Re: Hairy Upslug experience
- I think you should consider the revised method I put up at the wiki.
It disables the onboard RAM and then stacks the new RAM on top and uses the chip
selects. It does not require removing the existing RAM and is reversible if necessary. My
256M slug was built that way and the conversion was much faster. This unit runs 24/7 for
me and works great.
--- In firstname.lastname@example.org, "Rob Lockhart" <rlockhar@...> wrote:
> Ted, if you are going to do that, sounds like you already know the right
> process to do it. But without a scope, it's quite difficult without
> commercially-available tools (like a Metcal). Put another way, I wouldn't
> even think about doing this with anything that Radio Shack sells.
> With solder wick, it's quite difficult to remove the original memory. What
> isn't so hard to do is to buy some identical memory (8Mbitx16) chips and
> piggyback those on top (with bending all pins downward but the pin specified
> on the yahoo groups pictures - bend that up), and wire the chip-select to
> the specified resistor pack.
> I would not suggest solder wick with the SDRAM - you're more likely to
> damage the PCB. A guy at work, a former technician, said a good technique
> they used in the factory was to put an excessive amount of solder on one
> side of a TSOP and heat it up with a wide blade. That served to ensure
> uniform heating and that no pads were ripped off when removing the package.
> I normally remove four surrounding caps (record the polarity as they're
> electrolytic), the battery, and buzzer before even touching the RAM. The
> last time I did that, one pin of the battery gave me grief; it could have
> been I had a marginally functioning solder tip (they do go bad on the Metcal
> units). I apply a liberal amount of flux to the pins. Then, I use two very
> wide blades (two separate soldering irons), and wait for the pins to heat up
> and gradually "slide" the part vertically away from the pads. I do both
> memory modules this way. Then I go back with flux and solder wick, cleaning
> up the pads. I also clean the new memory pins (if necessary). I place the
> part as close to center x-y as I can see in the microscope. I solder one of
> the center pins on one side, then one pin on the other side, then I proceed
> to solder all the other pins. The other memory module is done the same way.
> Now, I find the pin to bend upwards, and bend that up. It's 9th from the
> corner in the pictures (technically pin 19). Then I bend the remaining pins
> together downwards with a flat object. The memory module is placed on top,
> centered, and two pins are soldered (as above, one on each side). I tilt
> the PCB at about 45 degrees and proceed to solder the pins from bottom to
> top (using very fine solder and only a tiny amount). I find it easier to
> work left to right, with solder in left hand, iron in right hand. Flipping
> the board around, I do the other side (also at 45 degrees). Then I do the
> other memory module the same way.
> I inspect two or three times and ensure no solder bridges. Next, the CS pin
> is wired. I use 30-ga wire, and the best way to do this is to cut it a
> little bit longer and trim later. Attach to resistor pack pad, then bend it
> over to the top of the memory modules. It's much easier if you just strip
> the wire instead of worrying about keeping the insulation on the wire.
> Finally, I replace the components (be sure to remember the polarity!): 4
> caps, battery, buzzer. I usually put on a serial header at the same time (I
> suggest this for any and all hardware mods).
> Hope that helps!
> On Sat, Jun 21, 2008 at 11:32 PM, Theodore Killmeyer <tedkiller@...>
> > wrote:
> > Thanks Rob,
> > I am a EE, but don't have the scope and soldering station that would make
> > the process easier, that's why I was looking for someone to sell me one or
> > just create one for me (i.e. for Money) .
> > I guess I need to dig out my needle soldering iron and the solder wick and
> > get busy.
> > Ted
> > From: email@example.com [mailto:firstname.lastname@example.org] On
> >> Behalf Of Rob Lockhart
> >> Sent: Sunday, June 22, 2008 12:30 AM
> >> To: email@example.com
> >> Subject: Re: [nslu2-linux] Hairy Upslug experience
> >> Ted,
> >> Those FatSlugs aren't "made" anywhere in any sort of assembly line. I've
> >> manually made a few, through blood/sweat/after-work and using my work's
> >> equipment and scrap SDRAM components. It takes me about 4 hours, if I'm
> >> careful, don't make any mistakes, and if I check and re-check my work. Then
> >> about another hour to manually do some simple memory peeks/pokes to ensure
> >> no stuck address/data lines.
> >> I'm sure that someone could set up some kind of operation for FarSlugs,
> >> but I'm not as the SLUG project is a hobby for all of us here. There are
> >> lots of instructions available, and I could update the Wiki (with more
> >> useful manufacturing information) if someone wants to take on the task of
> >> streamlining the FatSlug process. I'm sure the process can be done much
> >> more quickly than 4 hours by people that do this sort of task all day long.
> >> Best regards,
> >> -Rob
> >> On Fri, Jun 20, 2008 at 9:00 AM, Theodore Killmeyer <tedkiller@...>
> >> wrote:
> >>> Hey Rod,
> >>> I did a google for fatslug's and looked at Ebay, do you know anywhere I
> >>> could purchase a fatslug?