For Sale 8MB RAM to 68000 socket
- Thread starter mkl
- Start date
- Replies 122
- Views 31090
Message: If you have a ram board from me sold before september 14. 2018, your board probably has a malfunctioning RAM controller. It can be fixed by reflashing the CPLD using 6-pin pogo pins and whatever tools you use for Xilinx CPLDs. I have no guide for the process now.
Bumping
Special offer: MC68HC000P10 or MC68000P12 CPU with the board free.
0 pcs v69 available. Price 50€/board + shipping + possible paypal fees. . E.g. untracked 100g shipping 5€ Europe / 10€ world
3 pcs lot with registered mail shipping included 150€ edit:
Last edited:
Thank you for your interest, but I have no working boards and no parts (dram, cpld) for new boards. I could try to debug the few boards that behave bad. Or even buy new parts but the prices have increased, so I have hesitated to buy them so far.
When I have a board deemed working, price with shipping to UK is 53.65 €.
First board to fix, version 68. It had capacitor stuck between cpu socket pins. Some solder joints were retouched and socket pins are now soldered from the bottom side too.
It made errors,
but after cleaning solder flux it tested over 7 hours. LCD monitor was broken earlier, but it shows enough.
Then pressing on the board made errors again.
I am uncertain what is the cause. The side slot ZIF CPU socket had 1 bad contact, so I changed to another setup. This other side slot adapter has econo zif on top of cpu, and mainboard zif is brand new. No errors if I press on the ram board.
Rarely seen error on a few ram boards, including this one, is malfunction of autoconfig. I don't know the cause, contacts or marginal autoconfig logic on cpld?
This first "questionable" ram board under test has id Q1 and the cpld code version has user id 6x8d.
First board to fix, version 68. It had capacitor stuck between cpu socket pins. Some solder joints were retouched and socket pins are now soldered from the bottom side too.
It made errors,
but after cleaning solder flux it tested over 7 hours. LCD monitor was broken earlier, but it shows enough.
Then pressing on the board made errors again.
I am uncertain what is the cause. The side slot ZIF CPU socket had 1 bad contact, so I changed to another setup. This other side slot adapter has econo zif on top of cpu, and mainboard zif is brand new. No errors if I press on the ram board.
Rarely seen error on a few ram boards, including this one, is malfunction of autoconfig. I don't know the cause, contacts or marginal autoconfig logic on cpld?
This first "questionable" ram board under test has id Q1 and the cpld code version has user id 6x8d.
Last edited:
thebajaguy
Active member
For the illegal Zorro, try a fresh DiagROM boot, and read the $E8.0000 area bits with the memory read tool. Compare to the Zorro spec for an 8MB RAM card that is working (have a printed pic / screenshot to side-by-side compare with a working one). The first 64K is all that matters pre-AutoConfig. Also, 4-bit nibbles are what the OS reads (not the whole byte). Chucky has noted in past conversations that AutoConfig on the original DiagROM has some bugs/quirks., so reviewing the memory-read (bits) is a better option to see the incorrect detail. You could also possibly make the board not show up for AutoConfig, boot up with the MON tool, enable the board for AutoConfig somehow, and read $E8.0000, and even program the board with a write to the correct address offset on the card. Watched a certain engineer do that a few times 35 years ago at GVP - didn't understand it until much later.
I have an MEM68K16M4XC9572V40 - I saw your note about a likely RAM controller issue above, September 15, 2023. I have since pulled the 8M board in my A1000 some time back for an HC533 instead. As it's a spare at the moment, and I need some practice with my Xilinx programmer, is there a place to grab the suggested code? Edit: Saw the link in the first post. Will try it.
I have an MEM68K16M4XC9572V40 - I saw your note about a likely RAM controller issue above, September 15, 2023. I have since pulled the 8M board in my A1000 some time back for an HC533 instead. As it's a spare at the moment, and I need some practice with my Xilinx programmer, is there a place to grab the suggested code? Edit: Saw the link in the first post. Will try it.
Last edited:
Bajaguy: If you haven't reprogrammed the cpld yet, check the old user id on chip and mybe save the old configuration. The oldest cpld files may have different problems (such as autoconfig) than the abel codes with bad dram controller in the middle.
Board "Q1" was tested more than 7 hours without errors.
Case "Q2" board has bad cpld ic, it wouldn't autoconfig and couldn't detect databits 13 and 12 going low. I put a new non-autoconfig configuration on it and diagrom 2 said address error the first time but worked on the next mem check rounds.
Board "Q3". Touch up a few solder joints. Brushed a few darkened pins with fiberglass brush. Didn't solder pins from below, so solder won't cover any more gold surface. Tested overnight without errors. Because of loose sockets, MC68HC000P10 was soldered to sockets, making solid contacts. This makes the ram circuit unserviceable though. Green light between pins autoconfig done. 10 more ram test passes. The CPU is included in the price.
Q4 has loose sockets too. You could plug an adapter board with thick pins to it.
Last edited:
Cooldude0177
New member
Will be declaring interest in about 4 wks
"Q5" has pin 64 replaced and excess solder removed from a few pin contacts at the bottom. 100 passes of memory test:
"Q6" had some solder points retouched.
Some silicone lubrication on contacts is supposed resist oxidisation. Pins have tin surface, gold contacts inside sockets.
tested 100 memory test passes using Diagrom 2:
Same thing with "Q7":
"Q8" didn't find ram after autoconfig, because of wrong cpld configuration. After reflashing tests ok with 20 passes in about 4.5 hours. Socket grip tested with pin:
"Q9" tested for about 60 passes:
"Q10" was missing dram chip and regulator, they were removed may be because the board wasn't working. Now it made address errors. After checking cpld connections with multimeter it worked for 50 memory test passes. Checked the cpld pins attachment pushing them sideways, and the pin supposed to connect to A18 moved. Resoldered all cpld pins and board is tested with 100 ok passes.:
Q11 is version 67 board, without schmitt trigger for the 7MHz clock. Maybe you have an oscilloscope to see that your Amiga has "good enough" clock for the XC9572XL, what ever that may be.
Tiny buffer was backwards and replaced. 75 ram test passes.
q12 v69 test passes ok
"Q6" had some solder points retouched.
Some silicone lubrication on contacts is supposed resist oxidisation. Pins have tin surface, gold contacts inside sockets.
tested 100 memory test passes using Diagrom 2:
Same thing with "Q7":
"Q8" didn't find ram after autoconfig, because of wrong cpld configuration. After reflashing tests ok with 20 passes in about 4.5 hours. Socket grip tested with pin:
"Q9" tested for about 60 passes:
"Q10" was missing dram chip and regulator, they were removed may be because the board wasn't working. Now it made address errors. After checking cpld connections with multimeter it worked for 50 memory test passes. Checked the cpld pins attachment pushing them sideways, and the pin supposed to connect to A18 moved. Resoldered all cpld pins and board is tested with 100 ok passes.:
Q11 is version 67 board, without schmitt trigger for the 7MHz clock. Maybe you have an oscilloscope to see that your Amiga has "good enough" clock for the XC9572XL, what ever that may be.
Tiny buffer was backwards and replaced. 75 ram test passes.q12 v69 test passes ok
Last edited:
New offer for june 2024: While available, 3 pcs with shipping and 3 unused nos cpu:s MC68HC000P10 or MC68000P12 either plugged on boards or separately, total 150€. Registered mail to Europe, normal 'Priority' standard shipping outside Europe. Outside of EU, there will be an "untracked" code and bar code on the CN22 label.
units that are tested were listed in previous post
units that are tested were listed in previous post
McTrinsic
Member
Hi,
only little more than the actual CPU.
It is meant to go on one side of a ‚Garlic‘ CPU switch.
Do you think it will work in such a setup?
It needs to be fully non-autoconfig as there are other boards.
only little more than the actual CPU.
It is meant to go on one side of a ‚Garlic‘ CPU switch.
Do you think it will work in such a setup?
It needs to be fully non-autoconfig as there are other boards.
Hi. Special firmware can be programmed on the cpld, that does not use autoconfig. I will program before shipping. If the firmware is not on the www page, I can upload it here.
The address range where the CPU sees the RAM can be customised, and it can be also full 8 megs at 200000-9fffff hex
Some versions of program file "addmem" might not work?, so need to find a working program file.
The address range where the CPU sees the RAM can be customised, and it can be also full 8 megs at 200000-9fffff hex
Some versions of program file "addmem" might not work?, so need to find a working program file.
Similar threads
- Locked
