Spare Drippings

It's taken several years, but I finally filled a jar with bacon grease.

Sweet Kiss of Defeat

Pucker up, here comes the sledgehammer.

After all that discussion of ECM repair in the Oldsmobile, the girly reports erratic check engine light flickering.

Well, damn.

Black Swan Day in Boat Land

In yet another post in what is increasingly a chronicle of automotive repairs I've completely in my increasingly long and increasingly repair-focused life, I will discuss the replacement of the radiator in the ChevOldsmoBuiac*, and the annoying computer failures that cropped up at the same exact instant for no readily apparent reason. I try not to anthropomorphize inanimate objects, but I swear, that car sometimes seems to be trying to convince me to spend hours in the cold under it. It's like it's lonely and jealous of the Volvo. Please come turn my nuts, I feel so unloved. 
 
 Part of writing all this down is that the amount of information available on the 1990s GM cars is not as concentrated or as good as the information I have on the Volvo, nor on the BMW I spent a copious amounts of time under in my teenage years. People don't love GM cars the same way, so a-body.net and a few random blogs are about as good as it gets. It's nice to have good information and nice pictures, so this is my contribution.

Anyhow.

This is the subject of today's discussion:

The story starts with the Girly calling and complaining that the check engine light is on, and the car is running "oddly", with no speedometer needle motion, and perhaps with some steam or smoke coming out of the hood. That's bad... most pre OBD-II cars only turn on the check engine light when things go seriously south. They lack the plurality of sensors to guess well, and the designers lacked the business acuity to flip it on every time the odometer crossed an even multiple of the oil-service interval. In the boat, when that yellow light comes on, the car thinks it's in major pain. So, I had her bring the car home ASAP and park it.

When I got home and check it, I found the coolant was drained from the radiator and pooling in the front of the car. The engine trouble codes suggested hot running (though not enough to trip the overheat light, this was a cold day in december, thank goodness!). Autozone had a new radiator in the warehouse, available next day. I ordered it and slept well.

The next evening came, and I started the car to put it in the barn so I could swap the radiator with some light and protection from the elements. The car was not a happy camper.... only a few cylinders were firing, the check engine light was on. Not a happy camper. I feared damage from overheating: burnt head gasket? Burnt valves? Rings? The list went on in my head. However, the codes still read that the water temperature was high. That seemed odd, given that the engine was pretty much stone cold. Maybe the sensor liked being in coolant?

I opted to swap the radiator and see.

Pulling the radiator in that car pretty easy, if quite messy, process. The radiator is held in by a couple of claw shaped clamps at the top, including one that doubles as an engine mount. Over 24 years, the bolts have gotten to be pretty corroded and they didn't want to come free. The fan also provides some support, and you need to undo two bolts hold the fan and the radiator in at the bottom. Then, both the coolant and transmission fluid hoses (the car has an integral transmission fluid cooler! Yeah!) need to be removed, a job that is hard to do without making a mess. After about 30 minutes worth of unscrewing and pulling things, I was covered in green and red die, and had the old radiator out of the car.

Inspecting it, it's was quite obvious that failure was imminent, and that replacement was necessary. The external corrosion, particularly towards the bottom, was quite bad. Like everything else in the olds, rust was the primary component. You can see where it cracked, about halfway down.

A heavy duty radiator is available that is wider, providing more air contact for the cooling coils. I opted for that since it was like $70. It looks a lot better than the old one. Stupid redox chemistry ruining my car parts.

Somewhat sadly, once I had the new radiator installed and the fluids topped off, the Girly assessed the inside of the engine compartment and asked what had changed.

But, really sadly, the car still didn't run right, with the ECM code (Code 1-4, to be exact) still claimed that the engine was overheating. Though, strangely, the temperature warning light was off. I also noticed that the cooling fan was running constantly. The computer was acting a bit confused. I pondered if there was mechanical damage or if there were midgets in the ECM.

It was around this time that something truely bizarre happened. I was fiddling in the car with the engine running, and I got out and closed the door. The cooling fan stopped, the engine decided to use all 4 cylinders, and the check engine light when out. The car made it around the block before the light came back on.

Midgets. Had to be midgets.

The first, easiest thing to do was to change the coolant temperature sensor. It's right on the front of the engine, $20, and in stock at autozone. Perhaps running dry hurt it? Though the ohm meter test showed it to have a plausible resistance, maybe it was out of range. When I got the new sensor out of the box, my heart sank. Anything this simple, just a slug of brass, is unlikely to wear out or suffer much do to a small amount of overheating. Installation of the new part showed no difference.

I checked the wiring harness from where it starts at the alternator, then around the front of the engine, over the driver's side of the block, then around the back and up into the firewall. No amount of wiggling caused the idle attitude to change. I pulled out the glove box and wiggled the pigtail coming out to the computer. That did something: the engine returned to regular idle for a moment. Touching the computer did something. Pushing the backside of the case yielded great correlation. There was a loose connection in the computer.

Pulling the computer is really easy in this car.  You pull the ECM and engine control fuses (fun note: engine will still crank with the ECM fuse out).

 There are two big connectors (black and white) that connect the ECM to the car. I cleaned the contacts with emory cloth, thinking they might be corroded. (They weren't.)

The computer itself. This ECM has a 1995 date code, though the car is a 1990. The ecm was replaced several years ago after the old one gave up on reporting speed to the instrument cluster.

 
The computer has lots of different mounting lugs. I read that it's designed to work with many different GM models.

The connectors are pretty clean. Also, keyed to keep the connectors from getting switched.

ROM access panel. If you ever get a rebuilt ECM, make sure it has a ROM that works for your car, or that you put the ROM for your car in. A TBI Iron Duke isn't a 3.3 V6, after all.

 Protective covering removed. This picture is depressing in all kinds of ways. First, it's hard to imagine how any part of the board could be shorting against the case. The entire thing is suspended with a generous air gap, and it looks to be ringed with a ground plane. Also, every chip has been rebraned with a custom Delco part number. That makes backing out the functional units of the circuit tough, even if one were to diagram it.

 

This blurry photo only barely shows it, but the board was 'remanfuactured' by re flowing. The protective layer of the PCB is bubbled and many of the surface mount components have moved a bit. That's worrying. Maybe one of those joints was flakey? I tried isolating the board from the case by wrapping it in electrical tape. Sadly, while it seemed to work for a bit, soon, the car was back to running poorly.

 I pulled the ECM again, and carefully reviewed all the traces and solder joints. In the lower, right hand corner of the component side of the board, there was a large-ish device (perhaps a transistor or a regulator?) that had quite a bit of solder on it, was well as mismatched layers of lacquer. Maybe it had been moved or replaced during the re manufacturing process? On the bottom of the board, there was a bunch of flux and air holes in the solder joint, and the surface mount transistor just above it looked to have been vertically offset by the reflow. Using the soldering iron, I filled in the joints on the large discrete, and re-puddled the solder on the surface mount device.

Fingers crossed, I put it back together and fired it up. That obviously, without question, solved the problem. Without undue abuse, the board refused to fail.

In retrospect, I wish I had known to try that back when the speedometer was on the fritz. For several years, I had to use dead reckoning to obey the speed limits, and a costly series of part swaps to narrow it down to the computer (upshot: I have a spare vehicle speed sensor, and the instrument cluster has been rebuilt.)

There appears to be no correlation between the two failures. It just somehow happened that way, or through a mixture of forces I can't divine.

In conclusion, the symptoms of for an early 1990's GM a-body car. If the engine has a rough idle (rich mixture), on check engine light, off coolant light, on cooling fan. ECM Code 1-4 (overheat detected), working coolant sensor (~2k ohm resistance), and possible a nonworking speedometer, it might well be the ECM.


EDIT: After a few days of joy, the ECM again appears to be on the fritz.




*(1990 Oldsmobile Cutlass Ciera, for the uninitiated)

Panorama Purge

I must find some creative new name for these posts.

Back when I was still using the G4, stitching photographic panoramas was a long task, and I generally waited until I had a bunch to do, then rammed them through on a rainy, lazy sunday. As time has gone on, Lazy sundays are quite few, and that's somewhat contributed to the absence of panoramas in my work. I increasingly have a low patience for computer work that requires me to sit and wait on the machine to accomplish something. The problem either needs to be super fast so I can push the process forward (like writing), or super long so I can do something else while the machine grinds. Getting the images aligned required a nudge-wait-save loop that was about 10 seconds long on the G4. It was miserable.

However, that computer is retired, and the new computer is capable of panorama creation in a near real-time kind of way.

I found this out when I took the panorama of the back yard, and then used Hugin to stitch it. Effortless. Actually, way better than effortless. The more photos in a panorama, the more processing time and resources it takes to arrange the final product. The G4, with 1.5 gigs of ram, would start swapping after about 10-15 photos, becoming too slow to make meaningful progress. With 6 gigs of rams, the newer machine doesn't show the same kind of resistance. I'll have to take more complex panoramas.



The newfound freedom inspired me to clear out the backlog.

I had a set of photos from Hidden Beach in Northern California in 2010. The panorama really doesn't do the drift wood justice, though it does capture the remoteness. I wish there were beaches like this on the east coast.

I also had some stitched photos I took inside of some government buildings I've visited in my travels.

Suburban Marsh

Our house has many cracks in it that appear to grow and shrink over the course of the year.

I believe this to be correlated to the soil saturation. The county soil assessment lists our soil as highly impermeable, with a wildly varying water table of the course of a year.

When we moved in it was impossible to cross the backyard without finding oneself ankle deep in water at some point. The grass was just an illusion of solidity. Mowing, as a matter fact, was really something of a wading exercise until June.

Rain was predicted starting at about 0400. By the time I let the dog out this morning, the marsh had returned.

The dog wants nothing to do with it.

Things such as this intrigue me. We live in an area that 50 years ago was a remote agricultural community, removed from the city by several degrees, but are really now just a pseudo-self governing suburb of that city. The place is filled with city people who generally expect a safe, clean, padded existence, free from the arbitrary forces of nature. but there, right by my barn, is reminder of the true nature of reality.

Little Italy

There was an adjunct living the the graduate dorms. This was before the mechanics of academia were clear in my mind, and the adjuct=poor identity was obvious to me. It was odd as a student having one of the teachers living in the same institutional conditions with me. We had a few conversations over time. My boots, caked with Pariah river mud, sat on the floor of my room for many months. She called them the Van Gogh boots.

Several years later, I was walking to work, and found her sitting on a bench in a storm, getting ready for the day.

In retrospect, I'm also pretty sure that was the last time I saw her.

Free Treats

At first I thought a giant demon had gotten drunk and then missed when it attempted to puke into the trash barrel, but closer inspection revealed someone had dumped a few pounds of cat food, perhaps on the way to the trash barrel.

The dog wanted to clean it up.

The answer was "No."

Yerba Buena Gardens at Night

I have been to Moscone for a great many events now, and paid no head to the signs that pointed to Yerba Buena Gardens. I sort of figured it was just another portion of the Yerba Buena Mall, which seems to spread across several blocks.

The gardens are actually a very nice architectural and biological work, and they were quite empty compared to the surrounding holiday hubub.

I am adding them to my visit-again list.

Mechanical Scrubbing Actions

I make a sauce for mac'n cheese that binds the noodles together and gives it a thick, smooth texture. The sauce is made from 2 cups milk, 1/4 cup flour, butter, salt, and pepper.

This mixture burns easily and also has a tendency to boil over quickly. The last time I cooked the dish, several weeks ago, I burnt the pyrex pot I was cooking it in quite badly.

I let the pot sit in the sink, scrubbing it periodically for several weeks. I assumed that since it was pyrex -glass-, sooner or later the carbon-slag would lift free and the pot would be easy to scrub clean. After two weeks, it became obvious that no sane amount of brillowpad and soaking would work. I wanted the sink empty. I dried the pot and it's carbon padded interior, then produced the dremel with a stainless steel brush. I ran it up to half speed and carefully applied it to the burn, which vanished in a puff, leaving clear, unmarked glass behind. I pushed the brush in until it deformed slightly, and cleaned the pot in rhythmic 1/8" strokes.

The pot is once again clean.

That dremel is like a miracle with an on/off switch.

Blow Hole

One of the first shocks of homeownership was the first gas bill. We left the house unoccupied while we worked on it for the first few weeks, so I was quite suprised when a 300$ bill showed up for heating an unoccupied building.

It was obvious that the many air gaps in the structure were doing nothing for it's energy efficiency.

Since then, I have been on a hunt for airleaks in the building. In a house that is designed to be windows-up in the summer, that eliminates the summer months as a rational point to find small leaks; the house breathes through the open windows and there is little movement through other routes.

Now that winter has returned, the problems of last spring have again become obvious.

The current home repair goal is to eliminate drafts and heat loss for general comfort and to further buffer the upstairs temperature, which only seems to have a Goldilocks temperature for 4 weeks out of the year. In the past few weeks I have been weather stripping the doors and windows, insulating the steam pipes (subject of an upcoming post), and draft sealing the attic.

The foundation has a small portico structure that extends under, and to a degree, supports the front porch. The porch is decked with tongue-and-groove boards, so I thought it might be air tight, especially after the brickwork in the foundation was fixed. However, while insulating the steam pipes, a strong draft was still coming in through the portico.

I solved that little problem.

I had originally designed a more elaborate, removable plug that used large pieces of polyethylene sheeting as the air seal. However, when I started cutting the board and really examining the shape of the hole and the discomfort of crouching in the crawl space, I resolved to act on the more direct cut/tape/foam permanent installation. If there is ever need to access the interior of the portico, the foam is easily enough removed, and this approach is completely air tight.

While installing the plug, the velocity of the draft through the gaps in the side increased as I plugged them. The flux of air was actually substantial, like the exhaust of the shop vac or a wimpy leaf blower, probably driven by convection of warm air out of holes in the upstairs. A few hours later, The Girly commented that it seemed a lot more comfortable. That can be taken as progress.

ChevOldsmoBuiac: Front Half-Shafts, Wheel Bearings, and Halfshaft Seals

Some photos and notes for anyone looking to replace the halfshafts and transmission seals in their A-body Oldsmobile Cutlass Ciera.

Some bygone aphorism stated that every journey starts with a step. In the case of half shaft replacement on a 24 year old car, that step requires a 36" long wrench and a 24" cheater bar. The axle nut was originally put on with Loctite and 190 ft-lbs. Then it had all of time and every weaving frost-heaved winter road in New Hampshire to rust over.

Marco Polo, eat your heart out.

The main inspiration for replacing the seals and and halfshafts is that they were all leaking. The shafts had been leaking for years, and though they still seemed good and tight, were in the way of replacing the seals that were leaking transmission fluid more quickly than I was comfortable with. I replaced the bearings too, since they are easiest to remove while the shafts are out, and the were getting a little chatting in hard curves.

The wheel bearings are simple to remove on the Cutlass Ciera: They're held in by 3 star-drive bolts to the steering knuckle. No hydraulic press needed. The half-shafts came out of the bearings really easy. A half-hearted rap with the deadblow mallet broke them free.

Leaky tie rod ends: a project for another day.

The knuckle needs to be separated from the strut in order to pull the halfshaft out. I painted guidelines on the assembly so I could reinstall without throwing the alignment. The bolts holding the strut to the knuckle were fairly corroded, and a huge amount of work was required to break them free and unthread them.

The halfshaft is held onto a splined haft inside the transmission by a spring circlip. It takes a bit of force to pop it of. I ended up jacking the car up another few inches so I could put the breaker bar between the frame and the shaft housing, then hit it hard with the deadblow hammer. The housing looks like plastic, but is really cast aluminium, and it suffered no ill effect from the whack.

The exposed shaft in the transmission.

Getting the axle seal out of the transmission was not an easy job. Haynes called for a hammer and a cold chisel, but there is not enough room to get a good angle on the seal, or a good swing with the hammer, so the process of deforming the old seal required patience, several flatblade screwdrivers, and a needle nose vice grip. To get the seal out, I pulled little bits off until I could get between the seal and the transmission to really deform the seal casing. It took over an hour to finally dig it out.

Transmission without any seals. Drivers side, I believe.

Getting the new seal in was as hard as getting the old seal out. Finding a round object suitable to evenly push the seal in was not easy. I tried for another hour or two to use various scraps of PVC and the socket for the halfshaft nut, but the seal refused to go in flush. I eventually went to the hardware store and purchased a 2 foot piece of pipe that fit over the seal exactly. I also removed the strut to give myself a straight shot at the seal.

In the end, 10$ at the hardware store did in 10 minutes what I had spent several hours failing to accomplish.

Say hello to my little friend.

New seal, installed.

New and old halfshafts, compared. The new shaft slide right in and clicked into place without a lot of force.

Wheel bearing removed from knuckle. There is a dust seal on the back of the knuckle that matche a seal on the halfshaft. However, there is also a seal on the bearing itself. I still stuffed this with grease to keep anything from rusting together.

New wheel bearing.

Passenger side leaks, for posterity.

The passenger side halfshaft has a splined in the plugs into the transmission. I suppose it's impossible to put it together backwards, even if the different shaft lengths didn't give it away.

I pulled the bearing off the passenger side before removing the halfshaft to give myself more room to work. It only helps a bit, still a tight squeeze.

Installing the new shaft revealed that it came with a ABS sense ring. This oldsmobile doesn't have ABS, and, strangely, had a large iron ridge in the steering knuckle that almost seemed to exist to keep the ABS shaft from being installed correctly. The method of attachement for the abs ring wasn't readily apparent (glue? pressed? spotweld?) so I opted to grind of the ridge on the knuckle rather than try to remove the ring on the halfshaft.

After grinding, I had to clean up all the shavings to keep them from getting into the assembly. In retrospect, it took so much time to grind the knuckle that putting a notch in the ABS ring to see if it would easily come off would take a lot less time.

General notes:
-Every fastener seemed to be rusted together. This job could probably be done in less than 8 hours if nuts and bolts would spin freely.
-Paint guidelines on the interface between the struts and the knuckles to preserve the alignment.
-Remove the struts. It's just 3 nuts and it makes a lot more space in the wheel well.
-Jack up both sides of the car so it's easy to push/pull the rack without having to twist the steering wheel. Getting the end of the halfshaft out of the wheelbearing required a lot of experimentation to get the knuckle into just the right spot.
-Come prepared with lots of grease and threadloosening fluid
-Be patient and have lots of pointy tools for removing the axle seals in the transmission.

Blinkenlichten II

Once taken, twice removed, I suppose.


By which, I mean to say that it was fairly certain that fixing the broken christmas light timer was possible if all the electrical components were intact. We bought two last winter, and thereby instantly removed the bedtime "Who's going to unplug the lights?" discussion and also the equally awkward "Wow look at the december electric bill!" associated with running the lights 24/7 during visits with family. On at 4:50, off at 11:00 every night.

Then, sometime this summer, one of the timers stopped working. Screen blank. No clicking relay. Zilch. Reset button did nothing. Plug/unplug did nothing. Appeared dead. But of course, I had already opened it and determined that it was almost too bloody simple to die. As long as the ballast capacitor stays in spec, it should work about forever.

 I pried the timer open to see this again:

For reference, the left side PCB with all the components is an AC/DC converter, and an SSR driven by the transistor, which is in turn driven from the blue wire that leads to the right hand PCB, which is the timer microcontroller, the screen, and buttons (not shown). The major components on the timer board are few. On the back, there's a oscillator, a photoresitor ("Dusk"), and a battery (the green shrinkwrap thing). The timer charges from a 1.45 volt line through the red and yellow wires. I think the AC/DC conversion is actually 12 volts (for the relay), and then a seprate resistor drops it further to supply the controller. Though, that leaves the purpose of the big red capacitor unanswered. Regardless, I wasn't there to diagram circuits.

The controller battery was a) charging and b) charged. That narrowed the problem down to the controller and the screen. I removed the control board and noticed that one of the fingers that holds the screen to the timer PCB was loose. LCDs often use a silicon surface contact. I removed the screen and found the controller CPU underneath, along with test, reset, ISP and ground pads. I grounded the reset pad for 5 seconds cleaned the LCD contacts,and pressed it back together until it snapped. The screen started working, blinking time like a VCR.

Rock on green box, rock on.