TUNING PLAN FOR ENGINE WITH PREVIOUSLY FIXED SPARK ADVANCE

[BRUCEROE]

TUNING PLAN
improve mid range torque
lower possible idle speed
reduced fuel consumption at part throttle with more advance
ease starting with reduced cranking advance

MJLJ-V4 TUNING PLAN
use manual ADVANCE knob to shift entire map up to + or - 12 degrees
after best advance found, modify ignition map to put knob back at 0

set no load (idle) advance for low KPA
for best vacuum/peak rpm, for different throttle positions

use logging to measure instantaneous acceleration at WOT with heavy inertial load
accelerate from idle to max rpm while logging.
plot instantaneous acceleration against rpm
shift entire map forward & repeat for another curve on same plot
shift entire map backward & repeat for another curve on same plot
at each rpm, note which advance provided the best acceleration
modify the base map to reflect the best advance at each rpm area
continue to optimize till the unshifted base map has the highest curve

smooth out the advance map area between idle & WOT

while the engine is in use, use the manual ADVANCE knob to see if any area needs change

RESULTS OF INITIAL TUNING TRIALS
Increasing idle advance increased rpm by about 400, no throttle change. Reducing the idle screw allowed idle speed reduction from 1500 to below 1000. However the engine would need more flywheel for smoothness at 1000, so it was set to 1200. At no load 3600 rpm, more advance brought it to 4000 rpm. This would be even higher, but the governor cut in. Next the power take off clutch engaged a heavy inertial load. At idle the throttle was snapped to WOT and manually held to at least 4000 rpm. The governor wasn't allowed to interfere. Each acceleration run was done 3 times, with the 3 graphed lines giving an average position to minor variations. A good starting advance map was produced, subject to more fine tuning if indicated later.

PROBLEMS WITH LOGGING
Logging files were named, started, & stopped using FILE commands. The sampling interval was set to 100ms. The space bar is supposed to MARK DATALOG, but using the log pause or clear button changes the space bar to something else.

With the log file recorded, a line could be moved across time to give recorded values. However, the time on this line was only resolved to seconds, even with maximum zoom. The maximum width for a second was less than 1/2" on my screen, and the graph vertical time lines didn't seem to line up with anything particular. It was decided to open the file to actual recorded values, print this, and work through the areas of interest.

The printed data showed timestamps of around 0.22 or 0.34 second intervals, never the 0.1 sec requested. The period followed by 3 digits gave the timestamp milliseconds. But if that
field had leading zeros, they would be left out. So 2.082 sec would become 2.82 sec, a huge difference if resolution is needed.

Next I started plotting the RPM Accel field, the rate of rpm increase per second. That is, the increase in rpm since the previous time stamp, divided by the time since the last time stamp. For some reason RPM Accel seemed to increase almost linearly with rpm. Hey, gas engines have a roughly constant torque over the useful rpm range. Certainly NOT directly increasing with rpm. Acceleration is proportional to torque, so it should be roughly constant as well. Some log lines were even showing acceleration when rpm was dropping.

This was not acceptable, so I calculated RPM Accel myself for each timestamp. Take the increase in rpm, divide it by the increase in time. This gave the sort of results expected. I ran the whole logging & correction operation several times, and got a good starting advance map. Bruce Roe