Mass Hardware Salvage Methods?

gte024h writes: "I have been approached by a friend working at a mid-sized corporation about a plan to salvage hardware from their "dead" computers to make working computers which would then be donated to a local school. In the past I have salvaged old hardware from them, but only a few machines at a time. With this plan I will initially get about 40 identical machines with perhaps many more later (he says they have a warehouse full). I have no idea what condition these machines will be in but I am pretty sure that out of 40 non-functional machines I should be able to get 10-15 working. Does anyone have experience doing anything like this? Would it be faster to strip the machines and test the components individually, or to troubleshoot each machine whole? Any ideas for efficient testing and such are greatly appreciated."

I know that volunteers from LXNY recently did something similar with a whole boatload of donated computers; anyone with experience care to comment on how to best handle the mixed blessing of donated, but old, hardware?

Comments

What I'd do

[JediTrainer]

Get a few units working 100%. Using those units, test the components of all the other units in assembly-line fashion (first test all the hard drives in the working units at once, then the power sources etc). Doing it this way will save you from having to debug each individually and will let you add parts to the "good" and "Bad" piles much more efficiently, by using the same procedure for all of them at the same time. You will get into a rhythm which will speed things up this way.

What do you want from them?

[Patman]

The question is, do you want parts, or whole systems?
If you're looking for whole systems, boot each one
until you find one that works. Swap parts from ones that
don't boot into the ones that do until you find
which parts are damaged on the unbootables.

If you're going for parts, take all 40 apart. Build
one out of good parts, maybe, for testing purposes.
Test them en masse - hard drive after hard drive
after hard drive, for instance.

Either way, this could be a nice windfall. Don't forget
to save all the little things that come in handy, like
like expansion card covers, screws, power supplies, etc.

Try 'em out ...

[Pogue Mahone]

From personal experience, the parts that usually fail through age are (in order):

1. Monitors (especially the cheaper ones) - easy to check out, although sometimes they only start failing after an hour or more of use.

2. PSU and CPU fans - usually very noisy, but silent once totally dead. Watch out for this.

3. Mouse - yeah, well...

4. Disk drives - which might be recoverable if it's only bad blocks, but probably not worth the effort.

Motherboards and other solid-state bits very rarely fail through age (unless the CPU fried due to 2). Most failures on these bits happen early in their lifetime. It's called the bathtub curve, and PCs are unlikely to be on the upward slope (unless they're _really_ old, in which case it probably isn't worth the effort of even looking at them.)

So your best bet would be to test all the boxes first, using a known good monitor/keyboard/mouse. Those that don't work can be stripped and diagnosed individually.

--

Been there!

[pongo000]

I volunteered at a local high school to help strip down approximately 60 486's donated by the IRS. From the 60, we rebuilt approximately 40 (including a nice file server and web server). We didn't test each component individually, as we didn't have access to that kind of equipment. Instead, we tested each machine as a unit. Those that passed were put in one pile; those that didn't (for whatever reason) were immediately stripped down, and those parts were put in the appropriate pile

Then we determined how many working computers could be built from the remaining parts. It helped that we were also donated a large box of NICs (some worked, some didn't, but beggars can't complain). From the remaining parts, we built as many working computers as we could.

We initially tried to install RedHat, but we could figure out no way to save the install configuration so we could configure all the machines alike. SuSE works beautifully in this regard: Configure one machine, save the configuration. Plug all the machines in the network, boot from a SuSE install disk, select NFS, and fire up the installs. I believe we had three NFS servers running, and we could get installs going on 20 machines at a time at a fairly good clip.

I think what you're doing is fantastic. I only wish I could convince more schools where I live to do the same. I approached the so-called "technical administrator" of our school district about using some of the boxes I saw piled up in my son's elementary school to ask about getting a low-cost Linux-based network up and running for the kids. At the mention of "Linux," said "administrator" blanched and told me in no uncertain terms that nothing but Microsoft would ever be allowed on school district computers. I just love narrow-minded people.

Teach others to do the work

[ables]

Don't salvage working systems and then give them to a school, rather teach school kids (especially disadvantaged inner city ones) to do this work themselves.

Then handling the situation on mass becomes simply an exercise in collecting the systems and returning the useless parts to the company for disposal (yes, don't let them stick you or the school with disposal costs!)

CMOS Batteries!

[BigBlockMopar]

Lemme nominate a part which I would place between 2 and 3; give it number 2.5 on your list.

CMOS batteries. Especially on those nasty generic motherboards that are too cheap to have any provision for replacement. (ie. no battery socket, no 4 pin connector with the jumper cap over the middle two, etc.)

Either ditch the board or break out the soldering iron.

Salvaging.

[Mullen]

This can be tricky, for something that should be simple.

Somethings to be aware of:
1) Make sure your getting usable goods. Some companies will "cook the books" when it comes to hardware and dump something of nothing of value onto a school and then write off the donated hardware as a tax break. More common than you think.

2) Make sure you get enough people to help you. It sucks taking all that crap apart yourself.

I did something like this for a college lab. Here is what we did.
Nothing was going to be counted as a loss. Meaning, if the motherboard was dead, then we get the cache, CPU and memory off it. Keep all spare parts, you will need them.
So my suggestion is:

1) Have a disassembly line. One person takes out hardrives, another does motherboards and so on. After about 4 or 5 machines, a person is really quick at removing the part. You should be done real quick after that.

2) Put all the parts you yank out into seperate box or area. Make sure to dust them off and look for the obvious problems with the part. This might sound like a waste of time, but you need to clean the parts if they are dusty.

3) After your all done with the disassemble, figure out how many machines your should get, and start assembling them. Your not going to have a perfect match. Some stuff is not going to work, your going to be short of certain items.

4) Make all machines a simular as possible. I know for a fact this will save you lots of time in the future.

Good Luck.