Issue with the STM 4MB RAM UPGRADE

[Malichite]

I have a Rev H 520 STM still using the original TOS 1.0, but it was my understanding that all TOS version support a maximum of 4MB. I proceeded to remove the existing memory chips with a combination of a desoldering tool and an Atten heat gun, trying to make sure I only applied enough heat to get the chips out. I then prepped the RAS0 line with the cut trace and 68Ohm resitor and then proceed to install the STM 4MB RAM UPGRADE module. As I have the Rev H board I originally tried the alternate locations for RAS1,CA1H, CAS1L, but when I hit the issue I double checked the solder joint on the module and moved all four wires to the back side of the motherboard with no change. I've double checked the pins for the four lines and it appears correct as does the 68Ohm resistor on the RAS0. However, on boot I only get this strange black and white pattern show in the attachment.

Did I miss some other step that I need to change for the 520 STM? A TOS upgrade isn't required for this module to work is it? I'd appreciate any help. I believe that I picked up an ST diagnostic cart, but not sure where it is so I figured I'd ask here first to see if I missed something in the guide. Thanks,

520STM_Error.jpg

[rubber_jonnie]

First off, the instructions state that the resistor should be 33R-68R. You state you've used 680R which is out by a factor of 10. You might want to look at that first.

Other than that the usual checks for solder bridges or cold joints, it's just a question of rechecking your work for errors.

[tzok]

He wrote 68Ohm, so 68R - all fine.

Make sure you haven't damaged any traces between chips. Which exactly upgrade were you trying to install?

BTW, I still say that it is enough to cut one resistor to safely disable the on-board RAM on the ST.

[rubber_jonnie]

He wrote 68Ohm, so 68R - all fine.

Well, to me it could be a simple typo so worth double checking.

Make sure you haven't damaged any traces between chips. Which exactly upgrade were you trying to install?

Definitely a possibility that traces are damaged.

BTW, I still say that it is enough to cut one resistor to safely disable the on-board RAM on the ST.

The only upgrade in the store based on the OP's description is this: STM 4MB RAM Upgrade

Whilst I'd agree normally about lifting one resistor, this upgrade requires all the old RAM to be removed in order to fit it.

[Malichite]

tzok wrote: ↑18 Feb 2024 02:07
He wrote 68Ohm, so 68R - all fine.

Well, to me it could be a simple typo so worth double checking.

Yes, that is correct it's a 68R and I double checked the connection registers 68R across that new connection. I'll try to trace it to the MMU and just make sure it's got a good connection all the way, as this was removing a slight bit of solder mask and soldering to the trace. It appeared good as the copper trace was exposed, but I guess it could have had a teeny bit of solder mask still on it leading to a weak connection. Also I could try a lower value as Exxoss mentioned 33R - 68R and I have the full range of resistors between that I could swap in.

As far as the rest, aside from something related to heating the board with the heat gun the traces are all pristine and no scratches or damage and nothing other than the Memory was disturbed. The Exxoss STM 4M RAM upgrade module itself did have a few scratches from shipping, but they looked like the traces were intact and I covered them will UV curing resin just to be safe.

Whilst I'd agree normally about lifting one resistor, this upgrade requires all the old RAM to be removed in order to fit it.

As I mentioned before all of the RAM was removed already, but I'm not sure I understood the comment about lifting a resistor, did I miss a required step for the STM 4MB RAM Upgrades, as apart from adding the resistor on that trace near R134, I didn't lift any other resistors. Also I left all the caps related to the memory in place as Exxoss mentioned that it wasn't necessary to remove.

Sounds like I need to dig up that ST Diagnostic cart to see if I can get any further insight.

[rubber_jonnie]

Yes, that is correct it's a 68R and I double checked the connection registers 68R across that new connection. I'll try to trace it to the MMU and just make sure it's got a good connection all the way, as this was removing a slight bit of solder mask and soldering to the trace. It appeared good as the copper trace was exposed, but I guess it could have had a teeny bit of solder mask still on it leading to a weak connection. Also I could try a lower value as Exxoss mentioned 33R - 68R and I have the full range of resistors between that I could swap in.

No you didn't miss a step. lifting a resistor and tying to 5v is used on some machines to disable the RAM. You've already removed all your RAM, so it's not a problem.

[Steve]

@Malichite

It might be worth watching Adrian's digital basement's installation video (it's for agranlund's version, but it will be the same installation) mainly for his fault finding techniques. His particular STM had a previous 1mb RAM upgrade and because of this he needed to reverse a few broken traces to get it back to normal.

[Malichite]

I've double checked all the connections to the MMU and it seems like everything is connected, so I'm starting to think the module itself is just bad. I been able to continuity check all the address lines (MAD0-9), as well as CAS0/RAS0, CAS1H/L RAS1 to the MMU, so no damaged traces. Though I did notice that connecting MAD9 from the upgrade board ties MAD9 to RAS0 with a 68 Ohm resistance between, is that expected, was MAD9 just the easy way to access RAS0? I've also traced all the DINs from the ST board to inputs on the two memory packages on the upgrade board and everything checks out continuity wise.

I'm beginning to suspect the upgrade module itself as not only did it have a few scratches from shipping, but was also slightly bent. I didn't think anything of it at the time as the part that was bent was just traces to the middle, but now I'm beginning to wonder it it flexed the memory chips themselves and broke something internally. There are no fractures on the solder points for the memory modules and no cracks or anything obviously damaged, but I can't think of anything else aside from bad memory modules.

Also I did find the diagnostic cart, but this error takes precedence on boot so nothing proceed further with this memory issue. Guess I'll have to look at ordering another one as I really don't want to reinstall the original memory and then I'd still need the stacked approach to at least get the system up to 1M.

On side question, there is that 470uf 16V Axial capacitor at the far end of the memory row that also got heated up a bit removing the chip right next to it. What are the failure mechanism of a failed bypass cap? Nothing is obviously damaged, but it did get some heat and this board is almost 40 years old. Best to replace it anyway, but just curious if there's any chance that might affect this upgrade.

[DoG]

MAD9 should not be connected to RAS or anything else for that matter. Connect MAD9 from RAM board directly to MMU chip.

[Malichite]

MAD9 should not be connected to RAS or anything else for that matter. Connect MAD9 from RAM board directly to MMU chip.

All connections for MAD9,RAS1,CAS1L,CAS1H are made directly from the STM upgrade module to backside of the MMU. However, upon inspection of all the MMU pins, as I was ensuring there were no bad connections/disconnects, I noticed that when I connect MAD9 from the STM upgrade module to the MMU backside pin 64, there is a 68 Ohm path from between MMU pin 64 (MAD9) and MMU pin 8 (RAS0).

Obviously disconnecting this from the MAD9 pin on the STM upgrade module removes this connection between MMU pin 64 (MAD9) and MMU pin 8 (RAS0), so this is something coming from the STM upgrade module itself. If that isn't intentional, further evidence that something is up with the STM upgrade module.