Mega 4 ST RAM issues

[robbo007]

Hi guys,

I just picked up a Mega 4 ST and it seems sysinfo only see 2048 RAM. From what I understand the Mega 4 should have 4MB of ram onboard right? Or was this an add-on feature?

 

Bank 0 $0424 2048

Bank 1 $0424 0

Bank 0 $FF8001 2048

Bank 1 $FF8001 0

 

Does it only count bank 0? or could I have a IC issue?

Regards,

 

 

 

Edited May 9 by robbo007
send too soon

[TGB1718]

Looks like you either have some bad ram in bank 1 or the address decoding chips.

[robbo007]

ok thanks for confirming. I've found the schematics. I'm no electronic wiz. I might try to see if the solder joints are ok as this was shipped from one country to another. Can I do some sort of continuity test on bank 1 ram or the address decoding chips? Are both ram and decoding chips still available to buy?

[Forrest]

Atari sold a Mega2 ST with 2 MB of RAM and a Mega4 ST with 4 MB. Both use the same logic board, but half of the RAM chip locations are empty of the Mega2 ST. Have you looked at your logic board?

Edited May 10 by Forrest

[robbo007]

I did. 4 rows of 8 ICS. So its definitely the 4MB version. If Bank 1 is not working could it be the address decoder chip? Or if one Ram IC is bad would it short the entire 2MB in bank 1? Could it be the ceramic caps on each ram IC? Apart from the IC being warm what's the best way to test for a failed IC?

Edited May 10 by robbo007

[TGB1718]

4 hours ago, robbo007 said:

Could it be the ceramic caps on each ram IC?

Unlikely unless it's damaged.

 

If you have some version of BASIC, you could try writing a short loop to poke say 170 (Binary %10101010) into

memory above 2MB i.e. starting at 2097152 and peeking back the value, check if its 170, if it's 255 then likely

a decoding logic fault, any other value would indicate a chip fault, if it's a consistent value, then you could

determine which chip it is byte the failed "bit(s)"

[phoenixdownita]

Watch out that the RAM chips that mega st has are 1 bit wide, to read a word (16 bits) each chip in the bank returns one of the bits, so if a chip is bad all words in those 2MB would have a bad bit.

 

I am not sure how at startup TOS detects the extra RAM and size, but the suggestion to write a few bit patterns and read back at an address above 2MB should tell you if the chips work unless the ST mmu blocks it (we know for example you can’t write in the 128KB cart space, you’ll get a bus error I think if you try).

 

So you can use 2 byte patterns %10101010 (the 170 above) read it back then %01010101 (85) and read it back, or even just %11111111 (255) then read followed by %00000000 (0) and read.

Working at the byte level you would be able to isolate upper/lower half bank, the bits that do not match the written pattern upon read should indicate which chip is bad.

 

You need two complementary patterns because the bad bit(s) may be stuck high or low (bus can be both pull-up or pull down, not sure what Atari did) and using the 2 complementary patterns you’ll cover all cases.

 

Edited May 10 by phoenixdownita

[snarkdluG]

Try YAART. You have to tell YAART to test the specific ram though. 

Edited May 10 by snarkdluG

[TZJB]

8 hours ago, robbo007 said:

I did. 4 rows of 8 ICS. So its definitely the 4MB version. If Bank 1 is not working could it be the address decoder chip? Or if one Ram IC is bad would it short the entire 2MB in bank 1? Could it be the ceramic caps on each ram IC? Apart from the IC being warm what's the best way to test for a failed IC?

 

The board looks pretty clean and there is nothing obviously amiss. If ceramic capacitors go faulty the 5V line would short circuit and nothing would work. However they tend to physically break and go open circuit if anything.

 

M511000A Fast Page DRAM is still readily available in various speeds, yours being 100nS, but faster ones usually work. I hope that you are good with a de-soldering iron?

 

As DRAM Bank 0 is working and DRAM Bank 1 is missing as you have noted, it is unlikely to be a faulty MCU causing the missing 2MB Bank. DRAM Bank 1 has it's own signals called RAS2, CAS2H, CAS2L.

 

As well as the Bank 0 equivalents, these signals originate from U30 STMCU and go to U59 74LS244 which buffer the signals. This is a common point to look at, although again, it is shared by DRAM Bank 0. U59 then feeds buffer resistors R71, R72, R74 respectively. An oscilloscope will help with diagnostics. Do you have access to one and what other diagnostic tools do you have available?

 

Depending on how the DRAM has failed, one non-invasive test you could do, initially with a cold and powered off machine, is to place a drop of isopropanol onto each of the Bank 1 DRAM chips, which would be the two columns of eight DRAM chips U60 to U67 and U68 to U75, and observing their rate of evaporation on applying power. Faulty DRAM can get hotter and cause quicker evaporation (but not always).

 

If you don't already own one, I can recommend the SidecarT. It will allow you to run many online cartridge images including the Atari ST Diagnostics, without dedicating an actual cartridge to the job. It can also run a floppy disk emulator and has also evolved to emulate a hard drive in a beta firmware.

 

https://sidecartridge.com/

 

[robbo007]

Thanks for the pointers.

My is coming from Germany to stay for a few days and he is very good at electronics. He is bringing his electronics kit. So we will take a closer look this weekend. I notice Side car is a Spanish product. I'm located in Spain I've tested the Mastery hard disk emulator and that works well. I was thinking of getting a Mega 30 hard disk. I've seen some on ebay not too expensive.

 

Here is the sysinfo report:

 

Edited May 15 by robbo007
added photo

[robbo007]

On 5/10/2024 at 9:00 PM, TZJB said:

 

The board looks pretty clean and there is nothing obviously amiss. If ceramic capacitors go faulty the 5V line would short circuit and nothing would work. However they tend to physically break and go open circuit if anything.

 

M511000A Fast Page DRAM is still readily available in various speeds, yours being 100nS, but faster ones usually work. I hope that you are good with a de-soldering iron?

 

As DRAM Bank 0 is working and DRAM Bank 1 is missing as you have noted, it is unlikely to be a faulty MCU causing the missing 2MB Bank. DRAM Bank 1 has it's own signals called RAS2, CAS2H, CAS2L.

 

As well as the Bank 0 equivalents, these signals originate from U30 STMCU and go to U59 74LS244 which buffer the signals. This is a common point to look at, although again, it is shared by DRAM Bank 0. U59 then feeds buffer resistors R71, R72, R74 respectively. An oscilloscope will help with diagnostics. Do you have access to one and what other diagnostic tools do you have available?

 

Depending on how the DRAM has failed, one non-invasive test you could do, initially with a cold and powered off machine, is to place a drop of isopropanol onto each of the Bank 1 DRAM chips, which would be the two columns of eight DRAM chips U60 to U67 and U68 to U75, and observing their rate of evaporation on applying power. Faulty DRAM can get hotter and cause quicker evaporation (but not always).

 

If you don't already own one, I can recommend the SidecarT. It will allow you to run many online cartridge images including the Atari ST Diagnostics, without dedicating an actual cartridge to the job. It can also run a floppy disk emulator and has also evolved to emulate a hard drive in a beta firmware.

 

https://sidecartridge.com/

 

@TZJB

Thanks for the clues. So I did have a spare working 74LS244 and I swapped it out. Same problems. So the 74LS244 is not the culprit. After using a thermal camera I saw the DRAM U64 is completely cold and all other DRAM IC's have some heat showing on the camera. I will try and resolder the pins just in case there is some cracked solder there or a broken contact. If that does not work do you know where I can source the M511000A ICs? As I suspect this DRAM is broken. They are quite expensive on ebay.

 

If U64 is bad would that show the entire BANK 1 as empty?

 

[TZJB]

4 hours ago, robbo007 said:

@TZJB

Thanks for the clues. So I did have a spare working 74LS244 and I swapped it out. Same problems. So the 74LS244 is not the culprit. After using a thermal camera I saw the DRAM U64 is completely cold and all other DRAM IC's have some heat showing on the camera. I will try and resolder the pins just in case there is some cracked solder there or a broken contact. If that does not work do you know where I can source the M511000A ICs? As I suspect this DRAM is broken. They are quite expensive on ebay.

 

If U64 is bad would that show the entire BANK 1 as empty?

 

 

Great idea to use a thermal camera. You may well have found the problem, but I do not have personal experience of this issue so I unfortunately cannot answer your question regarding U64 and the empty bank.

 

Try the people at https://www.atari-forum.com/. They may have a resolution. I would also say try Exxosforum but I am having certificate issues at this time.

 

Regarding spare M511000A RAM chips, I can only suggest eBay as other vendors do not seem to stock them due to obsolescence.

 

As to hard disk drives, you may not be able to have both a @masteries ACSI drive and a Megafile 30 connected at the same time, besides which 30MB is tiny these days, and Megafile hard drives are liable to fail at any time due to age. Plus the old RLL hard drives in a Megafile pre-date even SCSI hard drives.

 

Personally I am hoping that the SidecarT hard disk emulation will evolve into a permanent solution as it eliminates any DMA problems associated with ACSI.

[robbo007]

@TZJB Would these work? TC511000AP-70. Is that 70ns?

[TZJB]

2 hours ago, robbo007 said:

@TZJB Would these work? TC511000AP-70. Is that 70ns?

 

As you suggest TC511000AP-70 does indicate a faster RAS access speed of up to 70nS.

 

These are similar to your current DRAM chips, which have the designation Toshiba TC511000AP-10 100mS.

 

These should be a good substitute for your faulty DRAM.

 

I found a datasheet for this product which I have attached. Obviously it's not so new now 🙂.

 

TC511000AP.pdf