Old Hard Drives

When you visit family, and someone hands you an old hard drive from 22 years ago and asks for a data recovery because it could have your moms letters on it, you don't question it.  Grab that hard drive and take it.

So I started a little journey.  I popped it into my PATA USB case, and plugged it in.  It did not register with the system (I'm using Linux so I should at least see a USB interface get logged, even if the hardware has failed).

It spun up.  There were no clicks (so the hardware had not failed).  I dug into the logs - nothing, indicating the interface board was likely the problem.

So, I took the interface board off.  It was a Western Digital WD400 from 2003 (I told you it was 22 years old).  I grabbed my Western Digital WD400 from my case and started to swap the interface boards temporarily to commence a recovery.

One screw changed between them.  There are associated with the WD400 product line, but actually have different model numbers.  I'd wager that if it wasn't for that screw being in the wrong place and not being able to be anchored down properly, it would have been perfect, yet this would not be.

I could have 3D printed a bracket that fit across and held the interface circuit board down, but I just put the original back in, and looked at the jumpers.  Old PATA/IDE hard drives had a jumper on them for "Master", "Slave", and sometimes one or two other options (such as "CS" or "Chip Select").  There were no more than two IDE drives on the same IDE cable because of the electronic design.  One would be designated the "master", and the other a "slave".  In later years, they added a "chip select" option to try and make it a little more plug-and-play.

I had the thought I should move the jumper from Master to CS, and suddenly the old interface loaded when it was reconnected.  Ah hah!

It loaded no files (an empty filesystem).  The next step is always to run a recovery, just in case a new filesystem was put over the top of an old one.  So, I ran :

photorec /d Documents/recovery /dev/sdc

This dumped a few web files (CSS, XML, and HTML), a few executables (EXE and DLL), a few TXT files (that turned out to be nothing more than configuration files, registry files and some text extracted from executables).  Two files, a WAB and an EDB, both stood out to me, because a WAB is supposedly a Web Address Book, and EDB is an Exchange Database.  Turns out, they were simply mislabeled data files for the executables.

 No, the drives did not have any letters from my mother.  But it's better to know than not to. 

Doh! A Delicious Dish of Dumb Dumb!

Alright.  I'm only posting this because of the blatant error in my own reverse engineering.  I had a Coleman pop-up canopy, with fancy "push button" locks and unlocks.  It's really not bad - it's survived for 10 years with only minimal issues.  (Two years ago, one of the locking mechanisms on one leg wouldn't release.)  Here it is in all it's glory (still in it's bag).
 

This year, that same locking mechanism released (perhaps a better term is it fell apart).  Since I had silver filament and these were gray, I thought, what could I lose?  I'll see if I can draft and model up a replacement part, and print it.

I started out by taking the few remaining parts off to see what was broken.  This is a pretty ingenious device.  I'm impressed with the design (but not so much the implementation).

The broken part is circled in red above.  A few more photos, just to make sure :

It is time to create the model.  I whipped this up in FreeCAD :

About 50% done (I'm doing 100% in fill so it is solid), I picked up one of the parts, and realized - I didn't look deep enough.  There are two locking tabs on each end (I got both of those), and... recesses in the mating part where there were supposed to be four more locking tabs on the sides.

Really?  I play the part of an idiot VERY well, sometimes.  I'm missing the ones in green here.  Notice, I had them listed as an alignment pin.  Looking closer to the original, and these locking tabs had broken off (three of the four were down to the plate, one was half height, and I built all four off of that one).

So, back to the drawing board. I simply copied one of the two tabs and positioned the four new instances.

Okay, we can try printing this once more. Here we go.  8 hours and 38 minutes.

Alright.  The overall length should be 2.904", but the final print came out at 2.890.  In Cura, it shows 73.6, which is about 2.898", so we're losing about 0.008" in the print, and Cura is also not getting sizes from the STL file with much accuracy from FreeCAD.  That should be just fine, really, as long as I remember to adjust the size by 100.48% before slicing.

It looks good, the clean up was good (the aborted print gave me an idea of how this was going to be cleaned up), and with some fine adjustments (hobby knife and a Dremel for some inside work), it works exactly as I'd hoped.

I went to install it, and it simply didn't lock together.  So, that led me on a re-do, and I added the toggle cover with a new base that screwed together (no tabs).  This particular variation uses screws to bring the cover and the main housing together, but uses the same toggle switch itself, the same spring, and the same spring assembly.

Note, be careful how you print this - if you print side-to-side instead of top-to-bottom, the locking tab that holds the housing to the actual metal upper leg will break, and then you have to superglue it into place (ask me how I know). 

I have my canopy back!