Connect with us

Memory investigation on an HP35665A and bad desoldering

Discussion in 'Troubleshooting and Repair' started by (*steve*), Aug 2, 2015.

Scroll to continue with content
  1. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    Jan 21, 2010
    (Yeah, this doesn't sound like a repair... but just wait.)

    I have a couple of HP35665A's, one with 2MB of RAM, the other with 4MB. Just out of interest I decided that I would try to upgrade the RAM on one of them to 8MB (the max). When these were sold, memory upgrades were a board swap. 2MB (Std), 4MB, and 8MB were available.

    The thing is, these are rather nice pieces of equipment, so I didn't really want to sacrifice on (no matter how good the cause).

    I managed to get another 2MB RAM board, and as it has newer firmware, I resolved to start work on my original 2MB card.

    IMG_7162b (Custom).JPG

    This is an original card. The RAM is 2 banks of 8 chips at the middle top of the card.

    These are 1Mx1 chips, a 4M card has a single bank of 4Mx1, and the 8M card has 2 banks of 4Mx1.

    The boards all have different part numbers, and there are differences other than just the count of RAM chips. However, my 4M board is much newer than my 2M board, so the differences could be due to factors other than function.

    My idea is to remove half the RAM and to try to get it to boot up displaying 1MB of RAM. If I can get that far, it seems reasonable that the board supports both 1 and 2 banks of RAM (as would seem likely because the banks are there!). It's also half way toward removing all the RAM so I can replace it.

    The difference between 1Mx1 and 4Mx1 is a single address pin, and that pin position, while absent on the 1Mx1 chips, exists on the board, and is also wired up, further evidence that the difference is likely only in the chips used to populate the board.

    I started this project when really sick with the flu. This really wasn't a great idea, and although people think I can solder by sense of smell alone, the same does not apply to desoldering.

    This board was one of the first HP produced with up to two tracks between 0.1 inch spaced pins. What I didn't realise was exactly how delicate the traces were :-(

    WP_20150801_15_06_06_Pro__highresb (Custom).JPG

    Here's the board with the "optional" RAM removed.

    WP_20150801_15_07_22_Pro__highresb (Custom).JPG

    And a closeup of the top side of the board.

    I really don't want to show you the other side because I would not be overly complementary of someone's skill had they shown me.

    WP_20150801_15_07_37_Pro__highresb (Custom).JPG

    That's nasty.

    I started from the right, and I only really looked at my work after doing 7 of them.

    Actually, the board looks quite a bit worse than it is. The board has a coating on it, so some amount of the nastiness is this being scratched off, but yeah, there's quite a few destroyed tracks.

    It turns out there were several problems:
    1. Don't start something like this when you're unwell.
    2. The solder appears to be a high temperature type (ROHS?)
    3. The traces are unbelievably delicate.
    4. The suction from the desoldering tool was poor.
    The last of these factors was arguably the most significant. I "fixed" the desoldering gun, and the leftmost set of pads are the ones done with a much more efficient tool.

    Here is a picture of a fixed desoldering tool:

    IMG_7161b (Custom).JPG

    As mine was delivered, the clear solder receptacle had an open spring pushing against a metal disc, which was then pushing against a filter pad which prevented solder getting into the vacuum tube.

    My fix was to replace the spring with a chunk of "solder sponge" , the disc, and finally some cotton wool.

    This provides better airflow and made the desoldering dramatically easier.

    And here is where the fun started. I had to track down and repair all the broken traces.

    IMG_7160b (Custom).JPG

    Aint wire wrap wire wonderful?

    To cut a long story short, after a few goes (and the discovery of a jumper -- marked 1M -- that seems to enable single bank) the result is this:

    IMG_7159b (Custom).JPG

    The device doesn't actually boot up, but that may be because it requires more than 1MB of ram to do so. But it is encouraging!
    Last edited: Aug 2, 2015
    rer likes this.
  2. dorke


    Jun 20, 2015
  3. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    Jan 21, 2010
    I also have one, and the fact I didn't think of using it is another testament of not trying to do stuff like this when you're sick. I might give that a go for the other bank.

  4. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    Jan 21, 2010
    Unfortunately the hot air scorches the board before the solder even hints at starting to melt.
  5. dorke


    Jun 20, 2015
    It sure can scorch the print,but that means you haven't used it correctly.
    You should use it at a distance of about 3-5cm ,
    and more importantly, keep moving it constantly along the pins till the solder melts.
    Adjust the temperature to the lowest needed,
    I try to go with 290c and with relatively high air flow.
    I use the 5 mm round nozzle mostly.
  6. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    Jan 21, 2010
    Unfortunately I was using it correctly. I may have to pull out my board pre-heater to cope with the ground layer that is sucking all the heat away.

    Although, I did go out and give it one more try (at 100% airflow this time)...

    At 290C, even if I leave the air pointed directly at a pad, the solder does not melt.

    At 335C it eventually starts to melt the solder, but you can't get the whole length of the chip to remain melted. At this temperature, the board also starts to discolour.

    And yes, I also have a 5mm round nozzle that remains almost permanently attached to the gun.
  7. dorke


    Jun 20, 2015
    those ground plains(and power plains) are a big problem.
    In this case maybe a combination of sucking only them first,
    and then using the hot gun for the rest to pull the chips out.

    Another "brutal", but effective way .
    is to cut the sip chips with a good cutter and pull the pins out one by one while desoldring.
  8. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    Jan 21, 2010
    Well it's been a while... I was building a memory board but there's less space than I realised.

    So I bought the required chips to upgrade the board and populated it. After fixing my poor desoldering again, it works but it doesn't see all the memory.

    I'm hoping it's a jumpering issue.
  9. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    Jan 21, 2010
    Woo hoo!

Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day