Mercurial > whatever / file revision

=================================
 HP 34970A Data Acquisition Unit
=================================

:Author: David Douard
:Category: Electronics
:Tags: HP, 34970A, DMM, repair, test equipment
:series: HP 34970A repair
:series_index: 1


I recently "scored" a HP 34970A with the DMM included and 2 plugin
cards (HP 34901A and HP 34907A) for less than 200€. The plugins should
be in working condition, but the main switch unit was almost "given"
with the plugin modules since it is non functional.

.. image:: {static}/images/hp34970a/front-view.jpg
   :alt: front view of the HP 34970A data acquisition unit

It's in very good shape, the front panel and the push buttons are
clean. The knob is nice but it has a rough step when turning it for a
complete revolution.

When powered up, however, things are a bit nastier... The VFD is, as
often with these equipment, very dim and the digits tend to puke on
the others. In fact, it's mostly unreadable, and the unit beeps
several times, signs of errors (which I cannot really read them see on
the display).

Internals
=========

The unit consists in 4 parts:

- the main logic/cpu board on the bottom left side,
- the DMM module, juste above the main board,
- the frontpanel and
- the back plane where plugin modules are connected.

.. image:: {static}/images/hp34970a/bottom-view.jpg
   :alt: bottom view of the HP 34970A

.. image:: {static}/images/hp34970a/top-view.jpg
   :alt: top view of the HP 34970A with DMM removed

.. image:: {static}/images/hp34970a/dmm.jpg
   :alt: DMM module of the HP 34970A

The front panel hold the keyboard, the rotary encoder and the diplay system.

.. image:: {static}/images/hp34970a/front-panel.jpg
   :alt: front panel of the HP 34970A

The front panel is managed by a 80C51 (a 87C51 actually)
microcontroller. It's responsible for handling the keyboard and
feeding the shift registers of the VFD drivers (a pair of
HV518PJ). The communication with the main CPU is a serial line with
opamps to adjust the signal levels.

The chips are PLCC44 located under the VFD are the drivers:

.. image:: {static}/images/hp34970a/front-panel-pcb.jpg
   :alt: front panel PCB of the HP 34970A

On my unit, the voltage levels for the VFD are fine. I suspected that
one of the 2 VFD drivers (the one responsible for driving the grids)
were partially failing, so I ordered a pair of new drivers, as well as
some PLCC sockets. It was my first real smd revork. I though that it
might be easier to cut the pins of the soldered chip, but it was not a
good idea, I peeled 2 pads on the PCB. Thanksfully, they were not
broken. So I've undoldered the other VFD with my new cheap hot air
soldering station, and it's been way easier and did not damage the PCB
at all... Lesson learned.

After that first real SMD unsoldering experiment came the first SMD
soldering: install the plastic made PLCC sockets... I was not sure
wether to use the hot air gun or my Weller soldering iron, nor what a
correct temperature for the hot air station whould be adequate. The
informations on the socket datasheet were not very clear to me.

So I gave a try (I think at around 280°C). The sockets seemed properly
soldered, but in fact, a few pins were not making contact. So I
decided to cut the bottom of the socket (which is essentially useless)
so I can more easily rework these "cold" pins with my soldering iron.

The tricky part of installing these PLCC sockets is that decoupling
capacitors (C6, C7, C9 and C10) are a bit too close, so I had to move
them.

.. image:: {static}/images/hp34970a/front-panel-sockets.jpg
   :alt: PLCC sockets for the VFD drivers of the HP 34970A


Now I had to try to rejuvenate the tungsten filaments. Like for old
CRT (and even for triodes), the filament, which is responsible for
emiting electrons toward the anode (coated with phosphor) tend to be
polluted and loose their emissive power. A common trick to "clean"
them is to apply a much higher voltage than rated so they shine for a
few seconds. One mst be careful not to burn them, otherwise the VFD is
definitively lost.

.. image:: {static}/images/hp34970a/vfd-back.jpg
   :alt: back side of the VFD (probably dead) module

So I tried such a rejuvenation on my VFD module. I wasn't sure what
color the filaments must be heated to (between a light orange to an
almost white yellow). I was doing this using my `Lambda PSU
<{static}/LPD422FM.rst>`_ raising the voltage while looking at the
filaments.

The results are very disappointing. The VFD is a little bit brighter,
but the spilling over between digits is now much worse...

I guess this VFD module cannot be saved, after all...

So I've been thinking of a plan B, and I think I'll try to replace the
VFD display with a blue 256x32 OLED module. It's smaller (I cannot
find a module near the size of the VFD display), but it should still
be quite readable. Indeed, that requires a bit of work: I plan to
build an arduino-based adapter (using an arduino micor) that will take
place of the VFD drivers (then be fed by the serial-to-shift registers
outputs of the 80C51). One other solution is to replace the 80C51 with
a newly programmed one, but I think reverse ingineering the serial
communication between the front-panel microcontroller and the main CPU
would require much more work. Since it's a 87C51, I doubt I can easily
dump it's ROM...


Meanwhile, next step is to take care of all these errors.