Worklog PCSX2 Device

[Lightning]

At the risk of seeming like a pleb, I think it's time I engage the community once again, because as we all know, everything we don't know someone else does.

So here's what I've been working on; Its a Windows machine, which will run PCSX2 (and whatever else it can handle), and every game that's compatible with it. I plan to use a COMe-CVR6 (e2), a JXD Singularity S12K, and a custom designed carrier board.

The Compact Express module:
https://www.kontron.com/products/bo...xpress/com-express-compact/come-cvr6-e2-.html

The mostly complete case with LT070ME05000 screen (and capacitive touch)
http://www.jxd.hk/game-console/s192-rk/

And finally the carrier board:

I guess it's worth while to start with a list of concerns and other knowledge that I am currently lacking, so here goes

-Currently I'm working on getting the HDMI converted to MIPI DSI for the screen, and was hoping someone may have a full datasheet for the LT070ME05000, a long shot, but I need more info in it.

-Also looking for a datasheet/schematic for the TC358870XBG chip, I've ordered a board that uses it to try and gain some insights, but reverse engineering can only tell you so much.

-I have been tossing around the idea of making it a wireless headphones only system, no integrated speakers or jacks, so would I still need to use a DAC like the ALC1150, or can I omit that component?

-The touch panel uses an i2c protocol to communicate, and I was going to use an FT200XD to interface it with one of the USB lanes, thoughts/problems about/with that?

-Using an ATMEGA32 for the controller, coupled with an MPU-6050 for 3 axis capability, also over one of the USB lanes

-As you can see I'm using an M.2 SSD, and about PCIE lanes, must they all be matched to the same length, or can the lanes all be different lengths as long as the pairs within them are the same length as each other?

-Do I need to use an inductor to separate/buffer the 5 and 3.3 volt rails from whatever loads are attached to them?

 

[Lightning]

Okay here's a question for the community:

How might one make a circuit to control the brightness of an LCD, basically just something that could, on the command of say.. a GPIO pin or two, vary a voltage between 3.3 and 5 so as to increase or decrease the brightness of a panel?

 

[MomSpaghetti]

Okay here's a question for the community:

How might one make a circuit to control the brightness of an LCD, basically just something that could, on the command of say.. a GPIO pin or two, vary a voltage between 3.3 and 5 so as to increase or decrease the brightness of a panel?

Wouldn't you use a potentiometer for this kind of application? variable resistance to control the current? or are you doing something else?

 

[Lightning]

I want to be able to do it within the system software, that same way you control the brightness on your phone screen (it basically is a phone screen). The backlight LED's brightness seems to depend on where to voltage is between 3.3 and 5 volts.

 

[cheese]

If you want to use a microcontroller you could just do PWM out (and use a capacitor to smooth the voltage). Then just use two gpio pins on the micro to increase/decrease a variable.

 

[Lightning]

So when it comes to building a buck converter, is it the size of the the inductor that determines how much energy it produces?

 

[Lightning]

So when it comes to building a buck converter, is it the size of the the inductor that determines how much energy it produces?

Why yes Lightning, yes it is!

Anyway, now that I finally have that answer, I can continue this project!! Does anyone else suffer from project ADD?? One of these days I may actually finish one, but I have to say, the advent of language models has really accelerated the process.. So mush so that I may actually be able to get one finished in the arbitrary time limit set by the Rube Goldberg machine I call a brain.

To that end, I have re-birthed my COM Express module based handheld in KiCad, and as one might expect, almost completely revamped, and improved the design. The first thing I decided to tackle was the power system. The BQ25713 is my buck-boost charger of choice. Using external FETs, it handles the higher current demand much better than a fully integrated package. Those battery cells are 21700s and according to the math at least, I should be able to get a good solid hour and a half of play under heavy load. That may not sound like much, but this Ryzen chip is actually pretty powerful. In any case, it should run for nearly 7 hours under minimal load, and I plan to use Atlas OS, or possibly some stripped down Linux distro to maximize resources.

I also took more than a few pages from the book of @Tschicki, the genius none of us saw coming! The RP2040 master at the center, communicating over I2C with two others on the left and right side, forms the Gamepad. The Right side will only handle buttons in one core, but the Left side will be doing buttons in it's first core, while using it's second core to inteface with the touch panel from the JXD Singularity. They actually make an excellent portable screen, as they're just an LT070ME05000 LCD, with a perfectly sized glass touch panel glued to it. Shame they're so expensive to buy lol.

The LCD is interfaced with using a TC358870XBG, controlled and configured by an STM32. I actually had quite a journey with that part. It's based on those HDMI converter boards you sometimes see on AliExpress, and Raspberry Pi projects. I dumped the LT070ME05000 specific configuration firmware off one of them (pictured below), and have slightly customized it, allowing for more fine brightness control, as the original only had 4 levels and used only one button. Funnily, I originally thought this approach of converting DDI disgnals into MIPI was a little hackneyed, but it turns out it's exactly how the Steam Deck works, so there you go! Not so weird afterall.

The ALC1150 is a farily simple way of getting sound from the HDA lines coming from the COM Express module. I have the left and right speakers hooked up to the FRONT signals. Further testing is likely required though. Finally, in another of the pages taken from @Tschicki, there is another STM32 acting as a System Controller, reporting battery charge to the system, and handling power button funtions. I wanted to use another RP2040 for this, and perhaps I'll find a way, for now though, it's an STM32.

Thankfully there's actually a lot of documentation about reverse engineering these boards, but as far as schematics go, there's a lot of discrepancies. Because of this, I was sort of forced to create my own little version of it, and I'll be uploading everything related to interfacing HDMI with an LT070ME05000 once I get my test board back, and confirm new firmware function. For now, this is the board I mean. The problem is that they're made to be used by several different MIPI screens, depending on firmware, so there are a LOT of unnecessary components, and figuring out what was needed and not needed has been very interesting. What you see above around the Orange is nearly all that's needed for the LT070ME05000.

Finally, just to give some sense of scale for the whole device, and to give perhaps some idea of how much I've been able to shrink everything down, here is the LT070ME05000 screen, with the cVR6-E2 module on top, and a Canadian quarter:

Well we'll see how far I get this time lol. Thanks for reading!