An Olds '98 "no-start cold" with a 3.8 engine was recently dropped off at my shop. The customer said he had been referred to us as his last hope. So I did some research, and like many of the 3.8s I read about, this one had been to a few different shops and a myriad of parts had been tried in attempts to remedy the problem, but it had continued to come back when the right conditions were met. The parts list included a battery, fuel pump, crank sensor, power control module (PCM), coil pack and ignition module in various combinations of new and used.

The customer said the car would run fine most of the year, but would not start sometimes in "real cold" weather. If he did manage to get it started after fooling around with it long enough, it would be fine all that day until it sat long enough to cool off again, then it may or may not restart with enough attempts. He also said it would be a certainty that it wasn't going to happen on the first try in subfreezing conditions.

Given the current weather conditions (Figure 1) and considering that my shop is not equipped with a "freezer bay," I decided to let the car sit outside overnight in single-digit temperatures. Prior to that, however, I pulled it in to the shop and installed a few wire-piercing banana clips in some pertinent circuits while they were still flexible, and a fuel pressure gage (Figure 2).

I then dropped the PCM down from under the glove box to gain access to the injector circuits with the current probe. Under these weather conditions, testing is challenging at best due to the fact that even the slightest amount of manipulation of a wiring harness becomes difficult because of its inflexibility due to the cold temperature. Vacuum hoses will snap like glass with very little flexing and plug wires can be extremely difficult to remove without damage. I take advantage of these low-temperature occasions to stock up on a few of the more common part numbers I use on certain model Cadillacs - my specialty - such as power supplies and body control modules (BCMs) from the local salvage yard. I stock up because they are easily removed (even with gloves on) without the need for the special sockets and dexterity often required to manipulate around in the sometimes-caved-in dashes. The brackets and mountings that encase these modules like little plastic prisons snap like clay without putting up much of a struggle.

With the car parked outside and the test leads installed, I just had to hope the weather didn't change drastically as it's prone to do here in the Midwest.

Fortunately, the nightly news meteorologist was right in predicting near zero.

On this day, the failure conditions were ideal. With the wind chill, it must have been -20 degrees Fahrenheit. I feel like I'm embarking on an adventure.

When testing for signals under circumstances such as these (did I mention the wind was blowing rather briskly?), I try to get as much information as I can in the least amount of cranking time and disturbance to the wiring (Figure 3).

There are many factors involved in getting an engine to fire when temperatures get this cold - not the least of which is getting it to turn over with the oil as thick as molasses. The battery is less chemically efficient in colder temperatures, causing a lower output voltage. With less voltage potential the starter is now being cranked at a slower speed. The fuel pump also is running slower, first because of the lower battery voltage and second because of the extra load from the starter. The injectors have a lower voltage available to them and in many of the 3.8 powertrains share their feed circuit with the ignition system and the fuel pump. The coils need to produce a usable arch across the spark plug electrodes while also receiving a lower ration of electrons.

Fuel volatility is another factor that adds to the difficulty of starting in cold temperatures. Any one of the aforementioned parts can be getting the job done in slightly warmer temperatures while being marginally borderline, but can't quite pull it off in single-digit temperatures. Ground connections and contact terminals can show their intermittency as they contract with the absence of the sun's warmth. With all of the intermittent possibilities available and most of the guessing done already, I felt it was necessary to try to do some pinpoint testing in failure conditions that would yield positive results. Bear in mind that once this car starts, it will keep starting and running fine until it sits again for a long period.

A quick key-on fuel pressure check eliminates one suspect.

With the test leads previously installed, now I need to connect the scope.

With my Fluke 99b trigger set for a single shot, I crank the engine for about a third of an engine revolution (that's about ... arrear ... arrear) and get the first cold cranking primary current signal and the crank sensor 3X (Figure 4).

I have now verified that I have a decent crank sensor signal and primary circuit current, which - at least for the present - leads me to think the module and crank sensor are doing their job.

The next item on my short list is the cold cranking first injector pulse (Figure 5), which tells me a lot about the PCM control system and allows me to shorten my suspect list even further.

From the injector pulse shown in Figure 5, with a few simple mathematical calculations, I know the PCM has power and grounds, sensor data input and injector control. I'm not in clear flood mode, the coolant temperature sensor circuit integrity is intact and reading correctly, and the injectors are all working and have the proper resistance. After collecting a few more cold cranking signals and current waveforms, I head inside to put my venti caramel machiato in the microwave for a minute or so, then sit down to analyze the situation. It's looking like all the tests are showing normal results. I'm suspicious now (especially after reading that these coils are a known defect) that even though the primary pattern looked good and the coil had been replaced that I may be dealing with a weak spark. I head back outside to take some readings on the coils with my Ohm meter - 23.6 26K and 123K. I can still get a good spark from the other coils but it won't hit a lick. No backfire, no cranking speed increase like it's even trying to start. Against my instincts I order a new coil.

The next morning the coil gets here and the cold weather is still holding.

It's an easy replacement, so I just change it where it sits so I can get some good "before and after" data (Figures 6 and 7).

I reset the scope trigger and crank the engine and this thing fires up like I just shut it off.

I'm flabbergasted! (See Figure 8.)

I rehooked the scope to the injector current, then waited a minute for the PCM to time out, and cranked again.

The injector current shows some pretty interesting things happening in the first few pulses, which I'm going to write about next time. Before I got through writing this column, I had another one of these 3.8s (this one in a '90 Buick) towed in that was quitting only after it got hot, then restarts after about an hour or so cooldown. This one left me stranded on the road with scope in hand and plenty of time to collect some nice waveforms (Figure 9).

The owner is pretty skeptical when I call to tell him his car is fixed and wants to wait until it sits overnight again to see if it will start in the morning.

He shows up about 8:30 a.m. and has to wait until I get there because I'm running a little behind. The temperature is cooperating so well that I'm crossing my fingers that his car will start.

Rrrear..rrrRooOOMMM, she's hummin'. I still can't believe how fast this thing is starting with the new coil. He smiles and says he hasn't had it start like that in cold weather since he's owned it. Now we go inside for the fun part. Ralph is in sales and he's a good talker.

He asked me how we found out what the problem was and I showed him a bunch of the waveforms I collected. He then asks if he can have copies of them.

"Sure," I say, "You can have these. I have them stored on my PC."

I can't help but smile as he picks up a little stack of my business cards off the desk and asks, "You don't mind if I take a few these do you?"

"That's why we're here," I say.

Jeff Bach is the owner of CRT Auto Electronics, an ASA-member shop in Batavia, Ohio. For more information on this topic, contact Bach at (515) 732-3965. His e-mail address is northstarguy@zoomtown.com

RATE THIS ARTICLE What do you think of this article? Your input will help AutoInc. develop additional articles on this subject. Share your thoughts! Your name Your e-mail address

MOST ACCESSED ARTICLES Fuel Injection Service, Not Just Cleaning The Art of Extraction EGR Systems: Operation and Diagnosis Proactive Target Marketing:_Rethinking Your Business Strategy Engine Performance: HO2S Diagnostics MOST E-MAILED ARTICLES Developing Employee Potential How Critical Thinking Can Help Your Business How to Diagnose the Ford Glow Plug What to Look for When Shopping for the Right Shop Management Software Putting a Price Tag on Complaints

AutoInc. Web Site | ASA Web Site | ASA Testifies Before SBA Regulatory Fairness Board | Today's Trends in Technical Education | VINs: How They Affect Your Shop | The Importance of a Customer Mailing List | ASA Debuts 'Members Only' Portal | Guest Editorial | Tech to Tech | Tech Tips | Shop Profile | Net Worth | Stat Corner | Chairman's Message