Yokogawa 2524 PF Meter - Teardown and Repair

[amc184]

I've wanted a meter like this for a while, and jumped on this one quick when it came up on my local eBay equivalent.  Got it for about USD70, described as powering on and showing something on the display, but beyond that untested.





Shown with the temporary wiring I used for basic functional testing

When I got it, before turning it on I took it apart (of course, right?).  Everything looked good except for one thing; the soldering.  This thing had some of the worst soldering I've ever seen.  It looks like the wave soldering was so-so, but the flux cleanup was inadequate and everything was heavily oxidised.  There were lots of cracked and dry joints.  The worst though was the direct wire to board soldering, which all had huge blobs of solder but were still hopelessly dry.  It didn't take any photos, but I wish I had, you could have used them to illustrate an entire article on what failed soldering looks like.

After that I powered it up and tried it out, thinking "I'll be really surprised if this works".  I wasn't surprised.  The display came up, but it displayed zero in all functions and ranges, and the under range indicator stayed on.  Repair time.

The first thing I tried was resoldering the worst of the failed joints; those the were obviously dry and cracked, as well as all the wire to board connections.  I also replaced all the electrolytic capacitors, none were leaking or bulging, but the datecodes reveal this unit is from 1983, so they're probably toast.

After that I got some life out of the unit, I'd start to get some voltage readings when the input voltage was about 200VAC, but nothing before that.  I checked the power supply voltages as well, all looked good.  At this stage I had another look at the soldering and sighed, there were still heaps of nasty looking joints.  On the first pass I had resoldered about 20% of the mainboard and 30% of the sub-board.  By the end of the second pass I had resoldered about 70% of the mainboard and 100% of the sub-board, it was that bad.

That was a hassle, but it was worth it, after the second pass it's more or less working.  I now get realistic voltage, current and apparent power readings on all ranges.  I'm not sure how good the power factor measurements are, and it gives significantly different readings for each of the two methods.  That's probably down to how I tested it though, I think the power factor is only specified for currents that are at least 30% of a range, I only ran 200mA through it (using the 1A range).

Anyway, here's some pictures:



Back of unit, showing (from left to right) the power transformer, voltage measurement transformer and current measurement transformer.  Very little in this unit is at mains potential, just the transformer primaries and the power switch, will be great if I need to do any in depth troubleshooting.  No isolation issues to worry about.  The rainbow ribbon cable is an analog output, presumably for some old school ATE integration.



The sub-board where some of the analog arithmetic takes place.  The bulk of this unit is large number of opamps (NEC uPC4558) and 4000 series glue logic (Toshiba TC4011 4016, 4014, 4049, 4051, 4053, 4071 and 4081) that feed a single ADC.  The switches on the front panel manually select the range and the output of which analog function is fed to the ADC.  There is an intimidating quantity of trimpots, especially when you don't have a service manual.



The mainboard under the sub-board, even more opamps and glue logic.  The ADC is a Motorola MC14433, a 3.5 digit dual slope converter with a BCD output, made exactly for voltmeter with seven segment display applications like this.



The PSU section of the mainboard.  Note the two stripes of missing soldermask in front of the two main filter capacitors.  They correspond to traces on the bottom side of the board that were intentionally exposed, and it looks like Yokogawa were too cheap to make a second photomask for the top, nice.  Apart from that though, it's a really nice board.  Double sided, through hole plated, soldermask both sides - that would have been sooooooo fancy in 1983.



Back of display, too cheap for soldermask here.  Still double sided with silkscreen though.

So, where to from here?  I guess I'll check the voltage and current calibration, though I won't be able to adjust the meter.  If anyone out there has got a service manual for a Yokogawa 2524, I'd really appreciate a copy!  I'll check the power factor calibration at some point as well, but that'll be a bit more of an exercise.  There's heaps of power meters at my workplace (Yokogawa WT230s and the like, modern versions of what I have), so it'll just be a matter of dragging it in there.

A question I have for any old component gurus out there; what are these (black components at the edge of the board)?



They're decoupling PSU rails, and almost certainly capacitors.  They're embossed with the Fujitsu logo, and have two green paint stripes, but no other markings.  They're probably ceramic, but the beveled edge looks like a polarity indicator, which doesn't make sense.  They're 12.4L x 4.3W x 6.3H mm in size with a 10mm pitch.  There are electrolytic, tantalum and film capacitors elsewhere in different packages, which also makes me think they're ceramic.

[SeanB]

Moulded case tantalum capacitors, I have seen them before, and the insides are typical of a tantalum capacitor. Some have 3 or 4 leads, which allows them to have a ferrite bead on the one positive lead to make a lossy LC power supply filter in a single package, with the 4 lead ones allowing you to have a low impedance ground by either using both leads together as a connection or passing the return curreent through the capacitor.

[moose]

long shot...but did you manage to get a copy of the service manual?