patsmx5

iTrader: (16)

Got the new trans. Fluid in the new one arguably looks better than the new stuff in the bottle! I was actually thinking the guy that sold it to me must have poured clean fluid in the converter to make it look that clean. Pulled the pan, it's pristine and same clean fluid. So that is nice!

Old trans had about 400 miles the fluid, and the fluid is so hammered... Thin, and very contaminated. Bad.

I decided I'm going to pull the valve body apart on the old one, see how hard it would be to adjust the line pressure regulator and shim the accumulators. If it's not too bad, I'm going to do that to the new one before installing it. It looks like there's a 100 bolts to get the valve body out.

stefanst

iTrader: (1)

@russian is investigating the option for a custom shift controller right now with rusefi: rusefi.com ? View topic - 2003 Miata Transmission Conroller maybe that would be of interest to you.

patsmx5

iTrader: (16)

Cool. To be honest I built so much for this trans to do the torque management and torque converter clutch, that also contorlling shifting at this point would take like 1-2 days tops to complete. I'm already pulling in VSS, I'd need to add TPS and then some algorithms similar to the chart I already posted here to tell it when to change gears. Truth is the stock one does this fine from my experience, so I'm not worried about it.

SO...........

I pulled the valve body off the old trans, and did some reverse engineering on some of it for the line pressure and accumulators. I ended up messing up that valve body by taking it apart and having pieces fall out of it that I did not expect to come loose. It could be fixed if I knew where they went.

Anyways so after going of over and the FSM on how to build this auto, and some supraforums research on a similar AISIN trans, I decided to just raise the line pressure the easy way*, and shim the accumulators. For the line pressure, there's actually an adjustment on the trans for 3 settings, it was in the middle by default. I put it to high, and then built a custom spacer to essentially create another "setting", so now it's 2 clicks higher than stock.

I have numbers I gotta go through to get an idea of how much this should help, but rough guess is about 16% increase in line pressure. Spec is 141-162, so new range would be 164-189 PSI. Middle of that range is 176, so hopefully I see something close to that on my line pressure test.

patsmx5

iTrader: (16)

Trans is assembled and in the car. It looks like it's going to work, but I'm having a small problem. The newer trans, while it looks identical to the 99 in just about every way, I found out when I put it in gear on jackstands, the spedometer reads way too high. like 3-4 times too high. Because of this, the car takes off in 1st, and then goes into 4th because now light throttle in 1st takes me to like 50mph.... According to the speedo. 2-3K in 4th was about 180, wrapped the spedometer a lot more than I ever thought it would go!

So with a flashlight, mirror, and neck ache I found the output shaft on the 99' trans has a 4 tooth wheel that the VSS picks up. The 05 has 12 teeth.... So VSS is 3x higher than actual.

Not sure if going to swap to to a 05 TCU, or just cut the wheel to be a 4 tooth. I THINK everything else will work if I fix this.

But to cut the wheel I gotta pull the boost pipe/exhaust/PPF/Driveshaft and then the tail housing, and then I'd need a new tailhousing gasket on trans which would prob take a few days to arrive from mazda. And drain the fluid too...

Will think about it and decide what to do tomorrow.

Can't wait to get this fixed and go for a test drive!

codrus

You can probably solve it with an Arduino, a few passive components, and about 20 lines of code. Read the signals from the tranny, count them, and generate new pulses on every third input pulse.

--Ian

patsmx5

iTrader: (16)

It's a VR sensor, any idea on an easy way to simulate that? I can do the arduino and code, not sure an easy way to replicate VR signal.

codrus

How picky is the TCU input about it actually looking like a VR wave form? VR signals are AC with both the frequency and the amplitude increasing with speed, right? Does it look at both of those, or just the frequency? What happens if you feed it a square wave?

--Ian

aidandj

iTrader: (5)

FWIW the gauge cluster which takes a VR input for speed is perfectly happy being fed a square wave.

stefanst

iTrader: (1)

If you need a way to condition your VR signal to meet the requirements for arduino, I believe Aidan did something like that earlier, using the MAX9926. I also have a few boards from Oshpark (for rusefi), not yet assembled, that do the same thing. LMK if you need something. Should be easy enough to figure out, if your TCM accepts a squarewave- just send in a fake squarewave signal while the car is running and see if it picks it up. My speedometer does.

Edit: Second thought: You can halve the frequency of a squarewave with a flip-flop. Two flip-flops might be even easier than one arduino.

x_25

I am thinking something that blocks two and then connects, then blocks the next two may be the way to go? A simole logic counter should do it.

patsmx5

iTrader: (16)

I think I'm just going to tear it apart and either swap the wheel to a 4 tooth if that's possible, or just cut the extra teeth off. I was able to get a gasket locally. Mazda is out of them till March 2017...

patsmx5

iTrader: (16)

Swapped output speed wheels, shifts right on jackstands now. About to test drive.

patsmx5

iTrader: (16)

WORKS!!!!!!!!!!!!!

No slipping at all at 20 PSI boost. Spins when it hits second, and it spins going into 3rd as well!!!!!! Gear changes are a lot faster. I would estimate them at about 0.15 seconds for a gear change, feels as fast or faster than my C63 which has a sweet auto in it. Before the 1-2 took 0.9 seconds (all slipping) and 2-3 took 0.25-0.30. Essentially there's no slip, and I cut my gear change times in half! I can't really tell in a log, as it spins with street tires on all the gear changes so you see the RPM's balloon as it spins.

There was a lot of traffic so I didn't get to beat on it as much as I wanted to, but I made two or three 1-2 pulls, several 2-3 pulls, and one 1-2-3 pull. On the 1-2-3 pull, it spun going into 2nd and just lite the tires loose and spun up to redline and then went into 3rd, where it found traction and RPMs dropped suddenly. This trans has no problem holding 20 PSI right now.

Probably going to drive it at 20 psi for a bit just to make sure everything is happy and normal, then turn up the boost. The way it drives now, I think it would take 30 no problem. Maybe more! And I don't have torque management on right now either, so it's dealing with full power on gear changes and not a hint of slip!

patsmx5

iTrader: (16)

Also many thanks to the suggestions on how to get the trans to work with the 12 tooth wheel via software. It was what I was going to do until I was able to get a gasket locally. Once I had that, I figured it would be best to do a hardware mod or swap vs adding more computers/electronics. If that VSS thing ever broke, the trans would go into limp mode which is 4th gear all the time. Can't do that now.

stefanst

iTrader: (1)

Wooooooohoooo!

patsmx5

iTrader: (16)

I'm so thrilled it's working and holding. I haven't done any auto trans work in a long time, and it was intimidating pulling apart a low mile trans that I was sure was in great shape, and risking messing it up! But it looks like it worked, and it's shifting great. I really hope this things holds up to some boost. I think I'm going to take it to the 1/8th mile tomorrow, we'll see.

Plan for now is to swap on the drag radials for traction to get rid of wheel spin, enable torque management (and program it to reduce torque for 1.5 seconds on each gear change, overkill) but only have it pull timing, no rev limiter. Then disable my torque converter control (let stock TCU do that) and then datalog a 1-2-3 pull. Then by reviewing the log to see where it pulls timing, I can see the delay between commanding a shift and it happening by the time between when it pulled timing and when RPMs began to drop, as well as the shift times for each gear. Once I have that data, I can program a delay and exact amount of torque managment time for each gear. Then it should be close to dialed in, and can fine tune off of logs in the future. Then all the boost and see what it will hold!

x_25

Betting on 32psi.

patsmx5

iTrader: (16)

Did all this. And took it to the 1/8th mile tonight.

With torque management on, it works ok. I think I need to make it more aggressive on pulling power during the gear change, I only had it reduce power some, not a whole lot.

On 25 PSI, it ran a 8.71 @ 91.91 mph, also ran a 8.87 @ 91.76. At 27 PSI it ran 8.92 @ 92.44mph.

60' for these runs were 2.43, 2.33, 2.45!

It is dog-slow out of the hole with a stock converter. Standing on the stall for 3 seconds, it makes 1.5-1.8 PSI boost @ about 2,600. Gutless on the launch. Once it gets into boost it's awesome.

So a converter is at the top of the shortlist, probably a 4,500-5,000 stall. So is a pressure sensor that goes to 250 or 300 PSI so I can datalog line pressure on the trans all the time. I stopped running the car after going to 27 and it not going any faster, looking at the logs it seems the first pass it was perfect on the 1-2 shift, but the 2nd and 3rd pass it slipped a bit on the 1-2 both times. My torque management is programmed to wait a set amount of time, then pull timing for a set amount of time for each gear change.

See picture for detailed info on how it ran at 25 PSI. Some quick stats from the pic:

0-40mph took 4.52s, avg increase in speed of 8.85mph/sec.
40-92mph took 4.35s, avg increase in speed of 11.95mph/sec.

Essentially it should accelerate a lot faster in 1st compared to 2nd or 3rd (from more aggressive gearing of course), but the average of 2nd and 3rd is 35% better acceleration than in first!!! That's because the stock torque converter is killing my launch. In fact 1st gear should show 69% more accel than second, 1st should show 245% more than 3rd. Yet my car pulls harder at 90mph than a 30! LOL

So I need a higher stall converter, and it will make the car a lot lot lot faster.

ramensoop

iTrader: (1)

I just read through most of this thread, awesome work.

what part of Houston are you in? I'm up north in Humble off BW8.

patsmx5

iTrader: (16)

Thanks! I'm in Spring, TX, so not too far away.