Embedded - 323: Snail Appnote

Episode Date: March 5, 2020

Carmen Parisi spoke with us about changing jobs from a semiconductor specialist at TI to an electrical engineering generalist at Wasatch Photonics.  Carmen was previously on Embedded 216: Bavarian Folk Metal and formerly was the host of  The Engineering Commons podcast  Carmen works at Wasatch Photonics making Ramen Spectrometers. Spudger

Welcome to Embedded. I am Elysia White. I am here with Christopher White. And we are welcoming back Carmen Parisi, whose life as an electrical engineer has suddenly been complicated by software engineers. Isn't that always the case? Hello, Carmen.

Hey, guys. Good to be back. Thank you. So, yes, things really did get complicated in the last month or so. And actually, before that, your life has changed a bit since we talked to you two years ago. Has it really been that long? Maybe three. Well, yes, that too. I am a dad now.

Oh, congratulations. Oh, thank you very much. It's exhausting, but so worth it. Could you tell us about yourself as though you were on a panel at a conference? Sure. Yeah. So for those of you guys who don't know me, my name is Karvan Parisi. And for the last eight and a half, almost nine years, I've been an analog and power electrical engineer specializing in the semiconductor industry. But a month ago, all that changed and I switched jobs to more of a generalist position. And that switch is actually one of the things I want to talk about. But first, we have lightning round.

Are you ready? I'm ready. I got my hand on the buzzer. Okay. Favorite element? Ooh, nickel. Favorite laser wavelength? 1064 nanometer. I figured as much. Favorite voltage level? plus minus 12 volts when differential uh which sesame street character best represents you

god is it bad if i say oscar no no it just makes you a good RF engineer. All right, good. Now I've got to learn RF. Free space or fiber? Fiber. Tip everyone should know. Spudgers, turns out, are actually really useful. I never really used one, but now that I'm elbow deep in various fixtures, I'm scraping wires and you know pulling them through holes prying connectors they're they're handy also required

equipment for opening max exactly yeah what's this budger it's like a it's like a like the thing you use to spread uh spackle on walls it's like a flat offset spatula no not not quite it's like a fiberglass non-conductive pointy stick oh that's different okay i had something else in mind i i fix it praises the spudger quite a bit and uh up until now i really only used it for getting wax off my car because it doesn't scrape i guess i do have one of those too but i didn't know what it was called yeah you can get them for like a buck on DigiKey. So I always throw one in every now and again. And finally, most annoying software engineer habit?

The installation of the tool chain is the hardest part. Make that easier. We're not responsible for that. We complain about it just as much. We complain about it more. Yeah. We're not responsible for that. We complain about it just as much. We complain about it more. Yeah, I spent more time in the last month sweating over software installations than I have anything else.

You've been doing, you installed like STM32's cube IDE or something? Yeah, and I haven't dove into that too much. It's more Xilinx I have a bone to pick with. Yes. Oh, that's like 8 gigs of stuff, if I recall correctly yeah it's up to 20 now actually geez when we last spoke you had a podcast yes i did i was

co-hosting the engineering commons with uh three other engineers and now you don't. Yes, that's true. What made you leave the glamorous and lucrative world of podcasting? How did you get out, Carmen? Tell us. Tell us how to get out. Unfortunately, there was no like big VH1 behind the moment where we had a screaming fit on air and just everything blew up. It was just a boring story. Life got in the way. Couldn't devote time. Episode ideas never came as easy as they used to. We fizzled. I mean, you fizzled after a good long run. Yeah, I think I did it right around 100 episodes. So yeah, it's a long time. And personally, for me at least, it had been about five years. And I think I just said

all I had to say. I felt like I was repeating myself a lot. That's unfortunate. I mean, because it is tough to keep things fresh. And yet, some people kind of need to hear those things over and over again. Oh, yeah, for sure. And I'm not saying I'm done with podcasts forever. Obviously, I'm on this one, but I don't know when I'll have time to do it again. But I would be interested. You've also changed jobs from TI to Wasatch? Wasatch, I think it is. That's what everyone says in the office. Yeah, so I parted ways with TI's about a a month ago and i went to the exact opposite company instead of being large and tens of thousands of people it's about 70.

so about seven years ago i worked at a small company doing consulting after working at a another small company doing consulting but but doing the same thing. And they were called Aorus Robotics. Now they're a big medical company doing devices. But at the time, we were doing OCT imaging for ophthalmology. And we were using Wasatch. Which way is it now? Whatever you want to say.

We were using your company's OCT system back then. It was a long time ago, so I think things have probably changed. But when I saw that in the notes, I was like, oh, wow, really? Oh, yeah. Yeah. Unfortunately, I'm not on the OCT side of things, though. I'm on the spectrometer side. But I remember being elbow deep in a big box of optics and things and and playing with all that so that's pretty cool yeah yeah it's awesome my my cube is right across from one of the lead software engineers on that project if it's the same person i probably would recognize the name or his initials js uh possibly yes right. I'll tell them you said hi. So this was actually a big change for you because at TI, you were doing semiconductors, you were doing power chips, which were then used by electrical engineers. So your customers were similar to yourself. Yeah. Yeah. I've made the big shift of all of a sudden not speaking to 100% power electronics engineers. That must be weird.

It is. Yeah, I can't assume anyone knows what I'm saying now. I'm talking to mechanical engineers now, software, firmware EEs, more generalist EEs, physicists, you name it. So, yeah, it's definitely a wider range multidisciplinary team. How did you convince them to hire you? I'm a sweet talker. No, I mean, what carried me through, I think, was kind of my passion. I showed up to the interview and said, you know, I'm a power guy, but I'm looking for a change. You know, I read all these app notes in my free time. I'm just looking for something new. I want to apply these other skills. I want to grow as an engineer. And then it helped that I was able to analyze all the circuits they threw at me. And you've gone from something that is marketed to electrical engineers to something that your work needs to work, but that isn't what the marketing team cares about at all. Yes. And that's been a huge change is one of our assembly techs or the mechanical guy will pull me

over and say, you know, here's, here's the spectrum we're getting. This is what we see on the screen. And then I have to relate that back to what's happening on the PC board instead of someone actually caring what the switch node is doing. They only care if it impacts the final product. How, how is that different at TI? At TI you were more involved with writing the app note. At various points in my career, I've played a hand in the whole stage of IC design from product definition to validation to the app notes and customer support's always in there.

But in that case, the IC was the product, so your customers would ask questions about voltages and currents and noise levels. And here I have to relate everything into spectroscopy terms, and I'm still learning how to do that. One of the things about the products that the W company makes is they're not high volume, right? You're making lab equipment and research equipment. So these are a lot of hand assembled things that is that have been a change for you or does that not impact your role that much not too much no at least not yet that i could say uh it is kind of funny though it shift in thinking you know when you make an ic you make one and you want to sell a billion of

here everything seems to be custom because you can outfit a spectrometer in a million different ways depending on what you're using it for. So everything's kind of like a one-off. Very few customers buy just our off-the-shelf stuff. And even the software, I remember, was custom for what we were doing

and we had back and forth with, oh, we wanted to do this, can we change this? Which is pretty unusual for normal products, yeah. Yeah, yeah. And I think we're trying to get a standard software, but I'm still learning all that. And I suspect you're still learning what spectrometers are and do. What can you share with us?

So I'll preface this whole thing with saying I'm not an expert. Don't at me. But do you guys remember from your physics class, you know, you have an atom and the electrons have various energy states? Okay, so there's stuff called matter. But you guys know how like electrons can sit in various energy states? Yeah. Yeah. um but you guys know how like uh electrons can sit in various energy states yeah yeah uh so the goal of spectroscopy at least our specific type which is called the brahman spectroscopy um you shine a laser on a sample and it could be you know pills or pure chemicals or elements and uh you shine a laser and electrons jump up in energy level and 98 99 some ridiculously

high percentage of the light falls back down with the same wavelength that it jumped from and that's called Rayleigh scattering and we don't care about that we care about the Raman spectrum so that small percentage of electrons that emit a photon when they fall back down that isn't the same wavelength, that's called the Raman pattern. And you can fingerprint essentially anything with that. So it's used in pharmaceuticals a lot to say, oh, is this aspirin that we made really aspirin? Are there any trace elements here we don't want? It's used in agriculture to check food products. It's used for explosive testing and security applications. And it's also used for counterfeit

detection with certain bills. If they put X you know, X percentage of some element in there as a check, you can use the spectrometer to figure out if the money is real. And it's also used in the high-end booze industry. So have you already signed up for that customer visit? My name's on the list, but you know, I'm the low man on the totem pole, so it's going to be a while. That check in whiskey? Yeah, yeah. So if're making a like a really high-end scotch that goes for a couple hundred dollars a bottle or more um and a bar is selling it by the glass you can do a spot check if you have a portable spectrometer uh that the the bar isn't refilling it with cheap whiskey and selling uh you know for the same price as your high-end product.

Portable. How portable are these? I mean, if you're in a lab, they're reasonably portable compared to like an electron microscope. But, you know, if you're trying to go bar to bar, whatever, they're not the best then. For my own personal use, it's not practical right now to go to my local bar and say, hey, I just did a spectrogram of Glenn Moranji, and I'm pretty sure this is the worst pub crawl ever. Yeah, no. So there is work on making them more portable because, you know, you can think if you're looking for explosives, you'd want to be on the move. Right. explosives you'd want to be on the move right so you say about raleigh scattering and atoms and electrons does that mean you can tell the difference between like two different sugars that are just carbons and hydrogens or do you only get the list of elements um i, not an expert, but I think it would depend on your database that you're

comparing against. You should be able to look at sugar versus salt, obviously, and see, oh, there's peaks at various wavelengths in one and not the other, but I don't know how far in depth you can get there. Could you tell the difference between sugar in the raw and regular white sugar? I think if I remember correctly, you can, because it's not just a single atom responding. It's the... It's the molecule?

The bonds between the molecules that vibrate differently. And so if you have different isomers and things, they're going to be different. Yeah, yeah. Next time I'm on, I'll give you a way better answer.'s fine that's why we have chris here there we go between the two of us we got our bases covered yeah so it's really cool because you know it's it's total whole new world to me so um you know like our our techs have all gone through like laser and optics programs and then you get people who've been designing these things for years.

And there's so much you can do with it. And it's, yeah, my eyes are just popping all the time. Hopefully behind laser safety glasses. Exactly. Yeah. Yeah. Safety first. After the hazing, of course, then you're like, hey, look in this tube.

You still got the other eye. Yeah. Yeah. So if I remember how these things are structured, it's, it's's you've got a laser, you've got a bunch of optics, you've got some sort of, I'm thinking of the OCT stuff, so this could be wrong, some sort of diffraction grading or something. Yep, yep. That's still in a spectrometer. And then a camera and then all the power and stuff to run all that. So what are you working on? Uh, so my, I, I started the detector that's

picking up the photons and then go back through from there up to the software layer, basically. Uh, so, cause I'm coming from the analog and power world, I've been doing, you know, those kinds of problems and chasing down noise and ground loops and understanding how the architecture works. Uh, but then I'm going to start picking up some more firmware projects as the time goes on. I thought you said software people have cooties. I still think that, but I want to diversify a little bit. That was part of the reason that drove my shift to get out of semiconductors. I took a look around one day and realized I was

getting really niche. And the days kind of got a little samey. I was always answering the same questions, solving roughly the same problems, and I wanted to do more. So I realized I should probably work on my programming skills a little bit. And, you know, let's explore some new options. What did you do to work on your programming skills so far not a heck of a lot okay um i've been brushing up on my python uh just trying to automate more stuff with my oscilloscope my test gear um you know talking to it over various libraries and taking measurements um so that's been a big help. And then sitting right next to me, I got an STM board and some soil sensors. And I figure I'll try to do something in my garden this

year as I plant it. Because listening to this wonderful podcast and several others, just getting something working and reading a sensor is the first step. So I'm going to try that approach. That makes sense. Does that matter for work? Or is work pretty happy with just keeping you electrical? For right now, they're just happy with the electrical stuff because there's a laundry list of things that have to get done. But definitely, they'll give me more work if I can do it. So yeah, being able to write at least a first pass or debug issues with the STM32s that we work with will make me more valuable. And then someday, eventually, I could do fpga stuff but i'm real new to that it's different and it's weird

because fpgas are sometimes sometimes they're considered hardware and sometimes they're considered software and neither hardware engineers nor software engineers really know what they're doing at least not at first right in there so yeah i've designed a lot of FPGA power supplies, but never anything past that. I've done some very small blocks, like we did an I2C master that we handed off to the electric engineer to incorporate a lot of other stuff that was going in the FPGA. But yeah, it's definitely a different way of thinking for us software people because it's like everything happens at once. Yeah. But the code doesn't look like that. It looks like our

normal codes. You read it and you go, okay, wait a minute. This happens, then this, and no, not at all. Yeah, the only FPGI stuff I did was in school, and we had to use the schematic editor. It wasn't Verilog or VHDL. Yeah.

That's different. Yeah. Yeah. I'm, I'm very behind the times, but yeah, no. So I'm going to start off in the analog world and,

you know, debug some PCBs, design my own. And then, uh, we also hired some firmware guys to not long ago and they're going to, um, port and,

uh, old code to new board or something like that i'm just getting the project specs now and part of my job will be you know debugging it and making sure it runs right and giving them feedback so they can fix things and kind of learn things that way too on top of uh checking when my plants need watering are Are you worried about that? I mean... My plants? Nah, I can get new ones. They're cheap. I meant working with the more interdisciplinary team, with the software engineers,

because we do tend to have our own worlds. No, I mean, it's been great so far. I like it. You know, different challenges every day. That's good. Yeah. I'm sure I'll have a list of frustrations at some point, but everyone's great so far.

And yeah, I'm really liking it. It's great to come in every day and not already kind of know what I'm doing. And they've been, you know, really helpful, you know,

answering my dumb questions. And they, we have a few different offices, but I'm the only hardware guy in the, um, the Raleigh office here. So they really like having an EE who can debug boards real fast for them. Cause they're only so familiar with the hardware side. So I can say like, yeah, this voltage is what it should be. And this circuit works and, you know, it's your code or whatever. It's a two-way street. Be careful with that. it's your code or whatever it's a two-way street be careful with that it's your code phrase that's true that's true that's it's all the inflection right it's your code

all right no i'm writing that down hold on hot tip don't work on work on tone no sarcasm i found uh they're working in the the optics, that whole region of the field was really cool because there was just so much to learn. Oh, for sure. There was a lot of different disciplines there, right? You were saying the optical people and the electrical people. And there was a lot of crossover. Like, I remember having discussions with those folks. And at other companies where the disciplines are more separated, it was like, well, there's the guy over there. And it was cool to have maybe input into something that I didn't really know necessarily the field very well, but people asked for opinion and stuff.

Yeah. Yeah, I've just sat back once in a while and let, like, you know know two spectroscopists talk and nerd out yeah yeah it's a whole nother wavelength and i just wow try to pick up what i can was that a spectrometry joke it was a it was a pun it was an attempt at one yeah okay just want to make sure do you end up talking to customers at all yet? A little bit. I can only get into that so much. We're doing a little bit of custom work, so there I have some insight.

But if you guys were to buy just an off-the-shelf part, no, I wouldn't talk to you. There's other support people. Because you would say, I'm not getting the spectrum IC. And I would say, huh, that's cool. What do you want me to do about it? But that's kind of different from your TI job where you ended up having to do some customer interfacing. I wish it was just some. It was a ton of customer interfacing. I think that touches almost every level of the IC design, you know, product flow. Just about everyone talks to customers at some point. Well, and it's harder. I mean, now

you're going to talk to a physical chemist? In what language? Exactly. Yeah. And that's why there's the, you know, the people trained for that. And I'll step into that if I have to one day. But yeah, right now, thankfully, I don't have to just, you know, jump into the deep end there. No trial by fire. I mean, you have to describe what a spectrometer is on a podcast. So, you know, there's a little fire. A little bit. Yeah, true.

Maybe I'll, maybe when I get that first booze customer I'll go do the pub crawl with them and sample. That was one of the things with interdisciplinary teams or mixed discipline teams, multidisciplinary teams, whatever the is, is that because there is some gap, sometimes there is a little bit of fire. Have you had any adventures so far? Thankfully, no. No fires yet, but it's possible. See, the problem with this stuff is these lasers are very low power.

Yeah, they're not kilowatt lasers. That's sad. I'm reliant that the lasers I had were 25 watt. You could have some fun with those. Fun. Fun. For various definitions. You could have some accidents with those, but these, yeah. You could do some damage, I think, if you shined certain wavelengths with your eyes for a long time.

You're not trying to cook the sugar that you're identifying and measuring. You're not turning it into caramel. Different product line. Yeah, yeah. You're not turning it into caramel. Different product line. Yeah, yeah. You should consider it, though. I'll look it up. I actually know, I think... Laser creme brulee.

EMSL should do... There's definitely a market for that. A laser plotter. Imagine creme brulee with, like, your name spelled in it. Ooh, that would be cool. I'll hijack some samples tomorrow when I go in and get a Supercon booth or something next year. Move aside, Dippin' Dots. So what do you want to be when you grow up?

I mean, this was a big shift. Yeah, it was a very big shift um i'm still trying to figure that out uh i have like the general picture is you know someone gives me a a large budget and says go play with stuff you know whether that's uh you know the spectroscopy lasers or some circuit I think is interesting. Yeah, just kind of having a position where I can look at the big picture from the customer view to product definition and R&D and spot a need and design a product or a circuit or whatever, come up with something new. And I thought I was going to get into that in the IC industry, you know, in power. You're like, oh, let's do a new control scheme or a whole new switching topology. And it doesn't happen at a very fast pace.

So that was part of the reason I jumped ship was to, you know, iterate more. And now you're a generalist instead of a specialist moving in that direction yeah yeah but it's good uh what's the valve structure where they they want you to have a you know one specific focus and then branch out it's like the t structure or something yeah yeah uh so yeah i can definitely say i dove pretty deep in power and then tangentially analog circuits from there. And now I'm spreading out. I know which end of the Dremel to hold and not hurt myself. I can use a screwdriver, righty-tighty, lefty-loosey, starting to do some software and, yeah, branch out.

For me, I think I went the other way. And I have gone back and forth. But I definitely started more on the breadth side, and it took me a while to get to depth. Yeah, I just kind of stumbled into the depth first because I liked semiconductors, and I read a lot of Jim Williams and Bob Pease. I always thought I'd get that generalist, work on all the circuits, and then fell into power. And it was really cool, but, you know, I still wanted to see the other side of things.

It's good to hear from somebody who's deciding to try other things because it's easy to get stuck and to think you can't jump ship. Yeah, it definitely made me nervous. And, you know, I'm only a month in, I i'm a little nervous but uh it's cool to you know push yourself and do something you didn't know you could do so if you had a podcast who who would you want to invite oh good call uh what's this podcast on anything i want or does it gotta be still engineering i can be on anything you want i don't know if you've heard of podcasts i actually just started listening to a podcast it was about origami it was very exciting to learn that there was podcasts about things i was interested in oh yeah there's podcasts about everything it's i i listened to a few shaving

podcasts back in the day actually yes i got really into the whole you know shave safety razor and everything and got into all that. And yeah, those podcasts got old pretty quick. There's only so much you can say about certain topics. Like, yeah, yeah. You're smelling soap on air.

It doesn't help anything. Yeah. No, seriously, if you were going to do another podcast and I, I don't know if you, you would be interested in doing engineering or would you want to do a story time for your kid? That would be something cool.

Yeah. Yeah. Maybe if we both got into D&D one day, we can do our adventures for the whole internet to hear. Don't tempt me. There's a lot of D&D podcasts. That's not, yeah. There's a lot of engineering podcasts now,. There's a lot of engineering podcasts now, too, and not all podcasts. I've been hearing more and more, yeah.

What else do you listen to? I listen to a ton of podcasts. Engineering, though, the big ones, you guys, Amp Hour, Macrofab. Altium, actually, the PCB software, I'll plug them real quick. Their podcast is actually pretty good, on track. They talk to Altium guys to plug different aspects of the software, but also signal power integrity experts. They had an actually surprisingly interesting podcast on different PCB finishes and what they're good for and why you'd want one over the other stuff i never would have thought about uh really

interesting yeah it's cool when the companies sometimes do that as long as they don't stick to marketing yeah this one i will say branches out quite a bit and um yeah as far as corporate podcasts and stuff this one's really good let's's see. Pulling out my phone here. What else do I listen to? Oh, yeah. Shout out to Alvaro and the Unnamed Reverse Engineering Podcast. Yes, yes. Let's see. Then a few D&D

podcasts. He can give me a list of those offline. All right. Fair enough. Oh, got a shout out to John Chigi and Causality. That one's awesome. I don't think I've heard of that one. No. You know who John Chigi is, right? No.

He's the guy who does the Causality podcast. I actually was on Causality a few times back in the day. Yeah. He goes through like engineering disasters and what caused them, what went wrong, what did they learn from them. If it was a big industry-wide thing like an aviation crash or something, what regulations came out of it. It's really interesting. He does a lot of research. It's awesome. Wow. Should book him.

He'd probably be more than happy to come on. Let's see. Non-engineering, what do I listen to? Oh, yeah. I read a lot of Serious Eats, the cooking website. So I listen to their podcast, Special Sauce. Oh. I listen to Splendid Table.

I should see about Special Sauce. It's really good. Let's see. What else do I listen to? When do you have time to listen to all these podcasts? Yeah. It depends on the week. I'm pretty behind right now.

If I'm doing something, I can kind of zen out and lay some PCB track down for simple stuff. I can get through a lot of podcasts. I had to rework a couple of boards, boards so sat down under the soldering iron and had my headphones in but if i actually have to think then i don't listen to podcasts something i find is missing from software is is sort of not menial labor but the the electronics there's always a time to go with a board and sit with the soldering iron and do a little bit and have a few minutes of kind of zen out but you're still working right yeah software it's you know that the reason i've been listening to more podcasts but but that's different that's that's that's not software that's i don't know what you're doing that's data collection i've been cleaning a lot

of data it's super boring and yet i cannot look away from the screen. Zoe, are you retagging your MP3 collection? No. This is for machine learning. Oh, okay. Why don't you just let the machine learn how to do it? I've tried to explain that. Let the machine clean its own room.

I've tried to explain that, but nobody will listen to me. How good is your son at cleaning his own room? He's only nine and a half months. He's not good at, he's good at making messes. Yeah, and my machine learning thing is only about a year old. It just doesn't know much. We need to train a different neural network to help train the other one to clean the,

this is easy. We just need four neural networks, each of which. Neural networks all the way down. Exactly. When do you think we'll get to the singularity? And do you think the robot overlords will be kind or just destroy us? It's a lightning round question. I know, but.

I don't know about the singularity. I'm just waiting for self-driving cars. I hate driving to and from work. Yes. The commuting sucks. I don't even have that bad a one. It's only 20, 25 minutes. Oh, and that's another place for podcasts.

Exactly, yeah. Any other podcasts you want to mention? I listen to a couple of the ones from the Back to Work Network. Or not the Back to Work Network, 5x5. So, yeah, Back to Work and then all the various ones Merlin and Dan do. Chris is nodding. He listens to some of those, too. I dropped off those, but I used to listen to a lot of those.

Yeah, I go through spurts, and I'd add and drop at various times. But in general, over the years, I've listened to a lot of that stuff. It was kind of what got me to suggest that we do this podcast. I was listening to way too many of those. I was like, we should try that. After a while, I'd have to just hit move on to the next one, mark this one as played. If they go too deep off into some

Mac app, I don't care. Or comics. Hit or miss. Depends on which one. I'm not too big into the Marvel, but I've actually gotten into the DC universe. Literally a podcast about everything. This one is, yeah. Everything but the kitchen sink. I wanted to mention a podcast, a story podcast, because I do sometimes listen to those.

And I couldn't remember the name of it. It's called Tides. And I remembered that there was this thing about snails in it so i typed podcast snails and you know there's a lot of stuff out there did you find some podcasts about snails i found some podcasts about snails oh great it's just well what is the number one snail podcast? Is that on iTunes? Do they have a separate category for snails? Ah, oh, this time I typed snail instead of snails, and I got a different list. This podcast is about just specific snails. The Snail Cast. I don't think that one's actually about snails.

If you're a true fan, you listen to that one at half speed. That's the other thing. Anyway, sorry. Renting it back on top of it. Yeah, yeah. You guys want to know a hot tip for learning about something new? Go for it. Yes. Alright, I've used this, you know, previous jobs and definitely this one where I'm new to just about everything. I find if you search, this is just a general Google search thing you want to learn about app note. So spectroscopy app note, um, uh, thermoelectric cooler app note,

labor, laser driving app note. Um, and you just, no, someone's probably written one. Um, so, you know, this churns up the expected stuff, you know, if it's a electronics one, you know, linear tech, TI, analog devices, all that stuff. Um, but if you just do a Google search for topic app note, you find a lot of random little companies you haven't heard of before. Um, And just start clicking on them and skim through them. And you'll find a lot of crap, you know, poor English, broken sentences, whatever. But eventually, if you get to the bottom, save the good ones, but always check the references. And after a few clicks, you'll start seeing stuff everybody's picking out. And then go look for that.

Oh, yeah. you'll start seeing stuff everybody's picking out uh and then go look for that oh yeah cross reference the references to find who's who's got the best reference and then just go there yeah i've gone down quite a few rabbit holes that way on various topics and yeah it's helped me come up to speed pretty quickly that's a great suggestion i i think i myself forget to look for the app notes for things i'm actually working with and already have yeah like oh here's a great suggestion. I think I myself forget to look for the app notes for things I'm actually working with and already have. Like, oh, here's a data sheet for something and display driver or something. But if you go to the app note, they tell you how to use it. Yeah. Yeah.

Tying it back to spectroscopy, the thermal electric coolers, you use them in Raman spectroscopy because if you remember, geez, before we got onto shaving and snail podcasts, you're looking for like less than 1% of the light that scatters off when you shoot the laser at your specimen. So it's a, you want a very low noise environment. And if you cool down your detector, you'll get a lot less detector noise. So you can get a better signal in certain applications. So I'd never really worked with thermoelectric coolers before, but it's pretty close to switching regulars, actually. And I never would have known that if I didn't start searching for app notes on how to drive them and how to use them. And those have to be like PID controlled, right? Yeah, usually they're PID.

How cool do you have to make things? It depends. Some barely cooler than room temperature, like room temperatures 23-ish Celsius. So some go down to like 15 Celsius and some go down to like minus 15 Celsius. It all depends on what wavelength you're looking at and what you're trying to measure. So, for example, if you're trying to measure, like, the color spectrum of LEDs or whatever, you wouldn't cool at all because the detector noise isn't your limiting factor. So, why would you pay extra to cool your detector down? Are these, like, Peltier's? How does this work? Yeah. Yeah, yeah. Peltier cools.

Coolers. Well, then I'm going to stop typing thermoelectric. You put electricity in, and it separates the hot from the cold atoms, and it shoves them to one side. I'm making this up. And it feeds a little demon. There's a little demon in Maxwell, and he's got little tweezers. Yeah.

You just pluck them out, and you only get the cold electrons on one side. I actually don't remember how they work at all. Yeah. You just pluck them out, and you only get the cold electrons on one side. I actually don't remember how they work at all. You know, you apply a voltage, and you get a hot and a cold, and you can switch the voltage to get a cold and a hot. You switch it fast enough, and you average to a temperature. Oh. Yeah.

Ow. I think that just hurt my brain. Did you hear the pop? I think there was a pop. I did. I did crack my knuckles. Maybe that's what you heard. I'm going to ask you a very specific question.

You may not know the answer. All right. With the detectors we were using at all these jobs, there was a thing called a transimpedance amplifier that everybody kept talking about. And I never bothered to look it up. It's so fake. It sounded really cool, but I don't know what it was. Do you know what it is and what it's for? Yeah, yeah. Trans impedance would be current in, voltage out,

because you get voltage over current for the transfer function. So it's an impedance, a resistance. There's also trans conductance, which is voltage in, current out. Oh, that's... Yeah. That's disappointingly simple. Yes, that's pretty good. So just based on the loose explanation yeah depending on what detector they were using it could have outputted a current signal versus a voltage signal so you just have to turn that

current into a voltage and then you can put it through an adc because you can't put current through an adc you can filtering and some game yeah See, I probably had watched too much Star Trek as a kid, so it sounded like transwarp drive to me. Yeah, you think it's going to be some cool space-age thing, but it's just like the photodiode op-amp circuit. That's the transimpedance amp outlier. All right. Okay, now I'm really disappointed.

Because your signal from the photodiode is a current, and it just goes through a resistor, and then on the other side, you have a voltage. But there's an op-amp in the middle. I have another question. All right. This one also sort of from a movie. So I saw in this movie about a flux capacitor. What exactly does that do?

That's time travel. That's a different area of physics. I'm not in that one. Question for you embedded software guys. I saw this movie. They wanted to hack the planet. And when they were hacking, you saw a city view of the computer. Is that what you guys do when you write firmware?

No, we're way past that now. Oh, okay. Yeah, it's all VR goggles. I'm in VR goggles nine hours a day now. Okay, are you talking to me in VR goggles? Give me nine minutes. Are you watching the audio files go right by you as we talk? Yeah. It drops in little ASCII things, and I can see things in the ASCII.

Oh, man, you're in the Matrix right now. You guys are on another level. I'm still using a GUI, and I'm way back in the 80s. No, man, I sit in front of terminal windows and look at Python all day long. If I'm lucky, if I'm unlucky, I'm writing shell scripts to automate the boring thing that I have to keep typing, which is a bunch of Python scripts. That's actually my Python book is automate the boring stuff with Python. But I'm automating Python with Bash. Is Python the boring stuff? Oh, oh man this book's a recursion it's a rough life i love python because i'm not a programmer so usually i just google python thing i want to do and just like my app note trick someone's already

written a library for me i i love python and i am a programmer. So I think, yeah, no, and I discovered the same thing. It's like, how do I do this? Oh, import. Yeah. Import the thing I want, usually, sometimes. Yeah, that's been great. That's been allowing me to do a lot of things. And then I kind of get to go back and understand the code and look at the documentation and, you know, what the hell does this function do again? But I will say, if you're doing a lot of that, moving to the firmware world will be a rude and very unpleasant awakening.

I can't just type in, like, spectrometer firmware, you know, open source and just copy and paste? Well, actually, there is an open source RAM and spectrometer. There's an open source RCT thing, too. Well, remember, he was one of the finalists for the Hackaday Prize. Right, right, right. Oh, right. Yeah, that was really cool. The RCT or the spectrometer?

The spectrometer. Oh, wow. That's cool. I'll have to look that up. It was a while ago. Yeah, they've been doing the Hackaday Prize for, what, four or five years? Something like that? At least, yeah, I think five five years? Something like that? At least, yeah,

I think five. Four. Something like that, yeah. Long time. But yeah, no, there are things that you can find, but they're... Sorry, his name was Mark Johnson. And the show title is

The First Time I Was Electrocuted. That's something I've never done, actually, in all my years, is electrocute myself. And you worked on power, so that's a pretty good record. It's all been low voltage, though. But I mean, you know, for processor power, it's still one volt at 250 amps. So there's some meat there. I was going to say for for uh finding open source things for firmware you can there's often a lot of stuff like that but it's i find it to be of

dubious quality usually or abandoned or significant amount of work to get it working on your particular board or micro or yeah worked on the previous version of the HAL or works without a HAL or works with a different HAL than the one you have. Awesome. So it's good to know there's still simple fixes and frustrations like there is in hardware. So I have a question for you guys then. I've been given tips like use a spudger and Google App Notes.

In hardware, everyone jokes the first question is, you know, is it plugged in, turn it off and on again. But a lot of times I do make those dumb mistakes. And I forgot to plug in or, you know, connect a wire. And it's not a crazy issue with the circuit. No one's ever seen before. It's something goofy. As I start learning firmware and software, and you know, I'm over here thinking I've overflowed the memory or something. I just should take a step back and what? What's the first dumb step to do when debugging firmware? It depends on the problem.

Are you sure you loaded the code you thought you loaded? That's the first one. Okay, check the file name. Check the file name. Check that you actually downloaded. Yeah. Check that it's actually compiling and you have a new code that you're actually

putting on the thing. And so you change something in the code. So instead of it printing out version 1.0, now you're printing out 1.0. Oh my God, is this actually working?

Okay, okay. Because I've spent, for various reasons, this happens, but there's, there's been times where I've spent an hour or two debugging a problem and I keep changing something to see, does this fix it? Or does this change the behavior and nothing happens? It's always the same. And then you realize after an hour that something went wrong and you weren't downloading any new code to the device, you were just rebooting it. And so of course you haven't solved the bug because

you've done nothing. All right. So make sure I'm actually reprogramming the device you were just rebooting it and so of course you haven't solved the bug because you've done nothing all right so make sure i'm actually reprogramming the device that's a good one which is very close to did you turn it on that is that is a did you turn it on a firmware and the other thing is it is the sensor you're talking to actually physically connected okay and not just i'm good at checking that one i usually I forget things. You'd be surprised. It's unreasonable that nothing works without power, and yet I forget the power plug like three times a week, at least. But now with the software, you also have to check that the I squared C lines are connected. My scope decodes the various buses on the fly, so I can do that pretty easily. What scope do you have?

All right, I'll shout it. I got a Rigel scope, a nice brand new one, maxed out, one that fit perfectly into our budget. That's what you've got, right? Not a high-end one. Yeah, I got a Rigel. I got a pretty mid-range one. Yeah, they were running some pretty good promotions, so I got a Rygo. I got a pretty mid-range one. Yeah, they were running some pretty good promotions. So I got a whole bunch of software packages and upgraded bandwidth.

So there's a lot of scope for not a lot of money. And for 98% of what I do, it's perfect. The SaliAs are nice, too, because it's on your computer. But then sometimes it's better not to have it on your computer. Particularly with power stuff. At least for me, I like having it tied in with the analog because I've used these kind of scopes, not Rigel, but scopes with the logic analyzer built in over the years. And when you need that, you need it to trigger on a specific command and watch some analog voltages. It's such a time saver when you need it.

It's hard, though, because sometimes to set up the logic analyzer part is a pain in the neck, especially on the non-PC-based scopes. And so you have to admit that this problem is big enough and hard enough to invest the time in configuring the logic analyzer to do what you want. In my case, it has been. Yeah. So, yeah, part of my job in a couple careers has been power for Intel processors. And they have their own special bus that talks to the voltage regulator that talks back. And depending on what the computer's doing, it may only send a command, you know, every couple

seconds or, you know, whatever. And if you're having a problem transitioning from one voltage state to another, you got to look for those commands and trying to do it with just analog channels is impossible. You'll waste a ton of time. So spending the, you know, 20 minutes or whatever it is to capture those waveforms triggered on the Intel bus is such a time saver. So do you have any more advice for. I'm thinking,

I'm thinking, I'm thinking, I mean, beyond that, it starts to get into what, what's the, what's the problem trying to do. But, uh'm thinking, I'm thinking. I mean, beyond that, it starts to get into what's the problem. What are you trying to do? You know, if you start having weird things happen, look for memory problems, stack overflows.

Isn't that the website that yells at you for asking a question? That's where you're copying all your Python code from. Yeah, that's where you get your Python code. Buffer overflows, that kind of thing. What else is really common dumb stuff? Well, I don't think that's it. I think it's, I think, what advice do we not normally think of? Or that we give?

I mean, I love map files files and i think that that's one of the things people don't know about especially when you have i can't say i know what a map file is map file is is basically a text file in a somewhat readable format that tells you where all the things are placed in memory and okay if you're having memory problems like you're overflowing a buffer um you may not know where something's coming from but you know what's clobbered and so you can look at this file and say okay this was clobbered what's right before it oh okay um and there's lots of other uses for it yeah i mean it tells you how much code space you're using and how much ram you're using, allocating.

And then you can start looking at where is it putting the stack? Where is it putting the heap? How big are these things? It's in the same directory as your executable. So your hex file or your bin file or whatever you're downloading should also have a map file. If you are doing GCC and you're setting it up yourself, you may have to add the dash-dash-map, dash-dash-list-map. I'm not quite sure what it's called.

But it's a wealth of information that I think is one of those things that embedded software engineers kind of need to know more about than regular software engineers. A regular software engineer probably only ever looks at it in a big crash, and even then, GDP is reading their object file for them. And so it isn't a big deal. But for embedded, you need that offline because you can't touch the memory once it's really wedged. That's handy. Yeah, I'll have to dig into that and see if some of our code now has some I can use or browse. And I think learn to use a debugger. Don't just do printfs. Noted, noted.

I think we have debugging ports on our board. I think I've seen them in a few spots. So I should be able to use that. And there are newer debug methodologies. I mean, most of it comes down to JTAG. But people who say, I can't fit the JTAG on the board are no longer getting any passes from me. Because there are many ways to miniaturize the JTAG connector now and to just use like a tag connect or something to convert it. I love those things.

Oh, like the edge, the board edge connected version? Yeah. You put all your JTA the on the castellated yeah so it's not like that big standard j tag connector anymore it's a that's a littler one it's a littler one that converts to the bigger one okay and they have pogo pin ones they have all kinds of miniature yeah we've used a bunch of different ones that are higher density than the old big ribbon. Interesting. Yeah, all this stuff I never had to think about before. Well, and when you have a scope, you know, it's a decent scope.

At some point, you're going to look at JTAG modules and try to decide if you want one with trace or not. Or if you want one that... or not or if you want one that okay that's the only feature I can think of don't don't buy a trace JTAG unless you have a specific reason you need it okay alright let me get my pencil

alright map file app note and no trace JTAG trace is really useful in certain situations but it's i it was it was absolutely necessary at my last job for at fitbit for certain things um but never before that have i used it and I don't have a reason since. But basically, Trace, if you don't know what it is, it

takes everything that's happening in the processor and just logs it. So it's very nice for debugging because you can see a complete history of what the processor was doing. It's like a little black box. Exactly. And you can run back and forth and do all kinds of crazy stuff

that you wouldn't normally be able to do with a normal debugger so like if something crashes you can just go okay well let's just walk it back let's see what happened and that's where it broke it's good for power optimization too

because you can figure out exactly which functions are taking the most amount of time okay yeah it's good but there's other ways to do profiling yes Because you can figure out exactly which functions are taking the most amount of time. Okay. Yeah. But there's other ways to do profiling. Yes.

You shouldn't need these. I mean, definitely you shouldn't need trace. Most people go their whole lives without ever having to deal with it. But it is kind of like what I was saying with the logic analyzer is once you, once you need it, you really need it. And until you need it, you think, oh, it's too much of a pain. I don't want to bother to set it up. And then you realize you were wrong the whole time. You should have done it from the start. Yeah. Once you're converted, you know, that's all you want. Yeah. So for your personal projects, what do you, what do you want to use as a platform? What are you looking at?

I have to find some time to do a lot of personal projects. But, no, I picked the STM32. I got a Nucleo board in front of me, just because I know we use some STM32s at work, so I figured I'd start there. Yep, the IBM of embedded microcontrollers. Yeah, because it's a personal project, and I want to automate some stuff in my garden. But also in my free time, I kind of want to do other stuff than circuits.

So I don't mind the garden being old school and not roboticized. But some kind of soil sensor and a big flashing light that says water these plants would be kind of cool yeah i don't think you're alone in that either although i wonder if when your kids a little bit older if it won't just be flashing lights and big buttons yeah toys i have to try and get him involved in these things. Kids toys make too much noise. Well, not if you make them yourself. That's true. That's true.

I'll get him in on it and he can make his own toys. Yeah, at some point he's just playing with a rock. So, you know. Right now he likes wrappers and boxes. And rocks. It's a pretty cheap date, yeah. Do you have a kid or a cat? Sometimes I wonder. He's really good at feeding the dog goldfish crackers.

He just learned that one the other day. I bet the dog's thrilled. Oh, she's, yeah, his best friend when it comes to be mealtime. Okay, no, you were asking us questions. I remember that. That's what was happening before after snail podcast. Yeah. Yet another tangent. Do you have any others? I'm sure I'll ping you once. Yeah. Once I start writing some code.

You actually did. Nothing off the top of my head. I did. Yeah. I was, I was amused. start writing some code you actually did nothing off the top of my head i did yeah i was i was amused it was like you said you were gonna start working at the new company soon and you might have questions and i was like oh sure and then you emailed and it was like you just started like today yeah i think it was on like literally day four maybe five and And yeah, I couldn't, again, the tool chain question, I couldn't figure out what the heck to download. STM cube IDE. But if you search, you do get a lot of choices.

It's a mess. And of course, ST gives you cubes separately and other junk, and you can choose all sorts of permutations. I have cube MX right now on my desktop. Oh, see, that's different. Oh, oh geez i screwed up already cube ide yeah so so cube mx is the thing that allows you to set up the processor and it'll generate the boilerplate startup code and stuff um cube ide has that embedded in it but it's also the whole development environment so you can do your code and the and the chip setup stuff in the same window kind of thing it's a little easier okay noted cube mx is

nice if you're using a different tool chain for for building your software so it can generate like a gcc make file and stuff if you're using gcc separately. CubeID still uses GCC, but it's under the covers. It's a whole new world. It's a mess. It's a real mess. Yeah. But the nice thing about the CubeID is it's from ST. It's updated. It's the vendor tool. They pick the version of

GCC and GDB that works and then they just include it. You don't have to worry about paths and where is it on my computer and stuff. The downside is it on my computer and stuff. The downside is it's kind of a lumbering beast. I'm an analog curmudgeon. I need all the help I can get. So if the vendor tool chain is the least amount of friction,

that's what I'll go with. Yep. That's what I've been doing with my personal project. Yeah. I suppose the analog would be, you know, you guys looking at a voltage regulator app note i'm like i'm not gonna use that inductor i'll do that's a stupid capacitor i'm gonna pick my own and some people just straight copy the app note yep some people keep staring at the app note going um okay if i do this and I get it wrong, how many fires will I make? Not just will I, but how many?

How many can I make? No, I don't usually get to the can I. Although I do sometimes have a thing. I have, you know, capacitors sometimes. It's fun to watch them blow up. I don't know. I've smoked a lot of capacitors. That sounds different

oh no you always get some real interesting sounds and smells like um if you switch the positive and negative on your power supply and it uh you have the current limit set real high because it's a high power application and it just dumps that all into the board reverse polarity and not good stuff i have never in my life blown up a capacitor i must not be doing enough oh what am i we have some in the garage i don't want to do it intentionally now it's more fun if it's an accident oh no you should we should do it it's totally fun all right live on air right now. Alicia and Chris are going to get their power supply.

It's cold out there right now. What is it in California? It's in the 50s here, man. It's really brutal. It was 70 the day before yesterday, so it's been a real change. I'm in North Carolina. I can't complain too much. But it does drop to the 20s overnight.

By the way, we were supposed to set up a lunch for you in Svack. I don't remember what. Oh, yeah. What day is good for you? Oh, we are. Actually, yeah. We're getting lunch next week. All right.

That's good. Yeah. We've had a few before, and we've gone out to dinner with Dave Vandenbott. I don't know. You guys haven't interviewed him, have you? I don't think so. Although it's possible. He's been on the Empire three, four times.

He's been on the Engineering Commons twice. He's done the podcast circuit. He's here in Raleigh, too. Okay, now we have two people to look up. Yep. All right, well, I think I heard our dinner bell go off. It's the dinner robot? I think the dinner robot has told us it's time to go eat.

Okay. All right, great. Yeah, I'll let you guys go. This has been awesome. Do you have any thoughts you'd like to leave us with? Nothing too profound. No, just, you know, if you're out there and you're wondering, should you make a big change? I say go for it.

Stay curious. And I wouldn't call myself an expert on anything, even though I have nine years in power. And somehow I'm managing. Yeah, I like that message. Our guest has been Carmen Parisi, senior electrical engineer at Wasatch Photonics and former host of the Embedded Commons podcast, which of course you can still listen to. Yep, everything's up online. Thanks, Carmen.

All right, thank you guys. Thank you to Christopher for producing and co-hosting, and thank you for listening. You can always contact us at show at embedded.fm or hit the contact link on embedded.fm. Or you can send us messages over Patreon if you're a patron. And now a quote to leave you with from Jorge Luis Burgas. So plant your own gardens and decorate your own soul instead of waiting for someone to bring you flowers. Embedded is an independently produced radio show that focuses on the many aspects of engineering. It is a production of Logical Elegance, an embedded software consulting company in California. If there are advertisements in the show, we did not put them there and do not receive money from them.

At this time, our sponsors are Logical Elegance and listeners like you.