12/21/2025 - PCB Design Research

Today, I started learning how do I incorporate ESP 32 into my PCB design. Because I need two cores, and a lot of power I have chosen to base my board on the ESP 32 S3 Wroom 1. I will be using the inbuilt Bluetooth antenna version, to maximize simplicity. After I found a datasheet, I have found Espressif to have REALLY good documentation.

This is my first time making a board without using breakout boards, which is a first for me. I realized that most chips have a recommended layout. This basically shows how to wire that component, it also showed extra components needed to support that chip.

This is the ESP 32 S3 Wroom 1 recommended layout.

Today I made a plan for how my ESP 32 will integrate with the other parts of the PCB.

I am NOT going to just copy this. This setup includes more functionality than I actually need. I have decided to include the UART interface for further simplicity. But I will NOT include the standard USB interface, because while testing with a dev board at home, its serial output is sketchy.

This means I will need the UART chip to use a USB with my board.

After this I spent time researching about PCB assembly. While I am really good at soldering, the chips for this board are SMD, and the pads are facing down, and I don’t want to waste any money. This means I have to stick to the JLC PCB catalog.

After this I started watching videos and reading sources about ESP 32, to get an idea of the project ahead of me.

https://www.youtube.com/watch?v=jiJGbWOSdMo&t=773s

https://docs.espressif.com/projects/esp-hardware-design-guidelines/en/latest/esp32/pcb-layout-design.html

I also started to work on a list of parts for my PCB. I will talk more about this in my next journal update.

Overall today, I mostly had to understand these topics, like finding out what parts of the recommended ESP 32 schematic were actually necessary. I was just finding decisions I had to make, researching, and making the decisions.

12/22/2025 - PCB Parts Research, And some other research

Its winter vacation, so I had nothing to do for the whole day. So while I worked the whole day, I think I pulled around 506 hours, of locked in time. Turns out for PCB assembly, JLC PCB, has a few tiers of parts. 1. are the basic components, these are dirt cheap, but are often simple. Then there are extended components. You need to pay 3$ per loading of each of these, cause a technician has to get a roll of these specially from the storage. The problem is the basic components are pretty limited, in fact, there are 0 capacitors for use. So my plan is that for complicated, hard to solder components like the IMU, Barometer, USB chip, and any small hard to solder component, I’m gonna get done using JLC PCB. But any basic component, like diodes, and 0805 components, like capacitors, and resistors, and the actual ESP 32 myself, I’ll just solder myself, to save money. Plus I will get to learn more.

I also want to make sure that all my parts are not discontinued, and have a long support life. This way if I want to remake this board in the future, I won’t have to redesign for newer components.

After doing this initial research, for about two hours, even talking to JLC PCB support, I have made a MAIN parts list. These are the parts, that are not common parts, and aren’t obvious. Things like resistors, capacitors, and voltage regulators are NOT on this list. To get this list, I had to do a lot of research, talk to a lot of people, and look at JLC PCB, and DigiKey, to check component availability, as JLC PCB actually has a larger collection of components, but a lot of them are discontinued, and won’t be restocked. DigiKey quickly gets rid of discontinued products, so if it’s available on DigiKey, it’s a lot better assurance the part is usable. This process of finding components, and finding alternatives to outdated components, took a staggering three hours.

Main Processor: ESP 32 S3 Wroom 1

Main IMU BMI 270. (only 2$ a piece in SMD form)

Main Barometer is BMP 390.

USB C, for power

CP2102N for UART communication (This chip barely fits my constraints, the original chip was not common, so I had to do a lot of research to get to this chip.)

L8050QLT1G for automatic reset, and other ESP 32 related functions

MCP16311 for converting the high voltage, 12 volts from the lithium-ion battery to 5 volts for the servos. I had to do a lot of searching for this one, because I need one that can handle high voltages, it needs to be efficient, and it needs to handle high amp loads, for high power servos.

TPS7933 for converting the 5 volts to 3.3 volts for the IMU. I chose this sensor because it’s really popular and easy to use.

These are the main components I used.

After this, I got sick and tired of PCB design.

My rocket is going to need some actuators to gimbal the engine. I don’t want to use cheap Amazon SG90 servos, I want good metal geared servos. I went down a rabbit hole on this one. While BPS.Space used those Bluebird servos, those are really expensive, and I don’t need that low profile.

These seem to be a similar profile to a cheap servo, but they are like 40$. Which is not ideal. After this, I spent time ACTUALLY reading datasheets, and other sources of information about different servos. The stronger the servo is, the less I have to gear it down. This means actuation time can be reduced. The problem is that good servos are significantly more expensive.

I have also decided for this rocket to do a unibody design. Instead of having many subassemblies, slid into a cardboard tube, I will have one piece that has everything. This means it can be quickly removed and serviced.

I need to lock in, because, right now with winter vaction, I can probably work alot per day, and complete the design in time. So I plan to work 5 hours a day, because, I am not going anywhere.

12/23/2025 - PCB Design Prep And Intital schematic

I started with working on importing every footprint for all my pcb. Most components don't have inbuilt libraries, in kicad. So I have to import them myself. It took a whole hour to import everything.

For every component, I had to search for a library that has them. Then I had to individualy, import every single symbol, and footprint, and match them together. This propved to be very annoying, and cumbersome.

Often some components, actually did not have a footprint. So I had to actualy, import the package, and then I had to make sure, each pin, correlated, to the actual pin on the chip.

And because there was a'lot of chips I had to do this for, this took me a whole hour. Then after that I started working on the actual Schematic.

I started by connecting, up the USB. I first started by following the recommended, periferal layout. But turns out USB C has some differences, from normal USB. Two of the data lines, have to be connected to ground, via 5.1k resistors. Thankfully, some people, in the rocketry discord pointed this out to me.

This is the completed USB Layout. As you can see, its really similar, to the recommended layout, though, it does have to work with USB C.

After this I started wiring the ESP 32. I have wired the output pins, for the periferals I need, for my rockets flight computer. Like SDA and SCL for my IMU.
I also things liek the auto reset and drt pins, but I also included bridge pads, to manualy perform these operations.

This is the wired up ESP 32.

**

These are the circuit, that automates, things like reset, to that the board is actualy usable. But I did need a way to manually perform these tasks, soo

I added these bridge pads, to allow to manually program the board.

After this, the last thing I wired up is the UART interface, as I had to look at several documents, to figure the best way to wire it, in order to alow the USB to work.

This is the completed UART interface. This is what connects the uart interface from the microcontroller, to the actual USB port. I needed this circuit, as I DELIBERTLY decided not to use the inbuilt usb.

As you can see with my screenshots, I am being pretty, generous, with my decoupling capacitors, as I want to reduce, the change, of having problems, with the FCC, if I ever want to sell these boards.

After this I spent time actually adding the parts, I will need into a BOM, and seeing how much this will cost me in dijikey. I have also decided, to NOT get this board assembled, and to solder it myself, with my schools maker space. This means I HAVE to use relatively big components like 0805, to make my life easier.

I added most of the components, and finding them through dijikey, and finding exact alternatives for some of the components, took some time. I also had to research, the difference, between multiple really similar ESP 32's.

The cost for this PCB alone will probably be in the 100 - 150 range. And for other stuff, that why this project is probably going to cost 300 - 400$.

Oh yeah, I have a name for this computer, RAM R2, the second revision for my Rocket Avionics Module.

Locking in for 4-5 hours every day is actually working pretty well, Hopefully I can keep this up for the whole winter vacation. Hopefully my journal, doesn't seem like i'm lying about the hours.

12/24/2025 - Finished the PCB schematic.

Today, I locked in for alot longer than norma,, hence, the schematic is done.

Here as you can see, the whole pcb has been complete. This step took longer than normal, where it definitely took 5 hours of work.

Today, my freind, warned me that the voltage regulators that I planned to use, would not suply enough amps to run the pcb. This is because the 12 volts, needs to be converted to 5 volts for servos, and that draws 3-4 amps. luckily the replacement's I found are not more expensive than the standard.

Voltage regulator schematic.

After this I moved on to the sensors.

At first I tried using the datasheet, for the BMI 270 IMU. In the end, they turned out to over complicate, it, and did not explain, things like how to enable the pin. This is probably my fault, and not the actual manufacturer. This is why I actually used spark funs breakout board schematic.

The BMP 390, turned out to be a'lot more simple to wire up, and I was able to get it over realy quickly, rather than the hour, it took, to understand how to use the bmi 270. If I had just used spark fun, I would have gotten it over in 10 min.

After this I started working on the Pyrotechnics circuit. I made sure to add pull up resistors to ensure that line noise, couldn't fire the explosives connected to the pyro port. I actually tried for a long time to continuity detection,, but in the end, to avoid wasting time, and since I am the only one using this board, I decided to remove continuity detection, in favor for a more simple layout.

Two things that are really import for me is to have reverse polarity protection for this board, and to have a easy way to switch of the computer. So I used screw terminals to easily switch it off, and three diodes in parallel for reverse polarity detection.

One change i did for the esp 32 circuit, is adding a'lot of contact pads for the pcb. I did this so that, in the future, if I ever need I2c access, or GPIO access, I can wire it easily. For example, lets say I need a gps module, this would solve the problem, as I could easily use these pins.

This flight computer is made to be a final flight computer, so I need to have the ability to expand my board for more functionality.

After this, I shared my board with people who actualy knew what they were doing. While I got a'lot of positive comments, I also found out, I had badly screwed up the BMI270 schematic, even after using SPARKFUNS sheet.

1. the names of the pins on the spark fun schematic, are DIFFERENT frtom the standard.

2. incorrect interpretation of schematic.

Turns out, I had done stuff, like trying too use both 12c modes, and incorrectly enabling i2c, so I had to spend about an hour, fixing a'lot of stuff. The screenshot provided is the final one I made.

After this I ran the DRC, so find any flaws. Never the less, there were 100 flaws. All the screenshots in this journal entry, are close to final, if not close to final. My first design, that I made, had a'lot of flaws. good thing I ran the DRC.

I know this amount of time that I am taking is a bit more than others. I would like to remind you, that this is my first custom SMD board, I only have experience with THT breakout boards, and I am spending a'lot of my time learning. And btw, 5 hours is not my whole time I spent. I have been working on this the whole day. 5 hours is just the time that I have been locked in.

After I completed the schematic I ended up having some annoying footprint problems, apparently I chose the wrong footprint to some of them, causing an error, I think I fixed all of those, after a bit of troubleshooting.

12/25/2025 - Placed all components and attempted routing, several times.

Today, I started routing my whole PCB. I started by making the board cuts. I decided on a 45 by 75 area for electronics, but an additional 6 mm for the ESP 32 pcb antenna. This means the board is actually 45 by 81. This is actually really small for a board this complicated.

After making the initial outline, I started placing all the components onto the board. I did not spend a'lot of time thinking where each component should go, and it was extremely unoptimized.

Here is the intial import components I added. After this I started adding capacitors, next to. components to decouple them.

For my first three routing attempts I tried using a two layer board to save money. The first attempt, had to stoped one hour into the process, as because the components were so unoptimized, it became imposable to rout any major, important connections.

As you can see, crucial lines like I2c, and other stuff, was not prioritized, and it was practically impossible to actually rout all the connections, while making good electrical decisions.

For my second attempt, I prioritized, important connections like I2C, and serial, and any IO connections

Those were the first things I routed. After that I started routing things like vcc, 5 volts and ground.

I made these thick busses, across the boards, that components can connect to. This simplified wiring, alot and clearly, I was going in the right direction. Bu in the end, it still became a rats nest, and after 2.5 hours , and only 30 traces remaining, it was pracitcly impossible to rout anything more. So I decided to start routing a third time.

This time, I started, by routing EVERY signal, not just the major signals. Then I routed ALL the power lines. Then I started continuing routing, but after like 1 hour in, I decided that its still really pad on a two layer board. 1st, I learned what a differential pair is. what I found is that I forgot to use them, for the usb port. so that signal is not usable, as the traces are completely different. and because these lines are like burred in, I would have to redo everything anyway. Also I realized, because the ground plane, was so broken up, since the second layer was necessary, a'lot of the decoupling caps would not be afective, since, the ground wiring was so screwed.

So at the end of this, I moved two a 4 layer pcb. Before I even made it, I starting working, on puting the smaller components in places that make sense, to make wiring a'lot easier. With the 4 layer board, I have planned the bottom plane to be ground. the middle, for power and signals, and the top for pads. This should allow me to give for important signals, space. and make sure my differential pairs work.

I also have decided i'm going to assemble this board myself. I will get all the parts through dijikey, and I will get a 10 cm by 10 cm stencil when I order my PCB. I will use a SMD reflow hot plate, to do the reflow, this will make sure I have even heating. This should save some money.

Hopefully, my allocated budget will include things like a heating plate, and soldering paste, and esentials.

12/26/2025 - Routing the whole 4 layer pcb.

Today I finished the whole PCB, and passed the DRC.

Ass you can see, its a 4 layer pcb. This makes signal routing so much easier, since you can devote 1-2 layers on crucial far signals, and power. The screenshot provided, is the final version, where the DRC has been passed. In reality, it took two attempts.

I did the same order as my third attempt on the two layer board, I routing, signals, then power. With my component placing being optimized, this was so easy. I eventually did finish the whole board, with really good routing. There was no crowding, of the routing, and important signals, are isolated, and wont encounter any problems, once manufactured.

But then, after checking with people, I found many issues. 1. my via sizing, was way to small for jlc pcb. I do not want to pay any more than I have too. . Turns out my settings, allowed for connections that are really close. This would cost me a'lot more money. Something I don't want to do, since this board, is probably, going to be paid for by hackcludb.

I also met about 50 errors, when I ran the DRC, due to STILL broken footprints, turns out I selected the wrong foot print, for some of the footprints,

After this, I decided to redo everything, witch is like another two hour timeframe of playing snake with kicad :( .

So I first fixed most of the problems, with the first one. I rechecked all my footprints, just to ensure no problems occur.

I also, read all of JLC's PCB's constraints, and made sure my kicad settings match up with the JLC PCB, constraints.

then I started the 2 hour routing process. After getting ANOTHER attempt, to rout, I was able to improve my routing even more, I decided this time, to define exactly what the middle two layers are for, instead of just calling them data and power. I decided the 2nd to the bottom will be power, and the second to the top, will be signals.

While I did not stay completely true to this decision, This actually did make routing even easier, and kept it more neat, this is probably the first time, I am actually proud of the routing for one of my pcb's.

While I did have a few small DRC errors, I was able to quickly solve them. d

Ive Imported this file into JLC PCB, and the actual manufacturing, of the 4 layer pcb, and the stencil, is 17$. but shipping and taxes, arn't free, so its actually going to be about 33$.

Along with all the components, assuming there is going to be enough for three boards, since im doing it myself, and I need backup, this is going to be about 130#. This will actually be significantly, cheaper, than if I used PCB assembly.

After this, I am going to start, researching, and looking at other, thrust vector control mounts for model rockets, while I do know the general gist for TVC rockets, I will be looking at them, and making my own modifications, to increase rigidity, and to increase the speed of the TVC mount.

So far, I think its going to be similar to the BPS.Space gimbal.

.

But, there are several things, like the pushrods, that I want to redesign, so ima just redesign the whole tvc mount from scratch, and incorporate, some other design changes.

12/27/2025 - Completed avionics subassembly.

This is the completed avionics assembly for my propulsive landing rocket. After this, all thats left, is the landing legs, witch I expect to take, one day, and then the software, witch I should be able to write in a day as well. PCB design is what I sucked at, everything else, im pretty cracked at, and I can do alot faster than the average. That is why this project, is going to be around 45 hours of work, though, with the amnount of effort I out in, I expect this to be tier one.

I started by looking at the BPS.space TVC mount. As this is my first time, developing my own TVC mount, I first researched the BPS.space tvc mount.

The BPS.space Gimbal uses a design that I will inspire from, the onion gimbal. Basically, this mount, has an outer gimbal, that rotates, an inner gimbal, that rotates the motor. since each layer, can move independent of the other, this allows the gimbal to rotate in both x and y. I will be using a similar approach. BUT, the outer gimbal, will me incorporated, into the avionics assembly unibody. This reduces, the amount of parts necessary. Less individual parts are lighter.

Some things I do not like about this gimbal, is the servo linkages. Those servo horns, and the metal push rods, have a'lot of flex. and because, its not a perfect fit, it has a'lot of slop. So I will not do this, I will 3d print my pushrods, out of markforged onyx. My school has an industrial printer, that can print ONYX, witch is a plastic, with similar characteristics to solid aluminum, plus that printer, can lay actual carbon fiber latices, into the part.

I also spent, a'lot of time, looking at many peopls, gimbals, to learn from, in fact, too many to discuss in this journal.

So, the first thing I designed is the TVC mount. It needs to fit in a 76mm aroframe, with reduced cutouts, so I started with the outer gimbal. I started, by sketching the maximum
diameter, Than I had to find a place to add the servo, without it clipping to the outside.

After this, I made the inner gimbal, and the outer gimbal. the hardest part, was again, figuring out where, the servo had to go. if its too inward, the gimbal loses its gimbal range, and if its to far out, it sticks out of the body tube. because of all these gimmicks, it did take my several hours to design the whole TVC mount.

After several hours I reached this model, that meets all my conditions.

After this, I designed the whole unibody, that connects, all the avionics together in one pieces, to reduce weight, and to make accessing the avionics easier. I started with adding the flight computer mount, into my avionics assembly. I had to spend time, making sure, all ports, are accessible.

Then I designed the reaction wheel.

Here, a motor, spins a heavy, wheel. This is for roll control. The plan is that, when the rocket, starts to roll, the wheel will spin in the opposite direction. This will cause the rocket, to correct its roll. Because, for landing a rocket, roll needs to be controlled, I added a reaction wheel here. I actually had to spend alot of time, researching good motors, that can handle the stall current that would be produced when rotating the rocket.

Overall, this took a day of work, so I have entered 7 hours, cause the realistic amount of time, I was actually focused and working.

12/30/2025 - CAD COMPLETE

The first thing I started was the landing legs. While the landing legs, sound like a simple, part, it took me like 6 hours, because, I am not familiar with any of the parts, so I had to research parts. Its also a complicated design, cause it has to extend.

I started with drawing out a single leg. I did this, because of the inherent complexity that comes with the legs.

As you can see, there are hinges, that allow the legs to start retracted. It does not showcase all the parts, but it does allow me to see, what parts I had to design. This is the final drawing. It actualy took several drwawings.

After this I first designed the base.

This is on the bottom of the rocket. It slides onto the body tube. This allows the force of landing to go into the whole body tube, and not rip a screw.

Then I cadded cf rods, and the hinges. Then I went into the asembly, and use revolve mates, to allow things to rotate.

After this, I designed the crucial piece, where the upper strut, pushes into the bottom strut.

This piece will have to be adjusted once, the leg is made, cause the upper strut, has to fall into this piece.

I did have to use a bit of trig, to find the ideal, opening, of the curved surface, basicly the angle that curved face is point to, relative to the body tube. Its realy hard to explain this to someone, who has not designed this own type of leg, so please, feel free to contact me for more insight.

Then I added them to the assembly.

As you can see, I also designed the upper hinges, for the upper strut, that connects to the syringe.

While it looks similar to the bottom, there was some problems, with my parametric cad, so I had to just recad it, witch took an amount of time.

I actualy, Almost imeaditly imporved the shaft of the locking mechanism.

This should make the upper struts sliding into the locking position esier.

Then I started working on the upper strut. The upper strut, will have a syringe, to absorb most of the impact.

So then I made a connector, that connects the imported syringe, to the actual landing legs.

.

After that, I designed the plunger that will also slot, into the bottom strut.

![image](/user-attachments/blobs/proxy/eyJfcmFpbHMiOnsiZGF0YSI6MzI2MzAsInB1ciI6ImJsb2JfaWQifX0=--d019e988a36baa94188efec5a5cc2b1fa6c49f4b/image.png

Right now its simple, later I will cad the part of it, that holds to the bottom strut

After that, I connected the syring asembly, to the upper mount, with 110 mm carbon rods, and the same hinge peices.

.

After that I designed the nub, that alows the laod to be transfered up the syringe.

After that I made some guide rails, to prevent the upper strut from slipping out.

Then I duplicated it across 4 legs.

Then I extruded the base to prevent the legs, from devloping past 60 deg.

After completling the legs, I added them to the main asembly

After that I designed a nose cone, I spent WAY to long, redoing the spline, to get the perfect shape.

Eventualy, I got to the COMPLETE cad