• Updated 2023-07-12: Hello, Guest! Welcome back, and be sure to check out this follow-up post about our outage a week or so ago.

My SE/30 hates 16mb ram sticks

Aeroform

Well-known member
I’m trying to upgrade my SE/30 with 16mb ram sims, but no matter what I try it refuses to boot as soon as I add them (jailbars and chime of death).

I have two different sets of 16mb ram sticks (a total of 8). One set from Siliconinsider here on the forums, and a second set from Keystron which others here on the forum had luck with. Neither seems to work in my machine, and I sincerely doubt both sets would be faulty.

I’ve tried every thing I can think of, tried both banks, in combination with other working sims. I’ve tested filling both banks with 16mb. One with 16 and the other with 4 or 1mb sims and so on.

I’ve tried using plastic shims for the slightly too thin modules, but the keystron set is exactly as thick as my “original” sims and still don’t work.

The only thing that makes my machine boot happily is using the 4mb + 1mb sets I have. Both banks seems to work fine and that memory tests just fine and runs without issues or instability.

Any ideas on what to test/investigate further? Does all revisions of the mb accept 16mb sticks? I seem to have a early rev. mb with socketed Cpu. It’s been recapped and works just fine otherwise.
 

chillin

Well-known member
I had similar issues once, vaguely remember no joy with 128MB, then tried 64MB, no good, and I kept at it and eventually I got all the RAM in, but I do not recall the solution. So I am guessing, Command-Option-P-R while boot for 3 chimes. I usually have to quickly lift and combo again after each chime.
 

Aeroform

Well-known member
I had similar issues once, vaguely remember no joy with 128MB, then tried 64MB, no good, and I kept at it and eventually I got all the RAM in, but I do not recall the solution. So I am guessing, Command-Option-P-R while boot for 3 chimes. I usually have to quickly lift and combo again after each chime.

Thanks for the suggestion, will try some pram resets! That’s one thing I actually haven’t tried!
 

trag

Well-known member
Could be that capacitor goo damaged/shorted the A12 line to the RAM slots. That address line wouldn't be used on 4MB and smaller SIMMs but is needed on 16MB.
 

Big Ben

Well-known member
Could also be some goo on other IC of the RAM bus. You might want to check that too.

When you have the sad mac did you have any error code that might help?

Also did you have any way to check if your ram sticks are fine on another machine?

Looking for a defective stick on the SE/30 is kinda complicated as it only accept booting if a bank is fully filled from what I understand. (Correct me if I’m wrong)
 

Aeroform

Well-known member
Could also be some goo on other IC of the RAM bus. You might want to check that too.

When you have the sad mac did you have any error code that might help?

Also did you have any way to check if your ram sticks are fine on another machine?

Looking for a defective stick on the SE/30 is kinda complicated as it only accept booting if a bank is fully filled from what I understand. (Correct me if I’m wrong)

Sadly no error codes, just horizontal lines and a chime of death. Will have to check the mb more closely. Both sets should be fine, or at least one as they’re both brand new.
 

Aeroform

Well-known member
the 74F258s will do this too if there is a bad/stuck line on one of them.

Thanks. Visually all chips looks good. Currently trying to figure out the schematics and figuring out what goes where on the MB to check the A12 line. Not easy haha, especially due to the very limited time with two small kids at home.

Can I use Bolle’s gerber files to more easily follow where traces go etc? I tried some online viewers but they didn’t allow following single traces etc. Just trying to figure out what and where to actually test on the board. Would I need a scope? Never used one 😅. Hoping my multimeter would be enough for some first checks. Would gladly take any pointers on what to test ”for dummies”.

Thanks for your input so far!
 

techknight

Well-known member
Thanks. Visually all chips looks good. Currently trying to figure out the schematics and figuring out what goes where on the MB to check the A12 line. Not easy haha, especially due to the very limited time with two small kids at home.

Can I use Bolle’s gerber files to more easily follow where traces go etc? I tried some online viewers but they didn’t allow following single traces etc. Just trying to figure out what and where to actually test on the board. Would I need a scope? Never used one 😅. Hoping my multimeter would be enough for some first checks. Would gladly take any pointers on what to test ”for dummies”.

Thanks for your input so far!

No. Bolle's version is an entirely different design. its routed completely differently. You can use the schematic, but not the board layout.
 

Aeroform

Well-known member
Can we see your 16MB RAM SIMMs? Can you post some images?

Here’s the two different sets I have, both with several confirmed successful uses in the SE/30, so don’t think anything would be wrong with these.

I’ve also attached some pic’s of the MB. It’s been cleaned locally around the replaced caps. I guess it could use another thorough cleaning but I’ve been holding it off as I don’t want to break anything haha. The F258’s looks good as far as I can see.

Also of note is the somewhat interesting factory botch wire.
 

Attachments

  • DBBD31EB-3156-4C07-9786-7FE8747CF547.jpeg
    DBBD31EB-3156-4C07-9786-7FE8747CF547.jpeg
    2.3 MB · Views: 90
  • FB0AB73B-1C24-4FD9-8D0A-F014A12D74AF.jpeg
    FB0AB73B-1C24-4FD9-8D0A-F014A12D74AF.jpeg
    2.6 MB · Views: 59
  • AD999EE1-5E19-400E-98FB-FBF0EE948AA9.jpeg
    AD999EE1-5E19-400E-98FB-FBF0EE948AA9.jpeg
    2.1 MB · Views: 41
  • C5601952-B4F5-4336-AE4E-AAA32CD1061C.jpeg
    C5601952-B4F5-4336-AE4E-AAA32CD1061C.jpeg
    2.2 MB · Views: 44

Aeroform

Well-known member
If I'm reading the schematics correctly (and haven't missed something), it seems the A12 line passes by the following chips on the way to the ram slots (other ram lines routes through the chips in parenthesis).

UI5
UI3 (+ UJ3, UI2, UJ2, UI4,UJ4,)
RP5 (+ RP4, RP6)
UI8
VIA1
VIA2

I'll start with cleaning and scrubbing thoroughly around those chips and report back!
 

Aeroform

Well-known member
Aaaand:


UK6 (+ UJ6)
UB8 (+ UA8, UC8, UD8)
UC6, UC7
UE8

and some others I missed as part of the video circuit. Gha basically the entire board haha. This area looks really poor from leaking caps, I guess I’ll have a good ol’ scrub of the entire board.
 

Aeroform

Well-known member
A proper cleaning didn’t do it. I suspect UE8 / UD8 as those are the only chips that look at all effected. If I’m not mistaken A12 passes through these (put as part of the video system?).

Is there any chance these chips could be the problem, or should I continue to look elsewhere?
 

Attachments

  • 189CECD7-BEF0-4807-8A6F-787FCF1CEC4F.jpeg
    189CECD7-BEF0-4807-8A6F-787FCF1CEC4F.jpeg
    2.4 MB · Views: 36

trag

Well-known member
and figuring out what goes where on the MB to check the A12 line.

Keep in mind it may be labeled A11 or A12, depending on whether the DRAM address bus starts numbers with A0 or A1. And this is the address line(s) between the memory controller and the DRAM SIMMs. Not the main CPU address bus.

The easiest way to find it is probably to get a pinout of the 30-pin SIMM standard (easy to find with Google) and identify which pin is A12 (or A11 if numbered from A0). Then trace backwards from there. There being the SIMM socket(s).

At some point, 32 bit (or 30 bit) addresses from the CPU get intercepted, interpreted as to whether they're pointed at the DRAM, and then multiplexed into two separate 12-bit chunks.

So while the SE/30 has a 32 bit (or 30) bit address bus, the addresses to DRAM are just 12 bits wide, maximum. But the address is sent to DRAM in two pieces after interpretation by the memory controller so up to 24 bits of address space per DRAM SIMM is supported. 24 bits => 16M.

Sorry I can't be more detailed, but I haven't read the SE/30 schematic.

Note: The 32 vs. 30 bit thing above is just me hedging, because, while the 68030 has a 32 bit address bus, as long as the software is addressing 4-bytes at a time, it only needs 30 bits. I'm not sure which way the SE/30 does it over all.

Here's a link to a decent pinout diagram. Pin 24 is the one you want to trace from. A little casual searching didn't turn up a good reference on which end of the SIMM to count from. Unfortunately, 5V and GND (easy to trace) are symmetrical on the 30-pin SIMM. 5V at 1 & 30. GND at 9 and 22. But I'm sure more thorough searching will answer that question.

http://www.interfacebus.com/Memory_Modules_30Pin_SIMM_PinOut.html
 
Last edited:

Aeroform

Well-known member
Keep in mind it may be labeled A11 or A12, depending on whether the DRAM address bus starts numbers with A0 or A1. And this is the address line(s) between the memory controller and the DRAM SIMMs. Not the main CPU address bus.

The easiest way to find it is probably to get a pinout of the 30-pin SIMM standard (easy to find with Google) and identify which pin is A12 (or A11 if numbered from A0). Then trace backwards from there. There being the SIMM socket(s).

At some point, 32 bit (or 30 bit) addresses from the CPU get intercepted, interpreted as to whether they're pointed at the DRAM, and then multiplexed into two separate 12-bit chunks.

So while the SE/30 has a 32 bit (or 30) bit address bus, the addresses to DRAM are just 12 bits wide, maximum. But the address is sent to DRAM in two pieces after interpretation by the memory controller so up to 24 bits of address space per DRAM SIMM is supported. 24 bits => 16M.

Sorry I can't be more detailed, but I haven't read the SE/30 schematic.

Note: The 32 vs. 30 bit thing above is just me hedging, because, while the 68030 has a 32 bit address bus, as long as the software is addressing 4-bytes at a time, it only needs 30 bits. I'm not sure which way the SE/30 does it over all.

Here's a link to a decent pinout diagram. Pin 24 is the one you want to trace from. A little casual searching didn't turn up a good reference on which end of the SIMM to count from. Unfortunately, 5V and GND (easy to trace) are symmetrical on the 30-pin SIMM. 5V at 1 & 30. GND at 9 and 22. But I'm sure more thorough searching will answer that question.

http://www.interfacebus.com/Memory_Modules_30Pin_SIMM_PinOut.html

Thank you very much for the detailed explanation. I believe that is the line I’ve followed on the schematics as it ends on pin 24 on the simm slot. But to be honest, it’s kind of a labyrinth with that signal being muxed and passes through a multitude of chips haha.

If you look on the schematics below, I guess I’m correct in that RABF(11) / RAAF(11) is the one I’m after. In that case I should start by having a look between RP6 and the simm slot(s) and working my way backwards from there?

For ref. the full schematics are here:

Or redrawn here:
 

Attachments

  • EA23B2D2-7291-41C3-AF29-4DE28DDB6388.png
    EA23B2D2-7291-41C3-AF29-4DE28DDB6388.png
    844.9 KB · Views: 22
  • 3E10D379-21B7-4C0E-B9C5-6096AE8E0860.png
    3E10D379-21B7-4C0E-B9C5-6096AE8E0860.png
    708.9 KB · Views: 25

Aeroform

Well-known member
Overall, that area really looks pristine, so I find it kind of hard to believe anything should be wrong there. UE8 / UD8 on the other hand..

2806EAEF-B0E4-455C-B0AD-C29DC368C9BA.jpeg
 

Attachments

  • B4E857DB-7243-47E6-A15E-81DE8A26ECEA.jpeg
    B4E857DB-7243-47E6-A15E-81DE8A26ECEA.jpeg
    2.1 MB · Views: 74

trag

Well-known member
UE8 / UD8 on the other hand..

Those do look iffy. Also, check those vias through the board to confirm they actually conduct from one side of the board to the other. My IIci had some vias chewed by capacitor goo that looked okay, but were actually corroded through.

Also, don't ignore Techkight's advice. He has far more experience at this than I do. While the solder joints of the F258s look good, it's possible one of the chips just failed.

I'm not sure how you test that without desoldering the lot, though.
 

techknight

Well-known member
I'm not sure how you test that without desoldering the lot, though.
You don't, you replace them all. That is what I do. I don't question it anymore as I got tired of messing with dodgy chips.

You can sit there for hours chasing your tail to figure out where the bad one is based on a logic analyzer or you can just replace them and save a whole bunch of time.
 
Top