512ke RAM upgrade woes

JeffC

Tinkerer
Sep 26, 2021
122
79
28
Seattle, WA
I'm having some trouble with a RAM upgrade on my 512ke, and thought maybe people would have some ideas.

I picked up this machine a few months back, it is equipped with a Sophisticated Circuits SCSI board and a Mac's-a-Million RAM expansion board. As you can see in the photo the RAM expansion has 3 socketed rows of 16 chips, allowing 2mb total (including the 512k on the logic board). When I purchased it only the first expansion bank was populated, and the machine worked fine and showed 1mb total RAM as expected. I decided to see if I could get it working with 2mb total, so I purchased 40 used (32 + 8 spares) RAM chips of the same specification as the originals. Unfortunately new old stock chips are prohibitively expensive.

Through a bunch of swapping I have confirmed that I have 32 good used chips (in addition to the 16 known-good chips that came with the board), all of which have functioned properly in the 1mb total configuration. I confirmed they are good by populating the first expansion bank with 16 used chips, and then removed those and swapped for another 16 used chips. When I say function properly I have not done any sort of RAM tests, I mean the chips passed the boot test and the machine was able to boot to the finder.

Of note, during my process of swapping the used chips in/out of the first bank, at one point I got a sad mac. I re-seated the last two chips I had swapped in and the sad mac went away, indicating to me that there is some sort of seating issue that can cause good chips to show an error. Also worth noting the board has some solder-blob "jumpers" that are used to set the total RAM - I know there is a setting for 1mb total and 2mb total, I'm not sure if the board can recognize 1.5mb. It's hard to see in the photo but in the bottom left corner there is a small box with a 1 and 2 next to it. Inside that box you can see two "arrows" which are un-jumpered, and below that is a solder blob that is jumpering two additional "arrows". When the 1 is jumpered as in the photo the board is configured for 1mb total (this is how it was when I got it), and based on the photos I have found online of this board as well as some information @JDW generously provided through a friend of his, the 2mb configuration is set by having both 1 and 2 un-jumpered.

So where I am now... I un-jumpered both 1 and 2, put the original known-good 16 chips in row A (as marked on the board, A is at the bottom, B & C above), and filled row B and C with 32 used chips, all of which had at one point worked in this machine. Powered on, sad Mac code 034000. I took all the chips out, sprayed all the sockets with DeoxIT, brushed all the chip legs with DeoxIT. Tried again, sad Mac code 030002. Dead Mac Scrolls indicates 034000 is bad RAM in a Mac Plus SIMM 2 or 4, and 030002 is SIMM 1 or 3. Since this isn't a plus, I don't know how those codes relate to my board, if at all. Also worth noting when I cleaned and replaced the chips, I didn't make sure that all the row B and C chips went back into the row they came out of, though I wish I would have. I've always kept the 16 known-good chips in row A.

Does anyone have any thoughts on troubleshooting further? I believe there are ways to get the machine to identify the exact chip that is bad, but I don't know if this is possible with the expansion board, I would suspect not. Based on the info @JDW found this board is probably a copy of the Beck-Tech boards of the day. Does anyone know if these boards are able to operate in a 1.5mb total configuration? If there is a jumper setting that allows 1.5mb, that would at least let me narrow down the number of chips I'm working with.


PXL_20210430_232829701_1.jpg
 

JeffC

Tinkerer
Sep 26, 2021
122
79
28
Seattle, WA
https://wiki.preterhuman.net/Mac's-a-Million#/media/File:8CEBC8C8-E6D4-4C86.jpeg might be a good starting point, visually - looks like if it's populated the jumper at JB1 needs to be OPEN rather than closed?

Yes I believe to be populated with 2mb, both jumpers at JB1 needs to be open as in the pic you linked. To be populated with 1mb total, based on how my board was set up when I got it, JB1 jumper 1 needs to be closed, and jumper 2 needs to be open.

My post was a bit unclear, when I tried the board populated with 2mb, when I got the 034000 and 030002 errors, this was with both JB1 jumpers open, same as the pic you linked.
 

JDW

Administrator
Staff member
Founder
Sep 2, 2021
1,578
1,376
113
53
Japan
youtube.com
Jeff,

I've written to Henry Spragens, who was a lead engineer at Beck-Tech and also one of the earliest people to join in our TinkerDifferent founding effort. He took a step back early on in the effort and is not on the current Admin team, but he did kindly provide us with some useful thoughts and advice pertaining to the new forum effort. He is a brilliant engineer who knows quite a lot about early Macintosh RAM upgrades, and he provided me some advice for you when you and I spoke about that upgrade privately by email. I sent Henry a link to this thread only moments ago, so let's see if he joins our discussion over the next day or so. If not, I will try to do more investigations on my end. I really want to help you get that upgrade working because I know full well how frustrating it is when you've spend hours and hours (and Dollars and Dollars) on something only to find it doesn't work!
 
  • Like
Reactions: JeffC

JDW

Administrator
Staff member
Founder
Sep 2, 2021
1,578
1,376
113
53
Japan
youtube.com
Jeff,

Henry has not yet registered here at TinkerDifferent but will at some point. He sent me an excellent response by email which I'm sure he won't mind if I now post for you here. Please keep in mind that what you read below is a 100% copy/paste from Henry's email to me, so he must get all the credit:

As I indicated previously, there is a 50% chance that your friend will be able to get 2MB to work. That many RAM chips in sockets is too much for the majority of Mac 128 / 512 motherboards to drive. Every single RAM chip and driver chip has to drive the tri-stated outputs of all other chips in the array, and sockets double the stray load capacitance. It takes fast, strong output drivers. I vaguely recall the preferred chips were NEC military ceramic types. 2MB without sockets was more do-able, and some socketed 1MB upgrades needed selected chips to be fast enough. The Mac Plus SIMMs have considerably less stray capacitance, and used faster chips.

He should know that just because a chip works in the 1/3 populated (1MB) board doesn’t mean it will be able to drive the fully populated board. The troubleshooting method is swap one chip at a time. If the error code changes, either the original or replacement chip is bad. If the error code shows more bad bits, it’s the replacement. If fewer bad bits, the original was bad. Slow going. IIR, the error code is the memory address which failed, and non-zero bits are bad, which may narrow the field if you know where the chips for which bit are located. ie. error code ending in 0001 would be a bad bit zero, 0002 bit one, 0003 bits zero and one both bad, etc. Since each chip drives one bit, and there are 4 chips to each bit, there is a map of bits to chips if you know the wiring of the board.

Somebody could make a modern RAM upgrade for these. Should only take a couple of chips today. Or why not a replacement motherboard with modern components? That would horrify the traditionalists, but Apple 1 clones made with gate arrays sell reasonably well.

Back in the day, guys who did upgrades kept the better motherboards on hand for larger upgrades, and upgraded average boards to 1MB. Since upgrades were usually done on a board swap basis - you give me your 128K mobo and money, and I give you a Very Fat Mac mobo - it was possible to predetermine which boards would work with 2MB of RAM. The trick was obtaining a stash of motherboards to work with. We got ours mostly from Apple employees. Bring us two 128K boards, and get back one 1MB board. I don’t know where the Apple folks got extra boards, but many of them needed fixing. Presumably production line rejects.
 

JeffC

Tinkerer
Sep 26, 2021
122
79
28
Seattle, WA
That was very informative, please thank Henry for putting that together. It's good to know that not all logic boards are up to driving 2mb, so it may end up just being a 1mb board, which is fine since it's still a piece of history. I need to come up with some sort of extension harness for the analog -> logic board cable, then I can take a shot at swapping the chips one by one and see if I can track down a bad chip, if there is one. Pulling the analog board out 32 times is a bit tedious and makes it harder to keep track of progress.
 
  • Like
Reactions: JDW

JDW

Administrator
Staff member
Founder
Sep 2, 2021
1,578
1,376
113
53
Japan
youtube.com
That was very informative, please thank Henry for putting that together. It's good to know that not all logic boards are up to driving 2mb, so it may end up just being a 1mb board, which is fine since it's still a piece of history. I need to come up with some sort of extension harness for the analog -> logic board cable, then I can take a shot at swapping the chips one by one and see if I can track down a bad chip, if there is one. Pulling the analog board out 32 times is a bit tedious and makes it harder to keep track of progress.
I really do wish there was an easy solution for the Plus and older like there is for the SE & SE/30, regarding an extension cable. ATX cables are so convenient on the SE series!
 

Elemenoh

Active Tinkerer
Oct 18, 2021
384
374
63
Bay Area
I really do wish there was an easy solution for the Plus and older like there is for the SE & SE/30, regarding an extension cable. ATX cables are so convenient on the SE series!
@JDW I think the cables on the 128K/512K are standard Molex KK 396 11-pin. It should be easy enough to make an extension cable. But to your point, more of a hassle than buying a readymade ATX extension.
 
  • Like
Reactions: JeffC and JDW

JeffC

Tinkerer
Sep 26, 2021
122
79
28
Seattle, WA
@JDW I think the cables on the 128K/512K are standard Molex KK 396 11-pin. It should be easy enough to make an extension cable. But to your point, more of a hassle than buying a readymade ATX extension.
Thanks Elemenoh, they do look similar, that would make it relatively simple.
 

Patrick

Tinkerer
Oct 26, 2021
434
1
224
43
@JDW I think the cables on the 128K/512K are standard Molex KK 396 11-pin. It should be easy enough to make an extension cable. But to your point, more of a hassle than buying a readymade ATX extension.
I've been looking into building one but I think the cramp tool is prohibitively expensive. Here is a mouser cart I put together. I'm still missing the wires on it.

mouser link ... But I would need the cramp tool JUST for this one or two cables and nothing else.

Also, I'm very much noob to this so I don't know if I put the correct things in that cart or not. so it seemed a bit $$ if I wasn't even sure I picked the correct things.
 

Stephen

BetterBit
Staff member
Founder
Sep 5, 2021
204
250
63
San Francisco
@JDW I think the cables on the 128K/512K are standard Molex KK 396 11-pin. It should be easy enough to make an extension cable. But to your point, more of a hassle than buying a readymade ATX extension.
I've used these in the past for the 128K / 512K. (mouser) (Molex 09-50-3111). I use one for my ATX to Compact logic board adapter which includes a breakout for RGBtoHDMI.
 

JDW

Administrator
Staff member
Founder
Sep 2, 2021
1,578
1,376
113
53
Japan
youtube.com
I've used these in the past for the 128K / 512K. (mouser) (Molex 09-50-3111). I use one for my ATX to Compact logic board adapter which includes a breakout for RGBtoHDMI.

Yes, but ATX (male & female ends) are so much easier overall on the SE & SE/30 because when you make a cable yourself, you not only need the right connectors at both ends, but the right terminals, the right crimper AND the right wire too! I'd go with 18AWG myself. But when you buy wire, it can be costly if you don't already have wire stock. And then are you really going to use the same color wire? Then again, buying the cable I proposed in my video could land you the wire you need, albeit all the same color. You would then need to cut and crimp. Crimping PROPERLY is easier said than done though.
 

Stephen

BetterBit
Staff member
Founder
Sep 5, 2021
204
250
63
San Francisco
Yes, but ATX (male & female ends) are so much easier overall on the SE & SE/30 because when you make a cable yourself, you not only need the right connectors at both ends, but the right terminals, the right crimper AND the right wire too! I'd go with 18AWG myself. But when you buy wire, it can be costly if you don't already have wire stock. And then are you really going to use the same color wire? Then again, buying the cable I proposed in my video could land you the wire you need, albeit all the same color. You would then need to cut and crimp. Crimping PROPERLY is easier said than done though.
To clarify, I mean converting an ATX power supply for use with a 128k/512k/Plus. As much as I understand, the SE & SE/30 use wholly different connectors on the logic board.
 

Elemenoh

Active Tinkerer
Oct 18, 2021
384
374
63
Bay Area
If you’re lucky enough to have an electronics surplus shop in your area you can find great deals on all the little bits and bobs. For instance, Excess Solutions in San Jose has the molex connectors, pins and wire extremely cheap. The $20 crimpers linked above or even regular needle nose pliers and patience would be good enough to make a serviceable extension cable. But the point still stands that’s a hassle compared to buying a ready made cable. Perhaps this is a little buisness opportunity for someone like @MacEffects
 
  • Like
Reactions: Patrick and JDW