@GTV reviews the Cosmic Fantasy 1-2 Switch collection by Edia, provides examples of the poor English editing/localization work. It's much worse for CF1. Rated "D" for disappointment, finding that TurboGrafx CF2 is better & while CF1's the real draw, Edia screwed it up...
Main Menu

PCE GT repair help (super hard!) needed please o_o

Started by segasonicfan, 08/04/2024, 06:16 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

segasonicfan

So my friend gave me his childhood PCE GT to fix.  Lots of sentimental value.  and man is it busted.  At some point the 5v rail I believe shorted and it fried the controller board IC.  ordered a new controller board, but I don't know what else it took with it.

I get a yellow and reddish screen, split in 2, with noise lines in the middle (attached).

IMG_20240804_033016.jpg

- Fully recapped
- Fully cleaned in Ultrasonic and manually with 99.9% IPA
- No audio (I can hear speaker hum, and volume works.  I'm confident audio amp output stage is OK.
- +5V rail is good (4.85v)
- +2.5v rail is good (for audio)
- 24mhz clock is good on both ASICs (followed guide from TG-16 service manual)
- H sync and V sync outputting correctly
- Backlight circuitry all tested good, 30V and 24v OK.  All oscope probes of transformer check out too.  Screen seems to be fine.

What I did in addition to above:
- swapped in full Hudson ASIC chipset from working CoreGrafx (white)
- probe (diode) tested every on the Hucard slot.  Compared with working unit, and values seem to be as expected / close (bus pins within range of each other, no obvious shorts).
- swapped in one of the VRAMs from the CoreGrafx as well

I'm at such a loss with this one...the board looks *so* clean.  anyone know of other stuff to look at?  I've poured so much time into this one it hurts.

I'd be so indebted to a helpful benefactor.  Willing to send thank you gifts!
<3

Thanks in advance!




Tak-MK

Hi there!

When I'm lost as you after doing way too many tests, what I do is to try to get a working one and start comparing every-fucking-pin of every-fucking-ic (or connector/etc :P) with an oscilloscope while running the same game until I find a difference.

Hope you can fix it!

segasonicfan

#2
Thanks for the suggestion Tak-

That's quite a last resort - sure to take hours.  I had a working one to compare, but the owner doesn't want to recap it yet, despite the speaker failing and leakage everywhere.  I had to give it back because I didn't want it to kick the bucket in my hands :/

The issue with oscope probing every pin, is that behaviors are different of course when a cart / game is running.  What I need is a better service manual, that gives more debugging info than "replace all ICs if clocks are bad." (basically what TG-16 service manual says).

I wish I knew:
  • - what is the boot up order? Which pins go high / low on startup (OE, WR, Read, etc)?
  • - does the metal shielding need to be in place for a proper boot?
  • - what pins / ICs are known to go bad with a +5V short?
  • - is there a source for the knobs (volume / brightness) to replace?  some of the plastic got melted on mine with chip removals  note to others - desolder those, dont even try to take out the main CPU (80 chip) with them in there.
  • - can a white CoreGrafx PCE chipset be swapped in?  Or are there revisions to the huson chips + VRAM and Work RAM?  (I actually swapped the *entire* chipset and results were the same).  I'm sorta guessing there's a slight variation.

Essentially what I have here is what appears to be a fully working GT (audio, screen, contrast, high voltage stuff), but a dead PCE chipset.  Would love help from someone who knows that architecture more.
I found this epic write up: (https://www.copetti.org/writings/consoles/pc-engine/) but it's not really repair / failure focused.

What I really don't understand is why the symptoms are the same no matter what I try.  No garbled graphics or sound.  I fear these 80's ASICs dont have the diode protections of 90s logic used in Sega consoles that I'm more familiar with, and just bite the dust at the first sign of a +5V short (such was the case with the controller IC, that's for sure).

I can verify from a working system that it will boot without controller PCB connected (just make sure pins 12 and 8 are bridged on the controller FFC plug if you want backlight on)

Some essential repair resources:
Main IC pinouts:
Hopefully all my debugging and suffering can benefit some other poor soul out there.
Attatched pics of my board, literally one of the cleanest you'll ever see.  Sigh.

Repair log:
  • - Removed all ICs (even RAM) from while PCE and replaced.  All pins checked under microscope.  No difference in symptoms
  • - Removed all electrolytic caps and rechecked all values and pads with DMM on cap setting
  • - Cleaned under all ASICs and RAM and caps, no trace damage I could find
  • - Ultrasonic cleaned the entire PCB in friends $1000k+ cleaner.  Cleaned uup after with 99.9% IPA after
  • - Began removing chipset again to fix my white PCE now, while doing that, melted some of the plastic on knobs :(  they still turn, though.

My postmortum on this guy:
  • - Was given to me for a recap, which was successful on first try
  • - Was given back years later for speaker not working. Likely broken headphone jack (testing seems to confirm this).  Wired up audio signals using retrosix guide and speaker "hiss" comes back
  • - +5v rail was shorted somehow in this test process.  Controller IC was getting *very* hot until it was removed.  Controller PCB replacement ordered
  • - Current symptom ever since then

To help others, I am attaching here:
  • - My board scans
  • - Blank board scans (courtesy of retrosix)
  • - TG-16 service manual
  • - ASIC chip pinouts
  • - Diode probe compare excel I made (comparing good unit FFC 40 pin cable with my bad unit)

Tak-MK

Many thanks for all the debugging info and photos! I feel bad for not being able to help after such work...
btw, just out of curiosity, how did you make the photos/scans of the boards? Plain normal scanner? Phone camera with lights? I'm searching for a better way of documenting mine but I'm always having some lightning issues...
P.S.: I've just read that they were done by retrosix :P I'll have to continue trying then.

Thanks a lot again!

segasonicfan

Quote from: Tak-MK on 08/11/2024, 09:19 AMMany thanks for all the debugging info and photos! I feel bad for not being able to help after such work...
btw, just out of curiosity, how did you make the photos/scans of the boards? Plain normal scanner? Phone camera with lights? I'm searching for a better way of documenting mine but I'm always having some lightning issues...
P.S.: I've just read that they were done by retrosix :P I'll have to continue trying then.

Thanks a lot again!

Yeah board scans were RetroSix: https://www.retrosix.wiki/board-scans-pc-engine-gt

To get good board "scans" I imagine they just used a camera light box.  Something like this:
https://www.etsy.com/listing/1656195528/

segasonicfan

#5
A further update on my painful journey of PCE repair....

So I swapped the full chipset back into my white CoreGrafx today.  Only now it's dead too :(
Next steps are:
a) recap everything (maybe caps died from hot air gun)
b) get a different game to try

I am worried that the Hudson ASICs are just so sensitive to anything they die with the slightest short, reverse voltage, or any similar issue.  Although I'm not sure what could have killed them in the swap, everything was triple-checked and both systems are rock solid V4.9V.
The Sega ASICs, made around the same time, are much more robust it seems.  It's a real bummer.

My white CoreGrafx just gives me a black screen.  Continuity checked the full ribbon cable and thats fine.  All pins for all chips check out under microscope and "push test" with tweezers.  One pad was damaged and jumped with inductor wire, not a bigge.

What *is* a biggie is that the PCB substrate began to warp and bubble during reflow.  These old 80s boards are just crap compared to the PCE GT / 90's stuff.  It seriously was bowing after a single chip removal.  The solder joints are all large enough that I don't believe it's what is causing the current issue though (plus I managed to minimize some of this).

Re; PC Engine GT, I removed the full chipset on Main PCB and cleaned it all.  But it already looked great.  Now it's a spotless mystery dead system.
Removing the volume knob is tricky btw - note there's a sorta hidden extra ground tab in the middle, so it actually has like 8 points to desolder.  It's a cool design to add ground below the resistive path, but man that was annoying cause I've never seen that before.

Tak-MK

Quote from: segasonicfan on 08/12/2024, 05:11 AMYeah board scans were RetroSix: https://www.retrosix.wiki/board-scans-pc-engine-gt

To get good board "scans" I imagine they just used a camera light box.  Something like this:
https://www.etsy.com/listing/1656195528/

Yep, I made a camera light box too but bigger to be able to photo other motherboards too, but it's not as good  :|

And regarding the GT and so, the PC just bending while using a hot air gun... either the temperature it's too high, or the board isn't that good as you said... I'd worry too