Help update Amiga-Hardware : Agnus

[alexh]

I've noticed that the information on Amiga-hardware about the different revisions of each of the Amiga Chips is sketchy at best, completely wrong at worse.

It comes from a lifetime of copying inaccurate information from all over the web without first checking that it is correct. I was hoping that some of us here on EAB could help Amiga-Hardware fill in the gaps and correct the mistakes for them?

If we take a look at the Agnus page.

http://www.amiga-hardware.com/showhardware.cgi?HARDID=1448

Agnus (512K) 8361 252125-01 A1000 (NTSC), A2000A (NTSC)
Agnus (512K) 8361 R3
Agnus (512K) 8367 252362-01 A1000 (PAL), A2000A (PAL)
Agnus (512K) 8370 318070-01 A500 (NTSC), A2000 (NTSC)
Agnus (512K) 8371 318071-01 A500 (PAL), A2000 (PAL)
Agnus (1 MB) 8372 318069-02 A500 (PAL), A2000 (PAL) Regular Type
Agnus (1 MB) 8372A 318069-029 A500 (PAL), A2000 (PAL) (changed after some damage)
Agnus (2 MB) 8372AB 318069-03 A3000 (PAL), A3000T (PAL) (came after 8375B)
Agnus (2 MB) 8372B
Agnus (2 MB) 8375 R0 318069-18 A500+ (PAL), A600 (PAL)
Agnus (1 MB) 8375VBB / 8375 R1 318069-16 A500 (PAL), A2000 (PAL)
Agnus (2 MB) 8375 R2 318069-10 A500+ (PAL), A600 (PAL)
Agnus (2 MB) 8375 318069-11
Agnus (1 MB) 8375 318069-17 A500 (NTSC), A2000 (NTSC)
Agnus (2 MB) 8375 318069-19 A3000 (NTSC), A3000T (NTSC)
Agnus (2 MB) 8375 390544-01 A500+ (PAL), A600 (PAL)
Agnus (2 MB) 8375 390544-02 A500+ (NTSC), A600 (NTSC)


The first thing that we could do with finding out is "Are the Agnus's for the A3000 unique to the A3000"? Were they 32-bit or something... I think not, but lets find out.

[alexh]

Agnus (2 MB) 8372AB 318069-03 A3000 (PAL), A3000T (PAL) (came after 8375B)

1) There is NO 8375B in the table (and there should there be!)

2) Look at the date codes on the three chips with the same number (318069-03)

8372AB 318069-03 = 4589, 5089, 5289 = 1989
8372R3 = 0291 = 1991
8372B 318069-03 = 4190, 3591 = 1990/1991
8375B 318069-03 = 5292 = 1992

Three examples of 8372AB
http://www.amiga-hardware.com/download_photos/a3000rev70.jpg
http://www.amiga-hardware.com/download_photos/megachip2000_1_big.jpg
http://www.amiga-hardware.com/download_photos/a3000rev61mb_1_big.jpg

8372B
http://www.amiga-hardware.com/download_photos/a3000rev9_big.jpg

8375B
http://www.amiga-hardware.com/download_photos/achipabig.jpg

In my mind that means that 8372AB came first, then 8372B and then 8375B, correct? Somewhere in the middle came the 8372R3!

An EAB webmaster has said that their 8372B 318069-03 (3791) came from an A3000D. Anyone else? This variant is missing a description.

What the crap is commodore doing having 3 chips with the same code, were they bonded out differently with the same silicon inside? Did they do different things?

[alexh]

Agnus (2 MB) 8375 318069-11

This has no description. I know myself that this is an NTSC variant used in the A500+ / A600. I have one myself in a DKB Megachip I stoopidly ordered from the US before I knew there were NTSC and PAL AGNUS's (or is that Agni?)

Here is some confirmation, an A600 with one 8375 318069-11 (4592) = 1992

http://www.amiga-hardware.com/download_photos/a600mbrev2b_1.jpg

[alexh]

Agnus (2 MB) 8375 R2 318069-10 A500+ (PAL), A600 (PAL)

There is no R2 silked on a 318060-10 Agnus, and hence having it in the table is meaningless.

8375 318069-10 = 0992, 3492, 4292, 4792, 4992 = 1992

http://www.amiga-hardware.com/download_photos/achipa_1_big.jpg
http://www.amiga-hardware.com/download_photos/a500plusrev8amb.jpg
http://www.amiga-hardware.com/download_photos/minimegichip_1_big.jpg
http://www.amiga-hardware.com/download_photos/a600mbrev13.jpg

It is the later version of the 8375 than those marked R0 but I plan to get some date codes in the table to make that obvious.

Need a picture of an agnus with an R2 silked on it to back up this claim, otherwise we should treat it as false.

[alexh]

Agnus (1 MB) 8375VBB / 8375 R1 318069-16 A500 (PAL), A2000 (PAL)

AFAIK VBB is not part of the chip designation. There are other chips with VBB on them. I am almost sure (but need confirmation) that it is something to do with the processing. Anyone confirm this?

I've only got one photo of this part 318069-16 and so I cannot deny 100% hat it never had R1 on it... anyone?

As you can see with a datecode of 1493 = 1993 this is a later agnus than the one in the table marked R2!! This makes much more sense as it has a later revision number.

Anyone any idea why this is marked as a 1Mbyte Agnus??

[alexh]

Agnus (2 MB) 8375 R0 318069-18 A500+ (PAL), A600 (PAL)

There is definately Agnus chips silked 8375R0, but the only picture I've been able to find is of the 390544-01

http://www.amiga-hardware.com/download_photos/3statechip2mb_1_big.jpg

As both these chips are designated PAL I think there has been a mixup, but more information is required to be 100% sure.

A date code of 0392 = 1992 is not consistant. There are earlier versions of the 390544-01 8375 WITHOUT the R0 dated 3591, 4291, 4391, 5291 = 1991. Need some more pictures of R0's to be sure!

http://www.amiga-hardware.com/download_photos/megichip_1_big.jpg
http://www.amiga-hardware.com/download_photos/a500_68010_big.jpg
http://www.amiga-hardware.com/download_photos/a600mbrev1_big.jpg

Hmm, I would assume that they introduced the R0 silk on later chips because they ran two lines? There are some 8375 318069-10 dated 0992 which is only 6 weeks different.

Can ANYONE provide a shot of a 318069-18? To be doubly sure?

P.S. Does anyone know how to work out the month from the Commodore date codes?

[keropi]

wow. I am confused...
nice quest Alexh, wish I could help though....

[alexh]

I've discovered that all chips including CBM use WWYY where WW stands for week 01 to 52 i guess.

Keropi, nothing to get confused over, Amiga-hardware has the wrong data and using their archive of pictures and everyone else out there's computers we should get the table ironed out in a day or so. I wonder if Ian will update it though. I have overwhelming evidence that his table is wrong.

You not got any OCS/ECS machines you could take a look at?

[alexh]

No-one has any input at all? Dont you lot have real amiga bits n bobs lying about? Surely you've not all got the screws in?

[amethyst]

I got my A500+ open right now, here are the markings on the Agnus:

C=CSG
318069-10
8375
3492 20VBB

VBB


Does that help any?

[alexh]

Yup, no R2 on it as I thought...

Need more examples of 8375 -16, 17, -18 and -19. Anyone help here?

Plus date codes from ANYONES Agnus to help work out some time lines.

P.S. Amythyst, while you've got your A500+ open, I'd take out the battery, or at least check it as they tend to leak!

[amethyst]

Funny that you should mention that. I replaced that a couple of weeks ago because I notice it was leaking with powdery corrosion. Cleaned it all up. Luckily it still works.

[lopos2000]

I have got a spare A500 board here:
Fat Agnus
C=CSG
8372A
318069-02
5290 23

[alexh]

lopos no sign of an extra "9" at the end of your 318069-029 perhaps much lighter / fainter?

Does that A500 have 512k chip or 1Mbyte chip?

[billy]

Sorry, I'm about 1200 miles away from my A2000's, 3000's and 4000T. Here's a pic of my A600 Aggie - UK Version.

10278
10279

[lopos2000]

No. The info I gave is correct.
It's 512k version.

[alexh]

Excellent, we need some more info on the RARER Agnus's the, -16 onwards.

Hope we can get some more info. The date codes are good though. I am rebuilding a new table now.

[alexh]

More information acquired.

All 8372 Agnus's were both PAL and NTSC unlike their predecessors.

For the 8375 series APPEAR that they went back to having PAL and NTSC versions, but I have yet to 100% confirm it (although the famous Dean K Brown thinks so).

Dean also said that the Angus's dont all have the same pinout and are not interchangable. That is worrying. I need to confirm that.

[NOB]

Here is a pic of an CDTV Agnus /Rev. 2.3 /1 MB Chip

[keropi]

nope sorry, only 1200 here...

[alexh]

Ian Chapman of The Big Book of Amiga Hardware has agreed to let me try to find out all the problems with the details on their website (regarding custom chip revisions) I hope that enough people contribute Agnus information so that I can then go on to update the other chips.

Agnus was the worse, because it had the most number of revisions by far.

[Zack]

I have a bunch of extra chips lying around here. The versions are as follows:

---
318069-10
8375
4992 20VBB

---
318069-10
8375
4192 20VBB

---
318069-10
8375
5192 20VBB

---
8372R3
318069-2
0291 23

---
8372A
318069-02
0491 23

---
8372A
318069-02
2291 23

---
8372A
318069-02
0291 23

---
318069-16
8375
1493 20VBB

---
318069-16
8375
4592 20VBB

[NOB]

Found a picture tha shows a Amiga 1000 Agnus:



MOS
8367R0
0986 S

[alexh]

@Zack

you da man!! Any chance of posting a photo of the 8372R3?

@Nob,

Cool... the A1000 Agnus was DIP rather than PLCC which is interesting. ALSO I've found a photo of the mysterious 8361 R3 on the Amiga-hardware website inside the earliest example of an A1000, and more importantly a date code. Another mystery cleared up.

[Zack]

Front and back of the 8372R3:

10319 10320

Btw. at least all similar sized chips have info on the back as well (Lisa, Gayle, Ramsey, DMAC, Alice, Agnus and Super Buster. Would this info be useful?

[alexh]

I dont think that the information on the back of the chips is important at the moment Zack. It might be in the odd situation where not all the information about the version is on the front of the chip. In this case I think it is.

Thanks for the pics. It's important you notice that you didnt 100% transfer the data from the image to your posting before...

8372R3
318069-02
0291 23

You missed out the "0" in "02" but you did such a great job and listed so many chips,we'll let you off this time ;)

Anyone else got Agnus chips and numbers, rare ones with the "Rx" next to them, or later ones -16 and later are what I'm missing most information.

Also if anyone has an 8361(R3) or 8367(R0) and can tell me how to distinguish a PAL one from an NTSC one I'd be greatful.

[alexh]

Picture of the new Agnus found.

Agnus (512K) 8367R0 ??????-?? A1000(PAL)

0986, 1486 = 1986

There is no commodore product code on A1000 chips, does anyone know if the product code was on the reverse of the chip?

[Overdoc]

Well, I currently got something like 50 A-500s, a few A-600 and and A-500+ and a 3000, but although I would like to help you I simply don't have enough time to open up all of them.
But I could have a look at the A-500+, because I'd like to check them anyway.

[breech]

Sorry to drag up an old thread. I added a note to the Wikipedia section on Agnus as I found that my b2000 R4 PAL Agnus happily addresses 1mb slow ram after modification.

MOS 8371
5827 21

I have not seen this chip listed anywhere. I am interested to see if this chip allows pal/ntsc switching with a hard hack.
cheers.

[Zetr0]

It might be best to point out that theres some error in the first post.

the A600 had two different 8375 Agnus

a 2MB 8375 3190544 (rev 1.0 type 1 *image from amiga-hardware* (http://www.amiga-hardware.com/download_photos/a600mbrev1_big.jpg))
a 2MB 8375 318069 (most common A600 rev 1.0 *type2* to 1.5)


from my investigations the 8375 318069-xx series is pin compatible with earlier models like the 8370/8371/8732 and the latters A/B and AB varients.

Other than the DIP based Agnus on the A1000 the 8375 3190544 series is exculisve to the A500+ and the first limmited release of the A600. the 8375 3190544 was also prevailent on a lot of 3rd party chip upgrades and is NOT pin compatible with the other 8375 revisions of Agnus.

[Toni Wilen]

Sorry to drag up an old thread. I added a note to the Wikipedia section on Agnus as I found that my b2000 R4 PAL Agnus happily addresses 1mb slow ram after modification.

MOS 8371
5827 21

I have not seen this chip listed anywhere. I am interested to see if this chip allows pal/ntsc switching with a hard hack.
cheers.

8371 is normal OCS Agnus. 512k chip, PAL hardwired. Second line is bogus datecode, nothing to do with chip type. Week "58" of year 1927 :)

[breech]

8371 is normal OCS Agnus. 512k chip, PAL hardwired. Second line is bogus datecode, nothing to do with chip type. Week "58" of year 1927 :)

Except mine is doing 1024k :cheese

[Toni Wilen]

Except mine is doing 1024k :cheese

Slow RAM? Slow ram does not require anything special from Agnus as long as there is circuitry that does memory mapping and refresh.

Also 1M hack can "work" with OCS Agnus, upper 512k of chip ram is still slow ram and it can't be accessed by Agnus DMA.

[alexh]

It might be best to point out that there's some error in the first post. the A600 had two different 8375 Agnus
There is no error. It lists both types of Agnus chips to A600. It was a quote from the then Agnus page on BBoAH.

Agnus (2 MB) 8375 R2 318069-10 A500+ (PAL), A600 (PAL)
Agnus (2 MB) 8375 390544-01 A500+ (PAL), A600 (PAL)

[desantii]

Here is an interesting tidbit, from my A500+

8375R0
390544-01
3291 20

THis is a PAL chip, but when fitted with a 3.1 kickstart (tried last night), you can select PAL/NTSC and works great.

I thought those chips were PAL only!

[Jope]

No, all ECS Agnuses (1M and 2M) are software switchable.

You're thinking of hardware switching (ground a certain leg to select the boot up screenmode), which some Agnuses indeed can't do.

[desantii]

I thought those were not switchable as there is a specific number for the PAL and NTSC

Agnus (2 MB) 8375 390544-01 A500+ (PAL), A600 (PAL)
Agnus (2 MB) 8375 390544-02 A500+ (NTSC), A600 (NTSC)

[alexh]

They all have the register inside them allowing them to change screen mode in software.

As Jope quite rightly said they don't have the pin which allows their default mode to be configured on the PCB and so are not interchangeable.

[desantii]

Ok, thanks for the input

[desantii]

Actually I question, are the 1mb 8375s pin compatible with the regular 8372a 1mb chips?

Disregard, I did not read the answer above!

[desantii]

FYI, not sure on all not being pin compatible. Just swaped this one (from an NTSC megachip)
Agnus (2 MB) 8375 318069-11


To this one (to try to create a PAL megachip) and it works great
Agnus (2 MB) 8375 390544-01

It might be best to point out that theres some error in the first post.

the A600 had two different 8375 Agnus

a 2MB 8375 3190544 (rev 1.0 type 1 *image from amiga-hardware* (http://www.amiga-hardware.com/download_photos/a600mbrev1_big.jpg))
a 2MB 8375 318069 (most common A600 rev 1.0 *type2* to 1.5)


from my investigations the 8375 318069-xx series is pin compatible with earlier models like the 8370/8371/8732 and the latters A/B and AB varients.

Other than the DIP based Agnus on the A1000 the 8375 3190544 series is exculisve to the A500+ and the first limmited release of the A600. the 8375 3190544 was also prevailent on a lot of 3rd party chip upgrades and is NOT pin compatible with the other 8375 revisions of Agnus.

[hexsane]

I have some confusion about this part:

Agnus (1 MB) 8372A 318069-029 A500 (PAL), A2000 (PAL) (changed after some damage)

I have a Rev 6.2 A2000

The agnus part # is 8372A 318069-029 (9 is floating and some distance from the -02)

Chip date is: 0990

Although you're reporting here this chip is 1MB SysInfo reports it as a 2MB.

Bug in SysInfo or newly discovered information?

Should I bother adding the additional 1MB of RAM to the motherboard or would it help confirm this as a 2MB chip?

-Matt

[cosmicfrog]

sysinfo reports a 2MB angus in my a2000 rev 6.2
still havent had time to read the number of angnus

if it is indeed true it makes doing the 2MB chipram mod easier

[alexh]

The 8372A 318069-029 _is_ a 1Mbyte part.

99% sure this is a bug / limitation of the SysInfo software.

It must not be possible to recognise that some 8372/8375 Agnus chips are 1Mbyte. Perhaps there is no way to differentiate? After all they are the same silicon, just bonded out differently.

You cannot use a 2Mbyte part directly in an A2000 motherboard without an adapter like a MiniMegi / MegAChip.

[rkauer]

You cannot use a 2Mbyte part directly in an A2000 motherboard without an adapter like a MiniMegi / MegAChip.

Oh, yeah?:p (http://www.mylinuxisp.com/%7Ejdbaker/oldsite/documents/2MBAgnushack_v1.4.txt)

[alexh]

Oh, yeah?:p (http://www.mylinuxisp.com/%7Ejdbaker/oldsite/documents/2MBAgnushack_v1.4.txt)
Yeah what?? That hack is a DIY equivalent to the MegAChip / MiniMega adapter.

[rkauer]

But the fact the user can hack the board dropping an 8375 Agnus (the only one that works!) directly on the Agnus socket of the A2000.

I still want to hack a rev6 board this way to validate the idea.

Thing is this Agnus is hard to get, even knowing A3000 Amigas use to die for nothing, leaving lots of spare parts...

[alexh]

Never seen an 8375 318069-19 in the flesh. I guess as you say they are very rare as A3000 came with a 2Mbyte Agnus so they probably made very few.

The 8372 318069-03 is probably easier to find?

[rkauer]

As far as I know, only the early A3000 Agnus is pin compatible with the A500/2000 Agnii, and even then not totally compatible as some signals are missing (read the hack on the link above).

But for the solely purpose of making an A500/2000 capable of 2Mb chip RAM without a plug-in board, then it is perfect.

So the model to search for still is the 8372AB, and maybe the 8375 318069-19.

Like I said, I need to find a pair to hack a pair of boards to validate this.

I know for sure this hack works on the rev6 A2000 as one Australian user did this to both A2000 he have.

[AndyLandy]

If it helps, my A3000 is a late-revision board (9.02) and has Agnus part:
C= CSG
8372B
2190 23

I'm led to believe that later-revision A2000s (r6.3-r6.5) have a 2MB Agnus on-board as standard, but only 1MB is populated on the board.

I'd also like to know how much ChipRAM is on r4.x A2000 boards? I've gotten conflicting reports as to whether it's 512k Chip/512k Ranger, or 1MB Chip.

Does the 8371 Agnus even support 1MB Chip? I don't think it does.

[8bitbubsy]

8371 supports 512kB chip RAM max, so 4.x boards were populated with 8371 and 512kB chip/512kB fake-fast. I think 4.5 (at least 5.x) came with 8372A which can address 1MB, but I am uncertain if the 4.5 ones were setup with 1MB or with 512/512.

[alexh]

I'm led to believe that later-revision A2000s (r6.3-r6.5) have a 2MB Agnus on-board as standard, but only 1MB is populated on the board.
I don't think this is true. Non of the B2000 motherboards I've looked at can take 2MB Agnus chip directly.

I've seen the very last B2000 motherboards came with 1MB 8375 Agnus chips which is perhaps where the confusion may come from?

[AndyLandy]

I don't think this is true. Non of the B2000 motherboards I've looked at can take 2MB Agnus chip directly.

I've seen the very last B2000 motherboards came with 1MB 8375 Agnus chips which is perhaps where the confusion may come from?

I didn't even know there was a 1MB 8375 Agnus! *sigh*, more C= inconsistencies, I thought it was just the 8372A that was 1MB.

The C2000 motherboards are sometimes claimed to come with "Full ECS Chipset", I've known people say they've had 2MB ChipRAM added to their board without changing the Agnus, but I cannot for the life of me find a photo of an r6.3+ A2000 board.

*edit* -- This photo is of Psodas' r6.3 from Amibay:
http://img710.imageshack.us/img710/6875/img3861o.jpg

C= CSG
318069-02
8372A
1692 23

So, they were still cutting 1MB 8372 Agnuses in early '92. Maybe it is a myth that late-A2000s have 2MB Agnuses.

[alexh]

I didn't even know there was a 1MB 8375 Agnus! *sigh*, more C= inconsistencies
No inconstancies. This is well documented. There are lots of 8375's including 1Mbyte versions.

I thought it was just the 8372A that was 1MB.
Nope, 8375 318069-16 & 318069-17 are both 1Mbyte

I've known people say they've had 2MB ChipRAM added to their board without changing the Agnus, but I cannot for the life of me find a photo of an r6.3+ A2000 board.
They must be using a MiniMegi chip.

Maybe it is a myth that late-A2000s have 2MB Agnuses.
Not a myth I've ever heard.

[rkauer]

There are some A2000 directly capable of 2Mb chip RAM. There a hack floating around guiding A2000 (rev6.x ONLY) to show how you can hack the board.

But it indeed requires an 8372AB Agnus.