Worklog Some PS2 Project

I found another component that although it's in stock, the BOM has a name for that component that I can't find anywhere outside of china, it's the 16.9344MHz oscillator, NX3225GA-16.9344M-STD-CRG-2, it's not in digikey, mouser, TME, etc..., it's actually under the name STDCRG2-16.9344M, it's the exact same component and it took about an hour to find it.
 
I found another component that although it's in stock, the BOM has a name for that component that I can't find anywhere outside of china, it's the 16.9344MHz oscillator, NX3225GA-16.9344M-STD-CRG-2, it's not in digikey, mouser, TME, etc..., it's actually under the name STDCRG2-16.9344M, it's the exact same component and it took about an hour to find it.
The part number of that crystal may have changed at some point (?), because all stock I ordered back then is marked with NX3225GA-16.9344M-STD-CRG-2. Searching for it in Digikey or Mouser now redirects to the STDCRG2-16.9344M, which is the correct part and fine to use. I will try to update that in the BOM today or tomorrow!


IMG_20260622_224354470.webp
 
I was going through the BOM to see if I could find cheaper alternatives to some components and I noticed that there's no caps in the BOM for the 32.768KHz oscillator, did you leave those out for some reason? I've never seen oscillators without caps, I'm going to use 10pf caps here as using the formulas in an adafruit blog post gave me around that value.
 
I was going through the BOM to see if I could find cheaper alternatives to some components and I noticed that there's no caps in the BOM for the 32.768KHz oscillator, did you leave those out for some reason? I've never seen oscillators without caps, I'm going to use 10pf caps here as using the formulas in an adafruit blog post gave me around that value.
C1501 and C1502 are marked DNP in the schematics and should ideally not be populated. If you compare with a 79k or 90k mainboard, they are also not populated there - most likely because they are integrated in the RTC.

The RTC will work in both cases (populated or not), but it will drift substantially more when they are populated (in the order of 30s to 1 minute per day). I tried populating them with 10pF or 15pF on the first revision, and that's why I decided to keep them DNP.

Without these caps the RTC still drifts by about 7s a day. If you are already looking for alternative components - this may be fixable using a crystal with lower capacitance and ideally lower ESR. I discovered this a bit too late, these PS2 models apparently use a Rohm BU9861 RTC+EEPROM. The exact datasheet isn't available, but checking Rohm's old datasheets for similar parts gives some hints about a better crystal selection for the RTC (BU9873 for example)
 
C1501 and C1502 are marked DNP in the schematics and should ideally not be populated. If you compare with a 79k or 90k mainboard, they are also not populated there - most likely because they are integrated in the RTC.

The RTC will work in both cases (populated or not), but it will drift substantially more when they are populated (in the order of 30s to 1 minute per day). I tried populating them with 10pF or 15pF on the first revision, and that's why I decided to keep them DNP.

Without these caps the RTC still drifts by about 7s a day. If you are already looking for alternative components - this may be fixable using a crystal with lower capacitance and ideally lower ESR. I discovered this a bit too late, these PS2 models apparently use a Rohm BU9861 RTC+EEPROM. The exact datasheet isn't available, but checking Rohm's old datasheets for similar parts gives some hints about a better crystal selection for the RTC (BU9873 for example)
I found this crystal ECS-.327-4-16R-TR, same footprint and frequency but a lower load capacitance of 4 pF and lower ESR of 60K Ohms, also Instead of the switch oled speakers I'm buying these as after measuring the STLs it looks like it will fit CMS-20143-158L200 and I'm going to solder them directly to the motherboard, also the joysticks I'm going with TMR instead of hall effect as I found an aliexpress listing for a lot of 10 at 2.40€ each, still looking for buttons and front glass, for the buttons I'm thinking of resin printing them, I know a guy that has a 3d resin printing farm, maybe transparent with the symbols embossed to then fill with colored resin, and the bottom rubbery part I could use TPU, what do you think of these changes?
 
Hey!

The specs of the new crystal and speakers look fine to me. Maybe the speakers require slight adjustments to the 3D printed covers, but that's no big deal (both .step and .sldprt are in the repo). The soldered connection is debatable, but in the end it's your build up to you to decide what's best for your use-case

For the TMR sticks please verify that they can operate at 3.3V, that's really important. Often the supplier won't give that info and for the sticks in the BOM I had to try myself. The 3DS XL screen lens is also available on aliexpress, I'd recommend the glass version over the plastic one.

I don't know how well 3D printed ps vita buttons would turn out, it certainly feels like a lot of effort compared to buying them on aliexpress. But you never know unless you give it a try, it sounds like a fun experiment for sure! Please report back if you give it a try!
 
I don't know how well 3D printed ps vita buttons would turn out, it certainly feels like a lot of effort compared to buying them on aliexpress. But you never know unless you give it a try, it sounds like a fun experiment for sure! Please report back if you give it a try!
Unfortunately buying buttons for the 15 consoles will cost me almost 200€, that's way too much for so little, I have an Idea to print the buttons in 2 parts, first I'll print the transparent part with a void in the back with the shape of the symbols to fill with white resin, then I print the back of the button and glue it with some resin, and to finish I print in TPU the rubber/silicone part of the button, or even make a mold in resin to make the part in silicone, this will make it significantly cheaper, although I have to find a PS Vita 2000 to measure the buttons, I only have a 1000 model.
 
@Tschicki
Hey! I just acquired 16 displays for the project but unfortunately 10 of the 16 displays ended up being 50pin, same signal, size and resolution but 50pin, is there any clearance in the finished product for a QSB to adapt the 40pin to 50pin in the motherboard? or even to solder directly on the LCD pins instead?

Edit: the 16 screens where a lot on ebay for cheap but on arrival 10 weren't as described and it was not visible in the photos.
 
@Tschicki
Hey! I just acquired 16 displays for the project but unfortunately 10 of the 16 displays ended up being 50pin, same signal, size and resolution but 50pin, is there any clearance in the finished product for a QSB to adapt the 40pin to 50pin in the motherboard? or even to solder directly on the LCD pins instead?

Edit: the 16 screens where a lot on ebay for cheap but on arrival 10 weren't as described and it was not visible in the photos.

Ah that's a bummer, I hope they were at least considerably cheaper than the buydisplay panel in the BOM!

The clearance from mainboard PCB to LCD is about 2.5mm, excluding all the 0402 components in that area (remove ~0.5mm).
Such a FFC connector is about 2mm tall -> it's a very tight fit, if you decide to have a QSB with 50pin connector.

I'd recommend to not solder the LCD to the mainboard permanently, it will make assembly and disassembly a real pain.

Did you also verify that the display timings and backlight are the same between your 40pin and 50pin panels? I have never seen these 5" displays with 50pin interface before, so I hope it's really the same panel. While it would be possible to adjust the current version of the video processor to other display timings, they are hardcoded and require a new bitstream.
 
I'd recommend to not solder the LCD to the mainboard permanently, it will make assembly and disassembly a real pain.
Maybe I did not explain properly, what I meant was instead of soldering a 40 pin connector to the board, I make a flexible QSB that adapts the 40 to 50 pin and solder a 50 pin connector in the QSB, so the QSB with the connector solders directly on the 40 pin pads on the board and then I bend the QSB so that the connector position is right above where the 40 pic connector would go. This is option 1.

For option 2 I was thinking of the same type of flexible QSB but to solder directly on the LCD flex, re-routing the necessary pins to a 40 pin connection, basically converting the LCD from 50 to 40 pins using a flexible pcb.

Did you also verify that the display timings and backlight are the same between your 40pin and 50pin panels? I have never seen these 5" displays with 50pin interface before, so I hope it's really the same panel.
No I did not, but I ordered an adaptor for the raspberry pi that has a 40 pin and a 50 pin connector in the same board to test, I already tested the 40pin displays, so if the 50pin work with the same output, then I can use the LCDs, also the 50Pin are all DXQ5D0015-CTP, can't find a full datasheet but from the information available, they use the same ST7262 controller.

I hope they were at least considerably cheaper than the buydisplay panel in the BOM!
With shipping and tax it's a little under 14€ each, but I contacted the seller and he gave me a partial refund so they ended up being around 10,50€ each, which ended up not being a bad deal, kind of, I still have to design a flexible pcb to convert the display or motherboard to 50pin.
 
Hey!

I did not plan for another update, but here we are! No worries, the portable is still doing fine - I’m about 50-60 games and ~1000h into my PS2 marathon and there is no end in sight! All during commutes of course, the portable already traveled over 40.000km in the last year!

It's now in use for quite a while and I noted some things that I would like to fix in the current design. More specifically:
  • The uSD2PSX doesn’t really look nice, and the contacts are quite worn down. It’s also a bit of a pain to assemble as it stands
  • While I took lots of precautions to prevent dust from entering the gap between the LCD and screen lens, over time it does manage to get in there
  • The screen lens constantly gets dirty when carrying the portable in the case - slightly recessing the LCD would improve this

uSD2PSX
The flex PCB was fun to play around with, and at the time the only solution I found to compress the design into an SD card format. The main issues: the glued assembly is still on the thick side of the SD tolerance range, the contacts wear down faster and it makes assembly more complicated.

This time I wanted to go another way, with a rigid PCB. But with a rigid PCB it’s difficult to achieve two important dimensions: the thickness from top surface to the contacts (~1.4mm) and the overall thickness (~2.1mm). I could go with a 1.2mm PCB, but then with components (e.g. micro SD card) the overall thickness is far above 2.1mm.
What I always wanted to try was stacking PCBs, and that’s the route I ultimately took. By stacking two 0.6mm PCBs I can achieve the required thickness; including a label and some headroom for soldering, I get 1.4mm at the contacts and there is enough room to fit the SD card and the components.

The plan was: have a base PCB which contains the whole circuit, and another PCB to add the contacts and some mechanical rigidity. Both would be connected via solder joints. Here it is:

IMG_20260911_104145844.webp


The nice thing: I can do the whole assembly in the reflow oven – no more awkward hand soldering due to a warping flex PCB, or lots of epoxy to hold everything together. I also designed a small cover to enclose the circuits, which may be adhered with one or two drops of glue. The test points are much bigger this time and have a 100mil pitch. I already ordered the PCB for a little needle probe to program and configure the uSD2PSX.

Looking at the thickness, I think next time I would go with a 0.6mm + 0.8mm stackup. It turned out that 0.2mm headroom for label & tolerances was overestimated, I actually had to apply 3 labels on top of each other to get to 1.4mm at the contacts!

Still, it’s much nicer and it fits the SD card slot on the portable much better due to the slightly smaller thickness! I’m still not sure how to integrate this into the repo, my preferred option would be to remove the uSD2PSX entirely and create a separate repo for it.

Screen lens
To fix the dust issue I wanted to try laminating the screen lens to the LCD. So far, I always tried to avoid this awful process and I understand that not everyone would be willing to do this over the old method. That’s why I would like to do minimal changes to the mechanical design for implementing this.

I did some research on LCD laminating and determined that the LOCA process would be best suited for this type of LCD. OCA sheets would be less messy, but I think that the metal housing of the LCD would interfere with the adhesive. I was looking for some quality LOCA glue options (e.g. APS-346), but it seems like they are pretty much unobtainable for me. The options left are the popular, generic ‘aliexpress specials’:
  • TP-1000
  • TP-2500
  • K-3022
The characteristics, according to what I could find:
  • TP-1000: Low viscosity when uncured; cures to a hard rubber material; very difficult to separate LCD and lens once cured
  • TP-2500(F): Higher viscosity than TP-1000 when uncured; cures to a silicone/gel-like material; ‘easier’ to separate LCD and lens once cured
  • K-3022: Viscosity comparable to TP-2500 when uncured; creates a very hard and permanent bond when cured; impossible to separate LCD and lens again
The requirements I set for lamination:
  • Viscosity should be as high as possible, as I’m concerned that the glue could get into the metal housing
  • The bond should be stable over time; I don’t want delamination, air bubbles or yellowing after weeks/months of use
I chose the TP-2500 for my first test, because of its higher viscosity. My only concern would be longevity; a weaker bond could cause delamination over time.

I did a couple of tries on PC sheets to find a good method for application and then tried it on a broken LCD with 3DSXL screen lens. While the first one or two tries still had some air bubbles, the third attempt had zero trapped air bubbles and no inclusions on the glue layer.

A couple things I noticed/learned:
  • The glue likes to creep into the metal housing and it will most likely never cure inside there.
  • Adding the right amount of glue is difficult. I like to apply more than needed, as excess glue spilling out the side is preferrable over repeating the process.
  • If excess LOCA glue gets inside the LCD, it will leave permanently visible spots/defects in the polarizer/diffuser layers. See the image below.
  • When cured, the TP-2500 glue leaves an oily residue on anything it touches.
  • Fixing the screen lens in 2 corners only is not enough to keep it in place during curing. It likes to ‘float’ away -> fixate all 4 corners
  • So far the best glue application was in this shape: |---< along the long side of the LCD. Then the lens can be aligned on the left short edge and slowly tilted in place (lowering it fast produces bubbles).
  • Applying too much pressure on the lens to spread the glue leads to air bubbles
  • The glue below the black border of the 3DSXL screen lens takes *much* longer to cure and needs UV exposure from all sides
  • The thickness of the glue layer may vary over the area of the lens, depending on how evenly it was pushed down
  • TP-2500 is definitely not removable without breaking the LCD!

I designed a fixture to hold & align the assembly during glue application and curing, then gave it a try to write down the preliminary process:
  1. If the LCD has no protective foil installed, add kapton tape around the border of the LCD (was the case for the LCD I tried this on)
  2. Open the metal casing of the LCD and add a line of glue around the LCD as sealing, making sure that there is no gap in the glue line.
  3. Close the LCD again, clean away the excess glue. This way, no LOCA glue can enter the LCD from the front. So far I did it with T7000 glue, but I’m not quite happy with it, the surface hardens way too fast.
  4. Place the LCD in the fixture, on the top edge fill the cavity between LCD and fixture with sealing material. There are special adhesives for this, but I just used modeling clay as sealant. This prevents LOCA glue from seeping into the LCD on the top edge.
  5. On the other three edges of the LCD, place kapton tape over the seam between LCD and fixture. This keeps the LCD border free from LOCA and also prevents it from entering the LCD on the other sides.
  6. Now the 4 edges and the front of the LCD should be glue-tight
  7. Close the fixture and tighten the 4 screws
  8. Remove the protective film of LCD & lens
  9. Making sure to keep dust away, apply LOCA glue in the |---< pattern
  10. Align the left edge of the screen lens with the fixture and slowly fold it into place (careful, orientation!). The slower the better.
  11. Apply slight pressure to distribute the glue over the LCD.
  12. When the whole area is filled with glue, check that the lens is flush with the top surface of the fixture on all 4 corners. If not, keep applying slight pressure where needed
  13. Inspect for air bubbles and inclusions. At this stage you can still remove the lens and clean everything to start again.
  14. Cure the glue under UV for 5-10 minutes
  15. Open the fixture, remove the excess glue and all the sealing material
  16. Final cure under UV
Adhering the front metal housing and sealing everything
IMG_20260918_230039809.webp
IMG_20260921_113429916_HDR.webp


Applying the glue and seating the lens
IMG_20260921_115044168_HDR.webp
IMG_20260921_125942096_HDR.webp


Improper sealing leads to defects. Mostly along the top edge:
IMG_20260918_180942445_HDR.webp


I the designed a little bezel to be adhered onto the lens, to take strain off the LCD and to permanently join screen lens and LCD (remember, TP-2500 is very soft when cured)! I tried a couple different ways of mounting it (without glue), but I always ended up having pressure spots on the LCD. That’s why the resulting assembly is still adhered to the front shell, similar to the old method.

The outcome is very good actually; I even prefer this new look compared to the old LCD assembly!
IMG_20261008_215134877.webp

One of these LCDs was already installed in my main portable a couple weeks ago, so far there are no air bubbles! A reference piece was left next to the window and so far showed no yellowing.

The process definitely has a lot of room for improvement, but it’s really cool to be able to laminate screens in very low quantities for such things. I don’t think I could go back to the air gap method! It should work on other LCDs too, with the right screen lens!


Another sidequest:
Last but not least I wanted to mention that I finally ran into a game that is NOT compatible at all with my portable. It’s EverQuest Online Adventures! Recently I found out about ps2online.com and I really wanted to play some online PS2 games!

Only one problem: Currently the design has no way to go online; no ethernet port and especially no wifi. I thought it would be fun to look into some wifi adapter accessory to play online on the go.

There are basically 3 questions between this idea and me implementing it:
  • Is it even possible to play online with this portable?
  • Can I make the accessory small and efficient enough to be able to power it from the portable?
  • Can I pull this off without making huge changes to the design?
The first two questions were already answered:
IMG_20260923_204915441_HDR.webp


This is still using a TP-Link TL-WR802N travel router as wifi-ethernet bridge. I was looking for smaller and possibly more efficient options:
Long story short:
  • The ESP32 router worked surprisingly well, but only when adding a static route to the PS2 in the upstream router -> not an option for a phone hotspot! That was needed because I simply could not get the NAT implementation working reliably
  • The Vocore2 works well too, but actually at a lower bandwidth than the ESP32 (still plenty for PS2 though). Setup was a bit of a pain, as that kind of bridge didn’t work out of the box.
I will probably go with the Vocore2, it’s the more flexible option of the two. In theory it would even be capable of hosting an SMB share. I’m currently working on question 3 and thinking about how to attach the ‘network adapter’ to the portable. It would be really fun if this adapter was removable, also to save power when not needed!
 
This is absolutely crazy!!
Laminated screens are a definite must have! For Wii portables, we source laminated screens from portable emulation console vendors (Anbernic for example). If you find a handheld that uses the same screen size as your portable, you can probably grab a laminated screen for much less hassle.
Looking forward to seeing the final implementation of online play!
 
Back
Top