802SHO 2010 Build

Disclaimer: Links on this page pointing to Amazon, eBay and other sites may include affiliate code. If you click them and make a purchase, we may earn a small commission.

802SHO

Boost > VE, MBT, Cams
Supporting Member
Joined
Oct 8, 2019
Messages
3,539
Reaction score
8,361
Location
Essex Junction Vermont
I have no clue how to ask this maybe im too much of a beginner i saw your post about mods and stuff i know for a fact im watching every post your made about your mods so im looking to follow your lead so my question is what is the epp your talking about im on there website and there is 2 differents thing i can buy and since im a beginner on the sho platform i have no clue what is the good one ( and yes sorry for my english grammar my main language is french)View attachment 98436View attachment 98437
It's been a while. I bought hot pipes and this intake 7 years ago I think. The rear turbo inlet silicone coupler is an itemized upgrade instead of included? I think I used this and it was included that long ago. So you can reuse the OEM inlet I guess or buy this at an additional cost. I'd buy it so you have a complete kit. If you have additional questions like this please ask me by sending me a direct message instead of here where I'm talking about my build and tuning progress. Look for the mail icon and start a conversation
 

802SHO

Boost > VE, MBT, Cams
Supporting Member
Joined
Oct 8, 2019
Messages
3,539
Reaction score
8,361
Location
Essex Junction Vermont
Rev8 is coming together. More response, a cleaner handoff, and faster recovery.

Rev7 finally gave me the confidence to start refining how this combination delivers its power. It pulled hard, the converter transition felt better, and light-load driving actually improved further after the pull. That makes adaptive learning part of the picture, so I’m giving the transmission time to settle before adding more converter application pressure or rate.

The biggest drivability change for Rev8 is the Driver Demand/OSS blending. Fifth and sixth still felt too lazy with light pedal, so the revised OSS map brings up those requests and blends them more smoothly into the stronger pedal regions. The full-pedal OSS multiplier stays at 1.00, and the upper Driver Demand cells stay at 500 lb-ft for this comparison. The improvement I’m after is a car that responds progressively and feels ready when I move my foot.

We also added focused user-defined controls through custom XDFs. Eight engine-speed protection parameters expose the base limit, shift offset, timing, ramp-down, hysteresis, and torque recovery. Another fifteen wastegate parameters expose feedback gains, filtering, dynamic feedforward, anti-windup, and air-delivery compensation. Having them available gives us more visibility into the strategy; we’re making targeted changes within that larger set.

The engine-speed settings carried forward are 6,300 rpm plus a 300-rpm shift offset, giving a nominal 6,600-rpm window while that offset is active. We still need the transmission to complete its handoff cleanly within that window.

The small engine changes and the targeted response tests are:

ControlRev8 change or planned test
Inverse-load clip, 5,000–6,500 rpm2.15 → 2.20
Borderline spark, 1.80-load row at 5,500/6,500 rpm9.5° → 10.5°
Wastegate TIP-error filter time constantPlanned 0.120 → 0.100 second
3–4 offgoing starting pressure, selected upper cellsPlanned 82 → 86 psi and 92 → 97 psi
Boost target remains unchanged, with Rev7 generally producing 20–21 psi. The upper-load preignition-limit cells remain at 16°. The next log will show whether the extra load allowance and borderline timing actually produce more delivered torque.

The WGDC-response edit reduces filtering of the pressure-error signal. We’re testing whether that helps the controller respond sooner after the shift. Rev7 took approximately 0.4 second after fourth settled to recover above 20 psi; the goal remains 0.2 second or better, with controlled boost and no oscillation. It didn’t fall below 17.5 psi boost though during the handoff

For the 3–4 flare, the proposed pressure adjustment is confined to the 300/350 lb-ft rows at 3,500–6,000 turbine rpm. It’s a small test of the offgoing clutch’s support during the handoff. I’ll be watching for less turbine-speed flare without introducing a bind or harsh transition.

The current TCC upshift apply delays are 0.10 second in third through sixth. Those stay, along with the OEM zeroed gear multipliers. Since the car continued improving afterward, I want another look at desired versus actual converter slip before increasing application rate or capacity.

The next logs need to show:

-Stronger, smoother light-pedal response, especially in fifth and sixth.
-Less 3–4 flare, with both the full shift sequence and active handoff measured.
-Quicker converter-slip and boost recovery.
-Lambda following the 0.80–0.82 commands, clean knock behavior, and rail pressure keeping up with demand.

Once the behavior is consistent, Dragy will tell us where this setup stands.
 

bpd1151

Lurking Around
Joined
Feb 21, 2010
Messages
3,142
Reaction score
6,696
Location
SHOcago 'Burbs
Flex fuel controls added!

The old 10-12 ECU’s hiding plentiful capabilities[B/]


The irony of this discovery (& all the others to date) makes me SMH at all those who've claimed they weren't there, or weren't accessible etc.

Particularly a vendor I no longer affiliate with, who claimed (repeatedly so) that they were God's gift to the platform and how they and only they, were getting access to "things" through FMC. Gag. Liars.

Pretty cool to witness your gains there Andrew. Across the entire board.

76ead55ec757a534c783616bb6382301

Sent from my SM-S928U using Tapatalk
 

802SHO

Boost > VE, MBT, Cams
Supporting Member
Joined
Oct 8, 2019
Messages
3,539
Reaction score
8,361
Location
Essex Junction Vermont
WE ARE BACK IN THE DARK HIDDEN CORNERS OF THE OS EXPLORING
Major increases in driver demand torque led to a very confused ACC apply rate. After Rev7, I made some changes for Rev8 that included more torque at low pedal positions in Driver Demand. Basically, I wanted the car to give me more response without having to push the pedal as far.

The response improved, but the torque converter clutch started taking longer to couple. We had been trying to shorten that delay, and I was feeling the opposite. I drove around for about an hour and never went wide open. When I tried pushing through the delay, the engine would rev, but boost struggled to build and the car felt like it was riding the converter. Once it finally coupled, the power came in. Rev8 ended up being a drivability log because I wasn’t happy enough with that behavior to do a full pull.

I still wanted more torque down low, so I made Rev8 1.2. Again, the converter seemed to want to stay open longer. Again, I held off on going wide open. The low pedal regions improved exactly as intended and it feels smoothly responsive, not jolty or harsh, just clean and enough power to make the car feel light and willing to fly but restrained to my foot. By Rev9, the pedal response was much closer to what I wanted, but that disconnected converter feeling was still there.

Still I eventually tried a brake-boosted launch after about 45 min of learning. I was surprised to reach over 3,000 rpm and at about 3,200 rpm thinking that was a potent first try (car felt like it had more rpm before wanting to walk the brakes) I let off the brakes and the car lurched forward, tires broke traction and the car pulled power hard.

That launch log showed the computer stepping in through a torque-control source labeled “TCase / Low Range / Driveshaft.” With my foot still in it, the requested torque dropped, the throttle closed down, and boost fell sharply. That points to a deliberate drivetrain-related torque reduction. The label alone doesn’t tell us exactly which limit caused it and I figured more hidden controls to find.
The important discovery in the normal driving logs was that the converter was being told to stay open. The requested slip went up and the reported clutch pressure dropped. When the computer finally requested low slip again, the converter generally followed it. Some of those open periods lasted seven to nine seconds.

That is why I kept telling Chuck that it felt like a confused converter. The extra low-pedal torque was making the car feel better in one way while making the converter behavior worse. Our working theory is that the stronger torque request is running into converter rules that were set up around a different response. More demand can tell the computer to release the clutch. While it’s open, the engine can also gain RPM faster than the transmission input, which can affect whether the computer allows the clutch to apply again. Other settings can then keep it waiting.

The computer may be following its rules exactly while producing a combination we don’t want. We still have to identify which of those rules is responsible here. That also explains why changing the visible apply-delay timer didn’t necessarily help. That timer is only one part of the decision. Making it shorter doesn’t cancel another setting that is still telling the converter to stay open.

This is why I asked Chuck to look through the A2L instead of continuing to adjust the same few controls in HP Tuners. For context an A2L is a definition file that describes settings inside the computer’s software. It’s Ford’s proprietary data and I was fortunate to get the file. It can identify controls that aren’t shown on the tuning screens we’ve been using and Chuck and build XDF files for HPTuners to expose them.

We found controls dealing with how much torque request causes an unlock, what happens when the pedal is pressed, how much slip is requested during that transition, how long it waits before trying to couple again, and what conditions have to be met first. It felt like it would delay on the 2-3 and hold delay, unlock and then delay again and then unlock again. Not a mechanical fault, it’s commanded to open its following desired.

Chuck built an XDF with 59 definitions so we can bring those settings into view. That doesn’t mean 59 settings need changing. It gives us a much better picture of what the converter is being instructed

Once those settings display their actual values, we can review the related controls together and decide which ones need changing to work with the new Driver Demand. The definitions identify the controls; they don’t supply the current values or automatically tell us the right replacements.

There was also a later fluid finding. I had to check the trans fluid level and found this: IMG 2422IMG 2425IMG 2424I had accidentally overfilled the transmission and found creamy-looking foam near the dipstick opening. It looked like paste but didn’t feel like paste. I had already done a drain and fill and made further level corrections, but an additional drain and fill mandatory now. That needs to be sorted out too. Aerated fluid can affect how the transmission responds, so we need the level correct before judging the changes. The foam doesn’t erase what the logs showed about the computer intentionally releasing the converter. That’s the important part…it was following orders it wasn’t bc it was faulty.

The BG fluid was in a commercial bag and I wasn’t prepared to have anything to dump it into except an empty 4 quart container I cleaned out. After the initial test of the used trans I did another drain and fill, I ended up taking fluid out bc I suspected I overfilled it…but after removing a little over a quart I rechecked and the dipstick was dry. Running and after cycling gears I added about 1/2 quart or so and then it was half full on the dipstick but only 120 degrees. Honestly I figured it would be overfilled to the intentional amount we’ve been doing as a community. But obviously there’s too much.

The controls we’re exposing cover what tells the converter clutch to release when I ask for more torque, how much slip it requests when I press the pedal, how long it waits before reapplying, and how it brings the clutch back in. They also reveal speed conditions that can keep it from coupling. We’ll evaluate those settings together so I can keep the stronger Driver Demand and get predictable TCC apply again. On the AWD side, we’re exposing torque limits at different road speeds, driveshaft-speed protection, and the backup torque limit used if AWD communication is lost. Those will help us investigate the launch power cut and decide whether a related setting needs correcting. We still need to read the actual values before choosing the changes.

We’ll get it.
 

802SHO

Boost > VE, MBT, Cams
Supporting Member
Joined
Oct 8, 2019
Messages
3,539
Reaction score
8,361
Location
Essex Junction Vermont
As the trans fluid drains Rev10 is ready.
For Rev10, I’m keeping the Rev9 Driver Demand and OSS changes because they gave me the pedal response I wanted. The converter calibration is being adjusted to work more coherently with those stronger low-pedal torque requests.

The important finding in the logs was that the prolonged slip was being requested. During the weak-response periods, desired slip was around 1,020 rpm and reported TCC pressure was near zero. Once the controller requested approximately 20 rpm slip, the converter generally followed that lower target much better.

Our working theory is that the increased torque requests moved the car into conditions where existing converter scheduling rules kept it open longer than intended. More pedal could increase engine speed ahead of turbine speed, lower the converter speed ratio, and potentially keep an unlock restriction active. Stronger acceleration could also shorten the predicted time before the next upshift, causing another condition to withhold lockup.

That would explain why shortening the ordinary postshift delay didn’t necessarily improve the actual time to couple. Other conditions still had to allow application.

The Rev10 changes are:
-Speed-lock minimum time: 10 → 2 seconds. This is the required predicted time remaining before the next upshift. It allows locking with a shorter opportunity between shifts while retaining some margin against applying immediately before another shift. It does not command a two-second delay or guarantee coupling within two seconds.
-Selected torque-dependent speed-ratio thresholds lowered. The 75, 100 and 150 lb-ft points go to 0.600; 200 goes to 0.620; 250 to 0.670; and 290 to 0.700. The 50 lb-ft point remains 0.600, and the 332 lb-ft point remains 0.720. This relaxes the restriction mainly at lower torque while tapering back to the existing higher-torque threshold.
-Speed-ratio hysteresis retained at 0.050. That separation between switching conditions helps discourage repeated lock/unlock cycling near the boundary.
-Additional relock holdoff: 0.75 → 0.25 second. Once the relevant unconditional-unlock restriction clears, this removes half a second of additional waiting.
-After-upshift lockup-rate outputs increased once by 25%. The original outputs of 25, 50, 55 and 60 become 31.25, 62.50, 68.75 and 75.00, with the axes preserved. This speeds the applicable clutch-capacity ramp after application is requested.
-Alternate torque-based converter scheduling remains disabled. Its involvement in the earlier delays is unproven, but we’re keeping that alternate path out of this candidate configuration.

What we exposed and why:

Using the Ford A2L definitions, we added access to converter scheduling controls that weren’t readily visible in the existing editor screens. Some definitions also cross-check controls already exposed by HP Tuners.

The relevant additions include the torque-reserve requirement for lockup, related modifiers, scheduling-mode enables, minimum pedal input for performance lock, and the predicted-time requirements before an upshift. These show that converter application depends on several permissions beyond the familiar apply-speed and postshift-delay tables.

The speed-lock time requirement was particularly useful: its original value was 10 seconds. We also confirmed that the converter’s configured WOT pedal reference is 585 counts. Those counts should not be read directly as pedal percent.

The proposed performance-lock minimum pedal setting is 150, down from the original 450. However, its per-gear performance-mode enables remain off, so that threshold change does not currently contribute to the coupling correction. The separate performance-mode time requirement stays at one second.

The torque-reserve curve is already exposed, but its input-axis meaning remains unidentified in the A2L. We’re leaving it unchanged. We also located live reserve results and a lock-permission flag, but those are RAM measurements; an Editor XDF does not automatically make them available in Scanner.

The engine torque model and AWD controls remain unchanged. The later overfill/foam finding is also being addressed with an additional drain/fill and correct fluid level.

The goal is to retain the stronger pedal response, allow the converter to begin coupling earlier when appropriate, and complete the application smoothly. This is a reasoned calibration hypothesis based on the logs and definitions; the exact scheduling condition responsible for the earlier delays has not been conclusively identified but that’s part of it. You have to solve for X……without knowing all the answers.

It’s really cool and really F annoying at the same time tuning this car. It feels like the engine is constantly vetoed by the TCM and it leads to detuning the engine in real time. The setup can’t make its power unless the TCM agrees. My assumption was a setup makes its power period …if it’s lacking you add x, y and z. Here I’m seeing repeated handshakes and permissions required to simply make its power. Very annoying but it’s also impressive that it’s this intricate. Over complicated? I’d say so. When you’re typing in the torque it’s allowed to have…that’s just f weird.
 

802SHO

Boost > VE, MBT, Cams
Supporting Member
Joined
Oct 8, 2019
Messages
3,539
Reaction score
8,361
Location
Essex Junction Vermont
Major find in the Rev9 brake-boosted launch investigation: we exposed an AWD final-drive/axle torque ceiling set to 4,645 lb-ft.
BEE33711 3E6F 45BC B286 D60AAA3BE84D

That number is referenced to torque downstream of the transmission. It doesn’t mean the engine makes 4,645 lb-ft. First gear and the final drive multiply engine torque while reducing rotational speed.

Immediately before the power cut, the ECU reported approximately 454.77 lb-ft of engine brake torque. Using 4.48 first gear and the expected 2.77 final drive:

454.77 × 4.48 × 2.77 = approximately 5,644 lb-ft of axle torque.

That simplified calculation is approximately 999 lb-ft—or 21.5%—above the displayed ceiling, before accounting for converter torque multiplication and drivetrain losses. The roughly 470 lb-ft engine request corresponds to approximately 5,833 lb-ft on the same basis.

Here’s what makes the connection significant: the log then switched to “TCase / Low Range / Driveshaft.” Pedal input stayed at approximately 65.5%, but requested engine torque dropped from roughly 470 to 254 lb-ft. Throttle angle fell from approximately 78 to 22 degrees, and boost dropped from about 15.4 to 2.3 psi.

We have a documented driveline torque intervention and a specific axle ceiling that the gearing calculation exceeds. That makes this ceiling a strong working explanation for the launch cut. The remaining uncertainty is whether this particular table or a separate AWD torque request supplied the controlling limit.

The planned Rev10 change is 4,645 → 8,000 lb-ft in all ten axle-limit output cells, preserving the axes and existing enable logic. That retains a finite ceiling with substantially more headroom. It doesn’t disable every AWD restriction or establish a new hardware capacity.

This gives us a concrete calibration target alongside the converter-scheduling changes already planned for Rev10.
 

802SHO

Boost > VE, MBT, Cams
Supporting Member
Joined
Oct 8, 2019
Messages
3,539
Reaction score
8,361
Location
Essex Junction Vermont
Rev10 update (more edits bc it’s still raining) more time to think so it’s welcome.

This revision brings the pedal request, gear selection, PNH converter model and fuel-pressure ramp together.

I liked the stronger pedal response in Rev9, but the increased demand coincided with less consistent converter coupling. That doesn’t establish Driver Demand as the sole cause, but it gives us a useful reason to return to the Rev7 request and get more of the response I want through gearing.

Rev10 restores Rev7 Driver Demand and its output-shaft-speed demand modifiers together. The other selected Rev10 changes stay in the package.

Under acceleration, that should feel less like an abrupt increase in requested torque and more like a controlled, progressive build. The pedal should still produce a strong response, but the transmission will provide more of the acceleration by keeping the engine in a useful gear rather than relying entirely on a larger torque request.

For the normal shift schedule, the new minimum speeds are:


ShiftMinimum scheduled speed
4→545 mph
5→442 mph
5→665 mph
6→560 mph

Only entries below those floors are raised. Existing higher-pedal thresholds and the separate WOT shift-RPM settings remain.

The intended acceleration sequence should look something like this:

-In fourth gear, the transmission should remain engaged longer instead of moving into fifth too early.
-As speed rises through the midrange, fourth should continue carrying the car until the higher shift requirement is satisfied.
-Once fifth is selected, it should remain available through a broader acceleration range instead of immediately stepping into sixth.
-Sixth should not be selected below approximately 65 mph under the normal schedule.
-If speed falls back below approximately 60 mph, the transmission should be more willing to return from sixth to fifth.
-If speed falls below approximately 42 mph, it should be more willing to leave fifth for fourth.

The result should be a more deliberate, connected acceleration path: fourth gear for the initial pull, fifth gear for continued midrange acceleration, and sixth only once road speed is high enough that the taller ratio will not dilute the response. The separation between upshift and downshift speeds should also reduce gear hunting when speed fluctuates around the thresholds.

This does not necessarily mean every acceleration event will produce a higher peak torque request. Instead, the car should feel more responsive because the transmission is less likely to place the engine in a tall gear before the acceleration demand is finished. The expected visual change in a log would be longer gear segments, fewer early 4→5 and 5→6 shifts, and more timely 6→5 or 5→4 downshifts as speed falls.

We also exposed and reviewed FNNEOPEN, the open-converter engine-speed model. The PNH SSAX is approximately 700 rpm over stock stall, and the existing low-speed logs suggest the original model underpredicts engine speed with this converter open.
IMG 2445

FNNEOPEN predicts engine RPM from torque and turbine RPM. We adjusted the implied slip portion of that prediction, using 90% of our estimated correction. The selected change covers 42 positive-torque cells, caps the added engine-speed prediction at 630 rpm in the lowest turbine-speed region, and tapers the correction away by 3,000 turbine rpm. Zero-torque and coast regions, along with the axis normalizers, stay intact.
IMG 2448

During acceleration before the converter is fully coupled, the controller should now expect engine speed to rise more in line with the installed converter’s actual behavior. Visually, that should appear as a more believable separation between engine speed and turbine speed during the initial pull, followed by a smoother reduction in slip as the converter applies or relocks.

The intended effect is not to create an artificial RPM flare. It is to prevent the controller from interpreting normal converter slip as an error caused by an engine-speed model that is too low. If the model is closer to reality, converter application should be more predictable, and the engine-speed trace should show a cleaner transition from open-converter acceleration to coupled operation.

That is a bounded model correction based on the evidence available. It doesn’t change the converter’s physical stall speed or establish a proven “perfect” calibration. The intent is to give the controller a better representation of the installed converter during application.

The converter package also retains the selected scheduling and relock changes, plus the single 25% after-upshift lockup-rate increase. Gear-clutch fill times return to the Rev7 values.

Under acceleration, the expected result is a quicker but more controlled return to coupling after an upshift. The 25% lockup-rate increase should reduce the time spent in an unnecessarily loose converter state, while the retained scheduling and relock changes should help the converter apply when the operating conditions support it.

The visual target is a clean sequence: the gear change occurs, engine speed settles without a prolonged flare, converter slip decreases, and turbine and engine speed move toward a stable relationship. The change is intended to improve connection, not to force lockup so aggressively that it creates a harsh event or interrupts the acceleration curve.

On fueling, we’re implementing Uwe’s guidance from XDI. His recommendation was roughly 2,200 psi around peak torque, ramping toward 2,800 psi at peak RPM, with lower desired pressure if pump demand approaches its limit. He specifically pointed out that there is plenty of injector capacity available.

After the intermittent near-100% HPFP duty readings, our Rev10 implementation reduces both high-fuel-mass cells at 4,500 rpm from 2,200 to 2,100 psi. The corresponding 6,500-rpm cells remain at 2,800 psi, with interpolation between them.

That modest reduction is intended to gain pump headroom through the affected range while retaining the upper-RPM pressure request. Lower pressure requires somewhat more injector on-time for the same fuel quantity, which is why Uwe’s assessment of the injector capacity matters here.

During acceleration, the fuel-pressure strategy should look like a controlled ramp rather than a pressure request that immediately pushes the pump toward its limit. Around the high-load, midrange area near 4,500 rpm, the lower 2,100-psi request should give the HPFP more operating margin. As RPM continues to rise, the request should build toward 2,800 psi by 6,500 rpm.

The intended visual result is a rail-pressure trace that follows the target more consistently, with fewer periods where pump duty approaches 100%. Lambda should remain on target, while injector on-time increases only as needed to compensate for the modest pressure reduction. If the strategy works as intended, acceleration should remain strong through peak torque without the fuel system becoming the limiting factor.

Taken together, Rev10 is intended to produce a more connected acceleration event:

-The pedal request builds torque progressively.
-Fourth and fifth gears remain available longer, keeping the engine in a more useful RPM range.
-The open-converter model better represents the expected engine-speed rise and slip.
-Converter relock should occur more cleanly after shifts.
-Fuel pressure should maintain the requested fueling with more pump headroom.
-The engine should continue pulling toward the upper RPM range without an early upshift, excessive converter flare, or fuel-pressure shortfall.

The next logs will show whether Rev10 delivers more consistent coupling, the gear-holding response I want, and less time at the pump’s duty limit while actual rail pressure and lambda follow their targets.

We’ve deliberately left part of the estimated converter correction on the table. The last 10% can earn its way in. Now we wait for dry weather and the car will Vote next.
 
Last edited:

802SHO

Boost > VE, MBT, Cams
Supporting Member
Joined
Oct 8, 2019
Messages
3,539
Reaction score
8,361
Location
Essex Junction Vermont
HPTuners, my strategy A2L, direct communication to Johnny Powers at HPTuners and using Chuck (ChatGPT Pro) has created an extremely rare case for me. I’d label myself as the least qualified with the most access. lol.

I can log my tune revision and immediately write a new tune. If something is strange and the exposed controls don’t make sense I can use the A2L and Chuck to expose exactly what I need to continue progessing the tune. When I’m able to see direct results from our exposed controls I will prepare an XDF for Johnny Powers and send it to him directly. No support tickets. No waiting. I also don’t need to wait for Johnny to expose anything, I can expose it myself with my User Defined Parameter license.

The loop is extremely efficient and I consider myself extremely fortunate to be able to do this.
 

802SHO

Boost > VE, MBT, Cams
Supporting Member
Joined
Oct 8, 2019
Messages
3,539
Reaction score
8,361
Location
Essex Junction Vermont
:hi5:Rev10 update. This old Green Oak ECU continues to be one of the most addicting, frustrating and rewarding puzzles I’ve ever screwed with. :typing:

The deeper Chuck and I get into the A2L, the more controls we’re finding that HP Tuners simply never exposed for this strategy. At this point Rev10 has 11 newly exposed UDP controls that we’re actually using. These aren’t random tables I found and maxed out. Every one connects to the current hardware or something we can point directly at in the logs.
Here are the 11:


  1. FNNEOPEN, open converter engine speed model. This is probably my favorite discovery. The PNH SSAX converter is roughly 700 rpm looser than stock, but the PCM was still modeling the behavior of the OEM converter. FNNEOPEN predicts engine RPM based on turbine RPM and converter torque while open. We used actual logs to correct the model 90% toward what the PNH physically does, intentionally leaving 10% out so the next logs can verify direction.
  2. FN_SRLK, converter speed ratio model. Adjusted to better represent the PNH’s coupling behavior instead of expecting an OEM converter speed relationship.
  3. MAXSLIP, maximum attainable commanded converter slip. Rev9 repeatedly showed exactly 1,020 rpm Desired TCC Slip while the car was riding the converter. Then we exposed MAXSLIP and found it was literally set to 1,020 rpm. :evilgrin: Rev10 is 800.
  4. SPD_LK_TM_MIN, speed-lock minimum time before the next upshift. Stock was 10 seconds. Rev10 is 2 seconds.
  5. CSHUUNLKDLY, relock delay after an unconditional unlock. Stock was .75 sec. Rev10 is .25 sec.
  6. OL_NO_RMP_TM, maximum open-to-lock mid-ramp time during a nonshift. Stock allowed 5 seconds. Rev10 is 2 seconds.
  7. OL_UP_RMP_TM, maximum open-to-lock mid-ramp time during an upshift. Stock allowed 2.5 seconds. Rev10 is 1 second.
  8. CCSCH_WOT_PP, the converter scheduler’s own definition of full WOT pedal. This matters because if I change when the engine recognizes WOT but the converter still thinks I’m at part throttle, I’ve created another disagreement between controllers. They’re now being coordinated.
  9. FNTQLM_AXL_AWD, the hidden AWD axle torque limit. Rev9 finally exposed this one the hard way. The stock limit was only 4,645 lb-ft. During the launch the log showed about 455 lb-ft Engine Brake Torque. With 4.48 first gear and the 2.77 final drive that’s roughly 5,650 lb-ft before losses. Right there Torque Source changed to TCase/Low Range/Driveshaft and the PCM murdered torque, throttle and boost. That ceiling is now moved out of the way.
  10. ESL_CTL_LMT, engine protection RPM limit. We moved the protection window from 6,400 to 6,600 rpm.
  11. ESL_CTL_OFS, transmission shift RPM offset. We moved the offset from 200 to 300 rpm so the transmission can use roughly a 6,300 rpm shift target while maintaining a real protection window above it.
The converter model is really what started this whole rabbit hole.

Historically a looser converter on this platform has been problematic because changing the hardware didn’t mean the PCM suddenly understood what was installed. (Circle D) With the PNH we can actually see the mismatch in the logs.

One example was around 1,500 turbine rpm and ~100 lb-ft converter torque. Actual engine speed was about 2,090 rpm. Our corrected FNNEOPEN model predicts about 2,097 rpm there. That’s not “add 700 rpm everywhere.” The correction changes with turbine speed and torque exactly like the actual converter does.

Ford also has a switch specifically enabling FNNEOPEN as part of the stroke-to-mid-ramp transition during converter application. So this isn’t just an informational model sitting unused. The PCM actually references it while deciding what the converter is doing.

The idea with Rev10 is to tell Ford the truth about the PNH converter, reduce the ridiculous amount of time it’s allowed to hang around open or transitioning, and otherwise leave Ford’s actual clutch control strategy alone.

I also recently had some direct communication with Brad from AJP Turbo, who tuned this car years ago when it went 11.06. His explanation of the torque model helped put some of what we’ve been doing into perspective.

His analogy was basically that the torque model needs enough range, like a speedometer that reads high enough that you don’t peg it. Ideally the model still makes sense. You can mangle one model and compensate for it with another, but that’s when you start chasing your tail.

That made me go back and take another hard look at what we’ve actually built with Driver Demand, indicated torque, inverse torque and the OSS modifier.

And surprisingly, the receipts in the logs say we’re actually pretty close to exactly what Brad is describing.

My current full-pedal Driver Demand tops at 500 lb-ft. That’s intentional. It used to be 550 and even over 600, but Green Oak actively tries to satisfy requested torque. If the model sees a torque deficit it doesn’t just shrug. It can ask for more air, more load, more spark and manipulate throttle trying to satisfy it.

So 500 is the torque I’m intentionally requesting.

But that does NOT mean the indicated side only needs to read to 500. Indicated torque includes the torque required before pumping/friction/losses are accounted for.

The Rev7 WOT log actually recorded approximately 750 lb-ft maximum Scheduled Torque while Engine Brake Torque peaked around 536 lb-ft. Desired Load reached 2.15, actual load essentially matched it at 2.145, Desired TIP reached exactly 72 inHg and throttle stayed at 82°. Torque Source spent most of that WOT pull reporting Engine Indicated Torque Limit.

Rev9 tells essentially the same story once it escaped the screwed-up launch. Scheduled Torque again reached roughly 750 lb-ft, Engine Brake Torque reached about 544, Desired Load reached 2.20, actual load reached about 2.12 and TIP again reached exactly 72 inHg.

Meanwhile my current indicated torque model has roughly 830 lb-ft of upper range.

So there’s Brad’s speedometer analogy in actual data. The PCM is legitimately operating around 740-750 lb-ft on the modeled/scheduled side and the model reads to ~830. It has range without being pegged.

The inverse table mirrors that extended range back into load. Driver Demand asks for the torque I actually want. The indicated/inverse model has enough room to represent what the engine and its losses require. Then I intentionally decide where the party ends.

-2.2 load.
-72 inHg TIP / roughly 21 psi boost.
-Maximum 16° WOT spark.
-Power Enrichment fueling under 5psi boost

Those are my guardrails.

So Rev10 isn’t about eliminating torque intervention. A torque-based ECU is always arbitrating something. I’m trying to decide which intervention I actually WANT to win.

I’d much rather have it run cleanly into my 21 psi, 2.2 load and 16° boundaries than have some hidden 4,645 lb-ft AWD limit suddenly decide in first gear that we’re making too much torque and close the throttle. :smash:

We’re also rolling some of the increasingly aggressive pedal/DD behavior back toward the better coordinated Rev7 setup. Instead of continually inflating Driver Demand to make the car feel responsive, we now have access to WOT recognition on both the engine and converter sides and can coordinate those directly.

All 11 UDP changes are now in their own XDF too. Since HPT Compare can’t highlight these custom definitions, I can load that one XDF against Rev7, Rev8, Rev9 or Rev10 and immediately see what every hidden control was doing in that exact file.

This shit is ridiculously addicting. One minute I’m ****** because changing one thing makes something completely unrelated look broken. Then two days later you expose another definition, go back through an old log and suddenly realize the PCM was telling you exactly what it was doing the entire time. You just couldn’t see the table controlling it.

Frustrating as ****. Also probably the most rewarding part of building this car.
 

802SHO

Boost > VE, MBT, Cams
Supporting Member
Joined
Oct 8, 2019
Messages
3,539
Reaction score
8,361
Location
Essex Junction Vermont
Aeromotive 3.5GPM TVS Brushless Spur-Gear LPFP Failure

Well, Rev10 testing ended a little earlier than planned.

The Aeromotive 3.5 GPM brushless spur-gear LPFP is coming out and going back to Aeromotive for a rebuild. The pump itself didn’t randomly fail though. I killed it, and unfortunately the explanation is considerably less impressive than anything involving Green Oak, the transmission or Rev10.

I ran the car out of gas.

I had filled the 12-gallon cell recently, and somehow “I just filled it” stayed planted in my head. Normally that would be plenty because there are usually days between tuning sessions. This time I had been completely locked into the car. Flash a revision, drive it, datalog it, come home, analyze it, expose more definitions, make another change and go right back out. I had taken the car out three separate times since filling it and this was the fourth.

The Tinker V2 normally gives me fuel level, but right now I’m constantly datalogging with the MPVI3 and both need the OBD-II port. The OEM cluster doesn’t give me useful fuel level with the cell, so without really thinking about it I had turned fuel level into something I was keeping track of from memory.

Apparently my memory was still on the previous calibration revision.

The best part is I drove past at least three gas stations. I wasn’t trying to stretch the fuel or thinking I was low. It genuinely never crossed my mind. I was in that flow state where all I wanted to do was calibrate the car, feel what it did, save the log and get home to analyze it. I’m thinking about TCC behavior, turbine speed, Driver Demand, shift learning and whether all the new Rev10 logic is finally coordinating correctly while the 12-gallon cell is quietly becoming a 0-gallon cell.

And Rev10 was feeling really good.

I spent the first part of the night getting the transmission warm, getting the fluid level right and continuing the relearn. I also found a little trick the 6F55 really seemed to like. I’d get up to speed, manually downshift 4-3-2, put a little load into it and then return it to Drive so it could perform the 2-3 itself. Repeating that gave it consistent events to learn from and the transmission just kept getting better.

I saved a couple drivability logs and started leaning into it a little harder. The car felt good enough that I was basically done screwing around. I was about two minutes from the road where I do my real testing and I was ready to finally see what Rev10 did at WOT.

Then I braked pretty hard and heard a new sound from the rear of the car. At first it was intermittent. Then while going downhill it got louder and started coming and going. Eventually the Aeromotive LPFP sounded like my dual VIAIR 444C compressors had turned on and were rattling the aluminum rear-seat-delete panel they’re mounted to.

Now I’m completely confused because I’m thinking Aeromotive LPFP failure, electrical problem, controller problem, anything except the obvious. I drove about another mile and a half at very low load because I was in the middle of nowhere and somehow managed to pull directly in front of a gas pump.
IMG 2463
Perfect.

Gas station closed.

And it only sold 91 anyway.
IMG 2461
Stefanie eventually brought me four gallons of 93 and the cell swallowed all four without hesitation. That was approximately when the highly sophisticated diagnostic process arrived at “oh, you ******* idiot.”

Unfortunately, by then the Aeromotive LPFP had spent enough time cavitating and intermittently catching fuel that adding gasoline didn’t make it happy again. It would make horrible noise, chatter, cycle, eventually load the 30-amp circuit enough to open the fuse, and even after adding more fuel it never returned to normal. At that point I was done experimenting with a very expensive LPFP.

The tow somehow became the best part of getting stranded. The driver who showed up was the exact same guy who had towed this car off the interstate before when I had the turbo oil-return/scavenge problem. He remembered the car immediately, knew it was low and knew I was going to be particular about loading it.

IMG 2464I pulled the front splitter/chassis pieces, raised the front on the air suspension and got inside to handle Neutral while he carefully loaded it. When we got back to my house, he lined the rollback up perfectly and unloaded the car directly inside my garage. It literally could not have gone smoother.

The first company I called wanted $350. This guy wanted $180, so I handed him $200 and told him he was a legend and probably the best tow-truck driver on Earth.

So that was the ridiculous part. The interesting part is that the Aeromotive LPFP died while I was datalogging.

Before that point, Rev10 looked really good. The strongest clean test I got was only about 62% pedal, so this is not a WOT validation, but the car reached roughly 4,000 rpm and 470 lb-ft Engine Brake Torque.

Desired Load was 1.794 and Actual Load was 1.796. Scheduled Torque was roughly 521 lb-ft, Desired Brake Torque around 483 and actual Engine Brake Torque about 471. Torque Source remained Driver Demand with No Limit Active.

That Desired Load versus Actual Load relationship is probably my favorite part of the log. After everything we’ve been doing with Driver Demand and indicated/inverse torque, Green Oak asked for 1.794 and got 1.796. At least through the load Rev10 actually saw, the torque model looks extremely coherent.

Air and fuel control were equally clean. Desired TIP was about 61.9 inHg against roughly 60.9 actual. Commanded lambda was .820 against approximately .815 measured. Desired DI rail pressure was 2100 psi and actual was around 2127, while spark remained primarily on Borderline with essentially no knock activity.

The car genuinely felt good.

The PNH data was encouraging too. Once Green Oak actually commanded the converter coupled at roughly 470 lb-ft, Desired TCC Slip was 20 rpm, Actual Slip was about 26 rpm, Speed Ratio was around .994 and TCC pressure was roughly 89 psi. Once the PCM asks that converter to hold, it holds.

Then later in the same log the Aeromotive LPFP falls off a cliff.

It’s screaming so I’m limping it. At only around 10% pedal, the PCM continued requesting normal rail pressure while actual rail pressure collapsed through the hundreds and essentially hit the floor. Lambda went dead lean and the engine lost power. Then fuel apparently reached the pickup again and rail pressure suddenly recovered, which perfectly matches what I felt when the car lost power for a second and came back.

Vehicle voltage remained healthy through the event. The PCM was still asking for fuel and the DI system simply stopped receiving a reliable low-side supply. The Aeromotive LPFP literally died while providing the datalog showing exactly how I killed it.

In hindsight, being two minutes away from the WOT road might have been the luckiest part of this entire stupid mistake. Had the pickup uncovered during a 21 psi, 2.2-load pull instead of low-load driving, I could be writing a very different post. A hurt Aeromotive LPFP is expensive. A hurt Aeromotive LPFP plus a hurt built engine would have been considerably less funny.

The other useful part is that Rev10 never gave me a reason to start tearing the calibration back apart. Everything it actually got asked to do looked encouraging. We didn’t get the WOT test, but the road leading to WOT looked better than it has in a while.

Since the car is going to be down for a little bit, Chuck and I have already started using the downtime to go even farther into the A2L. We’ve exposed the actual per-gear steady-state converter-slip targets, the per-gear lock/unlock schedules and more of the OPEN-to-controlled-slip state logic. We’re starting to separate exactly what determines when the converter is eligible to apply from what controls the application after it starts.

So Rev11 is probably going to benefit quite a bit from my temporary inability to drive the car.

I talked to Martin C. at Aeromotive tech support today and he set me up with an RGA number. I’m shipping the Aeromotive LPFP back for a rebuild, which should be somewhere around $350-$825 depending on how badly I hurt it. A new Aeromotive LPFP is now roughly $1,900, while the rebuild is essentially a new pump using my existing housing, so that takes a lot of the sting out of it.

Martin also went pretty easy on me when I told him exactly what happened. He said he’d done basically the same thing after putting a six-gallon cell on his motorcycle without a gauge. He couldn’t keep himself off the throttle, ran it down and learned the same lesson.

So apparently even the Aeromotive tech guy has been there. I still think “put gas in the ******* car” has earned a permanent spot on my pre-flight checklist. But for F sake LOL OMG
 

802SHO

Boost > VE, MBT, Cams
Supporting Member
Joined
Oct 8, 2019
Messages
3,539
Reaction score
8,361
Location
Essex Junction Vermont
I’ll be back up and running faster than I thought. I opted to slightly downgrade pumps bc the Eliminator (18389) is in stock and ready to ship. I’ll rebuild the bigger pump later. IMG 2474
That price it shows for my 3.5GPM with my options is actually over $1,900 and it’s 3-4 weeks build time. Rebuild is less than half but 3 weeks lead time. I am running out of weather here in Vermont. The eliminator was just under $1,600 and they didn’t charge me tax or shipping. Martin hooked me up best he could.
 

802SHO

Boost > VE, MBT, Cams
Supporting Member
Joined
Oct 8, 2019
Messages
3,539
Reaction score
8,361
Location
Essex Junction Vermont
I got held up at work and swapping over the fuel pumps took longer than expected. It went very smoothly and I flashed Rev11 and took and idle log. I literally need to go to bed so no driving I guess.

Removal was so easy. 8 gallons in this 12 gallon cell. Wasn’t much mess somehow.
IMG 2570
I revised my engine torque reduction for upshifts to .50 removing 50% torque. You must realize OEM made 365 tq and it used a 30% reduction. At my modest .58 that’s still way more torque remaining during the shift. .50 sounds aggressive but there’s still about 100 more torque than stock still remaining during the shifts. I’m not claiming this is the perfect amount but I will test this.

The pump that gave up the ghost IMG 2571
Eliminator is ready IMG 2572IMG 2573IMG 2574
Around 630 lph. The 3.5 was about 794 lph.
Max for stock location is DW’s new 430C, 430 lph.

First I turned the pump on and let it return. About 85 psi constant to the Aeromotive FPR. No leaks of any kind. I set up VCM Scanner, flashed Rev 11 and captured the cold start and idle warm up.

Rev11 first-start/idle log looks excellent.

Once the idle settled, it held essentially 600 rpm. Over the settled idle section the median was 600.5 rpm, with 90% of the samples between roughly 589 and 611 rpm.

Fuel rail pressure tracking was excellent. Desired rail pressure was about 238 psi and actual was about 243 psi median. Median error was only +5.6 psi, and 95% of the samples were within about 13.5 psi of target.

HPFP regulator duty was only about 13.75% at idle.

Both widebands were essentially dead on stoich:
Bank 1 median lambda: 1.00002
Bank 2 median lambda: 0.99978

Short-term fuel trims were also excellent:
Bank 1 median: -0.78%
Bank 2 median: +0.78%

Long-term trims were 0% following the fresh flash.

Knock retard was 0.0° throughout the settled idle section.

Charging voltage was extremely stable at about 14.54 V.

So for the first start on Rev11, idle quality, lambda, trims, rail-pressure control and knock behavior all look about as clean as I could ask for. Obviously idle doesn’t validate fuel delivery under load yet, but there is nothing in this log telling me to change the calibration before driving it.
 

802SHO

Boost > VE, MBT, Cams
Supporting Member
Joined
Oct 8, 2019
Messages
3,539
Reaction score
8,361
Location
Essex Junction Vermont
Switching pumps and looking at LPH ratings had me thinking. The new DW430c drop-in LPFP sounds pretty close to the roughly 630 LPH Aeromotive Eliminator on paper. But once you look past the headline flow number, the two systems really don’t have much in common.

The first difference is the pump itself. The Eliminator is a brushless TVS setup, so it has substantial fuel capacity without needing to run full speed all the time. Pump speed can follow demand, which reduces unnecessary current draw and fuel heating during normal driving while still giving me the pump’s full capability when the engine actually needs it.

Then there is the plumbing. My low-pressure system uses -10AN feed and return, an -8AN rollover vent, and an -8AN line from the regulator to the XDI60 HPFP. That is a very different environment from asking a stock-location drop-in pump to push its maximum output through a factory-style 3/8-inch returnless low-pressure system.

The pressure strategy is also completely different. I’m around 85 psi static with the engine off. The Aeromotive regulator is manifold-referenced 1:1, so engine vacuum pulls that down to roughly 75 psi at idle. At 21 psi boost, the regulator should be commanding roughly 106 psi on the low side.

That means the XDI60 isn’t simply being supplied above its minimum inlet-pressure requirement. The low-pressure system has to maintain very high inlet pressure while also maintaining the fuel volume the HPFP is consuming under load.

That is why comparing pumps strictly by LPH can be misleading. A pump’s useful output depends on the pressure it is working against, voltage, line restriction, fittings and the rest of the fuel system. The number that actually matters is how much fuel arrives at the HPFP inlet while maintaining the pressure the system is asking for.

From there the XDI60 takes over and raises fuel pressure into the direct-injection system. My current rail target is around 2200 psi through the midrange and ramps to 2800 psi near 6500 rpm. The XDI 2000 cc injectors are scaled in the calibration to 3000 cc per XDI’s strategy.

I do still use the OEM 3/8-inch hardline from the XDI60 to the rails, but that is on the high-pressure side after the mechanical HPFP. It is a short rigid line carrying fuel at thousands of psi, so it serves a completely different hydraulic purpose than a long 3/8-inch low-pressure supply line running from the tank.

So the comparison really isn’t:

430 LPH vs 630 LPH.

It is more like:

DW430c feeding a factory-style returnless low-pressure system

versus

TVS brushless Eliminator, -10AN feed and return, -8AN vent, -8AN regulated supply to the XDI60, 1:1 manifold-referenced fuel pressure, roughly 85 psi static and over 100 psi low-side pressure at 21 psi boost, followed by 2200-2800 psi direct-injection pressure.

The pump rating matters, but the entire system determines how much of that pump’s capacity is actually usable when the engine is asking for everything.

That was the point of building mine this way. I wanted the low-pressure side to have enough pressure and flow headroom that it would not become the first thing limiting the engine. It does not make 93 octane more knock resistant, but it gives the engine a much better chance of actually receiving the commanded fuel mass and rail pressure without needing methanol just to compensate for a fuel-delivery shortfall.
 

Forum statistics

Threads
107,294
Messages
1,183,253
Members
16,130
Latest member
raymo23

Members online

Back
Top