![[Pasted image 20250304135159.png]] ### Lab Update *2025-02-11 -> 2025-03-04 | [[#Pics|Skip to Pics]]* [[Lab Update - 2025-02-10|Previous Lab Update]] | [[Lab Update - 2025-04-07|Next Lab Update]] Biweekly Update was the goal, but life happens. Nearly all my effort has been going into the [[!Joule Thief]] lately. #### Highlights: - Set up a separate soldering table for SMD work, got some practice in. - Finished Joule Thief Proof of Concept - An Alkaline Cell is charging a Lithium Cell - Made some adjustments to improve efficiency and throughput - Started schematic & layout. Currently stuck in datasheet mines - Switched to OrcaSlicer, fine-tuned Rat Rig flow & support settings. - Replaced the hotend, nozzle, and LED Panels. - Printed, Cooked, and Shipped a ton of maglev orbs. --- #### Joule Thief ##### New Notes: - [[!Joule Thief]] 1. [[LTSpice LT3105 Deeper Dive]] 2. [[LTC3105 Better SPICE]] 3. [[DeepSeek Barebones Discrete BMS]] 4. [[Some Discrete BMS Builds & Parts]] 5. [[Digital Pots to Consider]] 6. [[Cheap MCU Research]] 7. [[Introduction to the ATTiny85]] 8. [[IRF9540N SPICE Model Generation]] 9. [[DIY Overdischarge Circuit]] 10. [[First Lithium Charge Logs]] 11. [[Alkaline Cell Pulsed Loading Tests]] 12. [[Old Inductor MPPC Settings + Larry]] 13. [[Joule Thief Path Review]] 14. [[New Inductor MPPC Testing + Larry]] 15. [[18650 MPPC Charging Tests]] 16. **DATASHEET:** [[INA219.pdf]] 17. [[INA219 Pop Quiz]] 18. [[INA219 Annotation]] ##### Option: Modeling Dynamic ESR - Considered taking a data-driven approach by getting a DAQ and getting some MATLAB going and modeling dynamic internal resistance of AA cells based on load current and time elapsed, recovery time, maximum power transfer. All in order to determine the exact optimal system for depleting a generic alkaline cell as quickly and efficiently as possible. After thinking about what would go into that for a while, and the exploration into the electrochemical limits and mechanics of the batteries (there are several factors composing the "ESR" of a battery), I decided to push that off as a "last 10%" effort ##### DIY BMS Design Research - Collected and watched some resources on DIY BMS Design so I could understand more of what goes into them. After a few hours of this, I realized that it's a bit simpler than I thought it was, and that I could implement some basic functions in discrete components in order to keep moving. I'm trying to avoid the whole "wait a week or two" for parts and "wait two or three weeks" for boards whenever possible. - Designing the discrete 4.3V overcharge disconnect circuit was a really good experience for me. It's kind of confusing trying to get voltage references to work when your supply voltage is changing. The whole thing felt like some sort of evil sudoku. But I started enjoying it after a while. I enjoy sudoku. - Took a look at the LTC4070, TP4056, DW01A - [[DeepSeek Barebones Discrete BMS]] - [[Some Discrete BMS Builds & Parts]] - Some Discrete BMS Builds: - https://youtu.be/OdH3wYeFJ3M - https://youtu.be/xY-adIeuMbc - https://youtu.be/x5YhSAEzBMg - https://youtu.be/Fj0XuYiE7HU - https://youtu.be/qRVEJjk5B_g - https://youtu.be/7AFbrKi0t_4 ##### Better understanding of the LTC3105 - After reading the datasheet, and playing with the LTC3105 on a breadboard for a while, I read the datasheet again and realized I was using the chip wrong in some ways in sim. Mainly, the LTC3105 can only take 400mA max input current. So if you want more power than that, you're going to have to increase the input voltage. [[LTC3105 Better SPICE]] - I fiddled around in sim for several hours seeing what the power in-power out looked like, how input voltage and current affected things. Efficiency, pulse time, total power delivery, battery discharge curves, etc. - Because of this, I was considering some sort of pre-amp stage between the battery and this boost chip, but I ultimately decided against it. - After making some adjustments based on my better understanding of the datasheet, I was able to get efficiencies >85% in sim, which was a big improvement and a green flag. ##### Getting LLMs to read datasheets and generate SPICE Models - Since I was trying to build a BMS using discrete components (that is, whatever I have laying around), I had to get creative. I was using a PMOS in sim and didn't have many PMOS on hand. The few I did have I didn't have a model for in LTSPice. So I got creative and pulled the datasheet, pulled SPICE models for similar PMOS from the same mfg, and then fed them into ChatGPT and DeepSeek. Letting them both take a crack at interpreting the datasheet and making informed assumptions to generate a new SPICE model for the FETs I had on hand. Then I had them split the difference between eachother. I don't really know how to go about validating that the SPICE models they generated are any good, but it was a really interesting exercise and I believe it has some promise for the future of ECAD. - [[IRF9540N SPICE Model Generation]] o1 ``` .model IRF9540N VDMOS(pchan Rg=3 Rd=70m Rs=47m Vto=-3 Kp=6 Cgdmax=0.64n Cgdmin=0.09n Cgs=1.1n Cjo=650p a=1.5 Is=50p Rb=30m N=1 ksubthres=0.1 mfg=International_Rectifier Vds=-100 Ron=0.117 Qg=97n) ``` ##### Getting stuck in Analog Sudoku Puzzles - I spent a LOT of time trying to get the DIY overdischarge cutoff circuit to work the way I wanted it to, but I wasn't really getting there and it was taking more time and effort than it was worth (seeing as I'm not really trying to DISCHARGE the Lithium cell right now anyway), so I decided to just skip it. If I need to drain the cells I have a USB BMS card I can plug my phone into or something. Not critical right now. It's hard to stop chasing my own tail like that sometimes. ##### Proof of Concept - 18650 charging from AA - I finally got the absolute bare-minimum proof of concept working on 2025-02-20. A 1.5V alkaline cell was being used to directly charge a 3.7V lithium cell. - It was charging very slowly and very inefficiently. Nearly 200mW out and only 58mW in. The 18650 was climbing at approx. 1 microvolt/second. Good enough for now, but not for long. - [[First Lithium Charge Logs]] ##### Finding ways to improve throughput and efficiency - Once I had the proof of concept down, it was time to start exploring other methods for getting more juice of the battery, and then other methods for increasing the efficiency/throughput of the boost circuit. - I wanted to try frequency-based or pulsed loading, since the ESR seems to steadily climb for as long as the alkaline cell is loaded. I thought maybe giving it room to breathe intermittently might help out. - In this case by "breathe" I mean disconnecting the load so the ion gradients can settle and the battery can return to its static state. - I wired up an alkaline cell with an INA219 and an NMOS to short it to ground. Connected the FET to my function generator and starting playing with all manner of square wave. - Generally, I found that frequency doesn't do much for this application. DC power extraction with the occasional rest period seems to be the optimal. No frequency/duty cycle seems to increase average power delivery. - I did find that I can get upwards of 600-800mW peak power with ~100Hz, 12.5% pulses though, which was pretty cool. ##### Wrong type of Inductor doing the switching - One big problem I kept running into was that the more I increased the supply power (by adjusting the MPPC pin), the lower the efficiency got. It seemed like the Joule Thief could only work properly in a "low and slow" mode, which ran counter to my intuition, so I started troubleshooting more. - Larry and I have been chatting a bit lately so we wound up talking about a USB Audio Card he was working on, and then reviewing my current Joule Thief circuit. I raised some concerns with efficiency and how I might approach solving it by controlling the LTC3105's MPPC pin. Then he took one look at the Breadboard and lowkey started roasting me for using variable inductors for a power electronics application LOL. This was definitely one of my *you don't know what you don't know* moments. - The variable inductors I got a few years ago for building a Theremin may be appropriate for tuning RF circuits & oscillators, but for power electronics applications like this they're very bad, since they have substantial DC Resistance. - After talking about it some more, I found some more appropriate radial choke inductors on Amazon and got Larry's blessing for them. - Once I got the new inductors and plugged them in, I started seeing dramatically better results. This is also the current circuit. Now I can get anywhere from 100-250mW out of the alkaline cell, with minimum 50%, and upwards of 90% efficiency storing the energy into the 18650. The previous circuit was also unable to provide more than ~50mW into the lithium cell, and I can pretty easily get 150mW now. So call it a 3-6x improvement - Additionally, the range of MPPC now behaves more intuitively, where I can choose to go low and slow at peak transfer efficiency, vs. a fast burn, where the efficiency drops but significantly more power is delivered to the lithium cell. ##### Moving into schematic & layout phase - After tuning up the prototype with those More Correct inductors, I decided it was time to start laying out the KiCAD schematic. Even if I'm not ready to order the board quite yet I should start getting some sort of draft going, so I can push for that milestone. - I realized I would have to wire up the INA219's peripheral circuitry myself now and not just rely on the breakout board. I mean I could, technically, but that's not really the intention here. - Which made me realize I haven't read the INA219's datasheet to know how to wire it up or how it works. ##### 3 sets of datasheets to failure - So I decided to make an exercise out of it and read the INA219 fully, annotated it, and took detours as necessary to make sure I understood everything I was looking at and working with when it comes to this chip. It was a pretty good experience for me. I haven't designed electronics professionally before, and I'm pretty inexperienced at board layout, so I haven't spent a lot of time just sitting there reading datasheets. Thankfully my Bluebeam skills translated nicely (I already edit/annotate pdfs all day at work). - Moving forward, I'll be reading and annotating datasheet pdfs more often, and probably doing little writeups and overviews on the chips. I think it's a worthwhile exercise both for the author and the reader. - Datasheet is at [[INA219.pdf]] - Then Larry Pop Quiz'd me about the ADC. Datasheets should have quiz material. - [[INA219 Pop Quiz]] ##### Next Steps - I'm going a lot slower than I expected to with the schematic and board layout for this first demo board. A combo of not having a lot of lab time & underestimating how time/focus-intensive it is to read datasheets when you have no clue what you're working with. But it's fine, since I'm learning so much so fast. I just have a bit a fire under me to get to a point where I feel comfortable applying to exhibit at Open Sauce. - After getting the better power throughput, the next question is really how to optimize power output at any given moment. I know the ESR of the battery steadily increases, the cell voltage steadily drops, as it remains loaded. I know that I'm going to have a microcontroller handling I2C anyway, so I currently think a digital pot that can manage something like 20k - 200kΩ will be the best option for allowing some sort of control/optimization algorithm dynamically adjust R_MPPC. But that just means more datasheets to read and more delays, to a certain extent :p - [[Digital Pots to Consider]] - [[Cheap MCU Research]] - [[Introduction to the ATTiny85]] #### Breadbox - It's been top of mind lately for me to basically hurry up and finish prototyping Joule Thief so I can start working on Breadbox in earnest. A lot of people have shown interest and I'd like to start serving them more directly as soon as I can. The main thing that comes to mind is setting up a number of different dimensions, formats, etc. so I can A/B test on twitter and see what people would like. - I'm not sure how I would go about collecting feedback on which modules people would want designed, but I'm sure if I just start building the ones I personally need people will show up and tell me what they want to see. It's a pretty interesting idea I want to pursue. I believe I can get something awesome ready in time for Open Sauce as well. - Jumperless: https://youtu.be/ECoV4VeyLQs #### Acetone Vapor Chamber - Another thing I've been excited to jump on next is properly designing and building the 500mm Acetone Vapor Chamber to take the Rat Rig and my print services to the next level. After going through the motions with Joule Thief, I feel a lot more confident in my ability to take on that system design. - Also, I've met so many very smart people that know so many diverse things, I really feel I could throw some of the problems I get stuck at onto the TL and get some very clever answers. It's really nice knowing that I finally have some minds to lean on, even hypothetically. - Larry had a great idea to use a magnetically coupled plastic fan to stir the flammable acetone vapors (like a chemistry stir) #### Rat Rig - Dialed in my support settings & got some Beautiful releases. - [[Good Support Settings]] - Had to replace the LED Panels since they burned out. They have been going for about a year. The whole process took about 5 hours start to finish with a new design. I'm pretty happy with it. - Currently considering getting a separate, dedicated Pi just for streaming high quality, higher fps video for the Rat Rig. I want to extract more aesthetic/content value out of it. - Also considering a space heater/chamber heater, since I currently have to preheat for around 75 minutes to prevent gantry warp/thermal expansion during printing. - I also recently bought a V1 Rapido instead of a V2 on accident, so I may shop around for newer heaters as well, budget allowing. ### Pics Lots of pics this time, so I put them in 3x3s. [[#Lab Update|Back to Top]] ![[Pasted image 20250304130344.png]] ![[Pasted image 20250304130351.png]] ![[Pasted image 20250304130400.png]] ![[Pasted image 20250304130504.png]] ![[Pasted image 20250304130410.png]] ![[Pasted image 20250304130420.png]] ![[Pasted image 20250304130430.png]] ![[Pasted image 20250304130457.png]] [[#Lab Update|Back to Top]]