Chip date formats?

Huge · Post by **Huge** » Sat Apr 28, 2012 4:28 pm

I'm writing a list of sega chips, and I decided to not list the chip date for every instance I can find, only the lowest and highest numbers per type of chip. Basically, to indicate between what periods was it manufactured.

However, I can't figure out some formats. For the Sega 315-5313 VDP chip, I see:
8Y 08 87 C (on a VA1 motherboard... probably USA)
94 06 63 C (on a VA2 USA motherboard)
95 05 61 C (on a VA2 USA motherboard)
9025 EBXD (on a VA3 USA motherboard)
9122 LAJE (on a VA6 USA motherboard)

For the first three entries, I guess the first character is year. The rest I can only guess with so few samples. They are all from 1990 at the most though, as later chips all use a different numbering system.

For the last two entries, 9025 and 9122 is year-week, so 1990 week 25 and 1991 week 22. No idea about the letters. This standard format has been used by the majority of other Sega chips later on, along with a standard 3 letter format on smaller chips.

Perhaps it would be better to just collect all the different dates I see on the chips for my list? It's more through, but seems a little redundant. Opinions?

Huge · Post by **Huge** » Sat Apr 28, 2012 4:33 pm

Another thing:
315-5433
this is listed in the wiki as "Megadrive I/O, Bus arbiter, Broken TMSS"

My PAL Megadrive has this chip. How is the TMSS broken? It displays fine at bootup, does it just not do any license checks?

ElBarto · Post by **ElBarto** » Sat Apr 28, 2012 4:46 pm

The program runs fine, but writing other thing than 'SEGA' into $A14000 doesn't lock the system when accessing the VDP after.

Why do you write a list when the wiki have almost (all ?) the chip listed ?

Huge · Post by **Huge** » Wed May 02, 2012 8:55 pm

Some more from a different but also Sega chip:
4Y 04 76 C
4X 20 78 C

ElBarto wrote:Why do you write a list when the wiki have almost (all ?) the chip listed ?

All 1200+ asics 22000+ proms?

Nemesis · Post by **Nemesis** » Wed May 02, 2012 11:30 pm

Are you wanting other people to contribute to your list? Because I have a very large set of hardware I could get numbers off. A lot of it is in storage, but there's a lot that I have accessible right now too.

Huge · Post by **Huge** » Thu May 03, 2012 12:03 am

Eventually yes, but first I need to get my shit together and create something worth contributing to. I'm notoriously lazy. Also I'd only add entries for which I have pictures of, board and all. Or at least mark if the info came from word-of-mouth.

I do have a running list for sega saturn serials though:
http://evilboris.sonic-cult.net/saturn/

ElBarto · Post by **ElBarto** » Thu May 03, 2012 3:39 am

Huge wrote:Some more from a different but also Sega chip:
4Y 04 76 C
4X 20 78 C

ElBarto wrote:Why do you write a list when the wiki have almost (all ?) the chip listed ?
All 1200+ asics 22000+ proms?

Oh I didn't know that you wanted to list ALL of them

But feel free to add them into the wiki if you want.

Just add a new table for each system in the Category:Chips page.

Huge · Post by **Huge** » Tue Jul 24, 2012 4:36 pm

Here's some info I got after comparing numbers, for Yamaha chips (which used the date format in my original post). It appears Yamaha switched dating formats around 1990 June.
There is possibly no exact date, as different plants may not have switched at the same time. But I have date codes corresponding to 1990 25th week in both formats.

Some samples, not sorted chronologically, from between 1984 to 1990:

4Y 04 76 C
4X 20 78 C
77 11 85 B
84 05 83 B
9112 71B
9416 51B
9613 518
8Z12 74 C
941451C
8Y 08 87 C
8Z 12 70 C
94 06 63 C
95 05 61 C

- First char is year, that's for sure.
- Second char could be month, since this format holds up with my samples: 1-9 for January to September, then X Y Z for October to December.
- Third and fourth char: day of the month
this is just my speculation, but it holds up, now that I've put more samples together. (part of the reason I'm collecting these values is so I can figure out the logic behind them)
- 5th and 6th characters :I am completely unsure about these. I doubt they collected the dates down to hour/minute, so they are probably production run numbers (such as codes for specific plants).
- The very last letter is likely the chip revision; which I'll mention below as it is also used in their later system.

And here's the second format:

9049 ABCC
9025 EBXD
9049 AEED
9122 LAJE
9419 ADPE

It goes two chars for year, and two for week of the year.
The meaning for the letters is unclear, they are likely production run codes again. I've compared a large amount of these per chip, and not managed to conclude anything solid, except for the last character, which seems to be the revision (respin?) count of the chip again.

Compare some dates for the well known Megadrive combined VDP/IO/YM/TMSS chip:

Code: Select all

315-5487    FC1004   9144 HALB
315-5487-01 FC1004   9240 ZCDE
315-5487-10 FC1004-X 9313 AAOI
315-5660    FC1004   9336 GBCM
315-5660    FC1004   9620 GAOM
315-5660-02 FC1004   9327 EBBM
315-5700    FF1004   9537 ACNH

All of them have the last character increasing over time and version.
The only odd one out is the last one, the 315-5700. This was used in late MD2s (seen it on a VA3 board), and on the Nomad. I believe it was Tiido who said that it is interchangeable with the 315-5660, which would mean that it is functionally identical, just not physically.

Note the other difference there, FC1004 vs FF1004. After comparing various components, the second letter here could possibly be a process node marker. Insert theory here about the lower node meaning lower power usage, directly leading to the idea of a handheld Megadrive (or the other way around, specifically designing the same chip with lower power usage, so they can do the Nomad).

I only have Sega related entries in my DB (who else did Yamaha manufacture custom ICs for in the 90s?), but it looks to me that it increased over years. For comparison, I've narrowed down my entries to all Yamaha chips, taking the earliest date I have recorded for each, and listing duplicates according to the last character of the date code:

Code: Select all

315-5066     2217     4Y 04 76 C (SG-1000 II VDP)
315-5124     2602B    77 11 85 B (Master system VDP)
315-5308     6045B    91 12 71 B (Megadrive bus arbiter)
315-5309     6046     8Z 12 74 C (Megadrive IO)
315-5313     7101     8Y 08 87 C (Megadrive VDP)
315-5313     7101     9025 EBXD  (Megadrive VDP)
315-5313     7101     9122 LAJE  (Megadrive VDP)
315-5313     FC1001   9151 LABC  (Megadrive VDP)
315-5313A    FC1001   9236 OAEG  (Megadrive VDP) - rev G ? this is on a machine I own...
315-5313A-01 FC1001   9223 ECPC  (Megadrive VDP) - reverts to rev C later...
315-5364     6045C    9049 ABBD  (Megadrive bus arbiter) (later revision of 315-5308!)
315-5487     FC1004   9144 HALB  (Megadrive VDP+IO+TMSS+YM)
315-5487-01  FC1004   9240 ZCDE  (Megadrive VDP+IO+TMSS+YM)
315-5487-10  FC1004-X 9313 AAOI  (Megadrive VDP+IO+TMSS+YM)
315-5660     FC1004   9336 GBCM  (Megadrive VDP+IO+TMSS+YM)
315-5660     FC1004   9620 GAOM  (Megadrive VDP+IO+TMSS+YM)
315-5660-02  FC1004   9327 EBBM  (Megadrive VDP+IO+TMSS+YM)
315-5700     FF1004   9537 ACNH  (Megadrive VDP+IO+TMSS+YM) Nomad and MD2
315-5687     YMF292-F 9432 LAME  (SCSP)
315-5687     YMF292-F 9551 AATF  (SCSP)
315-5688     FH3007   9434 0ZNG  (Saturn SCU)
315-5690     FH3006   9426 JZMC  (Saturn VDP2) from an extremely early console thought to be a prototype machine.
315-5690     FH3006   9501 LAZD  (Saturn VDP2)
315-5690-02  FH3006   9523 LAJD  (Saturn VDP2)
315-5890     FH3006B  9542 0APF  (Saturn VDP2)
315-5890-02  FH3006B  9601 ZAFF  (Saturn VDP2)
315-5960     FJ3002   9811 OAJF  (Megadrive VDP+IO+TMSS+YM+68k+z80) MD2 VA4 and Genesis 3 VA1
315-5964     FJB006   9628 ZAJD  (Saturn VDP2) dieshrink, so new design, and new respin/revision count
315-5965     YMF713-S 9635 EBDC  (SCSP+68k)
315-5966     FHB007   9621 OAPA  (Saturn SCU) dieshrink? respin? both?
315-5966-02  FHB007   9640 DANA  (Saturn SCU)
315-6119     FQ1003   9839 AAYD  (AICA)
315-6123     FQ8007   9830 ADYE  (Megadrive VDP+IO+TMSS+YM+68k+z80+ ..RAM ?) Genesis 3 VA2
315-6232     G21003   9944 ABQG  (AICA)

The increase in the Fx yyyy identification is clear, as components changed.

Note for 315-5313A: I own the machine with the G revision, so it's not an error for sure. But all other 315-5313A or 315-5313A-01 chips I've seen fall back to rev C though. Maybe it was the design with the best yields, so it was made default.

Note for Saturn SCU: last char in date code is G, therefore it could be as far as revision 6 by the console release. This could be possible as if you read the chip version register, it reads as 00 04. So this thing definitely went through many revivions!
It is my theory that when they wanted to make the machine "more powerful" in response of the Playstation tech demos, instead of a second SH2, they added the SCU DSP.
It's also the odd one out for the dieshrink theory, as it goes from FH3007 rev G to FHB007 rev A.

Note for last AICA entry: possible the node increase was so much that it overflowed (overflew?)? Well, it still increases, so my theory holds up.

... in conclusion: Have you ever noticed that just by figuring out how to properly phrase your question, you have already found half your answer? I originally wanted to asked about all this info, and it turned out that most of the answers were obvious from the example data I was about to post, 4 hours earlier.

I still need to figure out how to identify chips made by others than Yamaha or Hitachi though! Anyone knows a way to do that, if the manufacturer is not clearly stated on the chip? I think Toshiba was the other major manufacturer for Sega chips...

TmEE co.(TM) · Post by **TmEE co.(TM)** » Tue Jul 24, 2012 4:48 pm

Yamaha did not make chips themselves, they got them made by Sharp and NEC, and possibly others.

The 4 letter code is relating to where the chip is made or which machine or something. That is what I think, there's many variations.

SMS for example is almost entirely NEC parts.

315-5700 is pin compatible with 315-5487, 315-5660 and 315-5708, all of them are functionally identical and interchangeable. No idea what is the real difference. 5700 would possibly be low power version, due to it being used in Nomad (and nomad does not use any other ASIC than 5700)

Package and text style is what gives away the chip manufacturer.

antime · Post by **antime** » Tue Jul 24, 2012 5:52 pm

Huge wrote:Note for Saturn SCU: last char in date code is G, therefore it could be as far as revision 6 by the console release. This could be possible as if you read the chip version register, it reads as 00 04. So this thing definitely went through many revivions!
It is my theory that when they wanted to make the machine "more powerful" in response of the Playstation tech demos, instead of a second SH2, they added the SCU DSP.

The MANSYS.TXT file included in the SBL6 distribution package describes some of the changes in the SCU revisions, and they're mainly about interrupt handling. The earlier chip revisions were used in early development hardware. The document also gives the following version numbers and revision register values for the chip:
1.5 - 1
2.0 - 2
2.1 - 3
2.2 - 4

Huge · Post by **Huge** » Tue Jul 24, 2012 6:33 pm

TmEE co.(TM) wrote:Yamaha did not make chips themselves, they got them made by Sharp and NEC, and possibly others.

Where can I get this kind of info? It would be very interested in it.

The 4 letter code is relating to where the chip is made or which machine or something. That is what I think, there's many variations.

Well, the respin count was only one theory. It would make sense if they moved production through several facilities, or had several active at the same time making the same chips.
With the numbers I'm seeing, I still think that the last letter is a respin count, though. I guess I need more data to either verify or prove this.

Package and text style is what gives away the chip manufacturer.

Is there a guide somewhere for this, though? All guides I saw mention to identify chips by their logos.

antime wrote:The MANSYS.TXT file included in the SBL6 distribution package describes some of the changes in the SCU revisions, and they're mainly about interrupt handling. The earlier chip revisions were used in early development hardware. The document also gives the following version numbers and revision register values for the chip:
1.5 - 1
2.0 - 2
2.1 - 3
2.2 - 4

Nice catch. I think that kills my DSP theory. But I still think that the addition of a second SH2 was not an after thought, though. Unless we have some early source for that. I recall seeing descriptions from very early 1994, mentioning that the system has 2x SH2.