Ghosts on 315-5685 ASIC VA2.3 Model 2 Genesis composite output

Hi all,

I've got a VA2.3 Genesis with a KA2195D and 315-5685 ASIC. I was wondering if anyone else has seen this on their VA2.3 Genesis:
Attachment 13744

As this shows, there are faint 'bright' ghosts on the edges of the letters. It doesn't do this on all graphics; it only shows on certain graphics in certain games (like the big letters in Jungle Book, and around Batman/Robin in The Adventures of Batman and Robin). It did this from the moment I got it, I have not modified the video logic in any way. (Though I'll eventually get rid of the horrible KA2195D, I'm holding off on that for now.) I've got another VA2.3 with this ASIC (but a different encoder) and it has the same issue, but another VA2.3 and a VA2 I have that both have 315-5786 ASICs don't have this ghosting.

So on the 315-5786 based systems, there is no ghosting at all, only on the 315-5685 systems. Using the same cables and power adapter for all of them.

Any ideas on what might be causing this? (And if possible, a fix?)

Check c-sync. Try adding a filter to 'C8' (start with 47 pF).

Sadly, adding cap to C8 made no difference... I wish I had a scope to see if C-sync really is to blame.
Actually... I am starting to suspect a problem with the 315-5685's shadow/highlight mode support. I noticed this only shows itself on the 315-5685 systems when shadow/highlight mode is used. The shadow/highlight is being offset to the right by one pixel. On my VA2 / VA2.3s with 315-5786 it doesn't do this.

You can see this in the opening intro of Phelios, when it shows Athena chained to the wall. You notice the leftmost bit of some of the bricks is brighter than usual, and there's a bright 'outline' on the extreme left of the torch fires (this latter detail doesn't show up well on camera).

The VA2.3s in question with the 315-5685 ASIC:
Attachment 13746

My VA2 with the 315-5786 and VA3 with FC1004:
Attachment 13747

I do have a dead VA2.3 parts motherboard with a good 315-5786 ASIC, maybe I can try swapping it with that (and setting the appropriate 4.7K resistor pack to match) and see if that gets rid of the offset.

Update: Sorry to revive this old dog, but I've determined that the 315-5685 is at fault for this behavior. 315-5786 and other ASICs do not have this problem.
Sadly, this means that probably a good 1/3 to 2/3 of all VA2.3's made are subtly 'faulty' because the 315-5685 ASIC is the most common seen on these units. :( However, all VA2's I've seen yet seem to use the 315-5786 ASIC, so they are probably alright.

Another problem I discovered with some 315-5685 ASICs, is they have a bit of left-side corruption on some games that extensively use certain raster effects. On Ecco II, if you look closely at the extreme left side of the SEGA logo screen, you'll notice bars of solid color filling the areas just to the left of the wavy-gravy lines:
https://youtu.be/gRmrBaSWcrA
On other games, such as Shining Force II, you'll get tiles erratically wrapped, particularly when the DAO spell is used.

These issues do not occur on 315-5786 ASICs, or any other known ASIC to date. However, this latter issue does not seem to afflict all 315-5685's. (Maybe Toshiba and/or Sega quietly snuck a fix for this on a later run/stepping?) However, the 'off-by-one' shadow/highlight error is present on all 315-5685's I've seen to date.

As these are obviously complex logic flaws, the only way to fix these problems would be to replace the 315-5685 ASIC with a 315-5786, but this will probably also require swapping a few other parts, such as external resistors / capacitors, the resistor network, and maybe some other stuff. Unfortunately the only way to get 315-5786s as of now, would be to scrap them from dead VA2s or VA2.3s which have this ASIC... unless someone finds a source of them online.

Quote:

---

Originally Posted by Eep386

Another problem I discovered with some 315-5685 ASICs, is they have a bit of left-side corruption on some games that extensively use certain raster effects. On Ecco II, if you look closely at the extreme left side of the SEGA logo screen, you'll notice bars of solid color filling the areas just to the left of the wavy-gravy lines:
(picture pending - will update once I get back)

---

I have noticed that with Sonic & Knuckles's Sky Sanctuary Zone (when playing as Knuckles) with the 315-5685–equipped VA2.3 you sold me on eBay in 2017:

Attachment 14751

Can you please re-try attaching that file?

Yeah, I didn't know about these defects with the 315-5685 until the very end of 2017. Had I known about the chip's shadow/highlight problems and left-side corruption beforehand, I wouldn't have sold any Model 2's with it. For the longest time I wasn't entirely certain it was something I was doing wrong either, until I did some further testing on family members' VA2.3s and found that every single 315-5685 had that shadow/highlight flaw. Later on I discovered the left side raster effect glitches as well, which irritatingly seemed to affect only some 315-5685 chips. Suffice to say, I've stopped selling these faulty systems and 'retired' all 315-5685 systems in my possession, and they are basically parts machines far as I'm concerned.

It's pretty sad that one of the otherwise superior revisions of Model 2 Genesis has become extremely difficult to recommend, considering how hard it is to get one of these in the first place. Of course 315-5786 fitted VA2.3's are fine AFAIK, but they're even harder to find than the 'usual' 315-5685 VA2.3's are. I wound up with my '5786 VA2.3 from dumb luck, took a chance on a factory repaired system on eBay thinking it was a VA2, and I was pretty miffed when I discovered it was a VA2.3. But then I opened it up, and '315-5786' smiled back at me.

Quote:

---

Originally Posted by Eep386

Can you please re-try attaching that file?

---

Attachment 14752

Off-topic: Your replacement RESET button failed eventually, and I ended up replacing it with the Console5 blue clicky version. When trying to replace it with the one from my VA0 Genesis 2 (I was trying to switch them, I believe), a solder pad lifted. I ended up having to solder a wire from the replacement RESET button (the Console5 one this time) to where that solder pad had led.

It still serves as my main Genesis at the moment despite the buggy VDP.

Nope, that pic is still not working. Do you have something like photobucket?

Better still, do you have Discord? You can drag n' drop pictures with relative ease that way. I'm mrbleggerlarz on there (I am usually invisible, but I'll respond to replies received when I get time from work).

It's been many sold Segas ago, so please help me jog my memory: was that one of the ones I sold that I fitted with a CXA1145M, by chance? Also, out of curiousity, was the reset button I fitted clicky? I started using new, modified Arduino pushbuttons sometime mid-2018 I think, but it might have been sooner. The Arduino buttons are clicky.

This should work: https://i.imgur.com/6cjTIYK.jpg

• The video encoder is a CXA1645M, this unit's original according to you in the listing.
• Your replacement was one of the stock ones.
• Serial number is 440210490 in case that helps.

Yeaapp, that's the left side corruption flaw I was talking about. The 315-5685's are pretty busted, wish I knew all this before I sold some VA2.3s!!!!

Good to know it's not one of the new buttons I'm selling with my systems. So far I'm having pretty good luck with these arduino buttons.

Anyway check your private messages, I want to set this right.