PC Engine Duo Repair and Mod Advice

[Great Hierophant]

A friend of mine let me borrow his PC Engine Duo he received from an ebay seller from Japan called doujindance.  doujindance had replaced many of the surface mount capacitors and had installed an RGB mod.  He replaced the 5-pin DIN AV port with an 8-pin DIN port for the RGB lines.  There is a small board connected on one end to the DIN and on the other end to what I presume are the appropriate lines from the HuC6260A on the underside of the PCB.  I also understand that the CD-ROM laser assembly has been replaced.

Issue One : Freezes when playing ADPCM from CD

When I received it, I was able to test the HuCard slot with a Turbo EverDrive, and that works just fine.  However, I have been having serious problems with CD games crashing.  At first, I was only using CD-R backups, typically with high quality Taiyo Yuden CD-Rs burnt at 16-24x.  Many games would load correctly, but several would crash at certain identifiable points :

Ys Book I and II - Start a new game, game will freeze at the screen "In the Beginning"

Ys III - Game froze during the scene in beginning where Dogi speaks to the girl.

Ys IV - Game froze after Dogi speaks for the first time

Dracula X - Game freezes at loading FMV cutscenes.

All these instances have something in common, at these points the game will need to stream digitized ADPCM speech or sound effects from the CD.  I understand that the chip responsible for streaming ADPCM is the M5205 OKI ADPCM Speech Integrated Circuit.  At these points, I can reliably reproduce the crash.  Trying to narrow down the causes, I tried a genuine Ys Book I and II CD and it crashed at the same point.  These games will work if these areas are avoided (sometimes not possible) and CD audio will play fine.  I tried fiddling with the laser adjustment pots to no avail.   

There are other instances in some of these games where ADPCM is used and I never hear the speech or sound effect, but the game will not freeze.  Once or twice, ADPCM did work and Dracula X and Ys Book I and II did not crash at these points or any other point and played all speech and sound effects back correctly.  I could not figure out why the ADPCM would occasionally work but usually fail.  I suspect that this chip is failing and needs replacement, console5 offers them for a reasonable price.  Alternatively, I may need to replace the rest of the surface mount caps that have not been previously replaced.  Any advice would be appreciated. 

Issue Two : Jailbars in RGB Output

When I looked at the PC Engine's Duo's RGB output through a Framemeister, it generally looked great, but on some games, jailbars were quite noticeable.  These jailbars are not present on the native composite video output.  In this post, you can see pictures taken of my own screen :

http://nerdlypleasures.blogspot.com/2015/05/fun-with-framemeister.html

In the beginning Bonk's Adventure, they are very noticeable, with other games their visibility varies.  It is something that once it has been seen, it just cannot be unseen. 

Now the mod that doujindance installed looks to be using a tiny amplifier chip and some passive components.  It does not bring composite sync to the AV connector, there just aren't enough pins on the connector.  Sync comes from the composite video sync.  Do I need a different mod or do I need to bring composite sync to the Framemeister?

[thesteve]

The jailbar issue is the ground on the din8
Try steam hearts
See where/if it crashes

[ddoubledee]

Just an educated guess... But there could be a damaged trace(s) on the board from capacitors leaking in the past. It's been known to happen on traces going to the ADPCM stuff. Could have been overlooked when the capacitors were replaced.

Another good one to try would be Monster Lair. It's a regular 'ol CD-ROM2 game, so it only uses the 'stock' 64KB RAM... Except for one interesting thing... It uses the ADPCM memory for extra graphic storage...

[Keith Courage]

usually when ADPCM doesn't work it's a bad trace to the positive on this 22uF cap (circled blue). Voltage should be 3.8V


Also check that the VIAs off of the 22uf cap are getting power circled both in red.

These traces can go bad after an initial recap. They don't always show themselves ahead of time and can loose connection hours or weeks later sometimes.

[pnauts]

To have performed comparison pure Sync will not add you benefit to composite as sync imo.
TheSteeve, jailbars come from bad GND on DIN8 ? I often read here that some decoupling capacitors have to be used to fix it. I personally don't have the problem with mine.

[thesteve]

the original duo runs all the sin=gnals through chokes
the ground as well
this causes an inverted current pulse on the ground pin

[Great Hierophant]

I have read that a pair of tantalum 22uF capacitors on the 6260A may fix the issue.  Is there any other possible fix, or is it just inherent to the chip itself and nothing can be done?

[Keith Courage]

I've done the 22uf cap fix and it helps but doesn't get rid of the jail bars completely. Depends on the game being played of course and what TV you are using. I've noticed that newer flat screen TVs have a much smaller chance of displaying jail bars than older CRT TVs do. Not sure why.

[Great Hierophant]

the original duo runs all the sin=gnals through chokes
the ground as well
this causes an inverted current pulse on the ground pin

If the AV lines are run through a choke, would bypassing the choke fix the issue?  How can that be done?  By contrast to the composite video signal, the added RGB lines are not run through a choke, so would adding chokes to those lines balance out the issue and remove the jailbars?

[thesteve]

Bypassing the choke works
The issue with the chokes is your using a shared ground pin

[Great Hierophant]

Bypassing the choke works
The issue with the chokes is your using a shared ground pin

So, I should either desolder the ground pin or cut the trace to the ground pin on the DIN and use a patch wire to connect it to ground.  Where is a good place to get the proper ground?

[thesteve]

best place is the connector shell of the controller port