Replacement of a Mazda Automatic Transaxle July 30 – August 3, 2010

Soon after the end of his first year of university, my son Scott and I decided it was time to get him his first car. He would be working through the summer and would need a ride to and from work every day, since it would take too many bus transfers to be practical. And he would be able to drive himself the 550 km to university, not to mention visit his girlfriend about 190 km away. Since I already had a Mazda (my rally car), that I knew my way around pretty well, I pushed hard to find him an inexpensive used Mazda. They are all built using a similar architecture and I figured it would easiest to maintain that way. After a couple of false starts (both with Mercurys), we found an $800 1997 four door 626 model – with an automatic transmission. Even though Scott can drive a manual shift if necessary, he was pretty adamant that he would prefer an automatic. 

I wish I had researched the car and its transmission before we finalized the purchase. It has the Ford CD4E automatic, which seems to be the most trouble-prone automatic that Ford ever made. It is fully electronic and has a history of developing shifting problems, partly due to the components and partly due to the effects of excessive heat in the hydraulic fluid. Sure enough, after we fixed a few things to get the car certified for the road, some performance issues began to recur with the transmission. It got so bad that I looked for and found a used transaxle for $480 that we planned to use as a replacement. We scheduled the project a couple of weeks hence, believing that we had enough flexibility to do it at our convenience, some time before he would return to university. We should have done the swap immediately, as it turned out.


The replacement of the transaxle started earlier than planned. We had set aside Saturday to do it, but he called Friday morning while I was cutting the grass – fortunately I had stopped the mower to talk to Kipper. The car had died on his way to work and he couldn’t get it to move. He could start the engine and shift gears, but the transmission was not working. I told him to wait a few minutes for the oil to cool and then try again. He had pulled into a side street and partly into a driveway and I wanted to see if he could get it back on the street. He called me back to report that he’d succeeded in doing that, but the car wouldn’t move again.

I loaded my tools into the truck and hooked up the trailer, borrowing a chain from Kipper and taking all my tie-down straps. When I arrived at the car, it was positioned on a side street facing a busy road (Prince of Wales); so I had to back the trailer into the side street immediately in front of his car. We tried to push it onto the ramps and succeeded in driving it part way up, but it wouldn’t go any farther. I had a bright idea – to back the trailer under the car! We chocked the rear tires and I slowly reversed the truck, scraping the ramps on the road. This worked beautifully until the ramps reached the rear tires, when it wouldn’t go any further. I hooked up the straps and chain and started the tedious process of winching the car forward. Scott suggested trying to drive the car one more time and this time it worked! I drove it directly onto the trailer and we were home free!

We got the car home safely and I moved my Mazda and the Porsche out of the garage so we could back Scott’s Mazda inside and begin to work on it. We started the job around 10:00 AM, by jacking the car up to remove the front wheels. The first of many challenges immediately reared its head – my old cast iron floor jack wouldn’t lift the car high enough. I tried to bleed any air bubbles out of the fluid, without success, so I made the first of several trips down the street to Steve’s to borrow his good aluminum jack. While we were at it, we brought back the engine hoist, which we would need later on. I had intended to get Scott started on different steps in the process – which I had written out and posted on the wall – while I continued to cut the grass. But after I had cut the area between my house and Kipper’s, plus the west side of the back yard, the lawn care company guy showed up and sprayed the entire yard with fertilizer. I would have to wait a day to finish cutting the rest – which turned out to be a good thing. From then on, Scott needed my involvement in a big way.

The first step was to disconnect an oil line from the transmission to determine the direction of oil flow. That would tell me where and how to install the small radiator-style oil cooler and in-line filter I had bought. This test went off without a hitch, but when we tried to find a relay for the fuel pump we struck out. I was hoping to relieve the fuel system pressure in case we had to disconnect any fuel lines, but decided to just mop up any spills that might occur. We then began the serious disassembly process. While Scott removed the battery, air intake snorkel, air box and both oxygen sensors, I removed the crossmember under the rear of the engine, the exhaust down pipe, a bracket connecting the longitudinal engine support to the transaxle and the left splash shield.

Then we began the steps necessary to remove the drive axles. I separated the two tie rod ends after a quick trip to Steve’s to borrow a ball joint separator, since Gary still had mine. The tie rod ends have a castellated nut and cotter pin on the bottom and we could not remove the pin on the left side. Eventually, it broke into small pieces without coming out; so we had to use the air wrench to force the nut off. On examination, it was clear that both tie rod ends were damaged where the cotter pin hole had weakened them, so I called NAPA and ordered replacements. Then we loosened the two lower ball joint and tried to separate them. I used my spring compressor, a bottle jack to raise the hub, a crowbar, pickle fork (another trip to Steve’s), without success. I decided to abort this attempt and remove the wheel hub/steering knuckles from the struts instead.

This went fairly smoothly, but the removal of the main wheel centre nuts was another story. The right side came off fairly easily with a big socket and my breaker bar plus pipe extension, but the left side would not budge – even with the air compressor impact wrench. No matter how hard we tried to use the breaker bar plus heat from my propane torch, it simply wouldn’t move. In the process, I gave myself a cut under my jaw when the wrench slipped and I whacked my jaw on the top of the fender at the edge of the hood. Finally I decided to cut it off with the angle grinder. I cut into the nut on four facets, getting the blade as deep as possible into the rotor’s centre without cutting the rotor. With the cutting, plus the extra heat from the angle grinder, the nut came right off with the breaker bar – success! And the threads on the end of the axle weren’t damaged in the least. It’s a good thing I already had bought two new nuts. Then it took another few minutes with a hammer and various punches to get the axle to move inwards in the hub, eventually coming free. The splines on the axle were very rusty – there hadn’t been grease there for years!

After that effort, I thought it was time to remove the axles – wrong! We hadn’t yet drained the transmission fluid! With only a small puddle, we got the first axle back into place and properly drained the case. Then it was safe to remove the axles. I went under the car to remove the intake manifold support bracket, so I could get to the starter motor. Then we removed the longitudinal support under the transmission. At this point we hooked up the chain and positioned the engine hoist to support the weight of the engine, while putting the floor jack under the transmission. I measured the height of the car to ensure we could get the transmission out and discovered that the car wasn’t high enough. So we jacked it up even more on the left frame rail and moved the jack stand forward to maximize the height. I put the floor jack under the transmission and lowered the engine/transaxle assembly to get better access to the starter motor bolts.

Then we embarked on a long, slow process to remove the three bolts holding the starter in place. The one at the bottom was fairly easy, but the two at the top required us to relocate numerous portions of the wiring harness, just to get a wrench on the bolts. One of them came out easily enough, but the other one became harder to move the more we loosened it. The reason was that the end of the bolt had been exposed to the elements and was quite swollen with rust. In addition, it was evident that some bright lad had applied locktite to the bolt, making it extremely difficult to remove. Eventually it came out, but that process had to consume at least a half hour.

The next step was to remove motor mounts at both the front and rear of the engine. The front one had to come off so we could reach one of the big bolts holding the transaxle to the engine, even though the mount itself wasn’t attached to the car’s body. This is the one that rests on the longitudinal support, which by now had been removed. In order to reach the bottom two (of four) bolts, I had to contort myself under the car and over the legs of the engine hoist. The rear motor mount was far more difficult, because, once again, the two bolts fastening it to the transaxle were buried under the wiring harness and the heater hoses. I opened one the heater hose connections to get access and, of course, we got a big puddle of coolant for our efforts. We succeeded in getting one of the two bolts out of the motor mount, but the rear one posed a bigger problem. There wasn’t room to get a socket wrench on it, so I tried a box wrench with a dogleg at each end. That was the right tool, but we couldn’t get the bolt to move at all. I think the right tool is a shallow socket with a wobble extension and I decided to see whether Steve has such a thing in the morning. We quit for the night around 10:30, after working for 12 hours! I don’t think the original plan of doing the whole job on a Saturday would have worked!

Day Two

Saturday morning I awoke at 6:15, not exactly raring to go but knowing that I’d need the whole day to finish. I remembered that we still hadn’t disconnected the shift cable or speedometer wire, so that would have to come first. I also remembered that I have a shallow 17 mm socket in the small tool kit in the rally car, so I would try that on the motor mount. But first, I would clean up the mess on the floor and organize the tools so we could find everything again. At least we had used several clear plastic containers for all the nuts and bolts, so we should be able to find them. While Scott slept in, I also made a bracket for the new transmission oil cooler and mounted it in front of the air conditioning radiator, which is front of the normal radiator, surrounded by trombone coolers for the transmission and power steering fluid. What a mess of plumbing up there!

It’s a good thing we inadvertently chose a long weekend for this project, because it looked as though we’d need all of it. To remove the stubborn bolt from the rear motor mount, we tried all possible combinations of wobble joints, universal joints, shallow sockets with built-in universal joints, air wrench, etc. In the process I broke my own 3/8” drive universal joint and Steve’s 17 mm socket with built-in universal. Finally I decided that bolt was not going to come out without seriously improving our access to it. Even if I’d been able to unfasten the rear portion of the motor mount from the firewall, the transaxle portion was still too big to allow the transaxle to drop down and out of the car. Soon I decided that I could disassemble the universal joint and create some space between its pivot drum and the stubborn bolt. I was able to get a wrench on the left end of the motor mount pivot bolt from underneath and slowly loosened it. But when I tapped it out with a hammer, it collided with the hard power steering lines near the firewall and wouldn’t pass. After a few false starts I realized that if I raised the engine, the bolt would be able to clear the lines; so that’s what I did. Then we still had to get a wrench on the stubborn bolt to remove the transaxle half of the motor mount. To accomplish this and ultimately succeed in removing the stubborn bolt, I crawled under the car and used a 24” long socket extension to pry the transaxle away from the firewall far enough that Scott could get a socket directly on the bolt.

At this point my old friend Rob had come over to bleed his brakes and check on our progress. He helped us begin to lower and separate the engine and transaxle, so we could see the nuts which fasten the torque converter to the flywheel. It seems that these nuts are visible only through a ½” gap between the flywheel and the engine case, requiring some kind of special 14mm offset wrench to clear the flywheel’s teeth and sit squarely on the four critical nuts. Certainly I don’t have such a wrench and neither does Steve, so we decided to make on by cutting the head off a box wrench and welding it back on the handle parallel to its original position – creating an offset for 4-5 mm. By now it was 7 PM and Scott was going out. So I finished cutting the grass with the tractor and sat down to relax. Not much progress for a full day in Fearless Garage!

Day Three

Sunday morning I watched the Hungarian Grand Prix with my friends and had breakfast at the Swan. Afterwards I went to Canadian Tire to exchange two broken tools and to buy a sacrificial wrench for the flywheel nuts. I stopped at Steve’s on the way home to verify that he’d be able to weld the special tool that I wanted to make; then I went home and began to make it. I cut the head off a 14 mm box wrench about ¼” from the head and ground flat surfaces on the stub handle and on the main handle, clamping them in the desired position with Vise-Grips. Steve welded them together beautifully and I went back to try it out. By then Scott had returned from his night out and we got into position, with me under the car and him on the crankshaft pulley with a wrench, to rotate the flywheel for me. After a few minutes of trying to fit the new tool, I realized that I needed 15 mm, not 14 mm! How frustrating! Everything else on the car is either 10, 12, 14, 17 or 19 mm. But this Ford transaxle uses 15 mm nuts! So the sacrificial wrench has been sacrificed to no avail.

I found that I was able to use a 15 mm offset box wrench successfully, plus an articulated ratcheting box wrench I borrowed from Steve. After about 20 minutes of careful wrenching we had the four flywheel nuts removed. Then we carefully lowered the transaxle to a dolly, using a combination of the engine hoist and floor jack. It took a little while to find a position that allowed me to disconnect the speedometer wires from underneath our favourite motor mount, which is a really dumb place for an electrical connector. We had to raise the left side of the car a couple of inches to get enough clearance to remove the transaxle on the dolly, while supporting the engine with a bottle jack and piece of 4” x 4” wood. This was a major milestone – we were halfway home!

At this point, Scott pleaded his need for a nap, since he’d been up most of the night before. So I sent him to bed, while I finished cutting the grass with the hand mower, to pick up the previous day’s cuttings and trim the edges. I cleaned up the spilled oil and coolant from the floor and spread oil stain removing powder everywhere. Then I covered the wet area with a piece of vapour barrier, so I could lie on the floor without getting any dirtier than necessary. I reinstalled the troublesome motor mount and removed a redundant bracket from the new transaxle. Then I decided to begin the process of installing the replacement transaxle, since most of the work could be done using the hoist and jack. After a lot of manoeuvring, I positioned the transaxle correctly to allow the insertion of the main bolts which attach it to the engine.

Once we finally got all the bolts in position and partly threaded in, I was ready to start installing the flywheel-to-torque converter nuts. But I found two problems: first, I could see the studs on the torque converter, since it had slipped inwards (towards the transmission) and studs had become recessed from the rim of the bell housing, and second, the studs were not aligned with the holes in the flywheel.  So we had to remove the big bolts used to connect the transmission to the engine and move the transmission back so I could get a tool on the studs. I got a pair of pliers on one of the studs and jiggled the torque converter back and forth, while pulling it to get the studs closer to the flywheel. This worked really well – too well in fact. The torque converter came out too far and disconnected from the splined shafts at the transaxle end! No matter how I jiggled and pushed, I could get it centred correctly on the shafts.

This is a great illustration of why everything happens for a reason. Exactly at this moment, Morley came by to see how we were doing. He asked me whether I’d filled the torque converter with fluid before installing it. I hadn’t and perhaps I should have. This particular torque converter does not have an opening around its circumference to draw in fresh fluid – it’s a sealed unit. The only place to pour in fluid is the central splined receiver for the transaxle shafts. So I would have to lower the transaxle and remove the torque converter in order to fill it. Then I would have to insert it into the transaxle while fluid drips out and position it precisely on the splined shafts so it would mate correctly with the flywheel. Sounded like a job for Monday morning, so I sent Scott off on his date and went to bed. But first I posted a question to a Ford help web site, to see if I could get some expert help on how to install the flywheel nuts and to verify that I should fill the torque converter.

Day Four

By Monday morning I hadn’t heard back from a Ford expert, so I had to do a little more internet research on the installation procedure for the transaxle. I found a forum where someone had written that they’d put a quart of oil in their torque converter before installing it, but then I got a reply from Shawn – a twenty-year Ford master technician – who wrote that it wasn’t necessary. He also said there is an access hatch to allow for the installation of the torque converter to flywheel nuts, but didn’t say where it is located. Armed with this profound knowledge, I went to the garage.

I began by removing all of the large bolts that hold the two units together; then I lowered the transaxle almost to the floor so I could remove and partially fill the torque converter. After pouring a few ounces of fluid into the centre shaft opening, I turned it on its side and a small amount of oil came out. So I knew I could not add a full litre without making a real mess. While lying on my back, I lifted the forty pound converter into position and slid it onto the splined shaft. It took quite a bit of jiggling and pushing to get it to seat properly, after which I had no blood flow in either arm! But finally it snicked into place and I began to raise the transaxle into position next to the engine, using both the engine hoist and floor jack. 

Before raising the transaxle, I rotated the converter to position one of the four studs at the bottom, where I would try to match it with a hole in the flywheel. I discovered that I could use a screwdriver to move the flywheel one tooth at a time, instead of relying on my helper (who was still sleeping), to rotate the crankshaft pulley. Once I got the stud and hole close to alignment, I began raising the transaxle.  It took me a long time to find the right angle, clear a few obstructions and reach a point where I could start the main bolts into their holes. But eventually they were all threaded in properly and I tightened them enough to bring the studs on the torque converter into the holes on the flywheel. I used a bent coat hanger to determine whether my stud and hole at the bottom were aligned; then rotated the flywheel using the screwdriver and a convenient notch to pry it into position. Once I felt that it was as close as possible, I tightened all the main bolts so the gap between the engine and transaxle was less than 1/2”. Then I went to the crankshaft pulley and jiggled it a bit, using a socket wrench. I heard a distinct “click” and sure enough, the stud had seated into its hole.

I loosened the main bolts to widen the gap between the engine and transaxle and when I examined the stud I could see that it had barely entered the hole in the flywheel. I needed to find a way to bring the torque converter closer to the flywheel so I could start the nuts on their threads. Just then I noticed a hole in the bottom of the transaxle case, just about where the edge of the torque converter would be. I inserted a long thin screwdriver, made contact with the seam of the torque converter, and pried.  Presto! The torque converter slid easily towards the flywheel and I could see the full length of the stud! I yelled out “Yes!” and celebrated a small victory. It was then a simple, but tedious, matter of installing the four nuts using long needle nose pliers and tightening them with the 15 mm offset wrench. When Scott surfaced a little while later, I got him to hold the crankshaft pulley while I torqued the nuts as hard as possible. I never found Shawn’s access panel, but got the job done anyway.

After that, our assembly process went fairly smoothly, although not without small challenges along the way. We tightened all the big bolts, connected the speedometer wire, installed the starter motor and intake manifold support bracket, raised the engine/transaxle assembly, connected the motor mounts, installed the right drive axle and assembled the right steering knuckle/strut. By this time it was past the supper hour, so we stopped for the night. While Scott began to prepare supper, I disconnected the engine hoist and cleaned up a bit, confident that we would finish the job the next day.

Day Five

And I was right, but it took us until 4 PM to finish. Tuesday began sunny and cool, but quickly warmed up and got pretty humid. By 10 o’clock I was already drenched in sweat and I’d only been working on a wheel! I finished up the right side tie rod end and installed the left drive axle. But first I had to grind the end of the axle to remove the burr I had created while hammering it to remove it! Then I began to install the new lower ball joint on the left side, which I’d picked up from NAPA early that morning. I had connected the tie rod end and put the steering knuckle back in the strut, but found that I couldn’t get the old ball joint off with everything else in place. By now Scott had arrived and we disconnected the steering knuckle to get better access. It took quite a bit of prying and lifting to separate the control arm from the steering knuckle; then we supported the knuckle on a concrete block. I tried to use my ball joint separator (which Gary had returned), but the joint was too big; so I tried to hammer the joint out. Then I realized it had a C-clip holding it in place, so I removed that and the joint came apart pretty easily. From then on it was a straightforward process of installing the new ball joint and reconnecting the knuckle to the strut.

Then we began the process of reassembling the rest of the pieces that had come off first. I had already secured the many tentacles of the wiring harness with their original clips or zip ties, so we reconnected all of the electrical plugs. While Scott reassembled the air intake – four pieces – and the battery box, I was under the car installing the exhaust downpipe, oxygen sensor, front engine support beam, rear crossmember and splash shield. I added as much radiator coolant as I could and poured in 4-1/2 litres of transmission oil, plus a half bottle of Lucas conditioner. After a brief pause to go through mental checklists, I told Scott to start the engine.

It caught the first time and idled smoothly, with no strange sounds or smells. I checked the transmission fluid level and added two more litres. Then I got Scott to select different gears so I could observe the motion at the front hubs – the wheels were still not mounted and the car was still on the jack stands. After cycling through all of the gears, the brake rotors were turning in the right direction, although the first time Scott put it in Drive, he had his foot on the brake and didn’t tell me! I couldn’t figure out what was wrong! We quickly rectified that mental error and things seemed pretty good, except the transmission didn’t want to engage Reverse unless it had been put into Park first. Before installing the wheels, we measured side-to-side the distance between the front and rear of the brake rotors, to see whether we were in the neighbourhood of a decent toe-in setting. There was a 1/16” difference between front and rear and I had used white-out to mark the position of the tie rod ends, so it seemed safe enough to drive, if not good enough to leave that way.  Scott then installed the wheels and we were ready for a test drive.

After driving around the block a couple of times, I checked the transmission fluid level and it seemed good, but the engine temperature was too high. I added more antifreeze and measured the engine temp with my infrared thermometer, watching the temp drop to more acceptable levels. After a couple more laps around the block the temperature was extremely high, so I knew we had to burp the system to get air bubbles out. Pointing out that this is not recommended and not for the faint of heart, I carefully opened the hot radiator cap a few degrees with a rag and heavy glove. We could hear a lot of loud gurgling as the air bubbles escaped, so I knew we were on the right track. After adding more antifreeze, Scott did a few more laps and the temperature was good. I checked for more air bubbles and got nothing, so the cooling system seemed to be fine. I checked the temperature of the new transmission oil cooler and it was about 43C, so that seemed to working properly as well. After cleaning up the garage and himself, Scott was free to go to his girlfriend’s for supper. He thanked me profusely several times and agreed that he had learned a lot of useful things over the course of the five-day project.

Later that night, while I was relaxing with a book, the doorbell rang. It was unusually late for an unexpected visitor, but I went to the door and it was Morley. He had seen a small ¼” drive socket extension fall out of the car while Scott was doing his laps and had stopped by to return it. That gave us both a good laugh and I took the opportunity to thank him for his support and useful suggestions. I guess Scott had left it on top of the radiator moulding while he was assembling the air intake and forgotten to put it away. So the learning continues – but we had a great time, saved some money and learned a few things. It was a great project, but I won’t be doing it again soon! As a post script, we had one leftover bolt, the purpose of which has escaped me completely. We tried and tried to reconstruct the disassembly process, to remember where it had come from, but could not remember or find an empty hole where it should go. I am confident that everything important is well fastened, but it contuse to bother me that we missed something. Maybe someday I’ll have an inspiration and we’ll set it right.

Pictures are available here…

Leave a Reply

You must be logged in to post a comment.