ThinkPad 365XD CPU-card and RAM

[DK6400Brian]

Are the TP365XD cpu-cards made exclusively for this specific model only ?
Perhaps the card itself were made for other notebook manufactorers as well.
I'm on the hunt for an upgrade

Upon boot, the RAM-counter has 6 digits.
My 365XD have 72 MB installed.
The first digit is a zero, which indicates, that a 128 MB module actually would/could work.

Has anybody tried to install 128MB in a 365XD ?
The subject came up long ago.

Thank you.

[phool@round]

DK6400Brian, not absolutely sure because I don't own a 365 but possibly a processor from a 760/765 might work. At their peak they are 166MMX that can be clocked to 200Mhz safely. What I honestly don't know is if the cards are designed to fit the laptop or are they the same form throughout.

[leoblob]

The fastest CPU board I am aware of is this one @ 133 MHz: http://www.impactcomputers.com/82h8162.html

As for memory, I've never even seen a 128MB memory "card" with the right specs (144 pin SODIMM, EDO). Personally, I wouldn't read anything into that extra zero in the RAM counter

I don't recall ever reading on this board that someone has actually tried a 128.

[DK6400Brian]

Just have to ressurect this old thread from 2007, because my IBM Thinkpad 365XD is still in perfect working order and have received a fine upgrade since.

It was up and running this very evening, playing a couple of CD's, and just checking everything out on the machine.

Two upgrades was made in 2009.
I bought a tiny IDE<->CF-Card adapter thinking of giving the machine 8GB to play with when all of a sudden a magnificient deal came into my hands instead. An Israeli M-Systems 1 GB SSD with IDE-interface was up for sale on a local Danish auction site. Grabbed it of course for peanuts and gave the machine a fresh install of Windows NT Workstation 4.0.
Loved to be able to reinstall all drivers and the 6a Servicepack. Loved it

Also found a lot of modern programs in old versions to install. Boy it was fun.
Boot time, shutdown time, opening and closing programs are faster now and the machine ain't saying anything now. Quiet as a dead mouse. Beautiful.

A new IBM battery was bought and works pretty good, giving an hour of life.

Now. The reason for resurrecting this thread is because eBay by now (April 2016) has a lot of memory modules up for sale.
I stumbled upon a lot of 128 MB, EDO, 60ns, 144-pin, modules and are going to buy one of them soon.

Being cautious and suspicious, it just happens to be that one of them is meant to be used in a Compaq Armada 1130, a Pentium 120 MHz machine, which I think is within the same league as the 365XD.

What is left to say by now is, that it must be up to the BIOS if it will handle it in any way.
Time will tell and I will report back to you if the money is well spent or wasted. The eBay listing mentioning the Armada 1130 could be wrong.

Also, I will research and keep an eye out for any processorcards that would match the machine. A 166MMX CPU would be lovely if it would work.

Dear oh dear. This machine is almost 20 years old. It's a sweetie.

[DK6400Brian]

Recent developments. April 20, 2016.

Any thoughts of giving the 365XD a MMX-processor is out of question. Further research crushed any dreams in this regard.
All subjects, the 535, 380, 385, 760, 765-machines aren't compatible.
The CPU-card is strictly made for the 365X/XD.

However. A few Asian/Japanese 365X machines (2625-BJ9) were born with a 150 MHz CPU-card and digging deeper into this matter last night, one guy in Japan have tried to overclock that CPU-card to 166 MHz with some confirmed success.
It turns out, that on the CPU-card itself, the settings for FSB-clock (Called CPUCLK) and CPU-multiplier (Called COREFRQ) are screen printed onto the PCB.

So it's just a matter of placing one or more resistors on the right place/places in order to obtain any desired frequency between:

1.5*50 MHz= 75 MHz ==> 2.5*66 MHz= 166 MHz

Have a look (In Japanese, but with pictures):
http://www.geocities.jp/monkung1983/tp3 ... 65_cup.htm

Therefore, the settings for the 365X/XD machines leaving the factory are:
P100: 1.5*66 MHz or 2.0*50 MHz
P120: 2.0*60 MHz
P133: 2.0*66 MHz
P150: 2.5*60 MHz
...and overclocking to P166: 2.5*66 MHz
...underclocking to the lowest setting "P75": 1.5*50 MHz

There's actually room for two more experiment with this thing, because:
1: What will happen when [COREFRQ] has ASM on both R22 and R44 ?
Would the result be a multiplier of 1.0, 3.0 or nothing ?

2: What will happen when [CPUCLK] has no ASM on R13 and R5 ?
Would the result be 33 MHz, 75 MHz, nothing or some odd figure ?
Exiting


Upon hovering around on eBay today, I came across a picture of the memory-PCB (81H8192), which apparently is containing the original 8 MB onboard RAM and the SO-DIMM-slot.
But by looking at this picture closely, I can't help scratching my head and assume (hope) that the memory-PCB in reality is holding 16 MB, and that the switch is set for just using only half of it. The screen print sure is indicating this.
However. One reason could be, that the PCB is only prepared to be equipped with larger capacity chips.

I never have had any of my 365XD's completely disassembled before. Never had the guts to do so.
I'll practice with my other 365XD machine that is partly disassembled and I'll dig further into this in the coming days. It's way too interesting to let go. Setting both switch 2 and 4 to ON will clear this matter up and I will report back for further reference.

The memory-PCB can be seen at this eBay listing:
http://www.ebay.co.uk/itm/IBM-41H8192-8 ... SwZ8ZW8xXS

[theterminator93]

Hm, interesting. Mine has the 120 MHz CPU card - if I could overclock it (stably) to 150 MHz I'd be rather thrilled. Should be as simple as moving the resistor from R22 to R44 eh?

[DK6400Brian]

Apparently yes. I'd practice with a spare CPU-card first. Better safe than sorry.

I'm also having a 128 MB EDO 60ns 144-pin SO-DIMM coming up. Not expecting it to work, but giving it a tiny chance.

[DK6400Brian]

I'm also having a 128 MB EDO 60ns 144-pin SO-DIMM coming up. Not expecting it to work, but giving it a tiny chance.

The RAM-stick finally arrived and it replaced the 64MB-stick. It actually does work, to my surprise, but just 64MB is recognized. The machine adds the onboard soldered 8 MB, bringing it up to the ususal 72 MB.

It implies that the 128 MB stick has two banks and one of them is by default swapped out or not seen at all.

The Souhbridge however, lntel SB82371FB, looks from the documentary http://download.intel.com/design/intarc ... 055002.pdf, as if it would love to operate with up to 128 MB.
But in order to dig in to how and if ever the machine is capable of functioning with 128 MB RAM would mean dealing with pinlayouts, circuits, chips and their documentation...and I don't frankly have the time for reaching any/the final conclusions on this of whether some hardwiring, or changes with the components on the mainboard, might be possible, in order to connect the second memorybank to the machinery and/or whether there's actual connectivity onboard to the inside of the chips to function with 128 MB RAM. There should be, but it calls for real expertise to step in, if any utter final conclusion is to be dealt with.
So until further noticed and Hastalavista Baby: 72 MB RAM is the maximum for the 365X/XD.

[DK6400Brian]

So until further noticed and Hastalavista Baby: 72 MB RAM is the maximum for the 365X/XD.

But.....couldn't resist diggin' ..... Because the funny thing is, that the onboard 8 MB RAM is made up by four 16mbit RAM IC's. Researching further revealed, that a memorychip can be set up, controlled and used with different configurations:

Examples:

Eight 128 mbit chips = 128 MegaBytes
Four 128 mbit chips = 64 MegaBytes
Two 128 mbit chips = 32 MegaBytes
One 128 mbit chip = 16 MegaBytes

Eight 16 mbit chips = 16 MegaBytes
Four 16 mbit chips = 8 MegaBytes
Two 16 mbit chips = 4 MegaBytes
One 16 mbit chip = 2 MegaBytes

So I guess it's possible to desolder the four onboard 16 mbit RAM chips and exchange them with four 128 mbit RAM chips, since the control electronics is present on the mainboard. Only thing to determine is ot see if the corresponding caps and resistors have okay values for the 128 mbit chips to go in.

If this gives a pleasant result, the IBM ThinkPad 365XD will have 64 MB onboard RAM and a 64 MB stick in the SO-DIMM slot,
would bring this machine up to 128 MB RAM. The SB82371FB Intel chipset supports 128 MB.

I think it's a winner

I'm having other subjects coming up with the 365XD.
An IBM High Rate Wireless LAN (FRU 22P4592) have just arrived today and I've got the software for it in advance, so it's gonna be a good day. Yee-hah.

I've also got a couple of IBM PC Card Adapter/IBM Microdrive 1 GB on the bench, but the preliminary results sucks.
The thought was to have two Microdrives in the two PCMCIA-slots, partition both of them with 300 MB in the fastest 4.3 MB/sec zone, using a 64K allocation unit and the 700 MB in the remaining 2.7-4.0 MB/sec space.
Then create a Striped Set from the fast 300 MB partitions, 600 MB in total, and have programs and temp-directories run from there and non-critical data on the remaining slower partitions.

But it's not gonna happen.
First of all. The machine won't start with both drives in the PCMCIA-slots.
Second. With only one drive inserted in the PCMCIA-slot, the transfer rate is pathetic, 400 KB/sec to 1.7 MB/sec. High CPU%, which most certainly would imply the darn thing running in PIO-mode.
Two DMA channels are present, but theseare occupied by Auddrive and Floppy...and no workaround available to give the SSD (Disk-0 on IDE-0) or the Microdrive a DMA-channel.
I even tried to alter SCSI-settings in the Registry, but the changes are reverted back to original state when rebooting.

So these Microdrives will instead be NTFS-formatted with 64K allocation units and used for one giant ZIP-archive each for backups.
Further testing has to be done to compare transfer rates between the present FAT-16 partitions and the NTFS-64K Allocation Unit option.

That's it for today