Teardown, Repair and Analysis of an Agilent E3642A DC Power Supply

In this episode Shahriar attempts a repair of an Agilent E3642A DC Power Supply which is completely non-responsive. After presenting a teardown of the power supply, the GPIB interface is used to verify the functionality of the power supply. The problem is traced to the main display unit which communicated with the main power supply via a serial interface.

After disassembly of the display, it is revealed that the entire unit has suffered a catastrophic failure due to the VFD display drive IC. All components must be individually removed and replaced. Unfortunately the main processor is a Mask ROM IC version (80C51) and cannot be sourced. Can you help Shahriar find a replacement part?


  1. Wim Peeters says:

    Maybe jou find on other E3642A on ebay that does not work, but has another problem.
    If it is cheap, you could the use the display print of that.

  2. Volker Bosch says:


    you might like to try out the AT89S51. It is a flash based replacement of the ol’ i51. I once used one of these to replace a i51 for a redesign of an interface board.

    According to the data sheet the AT89S51 is available in PLCC housing and costs little money, e.g. http://www.reichelt.de/Atmel-C51-Controller/AT-89S51-PLCC/3/index.html?ACTION=3&GROUPID=2960&ARTICLE=50563&SEARCH=at89s51&OFFSET=16&WKID=0&

    Unfortunally you still need the firmware… 🙁

    Good luck!


  3. BobC says:

    My guess is the display board would be common to all instruments in the series, without changes. So the CPU’s primary job should key scanning and providing a host interface.

    I say this because the only upgradable firmware is on the main board: The display CPU should be generic (being a mask-ROM).

    From the schematic, the interface looks very I2C-like, with the same shift register being used for serial input and output.

    So, what is the main processor sending to find out the display is bad? Time to listen and decode!

    Oh, blast: Someone over on the EEVBlog forum already suggested pretty much the same thing.

  4. jgorsk says:

    Agilen’t didn’t use to set lock bits. The question is can you read back rom memory contents
    of N80C51 with a programmer. I’ve never tried that, not sure if that’s going to work on a mask rom 51.

    I had the same problem with a faulty display on the 34401A. It has the 87C51 micro. I could read the memory with no problem.

    • David Douard says:


      I am trying to build a replacement display for my 34970A which VFD has failed, but the 87C51 also died recently. Before the 87C51 died, I was writing the software for an NUCLEO board driving a blue OLED display just by sniffing the serial communication between the DP and the main board (thus keeping the original microcontroller handling the communication and the key pad). It was beginning to work quite well… until the controller died. So event if the code of the firmware of a 34401A may not be exactly the same, it sould be pretty close and it would help me a lot to have a copy of it. Would you be kind enough to send it to me?

      BTW, my repair “crusade” of the 34970A is described here https://whatever.sdfa3.org/hp-34970a-data-acquisition-unit.html (I’ve already reverse engineered most of the communication protocol, but it’s not complete yet).


  5. Pierre says:

    I’m a bit worried about the capacitors on this board, the tantalums on the 5V rail got 36V on them for a while…

  6. user says:

    Here’s a version of the manual, which contains some schematics, at least it has schematics for the frontpanel board.

  7. Franz Vollmaier says:

    Really great video!

  8. rasz_pl says:

    wouldnt count on ever getting your hands on the firmware, unless you send it to a decapping service 🙂 or find another working one _and_ HP didnt set protection fuse, they can be read in eprom programmer.

    google “Is a Sound Blaster Pro 2 worth decapping?”

    “easier” solution might be emulating it with another micro, thats 1-2 weeks of reverse engineering right there (unless you find another working one, then it should be trivial to decode protocol used between panel and main unit)

    • Windsor Chan says:

      Too bad that the manual wasn’t very detailed. I really like the old school manuals from
      Tektronix and HP.

      Perhaps instead of reverse engineering the protocal you can shoehorn a GPIB host into the 8051 and wire that into the bus 🙂


  9. Joe says:

    I am actually curious how you repaired the damaged VFD driver pad.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

  • Social Media

  • Administrator

  • Archives

This site is protected by Comment SPAM Wiper.