BGA vs. QFP – PCBA cost comparison
- Inspired by Tim Jager on the slack channel
- 48 pin QFP vs 56 pin BGA
- Escaping the package
- Layer count
- Increase in PCB tolerances
- Blind and buried vias required?
- Inspection costs
- Leaded packages like QFPs can easily be optically inspected
- BGAs typically require X-ray
- BGAs typically require small paste apertures and might require more expensive coatings and stencils
- Depending on lead spacing you may not need a stencil at your CM for a QFP
Using the MF platform to predict pricing
- Napkin math your PCB assembly cost the “easy” way
- Come up with a rough Bill of Materials
- Draw up a quick board outline that might be the size you require
- Enclosures dictate PCB sizes more often then not
- Upload to the platform
- Can change the bill of materials on the fly to make pricing considerations at prototype to high volume
- Ways to control pricing on your PCB Assembly
Special thanks to whixr over at Tymkrs for the intro and outro!
About The Hosts
Parker Dillmann is MacroFab's Co-Founder, and Lead ECE with backgrounds in Embedded System Design, and Digital Signal Processing. He got his start in 2005 by hacking Nintendo consoles into portable gaming units. He also runs the blog, longhornengineer.com, where he posts his personal projects, technical guides, and appnotes about board layout design and components. Parker graduated with a BS in Electrical and Computer Engineering from the University of Texas.
Stephen Kraig began his electronics career by building musical oriented circuits in 2003. Stephen is an avid guitar player and, in his down time, manufactures audio electronics including guitar amplifiers, pedals, and pro audio gear. Stephen graduated with a BS in Electrical Engineering from Texas A&M University.
Host 2 00:11
Welcome to the Mac fab engineering podcast. We're your hosts, Parker, Dolman.
Host 1 00:15
And Steven Greg,
Host 2 00:16
This is episode 290. So we got some very interesting discussion in the Slack channel today. And this was a question inspired by Tim Yeager. You pronounce up
Host 1 00:30
To me. By the way, this is totally an endorsement for joining our Slack channel. Because if you come and talk about cool stuff than we will probably talk about on the podcast.
Host 2 00:40
Yeah. So Tim asked about developing new products. And they were talking about, I guess, their microcontroller. The fact that he can, they could buy a microcontroller. And they're picking like between packages. Yeah, they have options. They have options. what's your what's the part number Tim? what's your
Host 1 01:01
What's the family? What who makes this magical thing?
Host 2 01:07
In stock Inc. So anyway, so Tim was asking this package of I want to assume microcontroller. This microcontroller is either in a BGA or q if P package and they were he was wondering about basically what what the pic in terms of like pricing and reliability and all this other stuff?
Host 1 01:30
Well, with with the caveat that the BGA was actually slightly cheaper than the qf P.
Host 2 01:37
Yeah, the raw component cost the BGA is less expensive. Which makes sense, it would be a easier package to manufacture on from an IC side.
Host 1 01:49
And it's a 48 pin qf p versus a 56. Pin BGA.
Host 2 01:56
I'm gonna guess the extra pins are probably grounds and powers and that kind of stuff?
Host 1 02:01
Probably. Yeah. Probably not a bunch of extra functionality pins.
Host 2 02:06
No. So I guess we'll talk about this question, because I thought it was really interesting to think about of, when you actually have options, what's the best way to go about with your product? For me, I'm just gonna come right out. And I'm just gonna say, go with leading packages, if you can.
Host 1 02:29
Compared to manufacturer like you,
Host 2 02:32
You manage that? Well, not just your manufacturer, like you, but your products gonna be cheaper to manufacture, and that kind of stuff. But why? Why would you? Why should you pick the CUPE over the BGA, even though the BGA is less expensive for the park components cost? I mean, there's a lot of factors behind that. So I guess a lot of factors. Yeah. So the first one escaping the package. I was hoping to have like an escape from LA or Escape from New York joke, but I've never got done the right one. So escaping the package. So what does that mean, Steven?
Host 1 03:08
Okay, so if you've got, if you got a surface mount part that has pads underneath it like a BGA, then to get traces out from underneath the BGA. That's called escaping. So when it comes to actually routing this part on a board, to get the the traces from the internal balls on a BGA can be significantly difficult. Now, with something like a 56, pin BGA, that's kind of an interesting number, I'm assuming it's a rectangular BGA. That's like a seven by eight pin style. So what that means is there's multiple rows of balls in this BGA. So to access the signals that are on some of the internal balls of the BGA can be tricky and can raise the assembly cost, just because you're picking one of these BGA styles. So with something like a qf P, all the pins are on the side, or the legs of the component. So typically, that's four sides of that component. So escaping it as in getting traces out from the part is considerably easier than a BGA. And most of the time cheaper. It's already technically done for you. It's already escaped, right?
Host 2 04:27
It's already escaped. Yeah. So on the BGA. So you actually even wrote an article?
Host 1 04:32
Yeah, actually wrote a pair of art articles. It's, uh, yeah, there's a, there's two. It's a series of two articles that talk about how to escape a BGA. And then some of the difficulties behind that and some of the considerations and the impacts on your on your PCB. So I think that's still on the macro fab blog. In fact, if you go to Google and you type in escaping BGA, I think that it shows up. Yes. So actually, so here's one of the rules. The thumb when it comes to BGA packages, if you, if you count the number of rows, or the number of columns of pads in a BGA, whichever one is smaller, most of the time BJs are squares. So your your rows and your columns are going to be the same. But in this case, like I said, I think it's a rectangle. So pick whichever one is smaller for every pair of rows or columns, that adds a layer on your PCB. So if you have four rows and columns, that you could get away with a two layer PCB, but as soon as you you're getting bigger than that, you need you need more than that. And most PGAS, you're never going to get away with a two layer PCB unless it's one of those little like four banger guys, you know. And those are only require one layer to get to escape. So BGA is required that your layer stack up in your PCB increases dramatically as soon as you add more pins to your BGA. Now, that's assuming you need to access all the pins on a BGA.
Host 2 06:00
Yeah, because it might be that most of the inners are grounds or powers and you can easily access those with just a little, little trace off with a via down, you know?
Host 1 06:12
Well, okay, so there's difficulty with that. Because if you if you think of a BGA one of the easiest ways to look at a BGA is just consider four adjacent balls, and then do the geometry in between those balls. And consider what's your largest via you can create with your largest annular ring around it, that fits within your manufacturers available tolerance and whatever you're willing to pay for, as soon as the pitch of the balls and in other words, the distance in between balls starts to shrink, which in BGA is that sort of the trend is things get smaller and smaller, your your cost starts to skyrocket because your your vias get way smaller, which which necessitates that your annular rings get smaller, which starts to become much more difficult to manufacture the actual PCB. And then you have to start getting into more I guess wild manufacturing processes like VI and pad and things like that when the pad is already four thousandths of an inch. Right? Yeah. Yeah. Basically,
Host 2 07:19
Your your your PCB tolerances start to go up, compared to the QFN QFC. I mean, it also depends on the QFN, arc or QP, as well, because you start getting basically, once you get below, like half a millimeter pitch, you start having to bump up your tolerances as well then make sure you know, are you getting enough? If you can? Are you getting solder dam on your silk on your on your solder mask? Or if you can't get that you know, are you are your pads basically the width of the of the of the J Li You know? Yeah.
Host 1 07:58
Well, okay, certainly depend with a BGA you kind of get hit in two ways. Whereas with a qf P, you only get hit once and what I mean by a hit is like your your board cost increases. So with a qf P, as the pitch decreases, you need to get tighter tolerances on your on your PCB. And that usually comes with adding cost and macros have I believe you guys call that extended? Right? Yeah, just extended manufacturing. Yeah. And, and what that typically means is just your tray size and your space. Yeah, right can get closer, whereas with a BGA, most of the time, you already have to be at your extended manufacturing for your tray size. And but then you also have to get beyond that smaller vias, so you pay two times over for that, or it's not 2x. It's just you pay for both manufacturing.
Host 2 08:51
And if you go and you go into blind buried vias, for small volume stuff, that's usually just like a blanket adder that the PCB fabricator will add on. But if you're in production, you basically pay per blind buried via, is how that works out. All right. So that's about the PCBA level. I think that's about all I can think about in the PCBA not PCB PCB level, like the actual printed circuit board, you'd have to worry about.
Host 1 09:20
Yeah, I mean, there's. So this is fairly standard now. But your surface finish Yeah, you pretty much have to get a flat surface finish. So E neg and a what OSP start to become but at the same time, those are kind of standard. So they don't necessarily add much cost, but they do guarantee that you need something flat like that.
Host 2 09:43
Correct. Great. Yeah, I can't think of anything else in the PCBA that are I keep seeing PSBA on the PCB that's different. But on let's say the assembly side or post assembly inspection costs. Because with lead ID packages like a qf You can easily just aeoi the whole board. And most most piece VA manufacturers a why everything?
Host 1 10:07
Hell half the time? You can do it with the naked eye? Yeah, yeah. I mean, like, obviously, like in a production sense, you wouldn't do that. But if there was an issue, you could almost easily see it just right away.
Host 2 10:18
Whereas the BGA pretty much requires X ray to see anything that's really wrong with it. They make like kind of like under scope, under scope, like a 90 degree bend, like prisms that you can like,
Host 1 10:30
Click, it's like a reverse Periscope. Yeah, now not up.
Host 2 10:36
But you can only really see like maybe the first two rows of the ball if you're lucky, if you're lucky. So yeah, you really need to get really high quality on your VGA, as you pretty much need X rays. Inspection, which adds cost.
Host 1 10:57
So let's let's, let's take a let's take a side note real quick and talk about the issues that come up with what what kind of issues do you see with manufacturing and BGA is like, what is a problem with the BGA?
Host 2 11:12
The big one I see a lot is basically under package shorts. To ball shorting together, cue balls shorting together or like a solder ball, a rogue solder ball like is underneath there.
Host 1 11:25
Rogue like that. Yeah, yeah. Usually, according to IPC 610, or whatever it is. Rogue solder balls are, they're unacceptable, but they're also a process indicator that that something is an issue, like, one little piece of solder in the paste came off and then in the oven, it's just random location, and then it solidifies somewhere on the board.
Host 2 11:53
Usually, that's from the process indicator on that is your ramp is too high or too fast on your solder paste, and your basically your paste was so you paste in your solder paste, that's kind of basically the flux that's binding your solder paste together. Got seated so quickly that it decided to explode little solder balls off of it. Basically what happened?
Host 1 12:20
There's there's a lot of different categories that that fits into. I can't remember the name of it right now, I'm sure we have plenty of people who have taken the IPC class, but it's like solder extensions, or, or like, basically, if your solder extends outside of the designed pad, or the like, I guess, if you were to take an envelope and say my component fits within this envelope, if you have any solder extending beyond that, that's a failure. And those balls technically fit within that. Definition.
Host 2 12:52
Depends on what the balls end up, though. So
Host 1 12:55
Well, you know, most of the time, when I see those balls, it's like underneath and away to five components on the sides of it, they just kind of like stick to the side of the component there. And that that ends up being a big problem in like high voltage situations, because now you've effectively significantly reduced your creepage. Between pad to pad right? Almost half to it basically. Right, right. And probably less than half right?
Host 2 13:21
Here, we talked about that. Like, what last week, was it two weeks ago?
Host 1 13:24
I don't remember. It was recent. Yeah.
Host 2 13:27
That's actually that actually is a good thing. Think about with cooperage is, is solder balls in your manufacturing, because they do happen.
Host 1 13:35
You know, actually, that's another thing that's important. Okay, so if you were to remove the solder balls from the underside of a BGA, it just has a bunch of circular pads on it, right? If you were to measure the distance between the edge of a circular pad to the edge of a circular pad on the bottom of a BGA that's not your clearance distance. Because if you look at the side of a BGA ball, they kind of pillow out. So your creepage from ball to ball is not the same as pad to pad. So you have to take that into account on a BGA because the balls extend outside the width proof bulge, I guess. Yeah, in a way. How often does Netfabb do BGA it's probably pretty often right every day.
Host 2 14:25
Yeah, yeah. Yeah. I don't know how many of that is high voltage BGA work. I actually don't know about any high voltage PGAS
Host 1 14:33
I'm sure I mean MOSFETs that do fairly high voltage but probably not like extreme high voltage.
Host 2 14:42
Are there are they are there BGA style MOSFETs Oh, yeah. I haven't seen any. Doesn't mean we have building I just haven't seen any. Um, okay. So that's inspection, which is basically you have to use an x ray The Oh, we were talking about manufacturing issues. Complaint arity, like during reflow. I haven't seen, like our fab having this issue, but I've seen other fabs have this issue a lot. Where they basically have an uneven like zone, usually either during ramp up or during reflow. So it's like, basically the zones too short.
Host 1 15:24
You reflow ovens, it's not big enough, is based not wide enough. Yeah, it's not wide enough, or big enough is half the BGA is in liquid and half is not, you're just screwed. Yeah.
Host 2 15:37
Though, that could be also a process indicator on your belt is going to slow through the ramp phase, because basically one end of your BGA is liquidus. And the other end is not. But usually, that means you need a bigger oven.
Host 1 15:56
So I found I found with with ovens, okay, so regardless of the oven itself, there there is you can follow the curve of a solder reflow. Like if you go look up, solder paste, they usually give a general reflow curve. And in fact, a lot of component data sheets give reflow curves themselves, especially you actually BJs do a lot. Yeah, yeah. And most ovens, it's not hard to actually hit those curves in terms of the length of time, most of the time, the it's the zone temperature that matters, and less the actual runtime of the of the conveyor, like in fact, you could probably put a component or put a board in an oven time it with the oven off and get your general conveyor speed and then adjust times on the hour. Oh, for sure. Todd's temperatures on the zones, because it's somewhere between like five and a half in eight minutes in the oven. I've found with the we have a Heller 1703. At work, it's like six and a half minutes is perfect for us with our sector.
Host 2 17:06
Though it depends on the PCB, it depends on the basis the thermal mass of the of the assembly,
Host 1 17:15
You know, actually come to think about that right there is another thing to take into consideration. If your PCB design requires a your manufacturer to change their oven profile, you might that may incur some cost that because that may be some engineering time that they have to go and figure out what is good for your board. So if you're using a BGA, and that requires a bunch of extra layers on there, that ends up making a bunch of extra thermal class, you may have to pay an NRA for them to go and figure out the oven cost. It depends on the manufacturer. Some manufacturers just eat that cost and some would make you pay for it.
Host 2 17:59
Next thing might be differences. The tooling between Q FP and BGA. BJs typically have really tiny pace apertures. And so you might need well you don't really have to worry about it but your CM will have to worry about this might mean a special stencil in terms of just smaller aperture or thinner or it might need a expensive coating on it like nano slick or there's like tons of trade names for that stuff.
Host 1 18:29
And thanks for watching. Yeah, because the
Host 2 18:31
Smaller aperture will have more paste stiction. And so it needs like a special cut or a special coating that allows it to release correctly. Now, that actually might still be a thing on the QP depends on pitch with like we were talking about earlier. If it's a really really fine QP you still need like a high end stencil but if it's like a normal point five millimeter pitch Q FP maybe your CM has got a Pacesetter, and then you don't have to pay anything for a stencil.
Host 1 19:07
So pace genders are godmode it's like a cheat sheet. They're just cheat codes for PCBs.
Host 2 19:15
So got two of them at Mac fab now so I'll see you really Yes, we got the my 700 now is even
Host 1 19:23
Better 200 Better than the my 500
Host 2 19:27
Sure. The big thing about it is it doesn't weigh as much as the old one did
Host 1 19:32
Because the old one was like a giant granite block.
Host 2 19:36
Yeah, so So the my suit is a machine made by my chronic and the my they also older machines are with m y than a number. So the My 500 was like their first generation Pacesetter, or is it my 405? It's 500 was the first generation Okay, and we have a 700 now as well so we Have you run both of them in parallel, but the 500 was like a four foot by four foot by four foot block of granite that the whole machine was like chassid. Inside, it was really weird how they constructed it?
Host 1 20:13
Well, okay, so the head as it moves has to go so fast and nothing has to be so rigid, that they just decided to make it a giant block of stone to make sure it doesn't vibrate
Host 2 20:26
Doesn't vibrate, and it still moves. Like you can feel the head still move inside there.
Host 1 20:31
That's a lot of mass to be like cranking around. Yeah.
Host 2 20:35
And so in the 700 it actually looks a lot like the platform tray feeder that we had on the GSM like tower that never freakin worked at all. Was a turd? Yeah, so machine was so bad. It looks like one of those except painted in the micronic colors. And so it's just like sheet metal box. And I'm like, There's no ways. So I think what they did is they made the gantry system just way lighter, and better, you know, acceleration and jerk control. And they don't need that all that mass anymore.
Host 1 21:11
You know, that reminds me like way early on in the podcast, probably in the first 50 episodes, we talked to a robotics company that makes robotic arms. And and the biggest thing was from the mechanical sides of things from the electrical side, you don't have to like, you have to consider all of these things. But they end up just being numbers from a microcontroller, like velocity, acceleration, stoppage, these kinds of things. But, but stiffness really matters a whole lot in mechanical systems to say like, I want to go from position x to position y. And I want to do it this fast. And I want to know that I actually got there and I want to stop and be exactly there. I think the very first micronic was just like, the way we do that is we just make it heavy as hell.
Host 2 22:00
Really, really heavy. Which of course made all the motors bigger, which makes everything else heavier. Yeah, yeah, everything scales, right. Yeah, I do really like their pick and place machines. So we have a mic 203 100 Now, anyways, the Mag Drive stuff. Yeah, cuz they're they're maglev that had rides on a maglev. And it's a that's like the Y or X gantry. And then the other gantry is the PCB going like this rack PCB just moves in and out of the Yeah, machine. Kind of like a Prusa 3d printer, I guess. Yeah. Where the bed moves. And that's that's the other gantry. And then z is just you know, the head going up and down. Right. I really liked those. Those. I mean, cuz they're so smooth. That's the thing is like, you barely even hear those machines are on.
Host 1 22:55
Oh, it's just like air moving by? Yeah. Just a Whoosh. Whoosh. And they can they can move that head fast. Really?
Host 2 23:03
That? Yeah. That mag Mag Drive has a lot of torque in it. I was put that way. Yeah. So I was wondering why they didn't do that for the pace genders.
Host 1 23:14
It also has a ton of positional accuracy. That's that's the thing that's most impressive. Because like getting a thing to, you know, hammer some current into a mag drive and get it to move. That's not necessarily the hardest part is getting it to Madrona getting it to there fast and getting it they're accurate. They do all of it. Yeah, that's expensive. We have a we have a Samsung sm 438 or something. And it just runs on ball screws, and it's still impressive.
Host 2 23:44
So it's like the GSM separate works. Yes. Good. So why were dot dogging on the GSM GSM was the is a pick and place manufacturer I think it's like actually universal is the manufacturer. Yeah, that's it. Mac fab used to have a universal GSM like ages ago like
Host 1 24:07
The very first big one you guys got it was like five, six years
Host 2 24:11
Six years ago at this point, I think. Yeah. And that was our first big boy pick and place machines. And it was a turd. I don't know how many times we've rebuilt it's so many it ran overs to warp but that was actually not its problem. Surprisingly enough, it just it was just worn out. That machine was just put just worn out and just it just everything needed rebuilding on it every so often.
Host 1 24:41
You know I wasn't there. I think you were the only one in the shop when someone came to buy it. Didn't they drop it onto their trailer like by a foot?
Host 2 24:49
Oh man. So yeah, so we moved to our current location, and let's the old thing and the old look we left when we At that point, we had already pretty much decommission the universal g7. We had some Micronics that we were running. Like, basically, we just made an entirely new line at the old fab. We call it old fab, which is actually not the oldest fab but old fab. So we loved it, that old fab. And this person came to pick it up. And they I think they're trucking it like they talked to, like halfway across the nation, but they came in a car. They had a car trailer. That was open car trailer. Okay, so like, it wasn't enclosed or anything like that. And yeah, like, they he tried to move it on like pallet movers and stuff, and just I think he dropped it like six inches a couple times, like, getting it to the trailer. And because I wasn't going to help him. I felt kind of douchey I'm like, Dude, I'm not helping you move this thing.
Host 1 25:58
Sounds like a you problem.
Host 2 26:01
And because it took a took, like a ginormous forklift to move it there and I'm like, Dude, if you don't bring a big forklift, I'm not going to help you move this thing. But he just had pallet jacks. pallet jacks in like, yeah, in shoving it wheeling in across the floor that floor like 100 feet to the loading dock. And and then he like winched it onto the onto the
Host 1 26:26
Trailer come along there. No way. Yeah, exactly. Hey, I guess you do what you got to do right?
Host 2 26:34
You gotta do what you got to do. But I'm like, how much on I'm reading chat record. I broke a hole in the trailer bed and we had a patch it when he got back. Oh, did you get that mercury one. GSM.
Host 1 26:52
We found the owner.
Host 2 26:54
Amazing. Yeah. Did you ever get that machine running? Right?
Host 1 27:01
I You see, I didn't even know like I just remember Parker come back to the office and be like it's gone. Never Sorry. Yeah. Apparently it didn't. It didn't end up working. Yeah,
Host 2 27:14
It was. We tried. We scrapped it from it. I feel so bad. Now. I guess. I'll put it where I put it all like Steve and I tried to make that machine run it. Actually, we didn't make a lot of boards with it. I think it did work. A few 1000. More NAD actually, I think we built like 10,000 Plus boards. So that machine, it just wasn't set up for what we wanted to do, which was high mix low volume. It was a it was a high mix high volume machine. Like you could put so many feeders on that machine. Yeah. It was like 64 feeders per side. And it was a dual gantry.
Host 1 27:53
Yeah, it was slow, though.
Host 2 27:56
Slow is relative. And we did 10,000 picks
Host 1 28:00
An hour. I also remember it putting diodes backwards on 700 boards.
Host 2 28:08
Because the person who programmed it programmed.
Host 1 28:11
They prepared that wrong. They knew they programmed it wrong. And they were like well, we'll just go forward with it. Yeah, that was the thing. That was a problem.
Host 2 28:20
Oh, we'll just go forward with it. And just like why
Host 1 28:23
It fix it in post. Oh, the good old days.
Host 2 28:29
Oh, man. Well, Mercury Inc. One. i It's unfortunate that it went to the scrap or went out for the GSM. Yeah. There's there was a lot of late nights of me dialing in the calibration of that machine. Could our actual calibration tools. It was pick up an O to A one and put it onto this dot and see how close you get
Host 1 28:56
Good old days. How did we get on to the we're talking about tooling we were talking about.
Host 2 29:04
Okay, so it's never gotten to the patient agenda. And we talked about machines.
Host 1 29:07
Host 2 29:11
So we're gonna we're gonna keep going down this rabbit hole. So before that, and I think was before Stephen worked at the fab. We had a Adele pick and place. Yeah, that was right before I showed up. And that was a aluminum extrusion pick and place that could barely do 300 picks an hour, which is like the speed a human can do. Now the great thing is, it's a machine so it doesn't complain. It doesn't need lunch breaks. So it can continuously do 300 Yeah, it can do 12 hours a day. 300 picks an hour. Actually had optical all that good stuff. We built the first. We built two products on it. We built the reload pros which was our first box build product? And we built like 300 or 400 pin hex pinball controllers on it like rev four I think of Penix old school. Yeah, really old school. And then and that was our reflow oven was that little three zone gold flow. So we were talking about reflow ovens. You could not reflow a BGA through this one because it was all a foot and a half long. Oh, yeah. All the zones fit in a foot. Yeah, all the zones fit in the foot and half. It did it did cure p because that board had i t q if p 100. On it, which was the pic 32 on it. It did that. Okay. You ever they could tell it was not refloated all at once, though.
Host 1 30:51
Yeah, it was it was having problems. Having to have Okay, talking about conveyor speed. If you're if your board needs to go through a foot and a half in six minutes, that conveyor speed is really slow. But if you talk about a big Heller, like I've got that's like 10 to 12 feet long, and it's got to go through six minutes as a hell of a lot faster. I remember I remember using that machine more for desoldering components where you'd put you put it in, you'd let it get into the liquid zone, lift the lid really quick and they'd go in with tweezers and snipe off the bar. Yeah, I showed you that trick. Oh my god. I was like, This guy's insane. But it works. It works.
Host 2 31:35
Yeah. So that was after we got the Bravo, which was our second reflow oven. Electro vert, electro vert, bravo, which was a, quote, lead free, unquote, machine.
Host 1 31:48
Yeah, only if you run it at like 120%.
Host 2 31:52
And so once we got the Bravo the gold flow went into hibernation, it just went into the corner. And I started using it to de solder s&t parts, because it was actually like, oh, you could just pop a board in there. And then when it got into the reflow zone, lift the lid, pull apart off good to go. And actually made it really good when you had to do like a lot of rework. Yeah. And yeah,
Host 1 32:16
I think steaming for multi legged components. It's great.
Host 2 32:20
Yeah, it worked really well for that. It's like using a a solder pot to D solder through hole parts. Well, you put the you would put the the board over the solder pot, and then it would indeed be enough jet I guess, to to liquefy all the legs at once and you can pull the part off. Saying same concept. This the safety
Host 1 32:45
Aspect might be a little lacking
Host 2 32:48
A little bit. Yeah. It's probably why they don't let me in the shop anymore.
Host 1 32:54
They know he's gonna do some crazy shit.
Host 2 33:00
All right, anything more about BGA verse Q, FP, we need to talk about, I think I think the
Host 1 33:06
Consensus in general, of course, all of this is in general, is that you're likely to save some money on your assembly and your raw PCB cost if you go q if P BGA. doesn't guarantee higher cost, but it trends towards higher cost. Just keep that in mind.
Host 2 33:30
I view it as even if the qf B cost more, you're going to save it on the tail end. Because one thing we've been talking about here is also rework BGA rework, you basically have to start over like either and reball in the BGA, or just scrapping the part and then getting a new BGA. Whereas the QP you can reuse a couple times before you kind of reach how many times you can reflow that part. But that would put this way it's like, if you didn't really work and you scrapped one BGA Well, there goes all your cost savings for that part. So,
Host 1 34:11
You know, at the same time, if you're developing a board with a BGA, put test points, you'll thank yourself like any any signal you have, make sure you can access it somehow.
Host 2 34:27
So I would say use a QP This is my role, basically use a queue of p if you have space on your board, only go leaderless when you don't if you one, you don't have any choice because the part only comes in the lists or you have space constraints. Or you know, I
Host 1 34:46
Would make that same argument if it was a que fn versus a qf p if you can get away with the q if P do it.
Host 2 34:52
Yeah, I'll for sure always that we actually switched. We only have one QFN style On The Pinball Controller, we switch it to a lead ID. So I see part just so that we can get rid of all lead lists on our board.
Host 1 35:09
Yeah, makes life way easier.
Host 2 35:13
Okay, I think that's it for BGA verse qf P, we're actually going to really just say that was like elitist versus non needed. Surface Mount component talk to
Host 1 35:25
I guess so. Yeah. I mean, we did specifically talk about BJs. But yeah, that does extend the kind of those two.
Host 2 35:31
Yeah. So what came out of this discussion on Slack was how to use the Mac FET platform to predict pricing.
Host 1 35:42
Right, because like, even even with everything Parker and I just said, it still might make sense financially to go with the more expensive route, depending on the part cost.
Host 2 35:54
And Chris Carter did kind of like a quick write up on the Slack channel of like, how to do this with the macro platform, like quickly. And this actually is probably something you should do for almost any kind of board, you're designing these for production or, or it's a product that you're developing with, because this is the thing, if it does X, Y, and Z, what it needs, what the product needs to do, what's the next post important thing per unit cost. And the earlier you can get that per unit cost to your your, the product designer, or, or your boss at work or whatever the the better, they're going to be at planning everything else.
Host 1 36:32
As soon as the bean counters Stop bugging you, you're doing it right?
Host 2 36:36
Yes. And so how to get how to get the bean counters off your back the fastest. So, most time you just do like napkin math, because you don't know everything that your PCB might need at this point, like capacitors and resistors, and passives. And that kind of stuff. But you know, like, Oh, I'm probably going to use this microcontroller, I need like eight MOSFETs, it's got, it's got these couple of special LEDs maybe, and it's got a analog to digital converter, front end with some icy protection. So you can build up a rough bill of materials just like that yours might be a couple dollars higher, a couple dollars low, but it's gonna be roughly where you need it to be. It's a good shotgun shot at the target. Yeah, exactly. And then you got to draw up a board outline that you need, what's the maximum that your board can be, and usually actually do have a good idea of what size that is, because a lot of times you're designing for an enclosure, or or you think your order, like the end product needs to only be this big. And so your PCB is dictated by that size. So you can just draw a rectangle that's that dimension, and just throw all your parts in there. And then you can throw it up onto the macro platform, and it will quote it out for you. And this is where you can go in and go, Oh, well, I want to see the difference between that qf P and BGA. And all you got to do is just change what that Bill material item is on on the bill material page. And it will just you don't even have to re upload a new no bill of material. Yeah, you just got to search for the new part, select it, and then see what the price differences and so show you the labor and the price difference between the components. And then for all the quantities one, want to make a million of them? Sure.
Host 1 38:28
So kind of to boil it down, what you're doing is you're just adding a dummy board that has an outline, you're throwing up a bill of materials that isn't necessarily 100% The end product, but it has all the major points on it. And then the the macro fab system will give you a ballpark price of your board is this big your bill of materials has these parts on it. And it'll give you ideas based on those the characteristics of the parts and the port.
Host 2 38:58
Correct. And then you can also like just change like the layer counts or like, oh, the BGA is going to require a six layer board first, the QP, which is four layer, and you can just do that change there to
Host 1 39:08
You know, okay, here's the thing, I think that's actually a tool that a lot of people don't know about. And I'm not just trying to sell macro fab here. That is a pretty good way. Like, let's say you have an idea for this new whiz bang product. And you need to go and create your initial findings document to provide to the accountants or whoever, your boss or whatever. This is a really great, great way to very quickly get an idea of the assembly cost the bill of materials, like everything, you can get your initial findings and you can be 10 steps ahead by saying like, Look, I've already have like, generally accurate pricing on this. So I can tell you, before we've even put pen to paper on a design, this is how much I think it's going to cost. And like you said there, you can change the characteristics about it and say Like, Oh, let's say we were running a four layer board, and we find out that we need to go BGA. And that needs to go from four to eight layers, you could already have that information in your presentation. So like, we know that in general, it's going to be a x percent increase.
Host 2 40:15
Yep. I did write an article about this a long time ago. I did do a quick Google search. I couldn't find it. So it's somewhere out there on the Mac fab blog as ought to find it. But it's about how to do this in Eagle. So you could probably expand that to other EDA tools. But yeah, I mean, I use it all the time, basically. Like, when I did the Pinball Controller, I based on like, okay, the board needs is going to be like, 10 by five, it was my goal. I ended up with 11 by five. That's I've cheated a little bit. But yeah, it's, uh, I just came up with the iPhone 10 by five as threw all the parts in the middle of the board, damn, and route anything and just uploaded that and is able to get I think it was within $4. And my estimate,
Host 1 41:05
Well, in terms of a percent, how much was that? It was probably pretty close. Right?
Host 2 41:10
That would be 3%.
Host 1 41:14
Okay. Yeah. So if you think about how awesome that would be, if before, if you're giving a presentation on new product development to your boss, and you can be within 3% of the target price. Like that's incredible, right? Yeah.
Host 2 41:29
I designed it. Yeah, right before
Host 1 41:31
You like before you've even opened a new schematic. So here's the thing I can tell you, from my experience, there's two major factors in in PCB assembly that are going to sort of adjust the price on your board, that is square area of your board, and then your pin count. And that would be you know, how, like, what's the total number of SMD pins. So like, you know, a resistor is going to have two and SLIC. Eight is going to have eight at tally those all up, the more pins you have, the more it's going to cost to assemble. And the more square area, the more the board's going to cost. So those are sort of like your, if you're thinking of like the very top level things, of what effect costs on a PCB, those are those are some of the big ones. Anywhere, not just Mac Feb. I know some places they they will go all the way down to adding assembly cost per component. I think that that ends up being a little onerous in a way.
Host 2 42:41
Yeah, it depends on basically how good their processes for pricing. So like if they're doing if you're if you're doing a BGA and they're not accustomed to doing BJs, they will usually have a price add or like that on their, on their tooling or something like that. Or what if you fit within their that CPMs Goldilocks zone of what they build, they can probably have a more formula driven way of pricing that out.
Host 1 43:15
You know, okay, so one of the big things where I think macro fab kind of knocks it out of the park with this isn't that you get an instantaneous price, because a lot of places will give you an instantaneous price. It's the fact that macro fab allows you to choose your quantity. And you can see that pricing for nearly any quantity. I mean, if you if you went online to macro fab, and you uploaded a board and you're like, I'm gonna make 15 million of these, you know, I don't I personally don't know, but I doubt that the price would be exactly what macro fab would want. You'd probably have to talk to somebody if you're talking about quantity. Quantity, they're like, You need to talk to us. Yeah, please call us right. But if you're talking about like, you want to look at the difference between five 105 100 Those numbers are going to be accurate.
Host 2 44:06
I actually I think it goes up to a couple 1000 before it starts to be like
Host 1 44:14
You're edge casing the algorithm if Yeah, hidden say about
Host 2 44:19
Pricing. Ambiguous is my that's the wrong word. But But yeah, because it because at that volume, a penny matters or like an extra step in the process matters. Actually,
Host 1 44:35
At that point, talking to someone you're probably going to get a better price than what the website tells you.
Host 2 44:41
So having run my own stuff, kind of like as a don't run them as like email@example.com and platform I run it as like my own stuff. It says like secret shopper, I guess in quotes Yeah, it When you get to a certain volume, contact for sure. Yeah, yeah, that tells you that's thing. The platform will say, Oh, you thought you need to contact us. Right? Well, man, I wish I could find that article. I'll find it later.
Host 1 45:18
It's it's interesting because the, the quest, I think the way that you had described in our Slack channel of, you could just upload a dummy board and change the bill of materials. I don't think that's necessarily exactly obvious from the get go is not actually I was that is such a useful tool. Yeah, I
Host 2 45:40
Actually was talking to church today about this. And I'm like, man, if we just had a front end calculator, I think it would help people out a lot. Because technically, that's what you're building for yourself is a calculator with a dummy board.
Host 1 45:57
You know, this is this is way far fetched. I really, really, really wish that I could have a calculator like macro fab, built into my EDA tool, such that I make decisions in my EDA tool, and it's like, Oh, your price just went up because you made that decision.
Host 2 46:14
You chose that BGA
Host 1 46:17
Oh, you want a one mil via well, your boards are gonna, you know, go up 100x and cost.
Host 2 46:24
Is that even possible? No, it's not. That's one. Yeah.
Host 1 46:27
Yeah, I think I think three mil vias are laser drills, right. And there's some of the smallest ones,
Host 2 46:34
Actually, I so I asked some PCB manufacturers about that, like what they consider laser drills. And most, I'd say most most PC, the ones that I've talked to, majority of them have will machine that still a three mil hole. Yeah, a three mil drill. Well, four mil, I think is the smallest side was talking to him to Okay, four mil, they will still have, some of them will laser them. But most of them nowadays have drills, their drum machines are accurate enough to hit that with a four mil drill. Oh, and it's, it's cheaper. So
Host 1 47:17
You know. So the slight side note, the CNC machine that I use at work, we actually just got a second one in today. I'm super stoked. So we've we've got I've got a killer machine shop at work. It's it's kind of a fun playground. But I learned not that long ago that this machine has become a standard in two industries. The PCB industry for drilling PCBs, mainly because it's so fast, it has a 60,000 RPM head, which helps out a lot when doing micro drilling. Because you can plunge faster, but it's also really used in the dental industry. For machining, dental implants.
Host 2 48:02
I'm just actually imagining your you're going to mount your head in a vise tomorrow and get some fillings
Host 1 48:08
Drilled out. I think they have like special vices for like dental implant vices and trunnions and stuff like that. They actually have this company, they they make our version and like, I don't know, 600 of their CNC versions, but they they have a specific one that's like a dentist CNC. So I don't know the industry is big enough, I guess. But But ya know, I've seen a bunch of PCB manufacturers use this CNC, mainly just because it has the spindle speed is so unbelievably fast. And it has enough tool space that it can carry a bunch of different sizes of bits, and it can drill really small holes with that.
Host 2 48:51
And you just use it to engrave aluminum all day every day. It does look great, though, so I'm not gonna knock it. Okay.
Host 1 48:59
So I've been asked so many times, why don't you laser because lasering is so much faster than engraving, right? You could just get, you know, the black laser coated material and then just blaster on. A handful of reasons are behind that. One. You still have to machine a lasered item like yeah, so gotta cut all the holes out for don't have to cut all the holes, right? People People are always like, why don't you just laser the holes? Well, okay, that's a completely different laser. You can't laser and laser mark on the same thing. Easily. You can do it, you can't do it easily and Okay, so a laser machine that makes something look good is different than a laser machine that cuts holes like us. Those are two different things and they're also two very different sizes of machine. So we opted to go with a unique system that allows us to both engrave and cut overhead holes in one process. So we put a sheet of aluminum down, we press go, we leave for the night, come back in the morning, and there's a whole army of parts done. And we never have to worry about alignment of anything
Host 2 50:15
That was about to say is if you have to switch tools, which like Xavier, let's say you you had to use two lasers or let's say a laser in a CNC. Yeah, to cut the holes and then do your engraving. How often have you got the scrap? Because one's not aligned correctly?
Host 1 50:34
Yeah, I mean, the thing about our CNC, there's never been, okay. If there's an alignment issue, it's because we designed it wrong. Yes, like you have to build misalignment into our stuff. So the only time we have ever scrapped material, which it happens, but the only time is when bits dough, that's it, and then you just go okay, well scrap those put a new bid in and press go. So I'm kind of stoked because yeah, we got our second our production has grown enough that like our the one we already have, is just not keeping up with production. So we got another machine. That's good. Yeah, so excellent problem to have good man. Okay, so I had to deal with some riggers this morning because we don't have a darkhei at at our
Host 2 51:17
They drop it six inches over and over and over again. No, these
Host 1 51:21
Guys were magical. Okay, I've used these guys three times. Now. This this rigging company is so awesome. Okay, so first of all, what's great is they allowed us to ship our machine to their location, and they received the machine, they put it on a truck, they drive it to our location with a forklift and offload it. Now, of course, you have to pay a premium for all of this. But if you don't have a darkhei to offload stuff, like that's what it cost. I watched these guys spin a 12 foot forklift in a six foot room. I don't know how they do it. These guys are so good. Like, okay, so we have like in our warehouse, we have other machines and we have tables and we have all this other stuff. I spent all day yesterday cleaning up our warehouse and like moving things around. But it was still like Man, if they have a big forklift, this is gonna be hard to get around. They can dance with a forklift. These guys are amazing. Like, and they've got a they've got a 2500 pound CNC on the end of this forklift and they're just swinging around like waltzing with this thing. Like, it was impressive. I have to say that I will use these guys in the future when I buy more machines for sure. Yeah.
Host 2 52:31
I just remember. Reminds me when we were unloading the Micronics. Yeah, or our my 500. And I bet you want to know from a tow truck. So we're not gonna get the last topic today. Because we're gonna have this story. We told this story like four times. I love we. Oh, yeah, totally gang, because it's still good. It's still good. So we we ordered on my chronic 200 And my chronic 500 The PACE cheddar from I think it came from Sweden, right. Yeah. So they show up at Downtown Houston. on a on a on a flatbed. Like semi truck. It was like an air ride truck. Yeah. But open. Like they're in crates. So it's like these two ginormous crates on the back of this semi truck. And the guy's like, Okay, how are you going to get them off? And we're like, no one thought no farther ahead than him showing up on the side of the road. Yeah. And so fortunate enough. One of our our neighbors is a it's a brewery. And they've got forklifts. And so we got one of those guys to come over with, like their biggest forklift. And we're actually able to pluck them off the trailer, the semi tractor trailer just fine. And so that way, you know, signed it, signed for it and driver drove away. And now we got to ginormous crates on the side of the road in downtown Houston. And we're like, okay, we need to get these. They can't stay outside overnight. The drivers like later, someone will try to steal them. Downtown Houston. So we got to figure out how to get these inside. And it's like, three o'clock at this point. And so we tried to like oh, just use the forklift. Well, the problem is lifting them up. They were not fit. The crates were so wide, they would not go through the dock high section.
Host 1 54:33
Also, also there was a ramp so you couldn't just pick it up on a forklift and drive it it was
Host 2 54:39
I was thing is like, well, we can just drive it up the ramp, you can't drive it up the ramp because it's so big, that if you lift it up to try to go through the ramp, the forks would hit the ramp as you were driving up, and it's like okay, how do we get so basically what we ended up doing is we got a flatbed tow truck to show up, loaded the crates onto that had the tow truck driver back into the shop up the ramp and then partially unload the ramp. So it was
Host 1 55:08
The tip the ramp, cars at like a what a 20 degree angle on a ramp and then flatten out the toach.
Host 2 55:19
Bed, flatten that thing out about roughly level. And then we were able to pick it off with the with the forklift and then yep was
Host 1 55:29
The forklift had to raise to like three quarters maximum height Yeah, with a $200,000 machine. Like more than enough that you could easily walk underneath the forks with this, like way higher than you should lift a machine like this.
Host 2 55:45
Yeah, exactly. And of course, it's like wobbling a bit. And we're like, well, the future of McWrap is right here.
Host 1 55:52
Everyone's just pinching real hard.
Host 2 55:54
Yeah, I think that was the last time we ever did a sketchy move. Because after that we we just hired riggers.
Host 1 56:01
Well, I when we bought the new oven at at the new shop, I rented a forklift. But we had a guy who actually knew how to drive a forklift. Yeah,
Host 2 56:11
You rented a forklift. And but you also rented a forklift that was big enough to easily carry
Host 1 56:16
Yeah, right. Like it got the real thing. Yeah. Which that was fun because we were supposed to have the forklift for a day and a half and they left it at the shop for like a week and a half. Yeah, so we all get to joy right in the back.
Host 2 56:28
So I think I think we talked about this story before too. But we were doing donuts in the back parking lot when it was raining for that big not forklift was huge. It was the forklift, I think you I think you basically got the biggest forklift they would let us rent to move that oven. Because that oven is big to
Host 1 56:49
The biggest thing was was that like okay, to get the long enough for
Host 2 56:52
Hopefully no one ensures us is listening to this podcast right now. No. Well, I
Host 1 56:57
Mean, this was this was a while ago as well. Statue limitations in order to get long enough forks to reach the entire crate length you had to get. So the the oven weighed something in the in the region of like 2500 to 3000 pounds. But I had to get like a 6000 pound forklift, just to get long enough forks on it. So it was like overkill. It was a beefy, forklift.
Host 2 57:22
Walkie Yeah, cuz, man we used when we got the Micronics off that tow truck, we had to use four extensions. And those are those are first of all, like, you talked about getting insurance being pissed off and you can't use No. So I think I think basically what we took learning from that is what we applied to when we got the oven, because the oven went piece of cake, no problem no matter what. So
Host 1 57:46
You know, okay, so you'd be surprised. Renting a forklift is not expensive. No, it's not on top of that. Like, we've never done anything dangerous with them. But they don't ask a lot of questions. It's not like renting a car. Like when you rent a car, like there's a lot more questions they ask than a forklift. Basically, with a forklift place. You call them up and be like, hey, I want a forklift. They're like, cool. It's this much money.
Host 2 58:10
Where do we need to drop it off at? Yeah, actually, all heavy equipments. Like because I had to rent a small like a bobcat, which was like a little little front end loader thing. front end loader. Yeah. It was like, I called him up and like, what date? And I'm like, here's my credit card number. And that was it showed up. I'm like, it's, I don't know. It's so weird. It doesn't seem like it should be that easy, right? No, and you're right, because like renting a car is like, they need everything in your like social security number and all this other stuff. Just to like rent a car that you would like you drive every day. Kind of weird.
Host 1 58:51
Buddy of mine visited Denver, and apparently renting cars is so unbelievably expensive right now, I haven't tried to do it. He rented a bobcat and drove. He rented a truck from Home Depot. And that was for like a week and just use that and it was cheaper. It's like what? $19 a day? Yeah.
Host 2 59:14
That's pretty inexpensive.
Host 1 59:15
It's really an expense. Yeah, yeah. Apparently you like if you run a U haul on a vacation, like go to the location, like fly out there run a U haul that's cheaper than running a car.
Host 2 59:27
I would. I couldn't believe that.
Host 1 59:32
No, another nostalgic cast.
Host 2 59:34
Yeah. So that was the backdrop engineering podcast. We host Parker
Host 1 59:39
Dolman and Steven Gregg. Let everyone take it easy.
Host 2 59:43
I hope we didn't. Our long term listeners enjoyed us regaling those stories we've told many of many of times,
Host 1 59:50
We have to indoctrinate all the new listeners. Oh, that's true. We need
Host 2 59:54
More stories. Well, me without the insurance people knowing
Transcribed by https://otter.ai