almost ready to submit!!!

Finished and I’m getting ready to submit the project for review! I took some new renders and also did the prices part for the project items. I have everything here now and it is looking nice.

Like I said before, I left some points to see and change better on the case when I have everything in my hands, so I can prototype 🤖 it better.

I don’t have much skill with soldering, so I really hope I can get the Top PCBA. This is all really cool and interesting that I was able to do, and I’m really happy about it.

At the end I also organized things a little more and changed stuff to track what is needed in the repo , help anyone that wants to look at it or reproduce the project somehow. The
System architecture section got really cool in the translation and organization, AI did that. I don’t really know how to make this type of diagram myself yet :/

I hope I didn’t forget anything and that the organization I made, and later asked to improve, documents the state well.

Basically I:

I finished organizing the repo, checking the prices, calculations, etc., taking screenshots , I kept iterating in the repo, editing, moving things, that kind of organization 🗂️

more docs things right these

Well, I tidied up the README and organized the folders and file names better. The repo in general .

The other image is because I think I didn’t show it properly here and I was reposting, checking items and analyzing anything left there on EasyEDA

its possible only four images…

I fiddled a bit more with these items, files 📁 , repo organization , stuff like that . Also, I took screenshots of some views from the project’s mechanical render…
I’m not really focusing on any one part, but I tinker a bit here and if I remember something I should have done or think could be better, I take care of it. Like parts of the firmware, code, repo organization, images, and the internal side of the apps: CAD and EDA. Basically, that’s it, to sum up.

I forgot to put the others Lapse

I’m done here! The two files we worked on are finished.

Now we have the initializer and use of the MCP23017 (I’ll just write MCP from now on in this log). The addresses have been mapped and the registers too. You can see what happens along with the logic of keeping the line in the LOW state and the rest HIGH. In this part, it’s necessary to use masks and bitwise operations to see how it works. It stays in a loop analyzing what was pressed and, of course, if something strange happens outside the configuration, it throws an error saying UNKNOWN.

I made a lot more progress on the firmware!!

As I said, the functions, declarations, and general use of the ESP-IDF are standardized, so it’s simple to build by consulting materials: docs, example codes, usage with some peripherals and components, AI, etc.

In this case, the use of I2C and SPI communication is also well organized. The drivers for the components I use don’t give you headaches for this.

Previously, we had already been working on the ‘macros’ part, or rather the project HAL, even without knowing what it was. In the second image, there’s the keymap, made as a 7x7 matrix for the calculator - what the keys do. That’s already connected to the names of each key. The scanning was done, a loop that identifies which keys were used, and the rendering of the items on the display . Taking the data buffers and transitioning the communication with the ESP about what was pressed and how it will relay that to the screen.

I think it’s interesting to mention:

Explaining how key reading works with the MCP: each key is assigned by the matrix indices - row x column. The columns are set up with internal pull-up resistors, so when inactive they measure 3.3V (level 1). The ESP pulls the row to 0V (level 0), and when a key is pressed, it knows it because it reads !=0 from the column and matches it with the row it’s accessing to get the unique identifier for that key. That way you know 100% for sure what was pressed. And this cycle keeps running every few milliseconds.

I got started at the base and analyzed the HAL (Hardware Abstraction Layer) of the project. I was studying some things, basically the functions and how to use the framework. I noticed that the functions are pretty standardized, like how to use registers, handle addressing, and use macros, directives/variables.

Basically, the headers I was going to use and docs and things like that I was checking out, now speaking exclusively about the Aliexpress display that was in Mandarin (some of the ones I saw while researching) and offline I was tinkering and looking at some things, asking AI to explain and check things to move forward. I got the variant I was going to use, and later on it got analyzed and the board_pins.h ended up fine.
I also started messing with the keypad, just defining it and later on how to monitor and use the identifiers of each key .

I started looking more closely at the firmware part! Actually, I’ve never worked with ESP-IDF, so I’m going to take another good look at it and refresh myself on things I might not be prepared enough for.

I grabbed it to check out and study with the example code "Blynk Led" —the one in the picture is a different one. It’s to control the internal LED with the BOOT button.

I’ve seen a bit, and RTOS seems really cool, here it’s FreeRTOS. But you need to pay attention and be aware, understand how things work. I feel like it’s going to be a long and tough road… but let’s go! The good thing is that we have AI to help us out and clarify points, test stuff, and so on.

I finished working on a few more things in the Assembly. I was aligning the cut positions in the view (remove extrude) and it took a while because the Mates, Fastened Mates created kind of restrictions that didn’t allow certain parts to move. Also put the 3D for the display. Then at the end I made the first commit to the repo!

Lapse

I just tweaked the parts a little more and then went to the Assembly . The PCB step was 50MB and took a while to load, and then when I tried to insert it, it didn’t load everything right after choosing a square option with a symbol similar to the Part Studio one, but eventually it worked. I used some fastened mates to join the items, and since this isn’t constant and there are hollow parts where you want to center/place in a specific spot, the Mate Connector doesn’t apply, so I used something like the transform to select the piece. That’s how I kept going until I got here! In the middle of the work, I had to edit the Part Studio because some points differed from the positioning and mounted parts of the PCB, like the USB and microSD.

That’s it—work, found mistakes, fix them, and move forward!

Lapse

What happened with the keypad being wrong was that in the sketch I dimensioned it incorrectly (swapped height for width), and that’s why some of the blocks went beyond the edges and weren’t symmetrical. I imagined that might be it and fixed it. I made a rectangle there thinking about maybe adding a simple lens protection for the camera. After that, a rough idea of the cover that is usually used to protect the front part. Besides that, just a simple fillet on the sides. I did that, but I admit I might change this idea and its overall shape. Once I have everything in hand, I’ll update it to have a proper enclosure.

Lapse

I finished the rear shell for the case. I made some bosses for the PCB fitting. It’s noticeable that the top right one is misaligned. I mentioned before that it’s because the traces and components were occupying that space, and when I remodeled it to 84x85, the hole slots didn’t go along—they got lost on the canvas. Later I remembered, organized, and figured out a way to put them like this. Only one ended up 'strange,' which is NOT a problem at all.
I made the shell 1.8mm thick and also cut the slot for the camera.
I left the keypad at 74x80, making a 7x7 feature pattern. This still bugs me because it doesn’t look right and the logic seems to change… I just need more experience and general practice to be ready for different scenarios. Some parts went out of place, but it’s just a matter of testing measurements until it works and maybe checking some guide videos on how to use this feature correctly.

Lapse

I had made a Variable Studio but I end up not using it and don’t remember much. I asked the AI for help since I’ve never really worked with this item and it seemed better for standardizing and putting something defined that is used constantly. But in the end, I remember the quota values and some calculations are simple. Here I did the rear and front parts of the case, I didn’t do the extrusion there but it’s more or less correct. I had been thinking, and as I said, I worked on an idea and previous version, so it was actually pretty easy for that.

I finished placing the holes on the PCB. Actually, I had forgotten… I was focused on finishing and trying to deliver something more professional in the routing part that when I finished I was like "finally did it? Can I export the production files now?" and the happiness and rush of the moment made it happen.
It wasn’t possible to perfectly align the holes and the thought I had because of the traces I needed to place on the PCB, so it stayed like this. Only the top-right one ended up ‘different’.
Now it’s just finishing the PCB and fixing the firmware . I was already checking and testing a bit offline.

I routed a lot more! I tried, tried to optimize so I wouldn’t have to run traces on Inner1, but it really wasn’t going to work. Partly because of the layout and distances of some ICs to the decoupling capacitors, practices for better signal quality. I forgot () and it was already almost done, so I came here to post up to this point

Lapse:
Lapse

Well, what happened. I went to check the price of the PCB – in this case, when exporting the GERBER it showed a price and it was pretty expensive, $33! The other 3 boards I made were within the 100x100 size and, if I’m not mistaken, the most expensive one was listed at $7. Here I’m kind of making different versions and testing, experimenting with what happens with certain configurations (size, traces, layers, type of components, price,…). I was putting it off and taking my time because I wanted to use what I had from the other one, but then I decided to start routing from scratch.

I’ve practically finished the new schematic for the raw camera version. I managed to tidy up and properly organize the layout for this new part. First, I looked over it to see what I could add and prepare in positioning the new components for routing. The camera I left is one that uses an FPC connector. I’ve already put this item in place, arranged it with LDOs, translators, and controllers.

Basically, it’s about handling the signals, how they occur, to make them compatible with the module I took and the rest of what I had already prepared. It got a bit messy because I really need to see how to place each item, and I wanted to make use of a large part of the general layout from the other PCB version…

lapse

I made some changes and was researching a raw version with the OV5640. To make it a more controlled and integrated work, and thinking about reducing it and making the aesthetics of the project look more like what it really is, reducing the thickness, I changed and added a lot of other components and I’m starting to organize them to be on the Bottom Layer. I need more LDOs, capacitors, and resistors.
Simple, right? The passives and ‘controllers’ for electricity, current, voltage, etc. I’m not really knowing much about the impedance part, so I use ChatGPT to guide me, teach things, and think about the structure of the project, the PCB…

Lapse:
Lapse

Well, things were already going pretty well. Basically, I messed around a bit with the routes since the camera pins would be placed behind the PCB, like on the bottom. Then I changed the header layer and rerouted it. Now I managed to remove the header from the 3D preview (since the header was used to make the holes and get an idea of the position of the Arducam pin rows). When I have it in hand, I’ll solder everything directly onto the board.

I was messing around a bit with other things too, in the firmware, doing research, checking data, etc. I used ChatGPT for that and even Antigravity. Then I moved forward with some things, checked, asked about the case. The thing is, I’m busy (or I should be) and couldn’t really do sessions and things of the project that I was working on to save. And then I really ended up forgetting to make the Lapses… a lot of times it was gradual work, bit by bit, and sometimes I’d go a while without touching it because I depended on my parents to call me or something like that, a pause. As far as I remember, I used to be able to edit the Lapses before, but I think it’s better to do it directly.

All of this wasn’t just 17 minutes! I realized that there was a WakaTime extension for EasyEDA and Onshape. At the time (~ about 5 months ago) I didn’t know about Lapse or anything like that and ended up using this. I ended up creating more than one project to test things out, run tests, and see other people’s implementations. And it turned out they came out with very similar names and all. I didn’t remember the name anymore and I think I synced a few incorrectly. I fragmented the time. BUT REALLY MOST OF IT WAS BECAUSE I DIDN’T RECORD, ETC. I think you can add like 3 more hours

Well, I tweaked the PCB design just a little more. Actually, very little—I was just seeing if there was a way to handle the USB traces since these two NETs were 170mm and had a lot of vias. The center already had the other components and trace routing, so I guess there wasn’t much to do. And honestly, I’m not experienced, so I don’t know how to make these things better.
Along with that, there’s roughly an idea of how the project case will look and the items I had already been working on.

Finally the errors disappeared… it turns out you had the custom DRC rules that I set. Then, like 2 or 3 NETs had thickness rules that couldn’t be followed because QFN components and the VSON package type have pads really close to each other (pitch), so there was no way to route 0.8mm traces on components with, like, 0.4mm pitch. Obviously, you had to reduce it to make it work, but I didn’t know by how much and whether it might be wrong at some point, cause problems, or lower the PCB’s lifespan. 😶

I figured out how to change the rules and get things fixed and done. Man, I think I should review it again and really check some automatic NETs from the editor. After all these days without touching this, I noticed there were these NETs without descriptive names. I don’t remember if it’s because I didn’t assign a NET, or if it’s just that connection between those pins or something like that. But overall it’s going to be easy.

I really hope SO MUCH!

Lapse
Lapse

I’m back… I won’t be able to be very active here due to classes and coursework.
I spent a good few hours working on this, but when I checked, it only showed 23 minutes… Maybe I linked the hours to the project incorrectly. Actually, there’s a file with the same name plus “_copy” at the end, and I thought I was already working on the right one.

Well, I actually finished the routing. It turned out there were some copper islands that weren’t connecting to the rest of the pour, so I had to route them too. But to me, some of the ones flagging routing errors (showing ratsnests) didn’t look isolated and should have been able to join the copper plane. Oh well. The center ended up looking messy; that’s where the components—and consequently the traces—are concentrated.
I’ll probably have to find a way to optimize and work on that. Plus, there are 39 DRC errors now…

I hope I can submit this project; I’ve already done two others, but the deadline unfortunately ran out while I was optimizing things, addressing review comments, and dealing with components and budget issues.

Oh, the way development work and hardware projects are recorded has also changed; now there are timelapses available here on Lapse:
Lapse
Lapse
Lapse
Lapse

I kept routing. I made the port expander matrix (the one on the FPC in the center left), the header for the camera, and practically all the ESP pins
I’m still thinking about tidying it up a bit to reduce the number of traces going into the antenna area. There have to be traces coming there, it’s already acceptable as it is, but I will optimize…

That’s kind of it.

I was researching, looking at some more DOCS and just tidying up the last bits to start routing. I used Perplexity for some research and to cross-check data and see the trace widths for the NETs.
I put DRC rules, so it’s much easier to work, because when you activate routing on a NET it automatically sets that thickness and rules in the project.
I routed these more critical parts, power, data, a bit of the display part and stuff like that. Even with plenty of space, it already got a bit messy. This is my third board, so I’m still not that experienced with it.
But this is the way to go. Most likely I won’t have to go back and change components. I just need to finish routing, think about the case and its mechanics

What annoys me the most is having to return/exchange a component or it not being what I expect. In the case of marketplace sales, stores like Aliexpress

What I changed

At the top, there’s the 2x11 header for the ArduCAM module—you just need to plug it in. I also added a few more devices (actually kind of switched around their positions and values).
I hadn’t shown it before, but I imported this 3D model and linked it to the display device.
I did a general cleanup and a lot more research, tests, t hings like that.

I’m doing kinda okay. I kept waiting for the sellers to respond to me. But I didn’t get much, I could only get one reply per day. I would send messages there, but it was like 11hrs over there, outside their working hours. A lot of time for nothing… Also, that display manual/datasheet didn’t come with the dimensions of the display assembly, like the total height and the screen size I needed. I found it on a website, and it said 6mm, so I left that to do the 3D design to import the STEP as a 3D model. Another seller said they didn’t have the manufacturer’s data I needed for the camera to check the dimensions and the bus.

I wanted one that was 3.3V, and usually it’s the common 22 or 24 pins. But random ones from Aliexpress can easily change something in the pinout or order, which breaks everything. Here I’m kind of developing the project blindly. It has to be a really precise version, and that’s difficult… I think I’m going to go with the ArduCam module. It’s not small or as modular as others, but for my case it’s probably okay. (until now YES)

As I said, I was going to send a message to the sellers, but I procrastinated, and by the time I sent it, it was already the weekend there. The sellers are Chinese, so their time zone is 11 hours ahead. Actually, it’s already Monday there now, so I think there’s no need to resend the message since mine is already in the queue (I hope…). Anyway, I messed around a bit more, organized the schematics a little better, added a few things, and thought about, more or less, how the items could/should be distributed on the PCB, so I did that with the images. So, this whole size isn’t totally necessary. It will help when routing, but there really aren’t any problems with that. What I might do is remove the part where the display goes since it will be securely framed in the future case.

They sent the link to the display manual

organizing schematics

Here are the schematic pages I’m organizing:
01_SYSTEM_MCU, 02_POWER_USB, 03_KEYPAD_IO, 04_DISPLAY_STORAGE,
Actually, I’m still missing one more schematic image: 05_CAMERA_TBD.
It seems only four images are allowed per devlog.
That’s it—trying to follow a logical organization, with a left-to-right flow and the power sections included :|

I added a few more components. The thing is, I’m preparing how I’m going to connect the keyboard/MCP of the keys to the circuit, and for that, I’m going to use an FPC for it and other LDO parts. Parts that deal with inductance/decoupling and protection of the circuit components.
I had to research and really look into things, and I’m already checking to find a module/version of the OV5640 with DVP FPC. I had to keep changing and reviewing the practical use of the components and kind of the electrical **‘needs’ **of the setup.

I switched the MCP MCP23017 to a QFN type. I realized that in the end, everything was most likely going to be pretty big. The final product thickness is around 15mm or more.

So now I’m studying and researching to change the components, organization, and then properly design the mechanics in CAD/case so everything stays stable. I didn’t really know the display dimensions. Some are kind of generic, with chip data/technical specifications giving a certain thickness, but most don’t have thickness. Looks like I’ll have to message the Aliexpress seller to get the exact data…

I feel way more lost than I should. It’s annoying because I had already thought a bit at the start and there wasn’t supposed to be so many reconstructions/pivots in the project.

Well, what I kept working on was the case. I was thinking that the right way is to make three separate parts: the back part where a phone case fits (thinking in this case), the front part where there are holes for the display, keys/buttons, etc., and the sliding cover with the tracks. This one, when using a normal calculator, usually fits on the back of the body and when not, it fits on the front, storing and protecting the screen. In my ideas, I asked Gemini to generate these images to help me clarify my thoughts and development. It turned out good!

Another thing is that I noticed the display PCB module isn’t centered relative to the screen at all. Because of this, due to the width of the PCB and the case ending up close to the display module, it can’t be centered since there’s little free area for horizontal movement and the board would already touch the case walls. I made this 3D model to see roughly how it would look, and it looked weird. Not too much, but… The thing is, I usually do some stylizations, drawings, stuff like that.

It would be possible to put a solar panel there, yeah, but I’m not going to go back much to that part of the EDA and rethink the system, mechanics, and circuit logic.

I was supposed to post yesterday but my uncles came over and I didn’t get back to it at night here

I’ve pretty much done the part that holds the PCB. The shell, I recessed inside it, and in the other picture you can see I made an Assembly to see how it was coming along and to have an idea of where the screw holes for the PCB are, and later what I should do on the case.
I haven’t done this yet, but you can see some Mate Connectors in the picture of the case alone (the light blue, slightly grayish part). I tried to make the holes, but the sketch planes and flat planes were on the top face and I couldn't make the holes. I spent some time trying that. I don’t know, but probably if I make a sketch on the bottom of the case I can make the holes.

That’s it, 3D for EasyEDA, I’ve already fixed it in Onshape and I’m doing well with the case part.

🫡🙂

I finished tweaking a few more things in the PCB schematic and getting it ready to export a STEP to use as a base in Onshape. Actually, I made the display way too big, standardized at 10 or 15mm, but it should be something like ~2.8mm. The thing is, when I was moving some items around, it ended up moving the Board Outline too, even though I didn’t select it… so I locked it to prevent that from happening, and I’ve already done the same in Onshape. Now I’ll start on the case and then come back to routing

As I said, I didn’t see the dimensions of the full PCB for the 2.8 inch version, and because of that, I was going to need to change the board (since I don’t want to increase the size of the final product). So I went after the dimensions of the 2.4 inch version and made the rectangle just to represent the space it would take up in the PCB schematic.
A strange thing is that when I went to change the 3D model of the device, instead of changing it, it added another part to it. I had to overlap them so it wouldn’t look so weird there in EasyEDA.
Here is the image of parameters

I was getting ready to start routing, but then I decided to take a quick look at the schematic again and check if the power supply and protection/best practices part of the circuit was all good. I consulted Perplexity and it suggested adding resistors and changing a capacitor on the EN pin of the ESP. I changed it from 100nF to 1uF and added a 10kΩ resistor along with it. I had already added resistors to the SDA and SCL pins and I don’t remember if I changed that..

I also looked up the data for the external components. The display and the battery. I wanted to make a 3D model to make it look better and to get an idea in the EasyEDA rendering. I searched for the display data on the site, on its manufacturer’s website but found nothing. So I googled it and the data came up on a site already known here and there it was fine, that must be it, I needed the thickness in this case ~15mm.

Ah, and I hadn’t noticed that the footprint I made for the display was only for the screen, not the complete module. Then I realized it’s going to be bigger than one of the PCB dimensions I had in mind… It’s supposed to be like that, I hadn’t noticed but in the image I attached with the green arrow it shows that it’s 50x86mm

more for PCB

Here I made some mistakes… I thought I had done the footprints, but after entering the 3D view/rendering of the PCB, I saw that there are only holes in the devices I made (these two in the attached images). I made some area that is seen as non-outline or something like that of the PCB. So I’ll have to fix that 😑

I also made an area on the board that will not have PCB material (FR-4) under the chip/module ESP antenna. This and the 15 mm keepout area around the antenna are to avoid interfering with Wi-Fi and Bluetooth connectivity.

I kept working on the PCB schematic. After tidying it up a bit, I decided to make a device and footprint for the battery I’m going to use. It’s kind of standardized with dimensions of 10x34x50, A type with 2000mAh. That’s it, I made a rectangle of 84 by 165mm and put some holes 5mm away from the edges. I still need to figure out what type I’ll use, whether it’s countersunk or not. But either way, between 2.2 and 3mm.

I’m still working on the schematic… I was finishing organizing things and then messing around with the PCB design at the end. I’m going to call it PCB SCHEMATIC. Earlier I changed the display to a 2.8 inch one because I realized it was going to be too big, and the style I’m going for is kind of like this Casio ClassWiz

Here in the attached images are the drawings of the device and footprint.

Here I tried to fix the NETs of the components and connections. I looked for the datasheets of the devices I’m using (it was easy since I just had to go back to LCSC with the Part Numbers, and on its page, the datasheet is already there). I downloaded them, some in English, others in Asian languages (I think if I remember correctly there was only Mandarin and Korean). I uploaded them to NotebookLM, and it gave me an overview of how things worked and pin data, that kind of stuff.

I saw that I had done some things wrong yesterday, so I had to fix the labels and that kind of stuff.

I created another device—just to simulate the 3.5 LCD display I found on AliExpress. After doing some research and looking at other options, I decided against using Seeed’s S3 Sense. It comes with an OV2640 camera, but I’m going to use an external OV5640 module instead—a 5MP one I bought along with the other two.

I did some research, checked out components and versions in the LCSC library, verified a few things with Perplexity, and added the items to the schematic. It’s still pretty messy, but I’ll clean it up, organize it, and get the NET’s sorted out later.

Well, I started this project here… the idea I had was to make a node that is a ‘calculator’ from scratch. I’m not exactly sure if I’ll include multiple buttons or common functions in it, because the difference with this one is having access to the AI API, and I’m also thinking about adding Wolfram Alpha. It’s going to have a camera to take a picture of the equation, send it to the API, and it returns an image of the solution. It connects to Wi-Fi and sends the data to a server where it’s processed, formatted in LaTeX, and then returns the image.

So I started by putting the components in the EasyEDA schematic and created a simple device that was supposed to be the Seeed Xiao ESP32 S3