Email or username:

Password:

Forgot your password?
Top-level
Tube🍂Time

the hard disk drive has a weird 60-pin AMP CHAMP connector (5176376-2) but that just gets wired to a 52-pin 2mm header. the pin wiring is 1:1 except for the last 8 pins which are not connected.

72 comments
Tube🍂Time replied to Tube🍂Time

it was surprisingly hard to figure out that AMP part number, btw. connector manufacturers constantly scrub all information about old products because who knows 🙃

Tube🍂Time replied to Tube🍂Time

and yes, the hard drive is vaguely like IDE but it uses the Micro Channel bus instead of ISA. IBM called it DBA (direct bus attach) ESDI. this drive uses the POS ID code DF9F.

Tube🍂Time replied to Tube🍂Time

i needed to figure out which interrupt lines were which, so i wrote a quick and dirty program to detect when an interrupt fires. then i manually triggered them by grounding each pin in turn.

Tube🍂Time replied to Tube🍂Time

on to a few physical repairs! this corner got really trashed by the hinges, so i need to figure out a way to restore it.

Tube🍂Time replied to Tube🍂Time

i've used a similar technique to repair an Amiga 500. the idea is to 3d print the missing piece.

Tube🍂Time replied to Tube🍂Time

it's a quick 20 minute print even with a 0.3mm nozzle, 0.12mm layer height, and 100% infill.

Tube🍂Time replied to Tube🍂Time

the next step is to carefully carve away at the plastic so that it fits well. small gaps are ok, there needs to be a little room for the glue.

Tube🍂Time replied to Tube🍂Time

for the glue i am using JB Weld. a little excess is ok since i can file and sand it down afterwards. the locking forceps are clamping the plastic part to the metal tab i installed a few weeks ago.

Tube🍂Time replied to Tube🍂Time

I'm also doing the other side as well where the damage was less extensive. unfortunately the DC-IN legend is damaged and the connector icon is half missing, so I need to figure out how to replace it. a water slide decal might work.

Cellivar replied to Tube🍂Time

@tubetime Paint pen and a steady hand? ;)

Your work is great to watch!

Darryl Ramm replied to Tube🍂Time

@tubetime All this is making me miss my old Thinkpad 701 Butterfly. Damn that keyboard design was so neat.

Tube🍂Time replied to Tube🍂Time

the glue has dried. I'm not crazy about the fit but it is much better than before. it should look pretty good after I paint it, someday.

Tube🍂Time replied to Tube🍂Time

the laptop was missing the modem cover so I'm making a new one. this design is supposed to slide in place, but I think the original was supposed to hinge down, and probably didn't have any openings in it.

Tube🍂Time replied to Tube🍂Time

oh wow i found documentation for the weird protocol used by the 2.5" ESDI hard drive!

Tube🍂Time replied to Tube🍂Time

a simple program should be able to retrieve these error codes. perhaps i can fix the hard drive?

SLeiBt replied to Tube🍂Time

@tubetime Fingers crossed it's not something that needs opening the enclosure to fix, as doing so outside of a clean room would most likely render it toast (bad sectors developing) a short while afterwards.

Tube🍂Time replied to Tube🍂Time

i did a little logic analyzer probing. it's waiting like 288ms for the HDD to respond, and when it reads the status register, it gets FAFF. this doesn't really make sense because upper bits should be 000 indicating the drive, not the controller.

FozzTexx replied to Tube🍂Time

@tubetime Other endian and the bits are inverted?

Tube🍂Time replied to Tube🍂Time

i needed to verify that the DMA arbitration bits were in the correct order on the docking port, so i've decided to use the Type 3 parallel port to trigger arbitrary DMA transfers. you can configure the arbitration level in the setup program.

Tube🍂Time replied to Tube🍂Time

this simple program configures the parallel port for Type 3 operation, turns on DMA, and starts a single DMA send transfer.

Tube🍂Time replied to Tube🍂Time

the operation won't start until the parallel port is ready--i had to pull the BUSY line low with a jumper, and now i instantly get this error! good luck figuring that one out
EDIT: i figured it out, 113 is a DMA arbitration error, presumably the "5" is the arbitration value. DMA controller going off into the weeds lol

Tube🍂Time replied to Tube🍂Time

ok, the machine crashed because we didn't set up the DMA controller, but that doesn't matter because the number '5' showed up on the arbitration bus! this experiment, along with a few others, let me confirm which arbitration pin is which.

Tube🍂Time replied to Tube🍂Time

i'm troubleshooting the weird DBA-ESDI drive (type DF9F) and it's very strange because it sets a bit indicating a power on self test failure, but when you try to retrieve the failure code (basic and advanced) it's all zeros.

Tube🍂Time replied to Tube🍂Time

here's the board. a bunch of proprietary IBM chips and an Intel 80C196 microcontroller running the show. maybe the problem is with the microcontroller.

Fritz Adalis replied to Tube🍂Time

@tubetime
I wonder if these have a serial diag port like newer drives do.

Tube🍂Time replied to Tube🍂Time

the microcontroller is the 80C196KW, which i can't find documented anywhere. but having the "80" prefix implies that it uses an external ROM, like the device nearby with the "4510" sticker. devices with ROM start with 83 and EPROM with 87.

Darryl Ramm replied to Tube🍂Time

@tubetime I'd really like the W variant to use external bubble memory just 'coz 1980s Intel. 🙂

But what the heck is it, only thing I can see is a Japan fab variant mentioned here: cpu-galaxy.at/cpu/Ram%20Rom%20 Even the 8XC196Lx Supplement PDF linked there does not mention it.

Tube🍂Time replied to Tube🍂Time

some parts arrived! this little adapter is supposed to break out the docking connector to a single Micro Channel slot.

Tube🍂Time replied to Tube🍂Time

lookie here! the MC Mechanic works! the two unlit green LEDs are because the adapter doesn't have the +/-12V DC-DC converters installed.

Tube🍂Time replied to Tube🍂Time

unfortunately the Snark Barker doesn't work. the card isn't detected by the BIOS, so that means it's not responding to the card setup signal. logic analyzer time!

Abe the Honest replied to Tube🍂Time

@tubetime *takes out a bucket of popcorn*
OH BOY this looks like my kind of movie!

Joe Pasqua replied to Tube🍂Time

@tubetime ok, now you’re just making up words ☺️

Tube🍂Time replied to Tube🍂Time

let's downgrade and try it out with my Micro Channel Ad Lib clone, the Plaid Bib.

Tube🍂Time replied to Tube🍂Time

oh this is exciting, the card has been detected!

Tube🍂Time replied to Tube🍂Time

and it plays music! i must be the first person to have built a 3rd party expansion for the 700/720 series Thinkpads in a very long time.

Tube🍂Time replied to Tube🍂Time

one major issue was that i got the polarity of the channel reset signal backwards.

Tube🍂Time replied to Tube🍂Time

I've instrumented the Snark Barker but somehow it boots without crashing and even plays back digital audio!

AMS replied to Tube🍂Time

@tubetime Some marginal timing that the added capacitance fixes?

Card wasn't seated last time?

Ian Scott :apple_inc: 🐙 replied to Tube🍂Time

@tubetime I don’t know why I’m surprised that IBM used MCA in ThinkPads, but it does make sense!

William D. Jones replied to Tube🍂Time

@tubetime Where's the code setting up the DMA channel so the parallel port doesn't read/send data from the weeds? :D

lopta replied to Tube🍂Time

@tubetime I've seen 3.5" ESDI but never 2.5"!

George R. M. 🇩🇪🇺🇦 replied to Tube🍂Time

@tubetime for a second there I forgot these use the aptly-nicknamed "ThinkPad 700 drive". What a cursed thing it was, DBA-ESDI 2.5".

Glen Akins replied to Tube🍂Time

@tubetime My guess would be acquisitions and not caring about all the product lines of the acquired company. Good news is now that TE owns all connector manufacturers*, they won't have this excuse going forward. (*slight exaggeration)

Brad Bell replied to Tube🍂Time

@tubetime I like how they keep making the connector but change the names and numbers

bitsavers.org replied to Tube🍂Time

@tubetime

I started archiving catalogs from connector companies just because of that problem.

Go Up