𞋴𝛂𝛋𝛆

  • 547 Posts
  • 5.63K Comments
Joined 2 years ago
cake
Cake day: June 9th, 2023

help-circle
  • Oh yeah, I’m sure. Cuba just came to mind as a common vacation destination for the Soviet block back in the day. I’m sure there are other places that are both tropical and within the CCP’s influence, though I do not know these offhand.

    That is not how semiconductors work in general, as far as I understand it. All the stories I have heard are of people that love to work for the purpose it brings to their life and are willing and want to work 24/7. Someone like this that takes a bunch of stuff is likely actually taking their own work with them after some corporate shit crushed their life’s work they have spent ages on and raped such a dedicated purpose. They are likely one of a handful of people that actually created all the value in the company but the corporate fuckwit is too stupid to comb through the details to understand the company.


  • Pleasure. There are a few things I want to better understand within the radio space.

    Talking over the air is not one if the things that interests me, but maybe some digital stuff could be fun. I need the time buffer built into text to collect my thoughts and find my voice through chronic pain issues.

    I would like to play around with antennae designs, acquire a better understanding of magnetics, build some discrete circuits, gain a fundamental understanding of the various active filter topologies, and several other little details.

    The recent solder smoke challenge 40m receiver was the first time I saw a diode mixer and really understood the theoretical framework of frequency division outside of LSI logic, like 4000 or 7400 series chips. Potatosemi makes such GHz capable logic, but doesn’t scratch the surface of the realm of the nude bearded virgin wizards and witches of radio.

    The superheterodyne stuff never made sense to me, or diode mixers, and most circuit blocks in radio. I built a little Manhattan style Colpitts oscillator that A2AEW shared on YT ages ago, that I use for testing crystals. I learned enough about guitar amps and effects that I understand most buffers and amplifiers, both discrete and op amp based. And I can at least identify that an op amp is configured as an active filter. I recently discovered discrete SAW filters.

    I had several in my miscellaneous crystals junk drawer, but did not know what they were. Now I really want to understand how those work and how to make one. There is definitely a gap between my knowledge of active filters and how these little devices work.

    It will probably never happen, but I find cavity resonators fascinating too. Building components stuff out of traces and empty space is some Jedi voodoo shit that tingles both my inner illogical miser… “(just beg borrow or steal a $20k network analyser…)”, and primal shamanistic dogma “(try and grow a beard… it will be better this time…)”. So yeah, mostly introverted stuff. It is probably about like me and car stuff. I’d rather paint and build motors than actually drive the thing stupid fast. Right now I’m in a really good spot to try receiving stuff over the Pacific too.







  • I can’t say what is what here from specific experience. Use some fine steel wool on the contact surfaces of every connection to polish it. Be VERY careful with steel wool around batteries as it will catch fire instantly if it bridges a large voltage potential, like from dropping it.

    When I am working with circuits and etching, steel wool is best for removing every bit of the oxide layer, ensuring a much better connection/consistent surface to work with.

    You generally need a 5+ digit digital multimeter to get into the range of measuring shunts directly. An ancient old shit brown ugly keithley 197 will generally work and can usually be found second hand for around $100. Then you can directly measure shunts with a 4 wire probe (current source and sense wires are separate to eliminate the wire resistance from the measurement).



  • Yeah, fast edges on the buttons. I think the whole design was originally intended to be 2 sided. The third layer has almost a complete ground plane, and the other two have sufficient infill and stitching to be a typical 2 sided design. Stuff would have needed to be moved around some to make room for the button contacts, but there was more than enough space.

    From what I recall of the guys doing the software hack, I think the STM32 H7 microcontroller was emulating the original 6502 based ROM. So maybe they were optimising the hardware as much as possible to avoid the typical timing issues present in emulation in challenging parts of the game.

    The coolest part of the hardware design is actually that little DC switch mode converter and battery manager. It is crazy efficient. Like, a full charge on the lithium cell will be at around half charged still after a couple of years, and while playing, it lasts abnormally long for such a device. It is not a particularly large cell either.











  • start with something more simple. Just use multiple choice in the background like an old choose your own adventure book but the AI makes the choices and pair that with a similar character complexity and scope. Use a multi model architecture where a small manager limits length and constrains to theme.

    Models are already developing an internal character profile for everyone in a prompt context. All you need to do is add some meta questions about the probable education level of the textual responses to determine age within reason. The model picks up on word choice and grammar well. There is a major component of dogma involved in all interactions where the model is trying to infer probable expectations. In fact, if you avoid describing characters and infer all within verbal dialog, it unlocks a lot more dynamic range of behavior.

    I do not think 3rd party junk is the way to go. Code it to own and utilize the local model ecosystem. Pitch it to sell bleeding edge hardware as a leading AAA. You will never get a consistent long term fine tune behavior from any API. You can fine tune train a local model to behave as expected 95+ percent of the time, which is better than any general API model.


  • It is ultimately the fine scratch texture. It is both holding onto some grime, and you are seeing the texture difference. The finish is quite thick in most cases. It will look better depending on the strength of how you clean it. Generally something like a magic eraser is pretty good. However, the best will come from polishing. It does not have to be anything elaborate. A machine polish with a heavy cut pad and compound like I would used to cut and buff automotive clear coat would be overkill, but are an example of what I mean. If I could get a polisher in there, it would take me 5-10 minutes with a cut pad and 3M PerfectIt 2. Toothpaste is technically a polishing compound although only around a medium cut. That with the inside of a sock as a polishing pad for an hour or more by hand will get nearly the same result. Keep the surface wet like a wet clay that feels barely warm the whole time. Never let it get dry or wetter. Keep the patterns random circles and never repeating rows or columns; keep it more random than that. Do something like a puffy cloud that is 2/3rds the total size moving around the surface. That will make it good as new… if you are able to defeat your inner impatience.