designing, you know fusion, all that, but also um he's quite good on the programming side. And so we started looking through all these Python scripts of like, okay, here's how we...
00:26:22
John S
you know, grab the image from the um webcam. And this stuff is over my, it's, I couldn't do this on my own, but it's also, it's not like he's writing the, all the code, he's using existing Python libraries to do this.
00:26:34
John S
It's like, okay, here's a webcam image. Let's find a human face. That's an existing plugin.
00:26:37
johngrimsmo
okay
00:26:38
John S
Or now we need to find find the coordinates of that in space. And I need to figure out how how that coordinate of that person in space tracks to the X, Y, Z of the eyeball. um And you start to break it down and you realize, no, this is really freaking cool.
00:26:53
John S
Um, so the, that was point one point two was like, it did get me fired up to think more about getting back into programming a little bit. And I'm not, I'm not good, but on the flip side, okay, start.
00:27:04
John S
And so in the, uh, downtime I wrote with the help of some Google searching and even some chat work, reach checking, I wrote a prime number generator in Python.
00:27:14
johngrimsmo
Ooh. Okay.
00:27:16
John S
Um,
00:27:17
johngrimsmo
It's like a design exercise, just like see...
00:27:20
John S
Well, so I Googled, to be honest, didn't know how you mathematically find prime numbers and you you take the square root of a number. So let's say what what we want to find out a number's from, what numbers from from zero to 100 are prime.
00:27:32
John S
So let's say we're now at number 98. What you do is you take the square root of 98, or let's do 81, square root 81 is nine. So then you do counting by twos, you would check to see if it's divisible by those numbers. So you, um in the Python code, and I won't bore you here, but the real quick,
00:27:48
John S
view is you say, what's the minimum number you want to search through? What's the range of numbers you want to search through? So like I want look from numbers 10 to 500. And then for each number 10 through 500, actually you should skip the even ones for obvious reasons, which I i could say, good thing I could do to edit that.
00:28:04
John S
Um, you get the square root of it and then you have to build a list of counting by twos up to the square root. Is it divisible by it? Which you use the mod thing to to determine whether it's a prime number. And if it is a prime number, you add it and into a prime number. I think it's an array, like a dot append command.
00:28:19
John S
Um, and again, I needed chat to proofread my code and fix a few things. So I did not, I could not have written this in isolation, but I also, I wrote it like I did it super fun.
00:28:23
johngrimsmo
Yeah.
00:28:28
johngrimsmo
Yeah. Oh, that's awesome.
00:28:31
John S
Yeah.
00:28:32
johngrimsmo
Good. Good for you.
00:28:32
John S
And then i'm like, why why am I doing this? Because like, then you asked chat, did you just write it from scratch? It was fine. And so I don't have the answers, but I think a lot of us are wondering about this.
00:28:41
johngrimsmo
Thank you.
00:28:43
John S
Like, why am I doing this? If I can just get a better thing from it and off the gap, Matt?
00:28:49
johngrimsmo
Yeah, it's that existential question of like, because I enjoy it and I want to understand it. And, you know, ChatGPT can spoon feed you all day long and it's great for a lot of things, but some things I just want to fundamentally understand.
00:28:54
John S
Yes.
00:29:02
johngrimsmo
And I'm very slow to do that. I take my time. I do a whole lot of research.
00:29:06
John S
Yeah. Yes.
00:29:07
johngrimsmo
I do a whole lot of testing and I want to figure it out myself. Like last night I picked up TIG welding um for for the first time in 15 years and I did not do any research, didn't watch a tutorial.
00:29:13
John S
I saw.
00:29:19
johngrimsmo
I did look up, do I put it on DC negative or DC positive? Just to make sure like,
00:29:23
John S
a
00:29:24
johngrimsmo
you know, basic settings for stainless steel and starting amperage and stuff. Other than that, I, I just picked it up and I just made a bunch of mistakes and I got the tungsten stuck to the piece and I, it worked.
00:29:31
John S
yeah
00:29:36
johngrimsmo
um But it was fun.
00:29:36
John S
Good for you.
00:29:37
johngrimsmo
It was super fun.
00:29:39
John S
Yeah, like the program that this guy's written for this webcam stuff is like six or seven different Python scripts. So that's, I think, I think where like chat GPT is not going to write you an operating system, right? It's not to do multi, don't think, multi-file architecture like this.
00:29:54
John S
So I do know that having some basis of knowing what to look for, what to ask, where Python libraries, when do you need the math library or OS library and
00:30:02
johngrimsmo
Yeah, yeah.
00:30:03
John S
um is going to help me be better at, and I think the the goal here to keep it relevant both to me professionally, but also this podcast is um using stuff like that when it comes to home assistant and UR script on how we're going to do the automation of the new one piece flow with the UR robots, with lights out manufacturing setup, with the brothers to say, hey, when the part comes off, you can use a web camera to see, i mean, you could use you use a web camera, John, to evaluate the state of your chip ah filters in the current tank. Now, you don't want reactive, you want proactive here, but...
00:30:39
John S
I love this idea of like, hey, I can just, tell I could tell when it needs to do a flush of chips in the machine or bingo.
00:30:40
johngrimsmo
yeah
00:30:47
johngrimsmo
yeah it's a feedback loop a visual feedback loop because the operator walking around going oh that needs to be replaced so how do you automate that kind of thing um
00:30:57
John S
Yes, right, right.
00:31:00
John S
Yep. Wasn't like that idea.
00:31:02
johngrimsmo
Well, the other thing I was gonna say with integrating ChatGPD or whatever into our ah development cycle, The more you know, as whether a programmer or machine designer or whatever, the better you will guide the chat to give you the answers you actually need.
00:31:18
John S
Yeah, right.
00:31:20
johngrimsmo
And you'll better be able to filter it and be like, no, that's stupid. That's not going work.
00:31:25
John S
Yeah.
00:31:26
johngrimsmo
ah Or like, let's try it this way. And that's the whole thing with working with ChatGPT is you are guiding it. It's not just it'll feed you an answer.
00:31:33
John S
Yes.
00:31:34
johngrimsmo
it doesn't make it right. ah we pseudo experts are feeding it a path and it's going down that thread because we tell it to.
00:31:43
John S
Yeah.
00:31:44
johngrimsmo
And I want to know intrinsically if that's a good thread to go down. you know
00:31:49
John S
A hundred percent. hundred percent. It was also nice that I was talking to home assistant, um, with my buddy and, um, he was kind of like, he gave me that little validation that I didn't need. Cause I was already confident, but like still nice to hear that he wasn't like, you're a lunatic. That's the wrong. He's like that. Yep.

Machinery Updates and Automation Plans

00:32:07
johngrimsmo
Yeah.
00:32:07
John S
Yep. He was kind of like, yep, you're on to something. Uh,
00:32:09
johngrimsmo
Nice. Cool.
00:32:11
John S
Yeah. It's like, I want when the machine e-stops because of the robot failed at 3 AM, I just want it to put the 10 minutes of footage in a folder, email me with the link to that folder.
00:32:21
John S
So it's in my inbox and I just, I don't search, don't scrub for the footage. It gives me a report in the video footage and I watch it like that's boss mode.
00:32:25
johngrimsmo
That would be sick.
00:32:30
John S
That is what we're doing.
00:32:30
johngrimsmo
Yep, yep, yep. If you can get a ah screenshot of the controls, you can see like what line of code and all that too, that could be awesome. Mm-hmm, mm-hmm.
00:32:37
John S
Absolutely. Yeah.
00:32:41
johngrimsmo
Because the current being an automated equipment, we're constantly debugging a failure mode. Like, you know, you come in on a Monday and you're like, oh, it's it's down. It's ah what happened, when it happened. Can I pick it up? Is something broken? is yeah what happened, when it happened? And we've gotten pretty good at it on the current with some of the systems I've put in. But for the most part, it's figuring it out.
00:33:03
John S
You ever seen that um Crossfire Plasma thing? I think it's from Langmeier.
00:33:12
johngrimsmo
Okay.
00:33:12
John S
it's just a plasma, but instead of it being a bridge style plasma table or gantry, it just has ah single piece of 8020 that goes out and another piece off of it. So it's like a XY coordinate that's only fed from one bar.
00:33:24
johngrimsmo
Okay.
00:33:24
John S
Sorry, that's way more complicated. What I got to thinking about was apparently the brothers are not as automation friendly as like a Haas is. So you can't have the brother load a new program, but then seeing what you can do with,
00:33:36
John S
visual camera systems and AI. It's just like, no, I can just train the AI. I can put a little finger, fake human finger on a XY grid over the control that's out of the way during the day, but it can actually come in and recognize the screen and load a new program.
00:33:52
John S
Like this is huge risk.
00:33:53
johngrimsmo
it could actually yeah i talking with um greg um luma labs he he says there is a next level of speedio not control but like offline control
00:33:54
John S
I'm well aware of this, but it's also like, heck yeah.
00:34:01
John S
Yeah. Luma, yeah.
00:34:09
John S
Oh.
00:34:10
johngrimsmo
that can do pretty much anything that like the big factories are using. um
00:34:14
John S
Interesting.
00:34:14
johngrimsmo
It basically has full control if if you choose to go there kind of thing, um which sounds super interesting, but I don't need it.
00:34:19
John S
Yeah.
00:34:22
johngrimsmo
But it's cool that it can.
00:34:23
John S
Yeah. Yeah.
00:34:23
johngrimsmo
Yep.
00:34:24
John S
we Exactly. I don't need this either, but it's also just like, I always laugh when they're like you can't do that. And I'm like, watch me like build a robot finger that's going to just do what it needs to do.
00:34:31
johngrimsmo
if you
00:34:33
johngrimsmo
ye
00:34:34
John S
Yeah.
00:34:35
johngrimsmo
Well, like on the current, because I can VNC into the Linux backend, can control everything except for cycle start, feed hold, e-stop reset, e-stop, things like that.
00:34:46
John S
Yeah.
00:34:46
johngrimsmo
But I can do a lot of other stuff. I can turn the coolant on and off.
00:34:49
John S
be
00:34:50
johngrimsmo
That's
00:34:51
John S
Which is awesome. It'd be also very easy to build an IOT inline switch that would hit your cycle start for you remotely.
00:34:57
johngrimsmo
good point.
00:34:58
John S
That's what Lights Out guys do. You just wire it in series.
00:35:01
johngrimsmo
No way.
00:35:02
John S
Yeah. And then it's what it literally, you cut your cycle start wire and wire the second switch in.
00:35:02
johngrimsmo
To the actual physical button.
00:35:08
John S
And then the pressing of either one of them is what triggers the cycle start. And then the other one is obviously pressed digitally.
00:35:13
johngrimsmo
Because those are momentary buttons, they're not hard switches.
00:35:17
John S
Bingo.
00:35:18
johngrimsmo
The e-stop is a hard physical switch, but huh.
00:35:20
John S
Yes. Well, e-stops are even easier. Like I think actually e-stops have to be designed to allow additional e-stops to be kind of wired in. So anything trips them. Like that's easy.
00:35:29
johngrimsmo
Yeah, the whole thing's an open loop string of e-stop, right?
00:35:32
John S
Yeah.
00:35:34
John S
Sure.
00:35:35
johngrimsmo
Yeah, whatever. Cool. So what's the speedio lights out update?
00:35:38
John S
Yeah, it's fun.
00:35:42
John S
Lighthouse update is we sold the lathe, which is awesome. So it's just got it.
00:35:46
johngrimsmo
Congrats.
00:35:47
John S
Thank you. It's awesome to see that another lathe leave the shop. It's on my way to heaven. Heaven being the shop without lathes.
00:35:53
johngrimsmo
maybe You had me tripped up the other day. were like, do not use Wordle. Do not do the Wordle word today.
00:35:59
John S
Okay.
00:36:00
johngrimsmo
And I don't play it, but I know my wife does. So i texted her at like 8 and I was like, what's the word? that Saunders is really messed up about this word on Wordle. And she doesn't do it till the evening. So I didn't get to know until like 9 p.m. when she did it. she sent me a screenshot. It's lathe. And I'm like, oh my gosh.
00:36:16
John S
Well, I was upset. Actually, it was funny that Lockwood texted me in like, because he saw my disclaimer. So it's kind of like, okay, like ah ah not numerous people on Instagram DM me screenshots that they got on the first try, which is just like warms my heart.
00:36:30
John S
And Rob and I both got it in like five.
00:36:33
johngrimsmo
yeah nice
00:36:33
John S
Like we're like latte, you know, loy like, and like, I got like, I got nothing. I got that. Like, I'm like, you gotta to be kidding me. But I guess hilarious.
00:36:44
johngrimsmo
nice
00:36:46
John S
Yep. Okay, so lathe goes out in about two weeks. I'm going to have the brother delivered, I believe, that same day. so we'll actually do two shifts passing the in the night.
00:36:52
johngrimsmo
me
00:36:55
John S
um And then we have a gentleman coming momentarily to look at another machine, which I'm quite hopeful could work out um because that's the last kind of piece of the puzzle. um So brother will be here around Thanksgiving is your short answer.
00:37:09
John S
I'm going out to Idaho to see the Lights Out guys in ah in eight days or something.
00:37:11
johngrimsmo
Cool.
00:37:14
John S
um They're working on our build. we have this we have the They have the Chicago robot. We have the Michigan robot here. Courtney's running it as we speak, like messing with it, like treating training it, doing stuff with it.
00:37:26
johngrimsmo
Cool.
00:37:27
John S
um Yeah, and it's awesome.
00:37:30
johngrimsmo
That's fun.
00:37:31
John S
Yeah, that's the good news. The downside is kind of like, ah, man, it's only November 5th, but the reality is... between those machines coming and going, training, holidays, Black Friday sales, all that stuff, it's gonna be the first year before we're really like focused back in on on some of this, which is fine.
00:37:47
johngrimsmo
yeah yeah does the ur addition that you're adding have to be bolted to the ground because the speedio does not
00:37:58
John S
The UR will be on a cart and that cart will be mounted to the ground. Does it have to be?
00:38:03
johngrimsmo
okay yeah really yeah
00:38:04
John S
Probably not if it were sufficiently heavy. um Yeah, we won't run our UR at full speed. you You get a lot more joint life if you run it at 70 or 50% for sure.
00:38:15
John S
And you get better. i think, don't know how to technically rate them, but your payload stability accuracy increases if you're not at 100%. Yeah. Yeah. yeah
00:38:25
johngrimsmo
Makes sense.
00:38:26
John S
yeah How about you? What's been top of mind or what's on your plate?
00:38:31
johngrimsmo
um Just those things. Did i quick foam thing, did I use the dynamic mixer on the last podcast?
00:38:40
John S
Yes, but you said you were still happy it you were going to keep playing. It was too...
00:38:44
johngrimsmo
I think I'm still there.
00:38:47
John S
Still there. no Okay.
00:38:47
johngrimsmo
It worked well. It mixed the stuff better. it What I'm learning is that I'm mixing the fluid really, really well. at I've got a good ratio now, but it's still curing with multiple big bubbles instead of a whole bunch of little tiny bubbles.
00:39:04
johngrimsmo
um
00:39:04
John S
Okay.
00:39:04
johngrimsmo
So the bubble density is variable, and I'm playing with that. And I think there's I've got a couple things to keep trying, but um my latest result is the best one I've had yet, which is kind of cool.
00:39:16
John S
Well, that's what you you said last week and I'm not going to give you a hard time for sure, but you were, you were starting to recognize needed to find, need to draw a line in the sand.
00:39:24
johngrimsmo
Yeah. And I don't think I've worked on it much since then, but, um, but yeah, progress.
00:39:25
John S
jesus Yeah, fair enough. So progress.
00:39:30
johngrimsmo
him
00:39:32
John S
Good. Good.
00:39:36
John S
um Um, and then you got to work on the current float sensor thing.
00:39:40
johngrimsmo
Yeah. I should have that wrapped up today. Like that should be back and running now.
00:39:42
John S
Ah, good.
00:39:44
johngrimsmo
now
00:39:46
johngrimsmo
Um, yep.
00:39:46
John S
Good.
00:39:46
John S
How

Halloween Product Launch Success

00:39:47
John S
are the holidays? How were the Halloween? how How was the reception to your Halloween products?
00:39:50
johngrimsmo
Oh, it was fantastic. Like everything sold out in minutes. I posted a thing that are like everything sold out in minutes and people are responding. They're like, no, it was seconds. Like I tried to get one three seconds later and I couldn't.
00:40:00
John S
Yeah.
00:40:03
johngrimsmo
And it was great. Huge response.
00:40:05
John S
Good for you.
00:40:05
johngrimsmo
Yeah. Very happy.
00:40:06
John S
That's awesome.
00:40:07
johngrimsmo
lot of hits to the website while people were like at midnight on a Thursday night was kind of great. um
00:40:13
John S
That's awesome.
00:40:14
johngrimsmo
Yeah, so it worked out really well. it was a lot of work for the team and everybody everybody was happy to do it. And it pushed us all to do kind of something new, something extra. um
00:40:23
John S
Sure. Sure.
00:40:24
johngrimsmo
It taught some of the guys some new skills they hadn't really pushed yet. So that was great. Like win, win, win, win. win
00:40:29
johngrimsmo
And other than filming a bunch of videos, I wasn't super involved in the whole project, which is epic.
00:40:29
John S
Good.
00:40:33
John S
Yeah. Good.
00:40:37
John S
You're going to do a, uh, you do a Katy Perry edition now.
00:40:40
johngrimsmo
For what?
00:40:42
John S
Isn't that who your PM is dating?
00:40:44
johngrimsmo
I don't know.
00:40:45
John S
Oh, I'm actually, i love you even more for not knowing that.
00:40:48
johngrimsmo
I have no idea.
00:40:52
John S
Yeah. Oh, good. The Halloween lives looked dial. Actually, it depends too, right?
00:40:56
johngrimsmo
Yeah. ye Yep, yep.
00:40:58
John S
yeah Yeah.
00:40:58
johngrimsmo
Yeah, we did 20 pens, 13 knives. um They look great. like Even Eric, as he's finishing them, he's like, these look good.
00:41:07
John S
You want to keep one for yourself?
00:41:07
johngrimsmo
Exactly.
00:41:09
John S
Yeah.
00:41:10
johngrimsmo
Yep.
00:41:11
John S
Sweet.
00:41:12
johngrimsmo
Let's see here. um
00:41:14
John S
see you See you next week then?
00:41:17
johngrimsmo
Sounds good. Yeah, you've got to go right now, right?
00:41:19
John S
Okay. Yeah, sort of. Yeah, sure.
00:41:21
johngrimsmo
That's fine.
00:41:23
John S
All right.
00:41:23
johngrimsmo
Perfect, yeah. All right, bud.
00:41:24
John S
you. Take care.
00:41:24
johngrimsmo
See you next week. Bye.