CIS to ITB EFI conversion

[scarceller]

I want to be sure I understand WOT ignition and what was tuned on dyno. Before dyno tuning at 5500RPM 100% load you had 35 deg but after tuning you found 27 deg provide best torque? If so that would mean you had far to much ignition at high RPM prior to tuning?

Originally Posted by dannichols1474

Hi Ben,

In the last post I said I thought a dyno session is good value for the money, then I reread your post and I want to add that if you are using the ECU to control the fuel injection only, then tuning the EFI with the big hills/foot on the brake technique is adequate if you are happy with your results.

I bought Al Kosmal / X-Faktory's Option#2 conversion package that added ECU ignition control to the ECU fuel injection control, so dyno tuning made sense for me to optimize the timing settings.

Here is a photo of the ignition table before dyno tuning using timing values from the characterization of the distributor mechanical advance curve before I started the CIS to ITB EFI conversion project:



Here is a photo of the ignition table after dyno tuning using timing values derived from timing sweeps to determine MBT (max brake torque) at various combinations of rpm and load:

[dannichols1474]

Originally Posted by dannichols1474

Part 7

With the mechanical work done it was time to start the software side of the project....

I also entered the ignition advance data I collected from the stock set up into the Ignition Table.
The VE, AFR and Ignition Timing tables are from after the dyno tuning.

Hi Sal,

The answer to your question is yes - the old mechanical advance had more timing than optimum at WOT and all high loads.

The above graph shows, amongst other things, the ignition table for the stock 2.7L motor with ITB EFI after dyno tuning.

The 2.7L stock engine has a CR of 8.5:1 and MBT at 6500rpm / WOT is 29.1 degrees.

The rebuilt 2.8L engine has a CR of 9.8:1 and MBT at 6500rpm / WOT is 28.8 degrees.

So as expected, the timing needed comes down as CR goes up. Also, timing needed goes down as load goes up in line with what I have read in tuning literature.

[dannichols1474]

While I am waiting for my next dyno tune / test session (this Wednesday) to verify 1.625" headers are better than 1.750" headers for my rebuilt 2.8L engine, I started wondering about procuring replacement parts for my X-Faktory CIS to ITB EFI option #2 conversion kit. I sent an email to Al Kosmal asking him where I could buy a replacement O2 sensor for my system when the need arises. Al pointed me to the 14pont7.com website and while there I saw under "Latest News" an article about maximizing accuracy (see photos).



Here is an excerpt of the calibration process from the article:





So after reading the above article, I decided to try performing the calibration procedure on my system.

I downloaded the Excel calibration worksheet. I started TunerStudio on my notebook and connected the serial / USB cable from the MegaSquirt2 ECU to my notebook, then I turned the ignition key to the "ON" position (key ON, engine OFF) and observed the Air Fuel Ratio gauge on TunerStudio - initially it read "13.20", then after 5 seconds it read "16.60". Then I entered the observed values above into the Excel calibration worksheet and the worksheet computed the Point 1 and Point 2 AFR values.



Then in TunerStudio from "Tools", I selected "Un/Lock calibrations" and unlocked calibrations then clicked on "Burn".



Then from "Tools", I selected "Calibrate AFR Table...". For "EGO Sensor", I selected "Custom Linear WB", then I input the Point 1 and Point 2 AFR values from the Excel calibration worksheet. When the values were entered, I selected "Write to Controller" then "Close".



Then from "Un/Lock calibrations" I locked calibrations then clicked on "Burn".



Then I shut down TunerStudio and disconnected the cable from the ECU.

[dannichols1474]

The final step in the calibration procedure is to repeat the key ON, engine OFF check of the Air Fuel Gauge reading during first 5 seconds and during the second 5 seconds. During the O2 sensor heat up cycle the controller outputs a voltage corresponding to an ideal AFR of 13.328 for the first 5 seconds and outputs a voltage corresponding to an ideal AFR of 16.66 for the second 5 seconds.

When I performed the calibration check after inputting 10.192 and 19.992 in the "Calibrate AFR Table..." the Air Fuel Gauge read 13.40 during the first 5 seconds and 16.70 during the second 5 seconds. When I input those values into the Excel calibration worksheet I got calibration set points of 9.895030303 for 0 Volts (Point 1) and 19.992 for 5 Volts (Point 2).

When I performed the calibration check after inputting 9.895030303 and 19.992 in the "Calibrate AFR Table..." the Air Fuel Gauge read 13.20 during the first 5 seconds and 16.60 during the second 5 seconds. When I input those values into the Excel calibration worksheet I got (for a second time) calibration set points of 10.192 for 0 Volts (Point 1) and 19.992 for 5 Volts (Point 2).

When I performed the calibration check after inputting (again) 10.192 and 19.992 in the "Calibrate AFR Table..." the Air Fuel Gauge read 13.40 during the first 5 seconds and 16.70 during the second 5 seconds. I stopped the calibration procedure as I had gotten as close to the ideal AFR values as I was going to get.

[dannichols1474]

So I was getting a displayed value of +/- 0.2 AFR on the lower calibration value and +/- 0.1 AFR on the upper calibration value - lower target = 13.328 & upper target = 16.66, and my MegaSquirt2 ECU is displaying AFR to the tenth of an AFR, so my closest displayed values would be 13.3 and 16.7 AFR.

I went back out to the garage with my notebook running the TunerStudio app and connected the notebook to the ECU and turned key ON, engine OFF. I unlocked the calibration tables and opened the AFR calibration table.

Now I started hacking. I took the average of the two Point 1 calibration values the Excel calibration worksheet had given me: (9.895030303 + 10.192) / 2 = 10.0435152 and entered that into Point 1 (0 Volts) of the AFR calibration table and wrote the values to the controller (pressed the "Write to Controller" button), turned the key OFF, waited 10 seconds, then turned the key ON, engine OFF and read the AFR gauge: 13.30 & 16.60.

Next, I increased the Point 2 (5 Volts) calibration value from 19.992 to 19.995, wrote the value to the controller, then cycled the key OFF to ON, engine OFF and read the AFR gauge: 13.30 & 16.60.

Again, I increased the Point 2 calibration value from 19.995 to 20.000, wrote the value to the controller, then cycled the key OFF to ON, engine OFF and read the AFR gauge: 13.30 & 16.60.

Finally, I increased the Point 2 calibration value from 20.000 to 20.100, wrote the value to the controller, then cycled the key OFF to ON, engine OFF and read the AFR gauge: 13.30 & 16.70.

I had now achieved the closed displayed AFR values to the ideal target AFR values thereby minimizing offset and linear errors and maximizing the accuracy of the AFR values the MegaSquirt2 ECU was reporting (and using for injector pulse width trimming) from the 14Point7 O2 sensor controller.

I locked the calibration tables, shutdown the TunerStudio app, disconnected the cable from the ECU and turned the key OFF.

[dannichols1474]

I decided I had better check out the Point 2 calibration value more closely.

I went back out to the garage with my notebook running the TunerStudio app and connected the notebook to the ECU and turned key ON, engine OFF. I unlocked the calibration tables and opened the AFR calibration table.

I decreased the Point 2 calibration value from 20.100 to 20.050, wrote the value to the controller, then cycled the key OFF to ON, engine OFF and read the AFR gauge: 13.30 & 16.60.

I increased the Point 2 calibration value from 20.050 to 20.075, wrote the value to the controller, then cycled the key OFF to ON, engine OFF and read the AFR gauge: 13.30 & 16.70.

I decreased the Point 2 calibration value from 20.075 to 20.0625, wrote the value to the controller, then cycled the key OFF to ON, engine OFF and read the AFR gauge: 13.30 & 16.70.

I decreased the Point 2 calibration value from 20.0625 to 20.055, wrote the value to the controller, then cycled the key OFF to ON, engine OFF and read the AFR gauge: 13.30 & 16.70.

I locked the calibration tables, shutdown the TunerStudio app, disconnected the cable from the ECU and turned the key OFF.

I was now very close to the Point 2 calibration value where the second 5 sec displayed value toggled between 16.60 and 16.70.

To recap: For my MegaSquirt2 ECU and 14Point7 Spartan 2 O2 sensor controller + Bosch LSU 4.9 wideband O2 sensor, I maximized the accuracy of the AFR reported value with the AFR calibration table using Custom Linear WB, Point 1 (0 Volts) = 10.0435152 and Point 2 (5 Volts) = 20.055 giving a displayed value during the first 5 sec reference check of 13.30 (ideal target = 13.328) and a displayed value during the second 5 sec reference check of 16.70 (ideal target = 16.66).

[rwest]

Dan,

Thanks for the informative write up. I have the 14 point 7 unit as well and will do this calibration too now.

Rutager

[dannichols1474]

So I got done doing an exhaust header size comparison between the Bursch 1.750" headers and the Bursch 1.625" headers (both without heat boxes) and the 1.625" headers won the shoot out on the dyno (as had been predicted by forum denizens).

Dyno data:



One thing the dyno doesn't show is the driveability and I had rebuilt my 2.7L engine into a high revving 2.8L with 911S MFI camshafts (WebCams p/n 05-062) which runs well on the highway, but is really challenging to drive around town or in stop and go highway traffic. The engine won't run below 2000rpm when cold, power from idle to 2000 rpm is non existent, etc.

Anyway, I was introduced to William Knight (knightrace), of the M1 cam fame for CIS engine owners, and he suggested his K45 camshaft and said it would give more torque across the rev range and be more driveable, so I ordered and and have received my K45 camshafts.

Now it's time for the camshafts swap. Engine / transaxle are out of the car and the engine is back on the engine stand for surgery.

[mikedsilva]

Those cams from web certainly are 'big'... and a very narrow lobe centre.
I think you'll be very happy with Williams cam. I've used it on 3.0 with itbs, and 3.5 with itbs and it drives like a stock car down low and revs all the way to redline.
I've even test run a 3.2SS with CIS and it actually idled 'ok'... this was only on the test stand and not driven.

Keep us posted.

[mb911]

Man you are on the search for max HP. Good luck

[dannichols1474]

Originally Posted by mb911

Man you are on the search for max HP. Good luck

Hi Ben,

It's not just about max HP. If it were, I would be done. But I am just not satisfied with the way the car drives - it makes me pissed off every time I come to a stop and have to fight with it to get going again.

I did the original CIS to ITB EFI conversion because I was looking for two things: one, I have always like the look of the 356's and early 911's with carburetors and wanted that look as opposed to the CIS look and two, I wanted the convenience of modern ECU control of the fuel injection and ignition timing (while retaining the distributor, the look).

So when I rebuilt the CIS 2.7L motor, sure I was after more HP, but I wanted to include more modern components for higher compression along with strengthening the case with stud inserts for all of the case studs (not just the head studs) plus balancing all the rotating and reciprocating parts so I could spin the engine faster and bigger camshafts (higher lift and longer duration) so I could take advantage of a better breathing, faster revving engine.

So when I finally experienced first hand what others had been trying to tell me about the driveability of the 911S MFI camshafts, I realized I must correct my mistake by installing a more modern design camshaft.

Anyway, back to the camshaft replacement.

This is a lot of work, but the it is giving me the opportunity to fix issues that I encountered from the first engine / transaxle removal, rebuild and installation.

One issue I had noted was the transmission mounting bolts (the ones with the pointed end, p/n 930-375-317-01) had worn / galled threads and the bolts weren't threading into the mounting points in the body smoothly, so I bought replacement bolts and used a thread conditioner (thread chaser), M12-1.50, on the threads of the blind hole mounting points. Then I applied some anti-seize to the new bolts and ran them into / out of the mounting points to verify smooth thread engagement.

Another issue was I used Locktite on the rocker arm shaft bolts. BIG MISTAKE. As I have found out while I have been removing the rocker arms, the Locktite squeezed off the bolt threads as I inserted the bolt through the bolt head side expanding cup and the rocker arm shaft. So when I went to remove the rocker arms and loosened the rocker arm shaft bolts, the bolt and the bolt head side expanding cup were glued / stuck in the rocker arm shaft. I got the rocker arms removed, but what should have taken me 20 minutes per rocker arm has extending to 40+ minutes as I have had to use Scotchbrite pads and a 90 degree o-ring pick to remove the Locktite residue from the bolt and nut threads - WHAT A PAIN.

If you are a DIYer and you're rebuilding your engine - DO NOT use Locktite on the rocker arm shaft bolts.

I found during the engine build that a 12 mm deep socket fits through the holes in the camshaft tower (carrier) rocker arm shaft flanges and is a good tool to drive the rocker arm shafts into and out of the camshaft tower. For the camshaft swap job, the 12 mm deep socket works great for driving out the rocker arm shafts (with a levering bar) on cylinders #2, 3, 5, and 6 (I am driving the shafts out toward the flywheel end of the motor). But the deep socket is too long to fit into the space provided for the cylinder #1 and 4 rocker arm shafts.

For cylinder #1 and 4, I am using three regular length sockets, one at a time stacked on another, to drive out the rocker arm shaft out of the cam tower. I made a socket holder "tool" out of bent copper wire to place a 12 mm regular length socket into the cam tower flange and use a levering bar to drive the shaft out partially. Then I use the "tool" to place an 11 mm regular length socket into the cam tower and use a levering bar to drive the rocker arm shaft out further. And finally, I use the "tool" to place a 10 mm regular length socket into the cam tower and use a levering bar to drive the rocker arm shaft completely out of cam tower flange.

Some photos:

[dannichols1474]

Old camshafts are out. After some R&R comes some gasket cleaning, then I can start installing the new camshafts.

Rocker arms cleaned and labeled by location plus associated hardware:



Front of engine without cams:



Timing chain sprockets and associated hardware plus new cams:

[dannichols1474]

New camshafts are installed and timed. The timing card for the K45 cams specifies 0.050" lift with 0.000" valve lash at 24 degrees BTDC on the overlapping stroke - no tolerance was given, so I figured if I ended up +/- 0.5 degrees that would good enough given the play between the camshaft sprocket, the camshaft flange and the dowel pin. When I adjust the crankshaft position and the camshaft lift both values are exactly 0.050" @ 24 degrees BTDC, but after the cam bolt is torqued down the values always shift a bit for me.

My crankshaft digital degree wheel is mounted on the flywheel flange of the crankshaft so when the crankshaft is rotated clockwise as seen from the front of the engine, the digital degree wheel rotates counter clockwise hence the negative numbers.

-336.5 degrees = 23.5 degrees BTDC



-335.4 degrees = 24.6 degrees BTDC

[dannichols1474]

Originally Posted by dannichols1474

Got up early this morning and got out to the garage straight away. Was I going to have sufficient clearance between the valves and pistons? The short answer turned out to be yes - the JE 9.5:1 pistons with the WebCam S-cams, 1mm deck height and 0.25mm shave of the sealing surface on the heads turned out alright.

One cylinder #1: the exhaust valve had 2+mm (greater than 2.5mm) of clearance from 60 degrees BTDC to 30 degrees ATDC on the overlapping stroke, and the intake valve had minimum clearance of 2.0mm to 2.25mm from 2.5 degrees ATDC to 10 degrees ATDC and 2+mm of clearance everywhere else.

Once I had the results from cylinder #1, I chose to check the intake valve clearance only on cylinder #4 from 10 degrees BTDC to 20 degrees ATDC on the overlapping stroke. The minimum clearance was 2.5mm from 5 degrees ATDC to 10 degrees ATDC and 2+mm clearance everywhere else.

Per Wayne D's 911 engine rebuilding book, the minimum clearance for intake valves is 1.5mm and for exhaust valves is 2.0mm, so my engine passed the piston to valve clearance check!

Test set up for clearance checking:



Data record sheet of piston to valve clearance:

This morning I got up early and got the #1 and 4 exhaust rocker arms installed so I could perform the piston to valve clearance check with the new K45 camshafts. The set up is the same as in the quoted passage above. The short answer on the results is the same, I have sufficient piston to valve clearance with the K45 camshafts. Where the 911S MFI camshafts had minimum clearance on the intake valves after top dead center, the K45 camshafts have minimum clearance on the exhaust valves before top dead center.

Data record sheet of piston to valve clearance for the K45 cams:

[dannichols1474]

Friday, I ran the 15 minute / 2000 rpm camshaft break-in procedure for the K45 camshaft. Right after, I changed the oil and oil filter.

Saturday, I set the idle speed and balanced the flow of the ITB's, then I took the car out for an autotune session and the car ran so well that I went for an extended break-in drive and covered 426 miles.

Sunday, I changed the oil and oil filter and re-adjusted the valve lash.

Today (Monday), I just got back from dyno tuning / testing with the K45 camshafts installed.

If you live only by dyno data graphs the 911S cam may look better as torque is higher, but the K45 carries further and makes almost the same max HP (233 / K45 vs 236 / 911S). But the dyno data doesn't capture driveablity and the K45 cam wins on that count hands down.

I am very satisfied with my decision to replace the 911S cams for the K45 cams. The car is much easier to drive in town or in stop and go traffic and goes like hell whenever I put my right foot down!

Dyno data 911S cams versus K45 cams - all else the same.

[dannichols1474]

Monday night after the dyno session, I got the Bursch 1.625" headers removed and installed the Knight 1.625" headers in preparation of getting the car towed to the fabrication shop on Wednesday. Turns out the fab shop needed to order material so we pushed the car tow until Monday.

Anyway, when I installed the Knight headers I saw that the #6 primary tube is in contact with the oil supply line directly beneath the oil cooler that sends oil from the oil tank to the oil pressure pump.



I didn't like the idea of the #6 primary exhaust tube heating, via conduction, the oil entering the engine even if that oil goes thru the cooler, if necessary, before it is distributed to the crankshaft bearings and the rest of the engine. So I applied insulating wrap to the oil supply line to reduce the exhaust tube heating.





Another thing I observed is the placement of the collector is very close to the exhaust valve covers such that I will need to remove the headers in order to perform a valve lash adjustment - this was not necessary with either the Bursch headers nor the SSI heat exchangers used previously.





Next up: Off to the fabrication shop to get the muffler connected to the Knight headers, then one more dyno session to see data comparing the Knight 1.625" headers to the Bursch 1.625" headers.

[mikedsilva]

I like this thread. Lots of really good info and process.

[scarceller]

What's the length of the primaries on the Knight headers? I have those Bursch headers with heat and they are 32-34" long primaries.

If the new Knight headers are shorter than 28" I suspect they may not do as well as the Bursch? We shall see.

But the Knight headers do have a favorable 2.25" exit on the collector, that could help if you have 2.25" pipe to the muffler but the runner length is the most important.

[scarceller]

The new torque curve in the green line is FAR better than before, this engine is night and day better now! Your increase in torque at 4000-5500RPM is huge, I'm not surprised you mention it's street behavior is vastly improved. Can't wait to see new numbers with Knight headers but I would not expect any miracles beyond what you already achieved. Nice work.

Originally Posted by dannichols1474

Friday, I ran the 15 minute / 2000 rpm camshaft break-in procedure for the K45 camshaft. Right after, I changed the oil and oil filter.

Saturday, I set the idle speed and balanced the flow of the ITB's, then I took the car out for an autotune session and the car ran so well that I went for an extended break-in drive and covered 426 miles.

Sunday, I changed the oil and oil filter and re-adjusted the valve lash.

Today (Monday), I just got back from dyno tuning / testing with the K45 camshafts installed.

If you live only by dyno data graphs the 911S cam may look better as torque is higher, but the K45 carries further and makes almost the same max HP (233 / K45 vs 236 / 911S). But the dyno data doesn't capture driveablity and the K45 cam wins on that count hands down.

I am very satisfied with my decision to replace the 911S cams for the K45 cams. The car is much easier to drive in town or in stop and go traffic and goes like hell whenever I put my right foot down!

Dyno data 911S cams versus K45 cams - all else the same.

[scarceller]

EDIT: First, do a pull with everything setup exactly like before. Because if you lower acceleration rate your results (Torque and WHP) will increase and you want to do side by side compare of the headers first. But after you do that maybe try lowering accel rate as outlined below.

For next dyno pull, I suggest you ask what the dyno acceleration rate is set at? I suggest 250-300RPM per second and no higher than that, this gives the engine stability during the pull. I can't tell from the screenshot how the dyno was setup.