Capacitors fail. It's what they do. Aluminum electrolytics caused my monitor power supply to fail. A failed capacitor took down Paul Rako's air conditioner. This is a story of how two burned capacitors and their associated Zener diodes caused an HP 34401A 6½-digit multimeter (DMM) to fail.

I've had this meter since 1992 and have brought it to a calibration lab a few times for a checkup. In October, I brought it to Transcat in Rochester, N.Y. while visiting the area. While at Transcat, I noticed the meter's AC voltage readings were far off, reading 8.5 VRMS with a 10 VRMS input. The DCV readings were well within HP's specifications. What happened?

After bringing the DMM back to my office, I powered it and found some rather odd behavior. On ACV, the meter would give no reading at all—even though the meter passed initial self-test—looking like there was no signal coming in. I was feeding it a 4 VP-P sine wave, which should have produced a 1.414 VRMS reading. To verify that the signal source was correct, I connected it to a true RMS handheld meter. It read 1.410 V (Figure 1). Close enough to 1.414.

AC voltage measurement
Figure 1. As expected, a handheld DMM reads 1.410 VRMS with a 4 VP-P sine wave. That's close enough.

Before opening the meter's case, I knew something was seriously wrong. My nose told me so, having detected the familiar scent of charred electronics. Now I knew to look for physical damage.

Fortunately, I have a copy of the original 1992 service manual, though Keysight makes the manual available. But, there's always the possibility that parts changed over the years.

The 34401A is held together by several screws that use a six-star-headed screwdriver. Mine was too big, but I found the proper size Allen wrench and opened the case. What I saw confirmed what my nose told me, something had burned (Figure 2).

text
Figure 2. Not only is the capacitor burned, but so is the device just above it. If you click to enlarge, you can almost smell the burned parts.

At this point, I had to identify the parts. That's where the service manual helped. Figure 3 shows the board layout. Capacitor C322 (22µF, 20 V) and Zener diode CR305 (3.3 V).

34401A board layout
Figure 3. The board layout identifies the damaged components.

Figure 4 shows the schematic section containing the burned parts. C322 and CR305 reduce an 18 V power source to 15 V, which drives the analog circuits for the AC measurements.
HP34401A power supply schematic
Figure 4. C322 and CR305 provide one of two 15-V sources to analog circuits, shown here as 15A. The components for the negative supplies showed no damage.

I then ordered five capacitors and three Zener diodes, which arrived the next day (free shipping, too). Good thing I ordered extra parts, as you'll see. Figure 5 shows the first stage of the analog circuit schematic, powered by schematic symbol 15A that originate in Fig. 4.

HP 34401A RMS converter schematic
Figure 5. First stage AC measurement circuits receive power from the line marked 15A.

Line 15A also connects to U304D, a DG411 analog switch (Figure 6) that connects to a shift register (not shown).

HP 34401A RMS converter schematic
Figure 6. The 15 V from line 15A powers the RMS converter circuit.

In addition to buying the parts, I also bought a magnifier that straps around my head so I could actually see the little bugs. Now it was time to replace the burned parts. That started with installing a pinpoint tip on my soldering iron. A friend suggested that I use solder with a thickness on 0.020 in., but if that's not available, to use my 0.031 solder and break off a piece by pulling on it. Doing so would stretch the solder, thinning the diameter. That helped.

In Figure 7, the new capacitor and Zener are in place. Soldering the positive end of the capacitor caused some superficial damage to the ribbon cable that connects the front-panel board to the main board.

capacitor Zener diode
Figure 7. New capacitor and Zener diode installed and ready to try.

Continue reading on EDN US to see what happened when I applied power to the meter.

Martin Rowe covers test and measurement for EDN and EE Times. Contact him at martin.rowe@AspenCore.com