EP#440: HMLV - High Mix Low Volume - Circuit Break - A MacroFab Podcast

Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Parker Dillmann (00:10):
Welcome to circuit break from MacroFab, a
weekly podcast about all thingsengineering, DIY projects,
manufacturing, industry news andHMLV. We're your hosts,
electrical engineers ParkerDillmann. And Stephen Kraig.
This is episode 440. So HMLVStephen.

Stephen Kraig (00:32):
What does that mean

to you Parker? Well, I would actually really
wish I came up with a reallyfunny acronym for it, But it
actually just means high mix lowvolume PCB manufacturing.

Which is so thrilling and exhilarating.

Yeah. Thrilling and exhilarating.

Actually, it is totally because that's what I do
on a daily basis. So I actuallyreally like it.

Yeah. And that's that's what Maccab
started as as well. Rememberback in the day?

Absolutely. Well, Maccab started as it, but also
you that's still your bread andbutter in a lot of ways.

Yeah. That's that's still what we do a lot
at, our HQ facility for sure. Sowhat high mix low volume is is
basically instead of, it's it'sbasically the exact opposite of,
like, what Apple does or whatFoxconn does, where you're
setting up a line that willbuild one thing.

And is really, really good at doing that one
thing.

Yeah. Very good at that one thing where, the
main thing with high mix lowvolume is mostly being able to
handle really high mix ofcomponents. That's what the high
mix part is. I guess you canalso say, like, high mix of
products, but I always viewed asjust like the component level is
handling lots of different SKUsof components, a lot of

(01:55):
different line items. And thenlow volume, meaning you're
building you know what?
What is low volume to you? Andyou

know what's funny? I was literally about to
ask you that same questionbecause low volume to me has
kind of shifted a little bit,because back in Macrofab, some
of the higher volume runs when Iwas working there were in the
higher 100, like 7, 8, 900 kindof thing. Whenever I was working
at WMD, my my previous job, highvolume for us represented, like,

(02:29):
4 to 600 and high volume whereI'm at now is, like, above 10.
So so it kinda shifts. It itjust depends on on what what
your requirements are.
But in general, I would I wouldprobably say, like, across the
board, let's say, 500 and aboveis high volume, and then 100 to

(02:53):
500 is medium volume andanything under a 100 is low
volume.

Yeah. High volume or low volume to me is I
guess I look at more like on thecomponent level and if you have
to use more than a reel of acertain component that starts
getting to I consider that pastlow volume status.

Okay. Yeah. I could I could see that.

So if you have like a board that has 10,000
resistors on it, and you needyou're building 2 of them.
That's and that that gets kindof weird, but that's how I start
viewing it as so I I most timethat cutoff though is funny
enough around 750 units. That'swhen that starts to play into
you now need to have multipletrays or multiple reels of a

(03:43):
component.

Okay. That's an interesting way of putting it
because you're basing it off ofcomponents and not units, not
finished goods. So, actually,one one other way of saying that
is if the kit of parts has toever be refilled, then it leaves
low volume for you.

And that's for me personally. Yeah. Okay. Okay.
Because I don't, like, theMacrofab platform itself doesn't
have this concept of high volumelow volume.
It used to back in the day,like, I'm back in 2017 era. It
had, like, a low volume buttonand a high volume button in the

(04:24):
platform.

Well and and you guys are way more of a spectrum
when it comes to ordering, yourquantities. And and if correct
me if I'm wrong, but theplatform will recommend
different manufacturing optionsbased off of your where you fit
in companies.

Yeah. So act actually, Donald, since we're
bringing up the platform, it'swhen so the platform has this
really cool graph where, it willshow you, like, your quantity
and the price breaks over, like,every unit you add on, it will
just calculate the price out foryou and just draw draw you a

(05:04):
graph. It's when it's when youknow what? I should come up
with, like, an actual, like,number for it because when the
rate of change on the graphlevels out, it's like when the
hockey stick levels out

Plateaus.

Yeah, plateaus, but still it always slightly
goes down usually. So there's arate of change, there's a
derivative of that graph when ithits under that number sure
that's high volume

Yeah. But but but it is there's never, like, a
exact defined point where itjust levels out. It it it it
just it tails off.

It tails off. It takes

a long time to get there. In other words,
there's not like this really,really sharp point where you go,
that's the best manufacturingpoint. That's my quantity right
there.

Yeah. Yeah. That is true because it's the
more you build all what youalways get price breaks or it
always because basically whathappens is you end up ordering
when you get big enough on youend up ordering directly from
the manufacturer, so you don'tgo through the distributor and
you get a price break there, butyou have to buy like boxes and
boxes of stuff from them.

You you know, actually, that that that point
we're talking about where itmight make a more sharp
transition, that's when youreally start to ask the
question, do I just buy my ownmanufacturing facility to build
my one thing? My volume is highenough that I could it might
make sense to hire a whole team,make an entire assembly line and

(06:29):
build my own thing.

Oh, that's way out there.

Yeah. That's like yeah. That's like you don't see
that part on the ground.

I mean, even even Apple doesn't even do that
anymore.

Yeah. Yeah. Yeah. Nintendo doesn't do that. Yeah.

They just go to Foxconn or whatever. Right. I
the only time you really do thatis when you have really
specialized stuff. Right.

Or or if you just have to have control in house.

Yeah. Yeah. When when increasing so I guess
I'll rephrase that high thebreak between low volume and
high volume for me is whenincreasing volume of units
doesn't significantly change theprice per unit anymore.

Mhmm. K.

So there's that there's that point where, like,
the the the graph kicks flat, orflattish, and that's where that
point is where I would considerlow volume, high volume,
because, and price and quantity.And funny enough, that is around
600 to, like, 800 units for mostboards. It's true. In that area,

which which falls in line with what I was saying
earlier when when I was workingat the fab. Those were kind of
the the bigger runs. I mean,there were runs even bigger than
that, but but those were, like,the bulk of what we would call
high volume.

Yeah. It's really funny now, where we have,
like, actual high volume stuffnow, and it's, like, you know,
people ordering boards and thengetting board shipped weekly for
years.

Yeah. That actually I would I would not
even call that high volumeanymore. I would just call that
continuous manufacturing.

It is continuous manufacturing like
Yeah. Oh, is this true? So along time ago, I was reading up
on different brewery brewerystrat. I know what time it's
High Vex low volume. I know whattime with that, but but
breweries, I think so this is astory I heard about Budweiser,

(08:34):
and they have a continuousstream of beer in their plants
for Budweiser beer.
Maybe it's for Bud Light. We're,like, the flow doesn't stop in
batches, like it does for mostbreweries.

Right. Well, correct. There is a Everything
is batch when it comes tobrewing. However, there I'm
trying to look it up right now.There there's a gentleman a long
time ago that patented a processof continuous brewing where the
output of fluid is met byincoming fluid and the yeast

(09:17):
continuously produces alcoholicbeverage and you just maintain
those input outputs and it justnever stops fermenting.
And so it's not a batch, It isjust a continuous process. I
don't I'm I'm failing to find itright now. Okay.

So this I didn't make that up.

You didn't make that up. I don't think the major
players use utilize thatprocess.

Okay. So I I was wrong with that then. Which
is fine. I'm always wrong.

Sorry. I'm not trying to just tell you you're
wrong.

So alright. Back to high mix low volume. So,
yeah, that's that's what lowvolume to me is in that that
range. And but what does thatmean for, like, for, like, you
as an engineer then? Like, whenwhen if you're out there looking

(10:10):
to go buy, boards, like, getyour PCBs assembled, Do people
actually search for high mix lowvolume?
Is that something peopleactually search for?

I don't think I've ever specifically searched
for that. Although, I I willtell you I have searched for
contract manufacturers and justwith a few glances at a website,
I can tell that thatmanufacturer is not right for
me. Because depending on howtheir their website is set up,
it it will tell you a bit aboutthat that Centimeters and maybe

(10:46):
that's this sounds a little bitunfair to just immediately write
something off. But I don't I'mnot gonna go to Foxconn and say,
can you make 2 boards for me?And if you go to the Foxconn
website or or or or biggerwebsites like that, you just
know this is large high volumemanufacturing.
And if I need a small run ofcheap boards, they're just gonna
tell me to pound sand.

But you've never been to Foxconn's website.

Yeah. There's there's there's certain
manufacturers that you couldjust tell that that's the case,
and, they're looking for thereally big orders.

So if you go to foxcon.com, It actually doesn't
say what they actually do on thefront page. The first thing you
see is technology for smartliving and then creation of a
better future together and thenthe next tab is ESG sustainable

(11:42):
developments. So I don't sayanything about actually building
boards and products throughtoday.

So does that scream, yeah, we'll make 2
boards for you, for $200. Notreally. Right?

They have something called 3 +3 equals
infinity. What is that?

If they're doing that kind of math, then, I can't
afford them.

Apparently, that's an event that they have.
3 +3equalsinfinityevent. Oh,well. Not talking about Foxconn.
But you're right.
You're right.

But that's that's an example of, but but what what
you were saying was actuallykind of the opposite. Like do
you go find a Centimeters that'slike we do the the low volume
stuff. And you know what'sfunny? The only one that I'm
really aware of is MacroFab,that that says, like, yeah,
like, we welcome this. Don't getme wrong, I've talked to plenty

(12:38):
of CMs who are like, yeah,absolutely we'll do that, but
but MacroFab screams it a lotmore or at least being able to
support it, I should say.
Because it's not necessarilyMacroFab's goal to just do the
the low the the, high mix, lowvolume stuff.

Yeah. That's that's what I'm just getting at
is, I talked about this before Iswitched to marketing full time
this year, at the beginning ofthe year and that that's one
thing I'm like trying to figureout. I'm like, what are
engineers actually searching forto go find things? Like, find
cms? Like, how do I market macrofab and find out, like, you

(13:22):
know, if someone's searching toget their their 50 let's say 500
units made of something, likehow do they even go about
finding a Centimeters for that?
And, I don't know. Are are theysearching HMLV? Is that, like,
that's not a that's not even aterm that is taught at all in
college. Right. So I don't isthat even a term people have

(13:43):
been searching for?
Or is it something more of likean educational topic on where,
like, you land on a site and youjust happen to see high mix, low
volume being on the sites?

Yeah. I I I think it it comes up in conversation
when you reach out to theCentimeters and you say, what
are your capabilities? Here'swhat I'm expecting. And they
just come back and be, like,yeah, that fits really well
within our process or no, that'sterrible. Sorry.
Go away.

I guess so. Yeah. I'm just trying to think
of how do I, would market a highmix low volume terminology.
Mhmm. I guess putting it maybeit doesn't, like, you don't
advertise HMLV because no one'ssearching for that.
Right? So it'd be really cheapit would be really cheap to
keyword that because no one'sactually searching for it.

But it wouldn't get you very far.

Yeah. But having it where getting across
to the person who lands on yourwebsite that you welcome high
mix low volume without actuallysaying high mix low volume
because the if you just sayHMLV, I bet you actually most
people won't even know what thatis if the engineers landing on

(14:56):
the site will see what thatmeans. So if you say we do high
mix, low volume as part of yourmarketing spiel to get someone
to click in, that might be abetter use of it.

Yeah. Actually, I think I think one way that might
work is showing or ordemonstrating somehow how you
fit in the design cycle for anengineering team. So in other
words, at the beginning of aproject, there's there's a lot
of churn, there's a lot of, newideas, there's a lot of design

(15:31):
work happening, there's mistakesbeing made, That's your high mix
low volume portion when you'redoing your first prototypes.
Like, nobody's expecting thatfirst prototype to be the the
final thing that you build. Soyou're building low quantities,
you're building a handful ofprototypes, you might go through
multiple cycles.
Then once you feel moreconfident in your design, you're

(15:52):
going and building a productionpreparation run. So maybe that's
50 units, maybe that's a 100units prepping for what is
actual manufacturing, and thenbeyond that, you have the next
run which is game day and that'syour actual flight run where
you're building, you know, a fewthousands of things
Demonstrating that you cansupport all 3 of those is in a

(16:16):
way showing HMLV. Just HMLV isthe very beginning of that
process.

And what's really interesting is we were
talking about actually, like,hardcore number units like, 6
100, 800 for this. I've I'vebeen talking to some of our
customers. So what they do isthey order their, like, core
hardware team will order 10.Yeah. Okay.
They order 10 units. And thenwhen they get those boards and

(16:46):
the hardware team validates it,they place another identical run
before a 1,000 and it's stillthey consider that prototyping
because that goes to all theirteams all over the world Okay.
For firmware development.

The okay. Wow. A thousand.

So a thousand units is like Yeah. That's just
that's just part of development.

Right. I guess it just depends on

the size of the company. Yeah. It depends on the
size of the company and whatproduct it is, but then, like,
because they've done, like, 3 ofthose so far. Wow. And I don't
know if it's that's exact numberI know they've done a couple but
so it's like yeah and then sothey do a revision they get 10
the hardware team validateseverything is good and then they

(17:32):
build a 1,000 more and that'sthat's that's for someone like
you and me, that's insane of aprototyping cycle.
Yes.

Yeah. That is. Yeah.

That must be the extreme.

But

that's that's so it's crazy to think about
that. If your project is bigenough has enough people work on
it. It's right though. Like ifif if you need, hardware in the
hands of whoever's working onthe firmware, and you got a big
team, you know. Thousand.
Yeah. Thousand. Yeah.

That's a lot. Yep. But but but so it just
depends on, I guess, two things.It depends on the size of your
company and it depends on whatyour design cycles look like.
Yeah.
And that kind of gets you out ofthe high mix low volume into the
higher volume production. Butthen again, maybe not because

(18:26):
your product like, I'll take myjob for example right now. We we
do space stuff. It's we're notsending thousands of them up
there. Highlight high volume forme is is tens right now.

The only only person that does space like that
is SpaceX with their satellites,their their Starlink stuff.

Right. Right. And even then and even then it's
not, like, a huge portion oftheir company.

Yeah. How many Starlinks are in orbit?

Oh, I looked that up a while ago. It's a lot.

How much in their constellation? 6,219 which
that's a lot of satellites.That's not high volume though.

That's 6 prototype runs for this company
you were talking about.

Not 6 prototypes for this one company.
Yeah. Alright. Alright. Let'sswitch gears on high mix low
volume then.
Sure. What because you becauseyou worked at WMD and I've been
working at MacFab for a longtime. What do you what's the
difference between a Centimetersthat can do high mix low volume
and one that can't? You weretalking about the ones that will

(19:33):
just say, you know, buzz off ifyou're trying to do something
like this.

Yeah. Okay. So in my experience, typically, what
that means is they havemanufacturing lines that are
very very focused. I'm I'mtalking about companies that
will tell you, sorry, go away.It they have they have machines
that are specific for highvolume and they have workforces
that are specific for highvolume.
And, and and typically, that'swhat drives them away from the

(20:02):
low volume stuff. That's Andthat's like wave soldering and
and and things of that sort.

Yeah. Because we've done some tours. This is
back when you were at the fab.We did tours down, some
facilities down in Mexico, andthose are definitely very
specialized facilities becausethey made the ones that we were
looking at, working with, theywere doing TVs. Mhmm.
And those were highlyspecialized equipments. The

(20:29):
built

in They had multiple lines, but they were
all identical. They had theirworkforce was trained
specifically on a single productand the intent was they could
bring that workforce in, theyknew exactly what they needed to
do, there was no they they hadsupervisors and and whatnot
going around, but it it it justit was intended for one thing

(20:54):
over and over and over. And itwas the I think one of the big
things is it's actually a hugeamount of work to get that kind
of a labor force to do somethingelse, because it takes so much
training, because it takes somuch setup, because it takes so
much extra effort. In fact, oneof the facilities we visited,

(21:15):
they we got to see their testingfacility. I mean, they had they
had rows upon rows of peoplewith identical testing, rigs
that, that were all, like, hyperfocused on a singular product.
And and even I remember walkingthrough the the storage they had
of the TVs, the storage facilityAnd and what it reminded me was,

(21:40):
in Indiana Jones when they whenthey put the Ark of the Covenant
in the box and there's just,like, the there's just boxes
upon boxes of relics in that onefacility. And, like, it looked
exactly like that, but they wereall the same TV.

I I think it's the that's the biggest
difference is is theflexibility. Because at, MacFab
at our HQ facility, we run mychronic picking places, and the
reason why we went because, butactually before then we had a
universal GSM, which is theexact opposite of that that
machine. That's a high volumemachine. Yes. And so basically

(22:17):
the main difference betweenthose those kinds of machines
are how many feeders does thehead have access to to pick in
place.
So, like, a high volume machinemight have multiple heads, but
it can only hit like, let's say32 feeders. Mhmm. Like,
actually, there's a lot morecrossover, like the more modern

(22:39):
machines have a lot morecrossover, in what they can do,
like, high mix and low volume.But, like, a Micronic is I think
it has something like a 128feeders its head has access to.
And it's got different differentformatting for how the head
picks stuff up.
So you have, like, one head thatmight have 6 nozzles on it or

(23:02):
something like that instead of adual gantry machine. Kind of
same thing with those, like,those turret style chip
shooters, which is a veryspecialized equipment that's
literally designed to only placeone type of components. So if
you need to switch that machineover, you gotta switch all those
nozzles manually, but it canplace that 10 k resistor perfect

(23:26):
at lightning speed. Right.

I've also noticed if you if you look at an
assembly line and they havemultiple pick and place machines
in line with each other, that'susually an indicator they're not
low volume. Because they'll haveone one of the machines set up
for all, like, the the thepopcorn parts, like, all your
r's and c's and things. And thenthe other one is typically set

(23:52):
up for your big parts like youryour ICs and connectors and
things of that sort.

Yeah. So actually, we'll roll back just a
little bit more on like when youstart considering the low volume
to high volume. When you got togo wave if you have through hole
parts, it's when you have to gowave solder. Yes.

Yep. Yep.

That's kind of the break point I think too.
Because one, you have to getspecialized tooling. Even though
the tooling is not that badanymore. Back in the day,
getting a wave solder palletwas, like, 1,000 and 1,000 of
dollars Yeah. And months oftime.
Now you can get one for a couple$100 and, you know, 2 weeks of

(24:34):
work. Yeah. To

get get there. They're not terrible.

They're not terrible anymore. I think it's
just But this is

Actually, you know, that right there, if if
any of your manufacturingrequires additional fixtures or
or jigs just for themanufacturing part, not for the
testing part, but just for themanufacturing part, that's a
good indicator you're in thehigh volume world. Right? Yeah.
Because because that's the wholething about, contract

(25:00):
manufacturing with low volumestuff is everything is so well
defined in terms of all themachines for the express purpose
of being quick with high volumechurn and, or sorry, high mix
churn. And and and so as soon asyou need things like wave
pallets or fixtures for holdingthings in whatever machine or

(25:22):
whatnot, the the only timeyou're willing to spend that
kind of money is if you'realready ready to go for high
volume stuff.

Mhmm. Yeah. And, but that also leads into
you're talking about theworkforce. That that's also a
big indicator is if a shopthat's high mix, low volume,
capable, their workforce isgonna be a lot more flexible
because they have to be. Theysee, like, let's give an example

(25:50):
of that TV factory.
They are building 1 SKU TV.Yeah. Okay. So if you were like,
okay, let's now build amicrowave. Just this is an
example.
They would probably have toretrain everyone.

Yes. Exactly.

Whereas at at MacFab, we don't build TVs or
build microwaves, but for theexample would be our our
workforce only needs, like, adocument to explain how it goes
together. They're how much morehighly skilled, line workers and
assemblers, where they can readdocumentation themselves. They
don't need to be fully trainedup every single time you need to

(26:31):
tear a line down. They're muchmore flexible. So that that's
that's where the skills that'sthat's the skill gap between
those two systems either, aswell.

And and that's not necessarily to just say that
someone in a in a large volumeis unskilled. It's just that
their skills are in maintainingtime as opposed to changing jobs
and being able to be flexible.

100%. Yeah. Didn't mean it as a dig or
anything like that at all.

Right. Right. Right. Right.

No. It's totally just a different skill
set. One is just making sure atask gets repeated correctly
with utmost precision and ontime Over

and over and over.

Over and over again. And the other one is
making sure that you're enactingprecision on changeovers, and
your changeover might be 4 or 5times a day. Right. And so you
and you have more time just inthe nature of the work to do
that changeover. Get getting a a

large facility with multiple lines, getting
that machine to turn over intosomething else is measured in
months, not in days. Let alone,like you said, somebody
switching jobs multiple times ina single day.

Yeah. I mean, it's hard for me to do and I
don't even do that work anymore.

Yeah. Yeah. It it does. Both of them require
special workforces, that arejust they're trained and they're
good at that. Actually, you knowwhat's funny is it's it's kind
of hard to even switch those 2workforces, for for those to
wear the others hat.
I mean, that's why you you hirespecifically for those positions

(28:10):
like for high volume or highmix. And both are valuable in
their own way.

Yeah. Now I don't have anything else to add
to to high mix low volume.

You know, the only thing I have to say is high
mix low volume comes with somereally interesting challenges
with parts handling. That'ssomething that I think is worth
noting because your inventorymanagement becomes really
critical for that. Because mostof the time when you talk about

(28:41):
your high mix stuff, you're notyou're not providing boxes upon
boxes of the same part for forthem to deal with. And most of
the time with high mix stuff,it's in and out of the shop as
soon as possible because that'swhat the customer wants and
that's also what the Centimeterswants because they get the job
done and they get paid for it.But that comes with challenges

(29:04):
because the same amount of careneeds to be applied to the parts
and inventory as the high volumestuff.

I I would say that is probably the biggest
benefit at working working withMac Crab is our inventory
system. Yeah. It's a class 1,class a feature set, I guess, in
our platform. It's so funny whenyou like because when you look
at our platform and you like,most users won't even touch the

(29:32):
inventory side. But how all thatworks on our back end is like
top of line.
It's it's not like we took ananother ERP system or whatever,
like a normal Centimeters.

Oh, it's homegrown.

Yeah. On our end, all our entire inventory
system's homegrown and fullyintegrated into our platform. So
it's it's funny. It's it's superspecialized in doing high mix
low volume handling thecomponents. So Yeah.
Yeah. That's a good point aboutabout inventory. Yeah. Because

(30:06):
that's in my opinion, that's themost important thing for a
contract manufacturer is andnumber 2 is making sure you're
is making sure the boards aregood. That's, you know, what you
actually deliver.
Right?

Well, I think I think yeah. I think inventory
can make and break make or breakthings.

Oh, it does for sure. Yeah.

Yeah. And it can also leave a really bad taste in
people's mouth or the theclient's mouth. So, just showing
that you care about the clients,not just their money, but their
their their their stuff.

Their stuff. Yeah.

That that that really goes far. Alright. So we
we we talked about

high mix low volume. I do have some macro fab
updates. Yeah. None of themactually have to deal with high
mix low volume.

Okay.

1, this is what I've been working on in
marketing for the past, like, 2weeks. I I went through our blog
and pulled all the old platformupdates and then all the new
ones. This this actually stemmedfrom, like, our product team
here wants to, like, market ourplatform changes better. Mhmm.

(31:16):
And so we came up with an ideabasically, and when whenever
they have an update, we make ablog post and we actually made,
its own, like, landing page thatthe blog hooks into, with its
own tag and stuff, which I knowit was, like, that sounds really
boring and, like, yeah, that'sblogs at Macrofab.
Right? And, so, yeah, Iorganized everything. So you can

(31:41):
go to I think it'smacfed.com/platformhyphen
updates and so you can see allour new features customer facing
features. Unfortunately, Ihaven't been able to convince
the product team to let merelease, like, what they do on
the back end for macro fab. Ilove to show that stuff too, but
so far I haven't convinced themyet.

Keep working on it.

Yeah. But I have all the customer facing
stuff there, and I got all theold stuff from back in the day.
It's really fun going downmemory lane of, like, back in
2013 13 and, like, rereading theblogs that, like, Chris, Church
and I wrote back then about theproduct. And there's, like, old

(32:26):
screenshots in there and a lotof those old blog posts were,
like, copy pasted, like, 4 or 5times through different, like,
blog hosts and stuff and so Ihad, like, spruce them up a
little bit. I, like, put, youknow, make stuff bold and put
headers on stuff and, like, fixcaptions or, like, images were

(32:50):
broken, so I fixed the images.
So yeah, it was a lot of fun,making the project work. I'm
pretty happy with it. I I don'tknow if that will sell us more
boards, but at least now like wehave a proper place for, you
know, product updates.

It's in one place.

Yeah. It's all in one place now And and reading
those old blog posts, it's it'salmost like, how did people give
us money back then?

There's always the early adopters. Right?

Yeah. It's because a lot of the it's very
interesting. I I highlyrecommend everyone's listening,
like, just go read some of theseold ones that we wrote because
that's very, like, our visionwas so grand. Oh, yeah. I love I
love reading it.
It reminds you it reminds you ofwhy you started it back in the

(33:40):
day. Like, what was goingthrough our brain back then? So
Right. Although, I kept keptactually, all those old blog
posts that, like, engineeringstuff that you and I wrote back
then are also up there, but it'sreally hard to find. So part of
my next initiative is to work onthat and make, like, all those

(34:01):
engineering articles easy tofind, all the technical stuff
that we wrote.
I wanna make all that reallyeasy. Same thing with like the
news, we have a lot of old news,Macfab news stuff that is just
hard to find and I'm just gonnamake it easy to find. So again,
I don't know if that's gonnasell more product, but

it makes me feel warm and fuzzy. I I still
occasionally run into articlesthat you and I wrote.

Yeah. That happens all the time.

We put, like, on Google. Just, like, oh, okay.

I recognize that. My my favorite is when I'm
googling something, and I alwaysput like, oh, I've I always put
because you see back in the day,you type something in Google,
and then you'd be like searchimages and then there would be a
discussion tab. Mhmm. Anddiscussion tab in Google, it

(34:52):
would just show, like, forms andstuff. People talking about that
stuff, and they got rid of itfor a long for a long time.
They actually just brought itback. Did they really? Yeah, so
I think it's called let mecheck. I think it is called if
you click more. No.
Oh, it's called forms. So it'simages, shopping, videos, news,

(35:13):
forms. So they basically broughtback discussion as a tab. Okay.
Well, because I what I was doingfor a while was I mean, for I
can say for a while, for, like,the last 8 years is I type in
something and then just put inform after it, and it would
filter only forms.
But now they added that as anofficial thing again, which is

(35:35):
nice. When I find I'm searchingfor something, click on a form
link, and they're referencingsomething that me or you wrote.

Yeah. That's cool.

And that happens not often, but it's it
happens more often than I wouldassume, given how the fast the
Internet is.

Yeah. Yeah.

Yeah. It's kind of scary that that people are
referencing stuff that we wrote.

Yeah. But a lot of the stuff we wrote doesn't
change.

That that is true. It doesn't change, and at
least it was correct enough thatno one corrected us. True. So we
know so at least it's correctenough. Is that going to be the
title of this episode?
Correct enough. Or, what did Isay before that?

You know, actually, it's funny because
before the podcast, we weretalking about, a handful of
things. And one of the thingsthat came up was the topic of
like good enough engineering. II feel like we have there's some
really cool episodes that arecoming around the corner here
with some really cool guestswhere we're gonna be discussing
this kind of stuff. I I I'mfeeling a trend show up here.

(36:43):
Yeah.
Exactly. Good enoughengineering.

Yep. Exactly. So yeah. The platform updates,
go check those out and then wejust launched a glossary on
electronics and PCBA terms.

This is actually really impressive.

Yeah. So it's maccab.com/glossary. Go check
that out. I put in an errorsomewhere, so go find it.

Oh, you purposefully put in an error.
Yeah. Is there a prize forfinding it? No. But if you find
it, go put it atforum.macrevab.com?

Yeah. Go post about it. Now, I'll click click
the little heart icon if whenyou post about it.

Yeah. We'll we'll give you a digital pat on the
back. No. By the way, this thisglossary is like 100 of terms
that all apply to manufacturingand electrical engineering and
PCAs and PCBs. And I was tellingParker when I was reviewing this
earlier, I was like, this thisthis is something that they

(37:50):
should do in college.
They should just give you thisand be like, these are the
things that are worth knowing.

Yeah. This is

I had to give this to to interns at work and
be like, if you have a questionabout something, go to this
first and then come back.

Yeah. I'm hoping it helps out some people.
So

It's at at a minimum, it is a really good
resource in one place. It's alsoprobably SEO for you guys.
Right?

I think that's why they originally wrote it.

Okay. What?

So because I didn't I actually didn't have
anything to deal with this toomuch. I'm just promoting it.

This is a lot of work whoever put this together.
Oh, yeah. Hats off to them, andthank you because this is
actually some really good stuff.

I I think it was, Mireille. I I think Pamela
worked on it too. I have tocheck. I know Mireille was
working on it a lot though.

Because this this is not like AI just crapping
stuff out. This is somebody,like, putting effort into

Oh, I bet your AI has already gobbled this
thing up though.

Oh, yeah. 100%. You guys are feeding the
monster.

Yeah. Some monster.

But at least you're feeding the monster with,
like, human created, like,accurate stuff.

Accurate stuff. Yeah. So go check that out. It's
maccred.com/glossary. Alright.
On a personal project updates,this is a rare double back to
back Steven Parker only episode.

Yeah.

So you actually get, like, what I've been
working on for the past week.

Wow. So it's car related.

It is car related. It is the PWM fan
controller. I actually now havea thread on our forum,
form.macro.com, for the PWM fancontroller. The schematic is for
all intents of persons, allintents and purposes. I gotta
slow down sometimes.

(39:43):
The schematic is complete. Andso I don't know if you want to
do a schematic review right now,or do you wanna do that at a
different time?

Oh, like we were talking about, like, many many
episodes ago?

Yeah. Well, I haven't picked the components
yet, but this is more of aschematic review to see if
there's anything I need tochange before I start picking
parts and putting the boardtogether.

Oh, live schematic review. I'm down for
whatever. Yeah. Let's do that.

Because there's 4 sections on the schematic
review I want to do, and Iactually outlined these in our
form. So if you go to our form,form.macro.comform.macro.comform.macro.comform.macro.com.
If you go there, you'll find thethread. It's gonna be at the
very top. I should just pin itso people can find it since

(40:39):
we're gonna be talking about alot.
I actually took excerpts of theschematic and posted it in
there, so we're gonna be talkingabout these four sections. It's
gonna be the PWM outputs, thedigital inputs, the thermistor
inputs, and the push pullcurrent drivers. Because the
rest of the circuit is all justnormal digital logic non

(41:01):
interesting stuff. I mean itmight be interesting to some
people, but to me I'm like yeah,I'm using a microcontroller I've
used before. I'm using a USBinterface I've used before.
Nothing too super excitingthere. So the the PWM output
which is the main reason tobuild this is to send a PWM

(41:22):
signal a 12 volt PWM signal toan electric fan. And this is not
even high current, this is justa PWM low, low current signal.
And so, what I have is a nchannel MOSFET that is the gate
of it is pulled down with a 10 kresistor to ground and then the

(41:46):
IO the 3.3 volt IO from themicrocontroller goes through a
100 ohm resistor. It's kind oflike a gate snubber because
we're only driving this with anIO pin, so you can't you can't
just unleash all the currents.
To get your your your IO driverson your mic controller will not
be happy if you do that for verylong. So put a little snubber

(42:08):
there and then the drain side ofthe FET is pulled up with a 100
ohms, which doesn't sound like alot, but I wanted to make sure
that this thing can switchquickly if needed. It's actually
only gonna be like a 100 hertz.It's the fans I've been testing
is a 100 hertz PWM signal whichis really slow, but I put a 100
there so you can actually switchit fast enough or quickly.

(42:32):
Actually I don't know how fastthis thing could switch until
you start getting too much slew.
I guess we can simulate it butthen yeah and then that drain
that pulled up drain is whatgets sent out into the world.
There is some ESD Protection onthe actual IO pin that goes back

(42:54):
to the microcontroller that goesthrough a it was the chip I said
last week which was the sp 721.It's like a, it's a silicon
controlled rectifier diode chipthing that, prevents over over
voltage and, transient voltagesand stuff like that. Pretty cool

(43:19):
little chip. I don't thinkthere's much.
I I mean, this is a prettysimple circuit. I don't know
there's anything else I canreally do here, that it could
make this better.

I I have one suggestion that I think it's
minor and I doubt in thiscircuit it would even apply, but
in general with, you got yourgate snubber on the other side
of your pull down resistor. Ithink it's better to put the
gate snubber on the front end ofthe, the pull your pull down

(43:52):
resistor. So in other words, thegate snubber should be right at
the gate. And right now, youhave it on the other side of the
resistor. I would I would justchange that one thing because
the whole purpose of a gatesnubber is to swamp or reduce
the parasitics right around thegate.
And, and so putting it as closeas possible to that, to the, to

(44:13):
the gate as you can makes themost sense. Now, this is just a
low side switch. It's not gonnaoscillate, like, it's just not
gonna be a problem. I'm justsaying like in general that's
good design practice.

Yeah. Yeah. What's interesting because
that's a that's a different endgoal with than what I was
designing it for. Because I'mmore worried about the the
driver in the microcontrollerbasically because it's in theory
basically when you turn 3.3volts on and you don't you
basically have a capacitor onthat gate and it's gonna wanna

(44:46):
gobble up as much current as itcan as fast as it can and which
can like instantaneous currentis really high and so you want
to slow that down. So puttingthat 100 ohms snubber on the
other side of that pull down isfine.
Yeah. The only thing is, Iguess, on the only thing is it

(45:07):
might slow down you closing theFETs. That's about it, and
that's not a lot.

Yeah, and I just given the purpose of what you're
doing here and the voltages thatare happening, that's not gonna
be an issue.

Alright. Let us know in the comments on
form.myfhead.com, which one youwould prefer. But you you are
right is flipping that around.

You also have a very slight voltage divider
here.

Very slight.

So it's you're losing 1%

of your voltage with this. Flip, I'm writing
this down. Flip gate snubber.

Yeah. Put basically the the gate snubber
touches the gate of the device.

Alright. So next circuit we're gonna look at
is the digital inputs. So thedigital inputs, are gonna be
active low.

Okay.

Active low, so this is like switch inputs. This
is looking at stuff like your,AC high pressure switch to see
if, like, you need to turn yourfan on to, because your AC is on
your air conditioner. And sothis is a, so yeah, active low
and so the, actual signal isgonna be basically that comes in

(46:27):
can be floating, and so we haveto pull that up to 12 volts. So
we have a 10 k resistor pulledup, and I do have, like, a point
1 microfarad part, capacitorthere just to kinda help smooth
stuff out because that thatcould be a wire that goes into
the engine bay and out intoinfinity. So it can pick up
noise, so we're gonna go, hey,let's just put a little just a

(46:49):
little capacitor there just tohelp smooth some stuff out.
Now but this has to get into our3.3 microcontroller so we we
basically next thing is we gothrough a voltage divider but we
are not only at 12 volts it's aautomotive system so it can be
technically max is like 30 voltslike peak transients, but you

(47:13):
don't have to worry about thatbecause it's very short time
period. 14.7 volts is about max.So basically my resistive
divider here is for the maxvoltage of the system, which
gives us like 3.25 volts comingout on the on the IO side. Okay.

(47:33):
To that voltage divider.

I guess I wait. I I guess I'm reading this
incorrectly. Am I is it right toleft or left to right? Your your
AC override in pin

Is coming from the world.

Oh, okay. So so I I was reading this left to
right. Yeah. So that so right isis your input and left is your
output.

Yeah. Left is the microcontroller.

Got it. Got it. Got it. Okay. That this circuit
makes a lot more sense now.

I can imagine so. Yeah. Yeah. And so, yes. So
our AC override in, which isfrom the outside world, goes
through that voltage divider.
And then halfway through it youknow on the voltage divider part
we go through a snubber and theninto the IO pin and then in the
middle of that voltage rail orvoltage divider, I do have a a,

(48:24):
a zener diode at 3.6 volts justto clamp anything that might go
over

Mhmm.

Just in case. And then there's also another
point 1 microfarad just sittingon the output, to help stabilize
the output of this circuit.Because again this circuit is
not gonna be switching reallyfast. This is going to be like a
0.1 Hertz max. Like, okay.

(48:53):
1st. I was about

to ask because you you are you are making a
little bit of a filter here.

It's a little bit of a it's a little bit of a
low pass filter here, becauseyou gotta think that the inputs
are actual switches. Like, it'scoming it's like a switch on a
pressure line or a switch onyour dash. That's that's what
these are for. So these are on,like, the fastest things gonna
switch is someone trying to hita toggle switch as fast as

(49:19):
possible for some reason.

Right. Right. So This is also not really because
you're you you have a capacitorinside of this divider. It's not
really a true low pass filter.It's like a low shelf kind of
thing.
Yes. Yeah. But but in terms ofwhat that in terms of the impact
of that on speed, this is prettyminimal.

Mhmm.

And and like you said, it you you don't care. You
you're not putting a PWM backinto this. No.

No. That's not what this is for. I say digital
input. It's a switch input.Yeah.
Okay. Physical switch. The fact

that you're pulling this up to 12 volts, but
your output can be plenty above12 volts.

It it Well, so that 12 volts that 12 volt rail
Yeah. Is connected to the system12 volts. So that 12 volt will
be like that 14.7 volts.

Sure. But you you said your output could spike up
to, like, 30 or something likethat.

Yeah. So the input of this whole system is is
snug too. Like, if you goactually go into the power
input, there's, like, a snubberto help prevent that 30 volt
from propagating anywhere elseto system. Okay.

Because, yeah, that was my next question is why
not just put a diode in seriesjust in case it back feeds?

Yeah. Yeah. Yeah. That's that's already
taken care of That's

taken care

of somewhere else. Stream. Yeah.

Got it. Okay.

Yep. Yeah. So this is this is literally
basically so you can hook up aswitch and then the other side
of the switch is to ground. Andso when you switch the switch
it, you know, pulls the groundand activates the circuit

or Is there any concerns with pulling, this
input up to 12? Any concerns forthe outside world?

If if you if you pulled it up to 12, it would
still be connected to thiscircuit except there'll be on
the other side of the fuse.

No. What I mean, you, the you are pulling it up
to 12 here. Does that affectanything on the input? Like
anything that's connected outthere. Everything's fine with
being pulled up to 12 is whatI'm getting at.

Right? You mean this AC override in?

Yeah. And you connect that That's what's okay
with being 12.

Yes. That's okay with being 12. And it's
also through that 10 k resistor.So, like It it yeah. It would
limit

the current.

It would limit the current. So if that wire
actually hit ground, it nothingbad is gonna happen. Right.

Right. Right. Right.

Yeah. Okay. I I was playing around with, like,
optocouplers and stuff to makethis work, and I'm like, that
sounds like a lot of work that acouple resistors and a diode can
do.

It depends on if you need isolation or not, which
doesn't sound like you dobecause the car is one system

and It's one system. Yep.

And and, you know, if you if you were really
really worried about, like, areally noisy environment, then
sure. But it doesn't sound likethis matters.

Yeah. The original prop fan had
optocouplers for that, and theyhave not really needed them. It
didn't matter because I that'swhat I thought is I thought like
isolation would be important. Itdoesn't matter because like you
your entire chassis ground mostautomotive stuff is not
isolated. Everything just sharesthe ground.

(52:31):
Everything's fine. All youreally have to worry about is
crazy voltage spikes becausesomething bad happened. Sure.
Like a battery got disconnectedor you got in a wreck or
something like that. Because yougotta remember you got 12 volt
lines on a battery that's huge.
So it it does like to smoothstuff out really well.

Yeah, it doesn't like change, right?

It doesn't like to change really fast because
they can it can just keepgobbling up that change. Yeah. A
lot. Oh, what's interestingabout that? Slight tangent.
So more modern so like oldschool alternators, I guess it's
probably up till the mid aughtsthey were this way. Like there

(53:18):
was an internal regulator onyour alternate, and then before
then 7. It's different like, youknow, back in the sixties. We
had generators, but anywaysalternators. You have a internal
regulator that has a sense wireso like sense wires on your
battery so it takes care of likevoltage drop and that kind of
stuff, but it's kind of like anall self contained system to

(53:41):
improve efficiency fuelefficiency because active you
know, your alternators pulling 4or 5 horsepower off your belt to
charge batteries and that kindof and run everything in your
car.

Right.

So to improve efficiencies, what I know this
is what GM does. I know a lot ofother manufacturers are doing
this is they they start doing,like, smart alternator control,
and so on, most of the time itonly charges your battery up to
80%, and then it just like freewheels the alternator and like

(54:19):
it basically lessens the fieldwindings, so it doesn't take as
much resistance. But theybasically found is like to
charge it up even more what itdoes is when you start braking
that regen breaks with thealternator too and helps slow
the engine down through that. Soit will increase the field
windings and charge your batteryup more. Wow.

(54:39):
So they got some fuel efficiencygains by basically when you're
cruising and you're over 80%state of charge they go well,
that's good enough to whereyou're gonna be able to start
the car several times withwhat's in that battery already.
We don't need to get up to a100%, and we can just charge up
higher when you're braking andnot actively using the engine.

(55:02):
So cool stuff. This is not doingthat, but I just thought it was
cool tangent because we'retalking about alternators and
charging and voltages.

Yeah. Yeah.

Cool cool stuff. Yeah. Yeah, that's that's
the digital input, if someonecan make a convincing case that
this circuit is bad, I'd love tohear because I actually came up
I bet you there's circuits likethis that exist, but I kind of
made this myself. So it's Sure.Probably got some weird edge
case that I'm not thinking aboutthat can go bad.

I mean, it's a it's a it's a voltage divider
with some voltage clamping andsome filtering. Yep. And

and a pull up. And a pull up. But, give it a
look. Let me know what I canprove. Don't say just put an
optocoupler because I'm notgonna put an optocoupler.
An optocoupler costs like 10xthe price, actually probably
more than that. It's probably a100x the price of these parts,
so.

Yeah. If you're trying to get this done as
cheaply as possible, you youprobably even have more
components than you need.

I probably have more components than I need.
Yeah. Make that work.

So Alright. Have you considered gosh. Let me
think about this. The lowvoltage condition. Because you
consider the high voltagecondition, but what happens if
your input is on the low end?
I guess, the the pull upcorrects that. Right?

Yeah. Yeah. I mean, it's like floating. I
mean, if you pull the switchdown, that's the that's the
application for that.

Right. Right. Okay. Never mind. You you you're
right.
You're right. Yeah. I'm so usedto at work. We we always have to
look at all edge cases, both topand bottom. And so that my mind
immediately went there I waslike wait, never

mind. Yeah.

The purpose is for this to go low.

Yeah. Purpose is to go low. Yeah. Yeah.

Well, I guess I guess the other the in this
situation the the potentiallythe low situation would be if
your 12 volt rail was on its lowside. So Yeah. Let's say the the
tolerance of the 12 volt railwas 11 volts, not 12. Will this
still function?

You you know, actually, that's an interesting
case is to do the math. I'mgonna write this down. Is do the
math for, like, 10 volts. So,like, if your battery is low,
you don't want it to turn onyour fan because that's gonna
drain your battery even more.

Correct.

So is is on the digital input is do do the maths
on low battery. Well, okay.

But but, actually, I I I think I would
make an argument that even ifyour battery was at 10 volts,
you would still want thiscircuit to function. You just
want to be able to read yourbattery voltage and your your
brain box says don't function ordon't turn on fan.

Exactly. What's, you know, what's
interesting is the I I don'thave it broken out because it's
just a voltage divider on the 12volt rail, but I do have where
it's reading the rail to seewhat the battery voltage is. So
you could be like, hey, if I'mreading a low battery and my
digital input is always low. I II need to do the math and

(58:04):
basically look at the what thetrigger point of the sandy 21 g
is Right. The microphone I'musing.
If you're like, if I'm at 10volts on this IO is that gonna
trigger as a higher low on myinput?

Are you in that gray zone where you don't know?

Oh, don't know. That's even worse.

It's way worse.

Because if it's way worse then I go, well, I'm
just gonna connect this to ananalog input now.

You you know, honestly, one thing that I think
would be fun with this is tothrow this circuit even though
it's just a simple voltagedivider, throw it into LT Spice,
put your tolerance of all yourcomponents in it and run a Monte
Carlo simulation. Do a 1,000simulations and see with all
just random tolerance on allyour components, what is your

(58:47):
range. Yeah. That would thatwould be fun as well.

Because I am going I'm using 1%, automotive
grade resistance.

I'm sure you're fine. No.

No. That's great because I actually wanna
know, like, if it's on if it'sgot a low battery and the
tolerances stack up is just theright way, am I not gonna be
able to trigger a high on thatinput pin? And so it's gonna
think that the AC needs to becooled down. The the AC,
condenser needs to be cooleddown, and goes, well, I need to
turn the fan on and pull 80 ampsnow. Yeah.

Right. I'm gonna throw I'm gonna throw one more
thing at you on that. Vary thisacross temperature as well. In
fact, just do the Monte Carlo atwhat's what's what's car
temperature, typical ranges?

Freezing to 200 degrees Fahrenheit.

I would actually say less than freezing. Cars
regularly driving lower thanfreezing.

Negative 40 Fahrenheit is the low end of the
parts for automotive grade.

So say negative 40 negative 40 c to 85 c,
something like that. See, willit work across that entire range
with varying tolerance as well.In in other words, you I think
you could you could run 2 MonteCarlos, one at negative 40, one
at 85 with your dollar shift.

Parker Dillmann (01:00:11):
Hey, you make me do a lot of work.

Stephen Kraig (01:00:13):
I know. I know. Make me do engineering work.
This is real engineering. That'sreal engineering.

You are right. That is that is

what it is. It's boring. I like this. Yeah. Now,
okay.

Go ahead. No. No. No.

You you get you have 1 or 2 more circuits to go?

I have 2 more circuits. Cool. Alright. Because
I know we're at time, but

No. No. No. This is great. Keep going.
Analog inputs. Yeah.

And so these are designed to read
thermistors. These are not likecurrent, driver like you know 4
to 20 milliamp or anything likethat or voltage output drivers
from sensors. These arethermistors so they are ntd, you
know, negative temperaturecoefficients temperature

(01:00:58):
sensors. It's a resistor thatjust changes under temperature.
And so these circuits Iliterally lifted these from
other engine control systems.
So these these this is what thebig players do too. And the
values are weird. I don't knowwhat the historical numb

(01:01:21):
reasoning for some of thesevalues are, especially the 2.49
k pull up, but everyone usesthose. Mhmm. Like you can go to
like automotive sensors like theOEM for the sensor, and they'll
give you a chart based off a 5volt pull up for 2.94 k.

Okay. Kind of weird. Just a classic value?

Classic value in automotive. Now I don't have
a 5 volt rail so I can't usethose charts. Actually,
technically, yes, I can. I canjust I what I did is I made a
spreadsheet to convert the 5volt pull up to 3.3 volt pull
up. Just to rerun the math forwhat the

voltage whatever the whatever the base baseline
current is yeah what

the baseline current will be because I'm
still doing a 2.49 k pull up,but I'm only pulling up to 3.3
volts because that's my rail Andthen there's basically a low
it's actually a low pass likepie filter where you have 2
capacitors and a and a I thinkit's a 2 point 2 k resistor in

(01:02:25):
there as like a as a filtering.That's all it is. And you got
thing is on the on the out onthe input of this is a,
thermistor to ground. Sobasically, it's a it creates a
the thermistor creates a low legof a voltage divider there. And
now of course on the IO sidethere's also that SRC protection

(01:02:51):
IC is is there too, so it'spreventing transients and
overcurrent situations on theout I don't think there's much
to talk about there unlesssomeone is like, that's bad idea
and his way to improve it.
That's because what this is whateveryone else uses. That's why I
went with this.

Yeah. I I guess just the the series 2.2 k
resistor that goes into theanalog pin. I get is that just
for filtering? It has a a 1microfarad cap. Yeah.
The ground

cap rate. Filtering. Yep. Okay. I think
that's what that's for.
It's just to help smooth outwhatever signal you're getting
off that that thermistor.

Do you know what the input impedance is of your
analog reading? Because justbecause you have 2.2 k is high
enough that if your inputimpedance on on your analog
inputs is low, that can affectyour readings.

I don't know what it is on the Sam d. Yeah.

I would say that's worth just checking
because that would justeffectively appear as an offset.
Right? For for readings or or ifthere's any leakage which there
should be very little. So Idon't think it's a huge ordeal.
It's just something to consider.

Let's see. I'm looking at samd21adcinput
impedance.

Like, I'm just throwing a number out there.
This is probably ridiculous, butif they had like a 10 k input
impedance and they needed to bebuffered, this this, 2.2 k in
series would would cause someissues. Yeah.

I don't know what the n p n pence is.

I think it's worth it's worth looking that up
just so that you have you're notjust cooking in inaccuracy.

Yeah. Yeah. I'm trying to see if there is a
impedance on the datasheet thatI can quickly find. I can't find
anything really quickly.

Let's see here. I'm just on a random website.
Input impedance appears to besomewhere in the

40 kilo ohms. Really? Yeah. According to Sam d
21 datasheet page 991. I didn'tknow that.
40 k.

40 k all the way down to DC. That a Okay. So, you
might you might start runninginto a little bit of inaccuracy
with that.

Right? Then literally, the next post is 3.5
k.

What? Okay. It sounds like there's some more
research that needs to be donehere, but regardless if it is
pretty low, yeah, then you needto either buffer it or consider
Mhmm. Some other kind ofsituation or circuit.

Yeah. I have to

see what the actual input opinions is on it.
So it might be worth puttinglike a voltage follower there.
Because because if it is justacting as a 40 k input
impedance, you have a 5% errorjust from having it that series
resistor.

Oh, that's about the error of the sensor
though.

Well, okay. But you're you're stacking that on
top of it.

Yeah. That's true. But it could it could
stack it in the oppositedirection.

Sure. Yeah. It's you're okay. So you you've tuned
your error.

Possibly. Who knows?

Yeah. Yeah. Yeah. So that sure. May maybe a little
op amp buffer would be helpfulthere.

That might be the right place. Put a, like a
dual pack because I have 2 ofthese inputs is put a dual
package op amp.

And just do a just do a a unity buffer and
you're getting.

Yeah. Just like I think I actually have
designed, like, a, what's what'sit called? A, instrumentational
instrumentational, amplifier.

Sure. That that's probably way more overkill than
you need. Yeah.

But it's already designed, so I can just
put on the board.

That's how Parker designs. But I already have it.

Already have actually, I think I have another
one that's just a generic op ampthat I can plop down down to.
That's like an issue.

Yeah. And I mean, the voltage range on this is
nothing.

So Yeah. 3.3 volt. Right. But,

it depends on how low you needed to go. Because if
you're trying to single supplythis and get it to go down to
ground.

No. There's a baseline.

Okay. So it's always above ground?

Yeah. It's always above ground. Cool. Yeah.
The the, resist the resistanceof the thermistor won't hit 0.

Okay. The resistance will never hit 0. Do
you ever need it to go all theway up to 3.3 volts? That would
be the situation where you havea sensor unplugged, basically.

Yeah. If your sensor's unplugged you actually
should be able to detect if it'sunplugged or not as well.
Because then you can throw anerror. It's like, hey, my
sensors unplugged. I'm gonna runat it like a baseline speed
because I have no idea what thetemperature can be.

Right. Right.

Yep.

Yeah. I guess you could just have the the brain
box know if it's this value. Ifit's above this value then
consider that an open probe.Yep. Yeah.
Okay. Yeah. I think I thinkthat's your best bet is just
throw in a little buffer.

I'm making a note. I put op amp bugger
instead of buffer. Okay. So inthe last circuit that we're
gonna talk about is I don't knowwhat the best way to call these.
It's a but it's a it's a pushpull current driver is what I'm

(01:08:35):
calling it.
It's an h bridge.

Well, half h bridge. Right?

Half h bridge. Yeah. And there's 2 of them,
though, but they're for 2different outputs. Right.
Technically, I guess you cancontrol a motor with it, but
that's not what it's for.
It's the control. It's thecontrol relays. It's the control
indicator lights anything likethat that's what's for and also
this is it's

just a circuit that can sync or source

Yeah. It's a circuit that can sync or source
so it can it can output 12 voltsor it can sync current from
something else. So you can put amotor on it. It's actually
designed to be like I think it'slike 4 amps at 12 volts. Mhmm.
And so I have a fuse on theoutput and, so that that way you

(01:09:24):
if it's sinking or sourcing itdoesn't overcurrent anything
It's using a p channel and an nchannel and then the p channel
is switched with an n channelFET. So the older logic is the
same. I think the only big thingwith this circuit that could be
improved is maybe like a logicgates on the IO pins that

(01:09:50):
prevent both from being activeat the same time.

Just being a dead short.

So you don't have a run through situation.

Yeah.

I on the software side, I'm going to put
in a check. So, like, I'm gonna,like, when I do the software,
I'm gonna, like, this part ofthe code will be its own like,
you will call a function thatwill set the PIN. Right? But
there'll be an actual, you know,function that will actually set
the registers, and that functionwill check to make sure that

(01:10:20):
these are not set the same.

Right. Right.

And if it is, then throw an error, and don't
do it. But we should probablyput just put some software lodge
or firmware. We should put somehardware logic right here just
to Double make sure and alsolike what if you have a failure
of a IO driver off yourmicrocontroller? So like a like

(01:10:48):
an I input a, a pin gets stuckhigh or something like that.
Mhmm.
So

I do see one thing that that that needs to be
swapped on this. So u twelve,your pFET.

It doesn't

need to be swapped, but it needs to be
flipped vertically because as ofright now, the the source and
drain are in the wrong positionsand the body diode will just
conduct straight through.

You are right. Yeah. Oh, I fucked that up bad,
didn't I?

Right now, it would just straight up turn on.

Yep. Flip the p channel. This is why we're doing
a review.

This is why we do it. Yeah. The other thing is the
similar to your first circuit,the gate stoppers, just move
them right to the gate. Yep.Technically, r 43, which is
your, like, your high leg pullup, that could constitute your
gate stopper for u twelve.

(01:11:50):
Just depends on how you lay itout.

Oh, yeah. Yeah. That was the plan.

Yep.

Because I'm I'm not worried about like so that
gate on u 12 is connectedthrough the drain on u 13, which
is the end channel that opens itup. Like, I want that to flow as
much current out as it can, soI'm like, I don't want any I
don't want a snubber betweenthat gate and the drain of u 13.

(01:12:16):
I want that to flow turn on asfast as you can. And those,
like, this is not designed a PWMor anything is literally like
the driver relay or somethinglike that, which you also don't
want to pulse. So You'reactually probably putting some
small capacitors on those gatesas well to kind of keep them

(01:12:37):
more stable might be worth ittoo.

You can always, just non pop just put a put a

pop up

down and not pop it.

Yeah. Just see if that is becomes like an EMC
issue. Sure. And be like, oh,I'm just gonna you know what,
actually, on that output, Ishould put clamping diodes on
that output. Unlike the,actually, I wonder should I do

(01:13:06):
it on the out of the fuse sideor actually, yeah, it should be.
On past the fuse before itleaves is where I should put
some clamping diodes. And whatI'm thinking there because this
what I'm using these for is todrive relays. Big relays too and
so if that if one that the relaydoesn't have its own snubbing

(01:13:28):
diode. Now I have them builtinto my circuit. Yeah.
On your flyback diodes. Butalso, if that fuse pops, that
relay is gonna wash it off.Right? And putting those putting
those diodes, let's say, closerto the FETs and on the other
side of the fuse doesn't helpthat relay at all. So putting

(01:13:51):
those diodes closer to theoutput might be the way to go.
I'll write that down as I needto put

Yeah. Yeah. Yeah. No. I like that.

I need to put an output, flyback diodes. I
mean, technically the the bodydiodes of the of the FETs do
that too, but that doesn't helpwith that fuse pops. I got some
stuff to change.

Yeah. I like your idea of putting, some hardware
logic in to prevent thesituation that they can ever
both be on. Even though there'snot a software condition that
should allow for that. I

I think

I think a a an honest to God, just some logic
gates in front of these things

Yep. To just force that. Yeah. Just to force
it. The part b a and gate.
Well, 2 and gates probably.

I think you can do it with an XOR and another An
XOR. Yeah. Yeah. Because an XORis gonna give you a 0 when
they're both high.

Yep.

And then, I don't know. There's a handful of ways
you could pull it off. But I Ithink I don't think there's
gonna you can't do it with 1logic, Kate. I think you need a
Correct.

Yeah. You need 2 XORs will work. Yeah. I think
that's right. Yep.
Yeah. That'll work. Well, a quadXOR package is what I need
because I have 2 of these.

Because you have 2 of these circuits.

Yeah. So I have 4 IO pins.

Yeah. Cool. I I I guess I'm a little bit curious.
So is this output supposed tojust be a general purpose
output?

Yeah. So because the design's open
source, so you can technicallywrite whatever software you want
on it. So, yeah, you can you canmake this drive whatever you
want. I'm gonna use it to turnour big chunky relay on, but you
can do whatever you want withit.

The the the only thing I'm I'm thinking, you do
have it fused, which is good,but right now there's nothing
like if you just took thisoutput and connected it to
ground and said go, this thingwould give you everything it
can. It would just puke its gutsuntil the fuse pops. Yeah. So

So the 12 volts that it's getting is Yeah. Fused
on the input Yeah. To the deviceand, there's that fuse there.
Right. And so that fuse is gonnabe a actual it's well, it's
gonna be a PTC fuse, but it'sgonna be lower than what the
FETs are rated for too.

(01:16:27):
So all the traces and everythingwe overspent.

Sure. Sure. Is that PTC faster? Will that work
faster than the the

FETs deciding to

Yeah. The source itself. Yep. Right. Right.
Exactly.

The GRU eating the FET?

I'm just wondering if it makes sense to
put a small amount of resistancein line just to act as a a a
really cheapo current limiter. Imean,

that's what a PTC, it already kinda is.

I I guess. Yeah. Yeah. I guess you're right.

Because I could put, like, an actual automotive
fuse there, but I don't reallywant to keep opening this thing
up if when, you know, if if thefuse was external, it'd be
totally fine, but the fact thatit's gonna be internal inside
this, like, box that'swatertight. Having to open it up
to replace the fuse is gonna bea pain

in the butt. Does it, I don't does it make sense
to have any kind of Here's somefun word for you, telemetry read
back. In other words, on theoutput, does it make sense to
read the output with themicrocontroller just to confirm
that it's doing what it needs todo. Like, yes, I'm outputting 12

(01:17:44):
volts. So, yes, this is 0 voltsor whatever.
Oh.

Just so that

you, like, you could read it back and know
that, like, you could you couldsense an open or a short.

Yeah. Yeah. Yeah. Like, on the other side of
the fuse.

Yeah. And then it could tell you, hey, this is
messed up

or not. Yeah. Just run that side of the fuse
through a voltage divider andthen put a TVS clamp on that
input. Yeah. Yeah.
I have plenty of of unused IOpins on the microcontroller to
do that.

Do that then and so you can you can you can you
know the help, like, youwhenever you turn this on, you
can confirm that it's on andthat there's not an error or an
issue.

Feedback on

on outputs. You could also potentially have it
where if it senses an error, itjust turns off your outputs and
like, I need to

Yeah. Something's wrong.

Pause.

Yeah. And you'll know if the FET the, fuse
blue or anything like that too.

That's all yeah. That's that's potentially only
true if you're sourcing voltagebecause if you're just pulling
it low, you can't tell anything.Right?

Well, no. You would see that pin if if the
fuse went and you're pullinglow, you'd actually see it go
high.

Oh, no. You you yeah. You're right. You're
right. You're right.
Yeah.

Because the current stops flowing to
whatever is on the outsideworld.

But what I'm what I'm saying is well, okay. So
this, that telemetry readbackwould only tell you if the fuse
has blown or not. It doesn'tactually tell you if current's
flowing. Correct. Which isprobably enough.

That's enough. Yeah. Yeah. Yeah. And and I
would use, like, some big, like,we're talking, like, a 100 k
plus voltage divider so thatYeah.

Yeah. Yeah. It's something real big.

So it can't drop that voltage. So, like,
let's say the case where, like,the, fuse goes and your sinking
current, you don't want to havelike a 1 k or a 100 ohm or
anything like that there whereit's gonna the current is gonna
decide to want to go in thereand drop Yeah.

You don't wanna load anything down. So, yeah,
you can just it's just thesniffer that that smells the
output line that says, is therevoltage here or not? Yep. I need
you to You you know that ifyou're sourcing, you should see
12 volts there. And if, youknow, if you're syncing, you
should not see 12 volts there.

Yeah. You should not see 12 volts there.
And you can check the oppositetoo. It's like if you're
sourcing, you should see 0there. And if you're, if you're,
syncing, you and not well, ifyou want to sync and you're not
syncing at that moment, youshould see 12.
Yeah. Yeah. Yeah. You should beable to you should be able to
read the opposite state as well.Scope creep.

(01:20:34):
Yep. Well, that's easy. That's 4resistors.

And if you have a bunch of extra IO lines
available I have

tons of extra. I'm like I think I've used, oh,
like, 8 out of, like, 30 Yeah.

Then then these are no brainers.

Yeah. I could pick a different
microcontroller, but I havetrays of these to use.

And you also already have it designed.

And I already have it designed.

Because that that matters

a lot. It does matter a lot, especially in
projects like this. Yeah.Spending 30 minutes designing a
part versus just plopping itdown and and spending that 30
minutes just doing the circuit.Alright.
I got some I got some work to doahead of me. Yeah. I like it. My
schematic was not done, but I'mgonna wait until, this episode

(01:21:22):
comes out. Oh, I'll probablymake these changes and then wait
till the episode comes out andthen everyone can dog pile in on
that thread and then, we'll workon that too.
I bet you there's some otherthings especially that digital
input. I think someone canprobably tune that a little bit
better. Especially also that PWMoutput circuit like Go to That's

(01:21:46):
a really rudimentary. Man. We'renot doing optos.
1, I don't want to design anopto for it, a part for it, and
2, it's just the prices. Yeah.Yeah. And it doesn't need to be
isolated. That's the thing.
We don't need isolation. It'sconnected to what it's connected
to on that other end as as apower source. So cool. Let's

(01:22:11):
wrap this thing up. Yeah.
Everyone got an extra 24 minutesof circuit break this week.

You're you're welcome?

So thank you for listening to circuit break
from Macrofab. We're your hosts,Parker Dillman.

And Steven Craig.

Later, everyone.

Take it easy.

Thank you. Yes. You breaker. You know, we need
to say breakers more during thepodcast. Like, when we when we
actually, like, ask people torip apart our schematics on our
form.macfab.com.
We should just refer toeveryone's breakers. For
downloading thank you, breaker,for downloading our podcast.
Tell your friends, coworkersabout our podcast circuit break

(01:22:53):
from MacroFab. If you have acool idea, project, or topic you
want us to discuss, let Stevenand I and the community of
Breakers know our communitywhere you can find personal
projects, discussions about thepodcast, engineering topics, and
news, and now recipes. We have awe're making a engineering
cookbook.
So you can find thatatform.macfab.com.

All Episodes

EP#440: HMLV - High Mix Low Volume

Episode Transcript

Popular Podcasts

On Purpose with Jay Shetty

Stuff You Should Know

Dateline NBC

.css-15opob5{left:0;position:absolute;top:0.8rem;} All Episodes

.css-14f5ked{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:2;overflow:hidden;}EP#440: HMLV - High Mix Low Volume