Amiga 600 Ressurection - Cap damage, RAM errors - Page 2

telemonster · 23 April 2020, 17:01

I took a few days break from it, I find that returning with a clear mind often leads to new discovery.

How common are failed customs? I'm getting worried that the issue is the Agnus or something -- but a the same time it's hard to tell. Also, looking at the PCB I think there are parts near a lot of the customs that would prevent adding sockets for them all.

The Amiga 600 worked fine, then the cap issue hit. I wouldn't think powering it up with bad caps would damage the customs? The 5v line should still have been fine unless somehow the bad caps for the +12/-12 caused issue but those tie to GND.

psoma · 25 April 2020, 11:20

I received 2 boards from someone to check out... both had DRD0 on Agnus shorting to GND, one also had Paula going to Earth on DRD0 and blown RAM chips. It happens, but it's not super common.

I'd buzz each of the data and address pins of Agnus to both GND and VCC just as a check. Everything should be in the kOhm -> MOhm range. If you see anything under that, mark it for investigation.

telemonster · 26 April 2020, 01:26

Thanks for the reply and info. I just buzzed through all the data and address lines from the Agnus (as well as some of them on the CPU side) and no shorts to ground or VCC.

I suppose swapping the DRAM ICs is an option but that would be more of a random act than troubleshooting a known issue.

cpiac64 · 26 April 2020, 16:55

you can add a601 or add ram with Victor Trucco mod _https://www.victortrucco.com/Commodore/ExpansaoA600/ExpansaoA600

and exchange ram internal with external

telemonster · 26 April 2020, 20:57

Quote:

Originally Posted by cpiac64

you can add a601 or add ram with Victor Trucco mod _https://www.victortrucco.com/Commodore/ExpansaoA600/ExpansaoA600

and exchange ram internal with external

Very cool! Thanks for the info!! One thing though, I would think that any expansion RAM card (or that add on) -- since it's tied to different Row/Column lines it's always going to be the upper side of the RAM. No way to make it lower "default" RAM?

Also, any kind of issues with any of the ICs on the Address or Data bus and it would cause issues with the trapdoor RAM... so that's a strike. I'm going to ask a local who might have an A600 if he has a RAM expansion kicking around.

I was thinking this morning I might use my bench supply to feed +12 and +5 on the floppy connector, just the rule out any issues with DC supply from the original PSU or the input filtering. I don't see any jitter on the scope but I'm using a DSO not the analog so...

telemonster · 26 April 2020, 22:44

No serial output using bench supply since missing the -12v or whatever, but colors flashed on screen mostly red so I think that is bad.

I found a video card in the garage (S3 Trio64) that has two memory ICs. I looked up datasheets on the one that was on the Amiga vs the one on the video card. Video card is faster, but looking up info it looks like 70ns DRAM vs original 80ns isn't an issue as the refresh window can go all the way out to 130 or 140ns or something.

I tried desoldering memory IC from video card with hot air reflow and that was a mess. Moved over to chipquick and that worked pretty okay. I've used the chipquick stuff with great success on NeoGeo and Konami arcade hardware, just have to make sure to get it off since it's brittle cold solder joint risk if left behind.

So I think I'm comfortable removing the RAM from the A600 motherboard and could sub in these other RAM ICs. I was thinking maybe if I desolder the originals, then just place the replacement on weighted down (to make sure alignment is spot on) and power it up -- maybe I can get by without soldering down the test chips?

Once it's time to solder down new chips -- that is where I get worried. I can order some Kapton tape and maybe some paste then go the hot reflow route to tack them down.

psoma · 27 April 2020, 00:11

Good stuff man - glad you are persisting with this. I used ChipQuik to remove the RAM and Paula and swear by it now... so much easier than the hot air (just don't drip it onto plastic)!

If it gives you hope - I re-soldered 2 x RAM, Paula and Agnus by hand. It wasn't easy, but it wasn't heart-in-mouth stuff either. I used a fine tip, decent (but not excessive) amount of solder on the pad and then a "heat and drag up" motion to bridge the pad to the pins. Yeah, I'd occasionally smear a couple of pins at once, but solder-wick sorted that out pretty quickly.

telemonster · 27 April 2020, 00:24

Cool! Yea, I watched a friend solder a jtag header on something they were ... extracting firmware off of by hand, using a portable microscope and nice iron. I think he wicked it off.

Hopefully my notes on here serve to help someone else.

Some sorta interesting things I guess. First I removed the top RAM IC, noticed nothing too different. I tried holding my spare on the pads, and when I got it to come up same error as before.

Then I removed the bottom two caps and the bottom RAM IC. At this point there is no RAM on board. Power it on, DiagROM basically puts out the same error. BUT... I notice something... when it's writing to all the addresses, when it's reading back the contents there are cases where it's showing a zero instead of a 1.

https://imgur.com/a/DcT3v9T

First photo is the board with the chips all removed. I chip quicked, then soldered down normal solder and removed it with a Hakko 808 desoldering gun twice. Then alcohol and paper towel to clean.

Plot thickens... badly.

Soooo. A friend and I went together on some Amiga hardware. There was a few chips in a bag, and just on a whim I just looked to see what they are. 2.04 Kickstart, and a 8372A with a yellow dot on it. It's a different agnus than what the A600 seems to have on it. The chip is in a bag, and the bag was opened. There is another baggie just like it, that is unopened. It has a sticker card that says "These are your serial numbers. Please place one on each of the following: The board, the manual, the registration card, the disk (if included.)

Oddly the numbers match the loose chip, but the chip markings look different even though it too is a 8372A. The new in package one has a C=CSG on it, the old one has thin line CSG on it.

Anyways, they're apparently Agnus chips but different model from the A600 one. Hmp.

psoma · 28 April 2020, 23:36

What's the part number on the Agnuses? I replaced a 318069-10 (yellow dot) successfully with a 390544-01... but both of those are 8375s.

demolition · 29 April 2020, 00:48

I found these pictures that fit the description regarding the difference in the logo:
https://www.analogic.co.uk/media/cat...18069-02_1.jpg
http://www.bigbookofamigahardware.co...os/agnus_3.jpg
Same part number, both 8372A. I'm assuming the reason for the different look may just be in the manufacturing location (or that they reworked the tools from 91 to 92).

telemonster · 29 April 2020, 17:57

Quote:

Originally Posted by psoma

What's the part number on the Agnuses? I replaced a 318069-10 (yellow dot) successfully with a 390544-01... but both of those are 8375s.

I will check when I get home. In this case they are definitely the wrong parts. The two spares I have are 1MB and for 500/2000 and the 600 uses a 2MB chip. Looking at wikipedia the pinouts are vastly different so no go there.

I should take another detailed look at any VIA's and traces near the caps and then if those are all good think about swapping some of the customs. From memory everything I found that was in a really bad area was related to video or audio. And to be honest, some of the pics I was seeing of other people's boards -- theirs looked worse.

The Agnus chip gets pretty warm during DiagROM but that might be normal.

I assume the 75 Agnuses are hard to get? I see a few on eBay and the prices are up there.

telemonster · 29 April 2020, 17:58

Quote:

Originally Posted by demolition

I found these pictures that fit the description regarding the difference in the logo:
https://www.analogic.co.uk/media/cat...18069-02_1.jpg
http://www.bigbookofamigahardware.co...os/agnus_3.jpg
Same part number, both 8372A. I'm assuming the reason for the different look may just be in the manufacturing location (or that they reworked the tools from 91 to 92).

Interesting! That looks similar to the two I have. I will take pics tonight.

demolition · 29 April 2020, 18:16

Quote:

Originally Posted by telemonster

I assume the 75 Agnuses are hard to get? I see a few on eBay and the prices are up there.

They are not hard to get, just somewhat expensive.

You can get NOS ones here (NTSC though):
https://icomp.de/shop-icomp/en/shop/...69-11-nos.html

Otherwise they are generally salvaged from 600 and 500+ boards. Both of those were relatively low volume, making the supply for 8375 chips low.

telemonster · 03 May 2020, 05:16

Okay, sorry for delay.

New in wrapper with serial number stickers Agnus = 318069-02 9 (8372A)

Opened wrapper yellow dot = CSG 318069-02(small mark) (8372A)

Amiga 600 of course has a NTSC 8375

Looking at the diagrom data... the reading of the RAM (even though they are missing) always seems to have problems on two specific bits... what looks like bits 15 and 16.

I might swap the chips at U21 and U22 just for sport. If the error moves that would indicate something is wrong with one of them. I have no idea what role problems with clock signals or other things could be playing.

demolition - Mine is an NTSC model, so that actually is a pretty decent deal and I would totally go for that to fix my machine. The downside at this point is it might not be that chip, but at the end of the day I would just have a spare out of it. Hmmm tempting....

stormahead · 27 July 2023, 09:56

Quote:

Originally Posted by telemonster

Okay, sorry for delay.

New in wrapper with serial number stickers Agnus = 318069-02 9 (8372A)

Opened wrapper yellow dot = CSG 318069-02(small mark) (8372A)

Amiga 600 of course has a NTSC 8375

Looking at the diagrom data... the reading of the RAM (even though they are missing) always seems to have problems on two specific bits... what looks like bits 15 and 16.

I might swap the chips at U21 and U22 just for sport. If the error moves that would indicate something is wrong with one of them. I have no idea what role problems with clock signals or other things could be playing.

demolition - Mine is an NTSC model, so that actually is a pretty decent deal and I would totally go for that to fix my machine. The downside at this point is it might not be that chip, but at the end of the day I would just have a spare out of it. Hmmm tempting....

Did you ever solve this? Same issue with a 1200 I'm working on

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Amiga 600 - Capacitor damage, PCB layout diagrams?	telemonster	support.Hardware	11	23 October 2018 22:49
Amiga 600: virtual RAM possible?	Supamax	support.Hardware	8	03 September 2012 23:59
amiga 600 (4mb ram) freeze	nixxj	project.WHDLoad	4	06 August 2012 23:02
Amiga 600 Ram?	Amiga Forever	support.Hardware	4	25 August 2011 20:28
WTB: Amiga 600 fast ram expansion	Photon	MarketPlace	9	13 December 2008 21:58

23 April 2020, 17:01	#21
telemonster Registered User Join Date: Apr 2020 Location: Northern Virginia Posts: 32	I took a few days break from it, I find that returning with a clear mind often leads to new discovery. How common are failed customs? I'm getting worried that the issue is the Agnus or something -- but a the same time it's hard to tell. Also, looking at the PCB I think there are parts near a lot of the customs that would prevent adding sockets for them all. The Amiga 600 worked fine, then the cap issue hit. I wouldn't think powering it up with bad caps would damage the customs? The 5v line should still have been fine unless somehow the bad caps for the +12/-12 caused issue but those tie to GND.

25 April 2020, 11:20	#22
psoma Registered User Join Date: Nov 2019 Location: Melbourne, Australia Posts: 193	I received 2 boards from someone to check out... both had DRD0 on Agnus shorting to GND, one also had Paula going to Earth on DRD0 and blown RAM chips. It happens, but it's not super common. I'd buzz each of the data and address pins of Agnus to both GND and VCC just as a check. Everything should be in the kOhm -> MOhm range. If you see anything under that, mark it for investigation.

26 April 2020, 01:26	#23
telemonster Registered User Join Date: Apr 2020 Location: Northern Virginia Posts: 32	Thanks for the reply and info. I just buzzed through all the data and address lines from the Agnus (as well as some of them on the CPU side) and no shorts to ground or VCC. I suppose swapping the DRAM ICs is an option but that would be more of a random act than troubleshooting a known issue.

26 April 2020, 16:55	#24
cpiac64 Registered User Join Date: Apr 2010 Location: Italy Posts: 1,136	you can add a601 or add ram with Victor Trucco mod _https://www.victortrucco.com/Commodore/ExpansaoA600/ExpansaoA600 and exchange ram internal with external

26 April 2020, 22:44	#26
telemonster Registered User Join Date: Apr 2020 Location: Northern Virginia Posts: 32	No serial output using bench supply since missing the -12v or whatever, but colors flashed on screen mostly red so I think that is bad. I found a video card in the garage (S3 Trio64) that has two memory ICs. I looked up datasheets on the one that was on the Amiga vs the one on the video card. Video card is faster, but looking up info it looks like 70ns DRAM vs original 80ns isn't an issue as the refresh window can go all the way out to 130 or 140ns or something. I tried desoldering memory IC from video card with hot air reflow and that was a mess. Moved over to chipquick and that worked pretty okay. I've used the chipquick stuff with great success on NeoGeo and Konami arcade hardware, just have to make sure to get it off since it's brittle cold solder joint risk if left behind. So I think I'm comfortable removing the RAM from the A600 motherboard and could sub in these other RAM ICs. I was thinking maybe if I desolder the originals, then just place the replacement on weighted down (to make sure alignment is spot on) and power it up -- maybe I can get by without soldering down the test chips? Once it's time to solder down new chips -- that is where I get worried. I can order some Kapton tape and maybe some paste then go the hot reflow route to tack them down.

27 April 2020, 00:11	#27
psoma Registered User Join Date: Nov 2019 Location: Melbourne, Australia Posts: 193	Good stuff man - glad you are persisting with this. I used ChipQuik to remove the RAM and Paula and swear by it now... so much easier than the hot air (just don't drip it onto plastic)! If it gives you hope - I re-soldered 2 x RAM, Paula and Agnus by hand. It wasn't easy, but it wasn't heart-in-mouth stuff either. I used a fine tip, decent (but not excessive) amount of solder on the pad and then a "heat and drag up" motion to bridge the pad to the pins. Yeah, I'd occasionally smear a couple of pins at once, but solder-wick sorted that out pretty quickly.

27 April 2020, 00:24	#28
telemonster Registered User Join Date: Apr 2020 Location: Northern Virginia Posts: 32	Cool! Yea, I watched a friend solder a jtag header on something they were ... extracting firmware off of by hand, using a portable microscope and nice iron. I think he wicked it off. Hopefully my notes on here serve to help someone else. Some sorta interesting things I guess. First I removed the top RAM IC, noticed nothing too different. I tried holding my spare on the pads, and when I got it to come up same error as before. Then I removed the bottom two caps and the bottom RAM IC. At this point there is no RAM on board. Power it on, DiagROM basically puts out the same error. BUT... I notice something... when it's writing to all the addresses, when it's reading back the contents there are cases where it's showing a zero instead of a 1. https://imgur.com/a/DcT3v9T First photo is the board with the chips all removed. I chip quicked, then soldered down normal solder and removed it with a Hakko 808 desoldering gun twice. Then alcohol and paper towel to clean. Plot thickens... badly. Soooo. A friend and I went together on some Amiga hardware. There was a few chips in a bag, and just on a whim I just looked to see what they are. 2.04 Kickstart, and a 8372A with a yellow dot on it. It's a different agnus than what the A600 seems to have on it. The chip is in a bag, and the bag was opened. There is another baggie just like it, that is unopened. It has a sticker card that says "These are your serial numbers. Please place one on each of the following: The board, the manual, the registration card, the disk (if included.) Oddly the numbers match the loose chip, but the chip markings look different even though it too is a 8372A. The new in package one has a C=CSG on it, the old one has thin line CSG on it. Anyways, they're apparently Agnus chips but different model from the A600 one. Hmp.

28 April 2020, 23:36	#29
psoma Registered User Join Date: Nov 2019 Location: Melbourne, Australia Posts: 193	What's the part number on the Agnuses? I replaced a 318069-10 (yellow dot) successfully with a 390544-01... but both of those are 8375s.

29 April 2020, 00:48	#30
demolition Unregistered User Join Date: Sep 2012 Location: Copenhagen / DK Age: 44 Posts: 4,190	I found these pictures that fit the description regarding the difference in the logo: https://www.analogic.co.uk/media/cat...18069-02_1.jpg http://www.bigbookofamigahardware.co...os/agnus_3.jpg Same part number, both 8372A. I'm assuming the reason for the different look may just be in the manufacturing location (or that they reworked the tools from 91 to 92).

03 May 2020, 05:16	#34
telemonster Registered User Join Date: Apr 2020 Location: Northern Virginia Posts: 32	Okay, sorry for delay. New in wrapper with serial number stickers Agnus = 318069-02 9 (8372A) Opened wrapper yellow dot = CSG 318069-02(small mark) (8372A) Amiga 600 of course has a NTSC 8375 Looking at the diagrom data... the reading of the RAM (even though they are missing) always seems to have problems on two specific bits... what looks like bits 15 and 16. I might swap the chips at U21 and U22 just for sport. If the error moves that would indicate something is wrong with one of them. I have no idea what role problems with clock signals or other things could be playing. demolition - Mine is an NTSC model, so that actually is a pretty decent deal and I would totally go for that to fix my machine. The downside at this point is it might not be that chip, but at the end of the day I would just have a spare out of it. Hmmm tempting....

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)