We often mention power supplies when we talk about the unsung heroes of enthusiast computing. But what about the humble power bar? Aside from the obvious task of providing multiple outlets for all of the gadgets you have around your computer, they also provide a central, common ground location for all of that equipment, minimizing ground loops. Most bars also give you some degree of surge suppression and power filtering.
In this picture story, we're taking a look at what goes on inside of a vintage APC Performance SurgeArrest (specifically, model PF11VT3-CN).
Here, in all of its curvy glory, is APC's Performance SurgeArrest with 11 outlets. Six are power brick pads, three are always-on, and the rest are controlled by the built-in switch. You get a two-meter power cable, phone line and coax cable surge protection, a 15 A breaker (on the left side), status indicator LEDs, a removable cable management clip, and 3200 Joules of surge suppression with $100,000 equipment protection warranty.
Given such a show of confidence from one of the top-rated brands in power products, you'd expect fairly good quality from this power bar, which used to sell for about $40. APC's nearest equivalent model today is the P11GTV, which has a similar outlet configuration, redesigned top-end layout, and a more blocky design overall. It retails for around $60.
The back of the unit features two mounting slots (which can accommodate three orientations), six screws holding the unit together, the wire management clip's attachment slot, and, of course, the electrical ratings and certifications sticker. We'll take a closer look at that next.
This unit was designed in the United States, but was assembled in the Philippines. It has all of the usual safety certifications, an FCC file number, and transient voltage suppression ratings of 330 V across all three power cable wire pairs.
Oddly, the French warning on the label allows for the use this power bar for anything, anywhere, as long as it stays dry, while the English warning only permits use indoors and makes an exception for aquariums.
Back when I originally purchased these bars (I own three), I thought the cable management clip was neat. Over time, I gave up on using it; invariably, the device I wanted to move ended up being the one with its cord nestled deepest into the clip. The plastic's somewhat sharp edges also contributed to my decision to quit using it.
Perhaps unsurprisingly, APC's updated SurgeArrest line-up dropped this feature in favor of cable straps.
Six screws later, the back cover comes off effortlessly, which is considerably more pleasant to deal with than the mess of ultrasound-, solvent- or induction-welded enclosures.
At a glance, everything looks nice, clean, and modular: there are shiny sheet-metal bus bars for power and ground distribution to outlets; separate modules for coax and phone surge suppression; and a larger board for the bar's core business—power filtering and surge suppression.
The only questionable design choice apparent in this picture is the daisy-chained grounds for the phone and coax modules. For surge suppression to be most effective, the lowest impedance possible between the surge suppression devices and ground is required. Ideally, APC should have connected all of its grounds to a central point within the bar, preferably where the main cable's ground connects to the PCB.
APC uses three different styles of connection strips:
- simple slit squares for all three ground bus bars
- offset stamped metal for the five center outlets' live and neutral connections
- more elaborate arced fingers for the power brick outlets
I see at least two problems here:
- The copper alloy that APC employs is very plastic. If anything happens to bend those contacts a little more than normal, they become loose. This is most evident in the ground strips, and I had never noticed how loose some of them had become until I tried plugging my lamps into another SurgeArrest bar for this photo shoot, only to discover that the center outlets were unusable with the thin plug blades.
- More daisy-chained grounds. APC could have at least connected the phone/cable surge suppression ground to the top strip, to which the ground connection from the main PCB connects.
Let's take a closer look at those connection strips.
Since APC uses exactly the same ground strips for the power bricks and the middle row—with the brick strips only having every other position available—we can clearly see the difference between virgin ground connection slots and those that have seen some traffic. The wear on the other types of contacts is not readily observable, but the metal color does suggest that the strips are made of the same malleable copper alloy. APC probably should have used a more springy or elastic alloy in these things.
With some of those ground slots bent beyond perpendicular, you can start wondering how well the ground contact with plugs might actually be after a few insert-remove cycles.
The rough child-proofing mechanism in front of the live and neutral connections does not help with smooth insertion. That likely contributed to the ground connections taking a worse beating than they should have.
Instead of soldering, screwing, or crimping the wires to the bars, APC opted for spot-welding, providing a quick (yet durable) connection.
Here is that little RF surge-suppression module, with all of its goodies hidden under a metal skirt.
Looking at the back of this PCB, it is obvious that there's little going on. Pins related to the F-connectors aside, there are at most two other components under there. One might be a gas discharge tube connected across the signal path and ground between the two F-connectors, and another, possibly a jumper, connected to the same trace at both ends.
APC cut corners by leaving the RF trace exposed on the back instead of using a double-sided board, so it could have put this on the top layer (facing inside the shield) with a solid ground layer on the bottom to complete the shielding.
Next comes the phone surge-suppression module. It's nothing fancy: a simple pair of fuses and two S14K175 MOVs.
One of those ground wires has a couple of strands splayed around the wire's entry point through the PCB, and the other has a lump of solder, presumably stuck to more stray strands. Was this unit assembled on a Monday morning or Friday afternoon?
On the rounded PCB edge, we see the wiring fault red LED, the overload yellow light with the neutral wire passing through a current-sensing transformer, and the Protection Working green LED.
Closer to the middle, there is a pair of beefy ferrite bar chokes and a 0.47 uF MPX-X2 providing some line noise filtering, along with the built-in switch for the eight corresponding outlets.
Along the right and bottom, we find nine MOVs providing surge protection. A 15 A built-in breaker and two thermal fuses are amongst them. The protection uses 20D201K parts (200 V) across live- and neutral-ground connections, while the other MOVs are 20D471K components (470 V) that go directly across live and neutral.
A 221R UMI thermal fuse (the thin, long, yellow device sandwiched between the four 20D201K on the right edge) rated for 102 C and 4 A breaks the ground connection should the surrounding MOVs start overheating. There is another thermal fuse heat-shrunk between the two MOVs above the breaker that is in series with the live connection directly from the power cord. It cuts off all power to the bar should those MOVs overheat.
At a glance, everything looks nice. Upon closer inspection, though...
Here is that back side again. See anything odd? Aside from the sloppy use of soldering flux, there are at least three issue that stick out—or fail to, in this case:
- The ground wire shows clear signs of a cold solder joint (bottom-left).
- The neutral wire does not come through the PCB (just above dead-center).
- Neither does the live wire (bottom-right).
This is surprisingly sloppy assembly, especially considering those are the three most important solder joints in the whole contraption. They all got botched up in this particular unit.
First up is the least-horrible of those bad connections: the neutral wire. It was clearly stripped long enough to go through the PCB and form a solid connection. But most of the exposed length is out on the PCB's component side, with little more than a dimple in the solder puddle from which it should have been protruding. This indicates that the wire was pulled back somewhat before the solder solidified or that the wire never got through (but did get hot enough to wick the solder off).
This isn't good. But at least the wire looks like it is thoroughly covered with solder, and it has the extra stiffness to prove it.
Second on my list of bad soldering jobs: the live wire. From the bottom of the PCB, it looks like a few strands are slightly protruding from the solder puddle. On the other side, the wire appears bridged to the PCB by a blob of solder coming through, with about half of the strands scattered around.
Again, you can only scratch your head and wonder how such a connection got past QC.
Third, there's the ground connection. And it takes the cake in my book. Of the three wires, this is the only one that passes cleanly through the PCB, yet it is also the only one that turned out to have a cold solder joint. Not enough heat was applied to the wire and solder failed to soak in.
If you ever took an electronics class with soldering involved, this is exactly how your solder joints should not look—particularly if it is both performance- and safety-critical. Like, say, the ground in a surge-suppression device.
I suppose tidiness is not too important, as long as the solder joints are good enough to pass automated testing. But this doesn't exactly inspire confidence.
Remember how this power bar is supposed to have a built-in 15 A breaker? Well, I got the inspiration for this story from an "accidental operational test" when a power supply I was trying to repair shorted out and blew up one of my SurgeArrest power bars.
- Up top is what that PCB area normally looks like
- Down below, you see my busted power bar's PCB
The first thing I noticed upon opening the enclosure was two pieces of thin, mostly transparent, green, paper-like material rolling around. Looking at the blown trace of similar width, I concluded that the pieces must have been the solder mask covering the trace before it blew up.
For the repair, I used a simple piece of solder wick to replace the blown trace, stiffened with solder to prevent it from flapping around and possibly shorting to the ground island in the middle. In the process, I started wondering why APC intentionally included such a weak point. Couldn't the company have simply soldered over those traces like it did almost everywhere else across the board? There are at least two reasons I can think of:
An intrinsic fuse: if all other current protections fail to cut power, a blown trace definitely will.
A sacrificial trace: In the event of a high-energy fault like a lightning strike, the MOVs will short enough current to ground to blow up the trace, thereby physically disconnecting the loads from the source. Because the arc this would create has to pass right next to the ground island, the arc should jump to that lower impedance path to ground, relieving the MOVs of their load.
Obviously, my repair voids those presumed extra safety features. But I'm still left wondering why bother with the cost of a built-in 15 A breaker if the traces within it blow up first when a faulty device is plugged in? Even the 6 A fuse inside the power supply doesn't have enough time to blow before the bar's trace. Does the built-in breaker even work? I had another go at fixing my defective PSU after repairing the power bar, and this time, the main breaker tripped first. The PSU's fuse burned out as well.
What could possibly go wrong in a name-brand power bar from a major manufacturer? A surprising number of things, apparently. This is what happens when quality control gives a sloppy manufacturing job a pass, thinking that nobody is going to look inside after the lid goes on anyway. The power bar with the blown trace was a P8T3-CN, and it had none of the sloppy wire soldering I saw in the PF11VT3-CN.
Originally, this story was supposed to be a simple look at what goes on inside of a quality power bar, but ended up turning into an unexpected cautionary tale—a reminder that even top-brand manufacturers have bad days on their assembly lines. After I finished the initial photo shoot, I jumped back in there and fixed the poor main cable solder jobs.