Fluke 5440B "Repair"

[wn1fju]

My Fluke 5440B started to act up after a period of dormancy.  The main symptoms were: 1) wouldn't hold calibration from day-to-day, 2) there
was a DC offset (at 0 V setting) of about 0.1 mV (went as high as 1 mV at one point), and 3) wouldn't return to the same voltage after I changed settings (which exercised relays). 

I pulled all 25 relays and found several contacts >10 ohms and a couple that were open.  Since Fluke paralleled most of the contacts, I thought
it might not be such a big deal.  And a lot of the relay sockets looked quite corroded to me.  So I burnished all of the relays and things seemed
better for a day or two, but then back to the original problems.

Faced with the prospect of spending something like $1,000 for 25 new relays, I decided to do the unthinkable!  I found a good deal (about $1
each) for a bunch of new DS2E-S-5VDC DPDT relays (the same ones used in the Fluke 5700-series calibrators).  I needed a total of 4x1 +
5x2 + 16x3 = 62 relays to cover the DPDT, 4PDT and 6PDTs in the 5450B.  Out came the old relays and sockets, and in went the new relays,
upside down with wiring to the board.  See the attached photo for a sample board.  I was a bit concerned about running 1000V in the 5440B,
but the relays are rated for that and the wire insulation was also rated for 1000V. 

So far, my problems seemed to have been solved and I don't notice any particular increase in noise.  The new relays are quite nice since they
draw about 1/4 of the power of the old ones and are, of course, much much quieter.  I hope they last!

Finally, in order to preserve the VFD, I decided that there was no real reason I needed the display turned on for example during the 4-hour
warmup period.  So I mounted a miniature DPDT switch on the front panel and interrupted the 29 and 5 volt lines on the ribbon cable
connecting the display board to the control board.  Now I can switch the VFD off and on as desired.

OK, tell me I'm an idiot and it will never work.

[TiN]

Perhaps making a "patch"-board for new relays would cost you less time then wiring all this..

[wn1fju]

"Perhaps making a "patch"-board for new relays would cost you less time then wiring all this.."

I did give serious thought to that and actually attempted to lay it all out.  It was a little complicated because in my 5440B, Fluke did the
4PDT relays by using a 6PDT socket and then using a non-standard arrangement where rows 1, 2, 4 and 5 were connected, rather than
the expected 1, 2, 3, 4 (or 3, 4, 5, 6).  Also, the replacement relays used a polarized coil (unlike the originals) and Fluke was not consistent
which pin (left or right) was positive.  So I needed to account for all of this.

To get all the combinations on a single board without having to design and procure multiple boards was problematic.  More problematic,
however, was that it was a real tight fit, particularly on the sample string board shown in the picture, and was starting to look impossible.

The whole relay replacement took about 20 hours of work, including ripping out all the old sockets and testing all the new relays.  I only
managed to destroy 2 relays during the whole process.

I wouldn't wish this task on my worst enemy!