By Jeff Bach
One of my favorite problems to stalk is the ever-elusive intermittent short circuit. Shorts can come in a variety of manners and descriptions. The descriptions given by customers are sometimes so vague it is a challenge just to figure out which circuit is malfunctioning to the point that the customer believes a short is involved. I just love it when a customer comes in with a 10-year-old American luxury car exhibiting multiple unrelated electrical problems that have happened over a long period of time and somehow the customer is relating them to a single “one money” short causing all of the problems on a list that looks as long as the “honey do” chart on your day off.
I recently had a discussion with a customer about a short in the taillight circuit that they had themselves convinced was due to a trailer wiring harness that had been recently installed. The customer was fixated on the fact that the truck didn't have a problem with the taillight fuse blowing before they had the trailer harness installed and basically wanted us to find out what the rental company did wrong to their wiring in order to assign culpability.
Since this intermittent short was actually “acting up” in the stall, we installed a current-limited pulse width modulated device in place of the blown fuse, which sends a current signal to the affected circuit that can then be traced using a lab scope and a magnetically sensitive probe. The closer the probe is placed to the affected circuit, the stronger the signal.
In a short time, we traced the short to the wire crossing from the left taillight over to the right. This wire was pinched between the bed and the frame - probably at the factory, but it did not show up as a short until recently.
Figure 1 shows the wire from the outside of the frame next to the bed mount. I understand now why the customer assumed the short occurred after the installation of the trailer harness.
Figure 2 shows the deteriorated mount from the inside of the frame.
The additional weight of the trailer being pulled caused the crumbling mount to allow the bed to smash the wire against the frame. The rust, not being the best conductor, allowed the short to become intermittent.
Another reason I like intermittent shorts is that their stories make memorable impressions in the minds of technicians ... stories worthy of retelling whenever the subject comes up.
One such short came to me as a referral from a transmission shop with the instruction that the electronic transmission on this '91 Caddy was stuck in “limp mode” (second gear), but would shift with the “box.” This means the internal transmission is OK and the problem has to be computer- or electrically related. After checking the shift solenoid's feed circuit fuse with a test light, I used the car's on-board computer to command the operation of the A and B shift solenoids while monitoring their current patterns with a current probe and a lab scope. No current was evident on either solenoid. I tested for power at the transmission connector and found none. This meant there was an open between the fuse block and the transmission connector. An initial visual inspection revealed no obvious signs of mispositioned, chaffed or cut wiring harnesses, or add-on electrical accessories. I traced the open to the back of the fuse panel where I found the image shown in Figure 3.
The burned-away section of wire - an obvious sign of over amperage in the circuit - shouldn't have been able to occur had the proper circuit protection been maintained. But during my exploration and attempt to document my findings, I discovered the evidence shown in Figure 4 - a melted 30 amps fuse - leading me to believe that once the wiring circuit was restored to proper operation, again I would begin my quest for the real problem.
The multitude of circuits fed by the 15-amp No. 5 fuse bears a striking resemblance to that of the offshoots of a spider plant.
Beginning with the transmission control solenoids (see Figure 6), I proceeded to operate the individual circuits using the on-board computer's output cycling function. With a current probe connected around a fused adapter plugged in place of fuse No. 5 and hooked to my lab scope, I proceeded to record waveforms for each affected circuit.
The waveforms for the shift solenoids (Figure 7) showed normal current draw.
The brake switch is wired inline with the VCC solenoid (Figure 8), so its circuit was tested during the output cycling of the VCC, producing the standard gull waveform shown in (Figure 9).
The EGR solenoid was next on the list. This solenoid gets current flow at key on and also looked normal when I cycled it (Figure 10).
Next, I tested the canister purge solenoid (Figure 11). It has been known to cause intermittent shorting problems, especially when wet, due to its location in the front fender well where it is exposed to the elements, particularly particles of the sodium family used during our Ohio winters to clear the roads.
The power steering pressure switch circuit (Figure 12) sends a data signal to the PCM during parking maneuvers to indicate additional engine load, hence no current waveform. Testing it consisted of manipulating the harness while having the scope monitor the circuit for glitches. None were found.
The last item in the circuit was the A/C relay. Notice in Figure 13 that the A/C has two power feeds from fuse No. 5. The first one is for the relay coil, and the second for the A/C clutch coil. The waveform in Figure 14 is the first of a two-part test that illustrates the current for the relay coil up to the instant the relay contacts close.
At this point, the current goes up off the scale. This is normal due to the scale setting the probe has been on to read the relay and solenoid currents I've been testing up to now. These currents have been in the 300-350 mA range. The pertinent information held in the typical relay or low current solenoid can be displayed in a 20mS time frame. A magnetic field with enough strength to pull-in and hold the A/C clutch plate, however, requires substantially more time and current to build. To capture and view this event taking place, the current probe needs to be switched to a higher scale and the scope's time base changed to bring the clutch engagement waveform into view. Only 60 to 80 mS of waveform will tell you all you need on an A/C clutch coil. A scope screen capable of 5 amps should capture the pertinent vertical portion of the wave. On my scope, this is the 5 and 50 setting - milliseconds and millivolts, that is. I think of it as the old “emissions warranty screen.”
Figure 15 is what an A/C clutch current waveform from a '91 Deville should look like with these settings during output cycling.
Anticipating a waveform to fill up the screen with the settings I had on my scope and probe, imagine the surprise I felt when Figure 16 appeared.
All I can say is that I instantly got one of those feelings that happens when, say, you're fishing and a nice largemouth bass breaks the water, dancing on top of the lake with his tail giving you full view of your wet line - glistening in the early morning sun - making a beeline to the corner of his mouth. “Gotcha” is what you instinctively say to yourself, and you know you're right.
I had to adjust the scope up two more scales (from 50mV-200mV) to get the voltage level to show on the screen (Figure 17).
The sharp rise upward in the beginning of the pattern indicates a short in the A/C clutch coil circuit. Though the fuse was not blowing and the A/C was working fine, I knew I had found the problem. I took another pattern after replacing the clutch coil. The result, shown in Figure 18, shows a classical (IMHO) example of the expected "Gull" waveform.
Hard to believe one can get these gratifying feelings from pictures on a scope screen, but ... that's the way it is with us scope junkies.
