Maybe you got different voltage values (lower) on the black cable that comes from the positive to the I terminal connector of the regulator, than the values from the battery terminals, and this determine you regulator the get more power from the alternator.
I qoute from the learnings of@TwoTired
Charging system methodical verification checks, CB750, CB550, CB500, CB400, and CB350.
Begin with problem verification and characterization with recorded data.
A -- Fully charge a known good battery. Let it rest for 2 hours, off the charger, and measure the battery voltage. (Target is 12.6-12.8V.)
B -- Start the bike and measure the battery voltage at idle, 2000, 3000, 4000, and 5000 rpm.
C -- Repeat the measurements of B with lighting off.
The above tests identify charging system success, failure, or degree of "faulty". The success voltages are listed in the Shop manual.
D -- Assuming the above indicates faulty, do check the RECTIFIER diodes with a diode tester or ohmmeter capable of testing diodes and uses more than .7 volts to make the test(s).
Of the twelve test made in D, six must read low ohms and six must read very high ohms.
E- assuming no faults were found in D, Measure the white and green wires disconnected from the REGULATOR. CB750s should 6.8 ohms - ish, CB550s/350s/ and 400s should read 4.9 ohms- ish.
F- If there are no bullet holes or road rash/divots on the alternator case, the stator is probably good. But, you can check for yellow to yellow continuity (.35 ohms) among all the wires, and that no yellow wire has continuity to the engine case.
G- Assuming no faults found in D, E, and F, measure the disconnected terminals of the REGULATOR. The black and white terminals should measure zero ohms (subtract meter error if there is any). Higher than Zero ohms, indicates internal contact contamination needing cleaning and attention per shop manual.
H - Assuming D, E, F, and G have not found faults. We can verify all the of the charging system minus the regulator is functioning correctly, by using a temporary jumper to connect the disconnected white wire (normally attached to the REGULATOR) and connecting the White directly to the the battery POS terminal. Repeat the B and C tests. However, if at any time the battery voltage rises above 15V, stop the test. Such an indication would prove the charging system capable of maintaining a known good battery. If this test never achieves 15V, then there is a wire/connector issue in either the ground path leading back to the battery NEG terminal, a wiring/ connector issue withe the rectifier RED path to the battery POS terminal, or you made a mistake in D through G.
I - (not used, can be confused with L)
J - The only parts that remain to prove or expose are the REGULATOR (in active mode) and the electrical path between the battery POS terminal and the black wire that connects to the REGULATOR.
K - Lying to and starving the regulator
The regulator can only do its job correctly if it gets a proper voltage report of true battery voltage status. The Vreg monitors the Black wire for this status. Measuring the voltage lost between the Battery terminals and the Vreg connections identifies problems that are not really the charging system's fault.
Two connection paths must be checked, the Battery POS terminal to the Black wire connection at the VReg, and the Battery NEG terminal to the Green wire connection to the Vreg. A volt meter can measure these losses directly by placing a probe between the two identified points, Black path and then the green path. The numbers are summed and the error seen by the Vreg quantified. Anything over .5V loss is cause for concern and anything over 1V is a certain issue to be corrected. Each connector, terminal, fuse clip, or switch in the pathway can cause voltage reporting loss.
The regulator also passes the received voltage on to to the Alternator field coil to create a magnetic field within the alternator, The voltage level determines the strength of the magnetic field and the maximum output capability of the alternator. Therefore, starving the Vreg of true battery voltage leads to reduced max output capability of the alternator.
L - regulator operation/verification.
The Vreg sends voltage to the alternator field in response to measured voltage which is battery state of charge. Any voltage at the battery of less than 13.5V sends full black wire voltage to the alternator's white wire. The alternator output will vary with RPM, even if "told" to produce max power by the Vreg. If the alternator has enough RPM to overcome system load, any excess power is routed to the battery which will raise the battery voltage (slowly if depleted and rapidly if nearly full). When the battery reaches 14.5V, the regulator reduces the voltage to the alternator, reducing output strength and preventing battery overcharge. If the battery exceeds 14.7V, the regulator clamps the alternator field coil power to zero (0V), effective shutting off the alternator.
Because, there is electrical load from the system, an alternator that is not producing power allows the battery to deplete and the voltage falls. The Vreg responds by turning the alternator back on in accordance with battery state/ charge level.
The Vreg state changes can be monitored/verified by observing the battery voltage state, and the White wire to the alternator field. (Two meters are handy for this.) The "trip" voltages can be adjusted with the adjust screw, while changing engine RPM and electrical load that the bike presents to the battery/charging system to "make" the battery reach the voltage levels need for the set trip points. IE. with load reduced (lighting off) and the engine above 2500 RPM, a charged battery will attain 14.5 V. Anything above that and the adjust screw needs to be backed out to keep the battery safe from harm.
The shop manual outlines bench set up mechanical adjustments that should be performed on unknown or tampered units. These should be resolved before final trip point adjust tuning.
Note that while the system is working, the Vreg can change states rapidly before your very eyes., changing 5 times or more while you blink. Therefore, you may have to mentally average values measured on the White wire if your selected meter doesn't do that for you.
