This is rather long, so please feel free to ignore if you're not interested.
I’ve had a problem with starting my Yanmar 2GM20F for many years now (maybe 12-15?). Basically, it always starts the first time at the dock. But after sailing with the engine off, it might start on the 1st time, but it might take 2-4 tries to turn the engine over. It's never failed entirely. I’ve been aware of info on this problem on the forum for many years, but it hadn’t risen to the top of my boat to-do list until now.
So today I took my computer (with chatGPT) and the fancy Fluke multimeter I inherited some years ago down to the boat, and asked one what to do with the other. The obvious answer was “you have the same problem everyone else has already described in detail”, but I still needed to trace the wiring and figure out if I wanted to do the relay fix or the new wire fix, so I did some tests.
Several mildly sweaty hours later, I have some interesting data on the problem. First, I seem to have the A-type, new-type Yanmar panel for my 2GM20F, based on my engine serial number and the Yanmar manual. This tells me which wiring diagram to use. I found the starter, and traced the wires as well as I could. The starter has 2 large lugs: one is hard wired to the starter with bare wire, and the other has a heavy gauge red lead from the battery switch (presumably) and a couple of smaller gauge red wires on it. There is also a small lug with a small gauge white wire on it. This seems to agree with the wiring diagram;
View attachment 239702 View attachment 239703 View attachment 239704
The red terminal was hot, and the white seemed grounded. ChatGPT told me that "The large lug with the huge red battery cable is the B+ terminal. It should have approximately battery voltage all the time that the battery switch is on. The small red wires attached there are using that same terminal as a convenient source of unswitched battery +12 V; one of those feeds the engine/instrument-panel circuitry."
It advised me to check the voltage between the B+ terminal of the starter, and the smaller white wire terminal while trying to crank the engine. The idea was that if the difference between the 2 was large, this would be evidence of a loss of voltage to the white wire that energizes the solenoid. It has me use the min/max feature of the Fluke meter to measure this voltage drop. My first test gave:
initial voltage = 12.82V
first crank = min voltage of 1.52
2nd crank = min voltage of 5.62
3rd grand = min voltage of 5.58
So in theory, this said I was losing quite a lot of voltage when cranking.
GPT suggested that this might not be the best test b/c there's the initial cranking voltage, and then the engine catches... anyway, it didn't trust the readings. So it suggested pulling the fuel stop cable (so the engine couldn't catch) and cranking for 2-3 seconds so that the meter would have time to settle. And also measuring white wire to engine block ground, instead of white wire to red starter terminal. This time, I was lucky enough to replicate the "no crank" condition. My measurements white-to-ground were:
1st crank = 8.6V. The engine turned over normally.
2nd crank = 6.2V. The engine did not turn over, so the solenoid didn't get thrown.
3rd crank = 6.2V. The engine did not turn over.
In other words, 8.6V was just barely enough voltage to pull the solenoid in, but 6.2 wasn't. I then tried the B+ starter terminal to ground:
at rest = 12.68 volts
several cranking attempts = 10.2 volts, but it cranked each time, so I was unable to replicate the no-crank situation.
Even though this seemed to confirm that the starting circuit (small red wire to panel key switch to white wire) had excessive voltage drop, chatGPT wanted me to do more tests "for intellectual curiosity" (who knew?). I played along to a point: I did try to measure the voltage drop across the red-wire inline fuse near the starter while cranking. It was essentially 0V (and it looked totally clean and un-corroded). I couldn't find any way (without stripping insulation) to measure the voltage drop across the 2 connectors for the wiring harness (both ends of the "3m extension cable"), so I didn't follow instructions on that test.
Since I was still trying to decide whether or not to do the new wire solution, or the relay solution, I managed to get to the back of the Yanmar engine panel, and tried voltage drop across the key switch. It was 0V while cranking, which means it wasn't the problem. At first, I was confused to find another fuse next to the panel (moderately corroded, actually), but quickly figured out that it was a fuse to the power supply (going off the key switch) to the engine fire suppression circuit. It had nothing to do with starting, and in fact, the engine cranked even with this fuse removed.
One interesting test: engine block to red wire on the start key switch was 12.67 at rest, 12.0 with the key on (buzzer sounding), and 9.3 while cranking the engine. ChatGPT seemed to think (and I agree) that losing 0.67 volts just to turn the engine panel on seems excessive. So it tends to imply some serious voltage loss in the start circuit.
Anyway, given that I managed to find the wire chase leading from the engine up to the panel, and I got the panel off for access to the back side, I'm leaning towards just doing a run of 10 gauge tinned marine wire, red and white, from the starter to the key switch on the panel. Plus a fuse holder in the red wire run, of course. To me, this seems like less of a kludge than doing the relay for a relay. (No offense intended to those that chose this route. I'm not proud: I'd do the relay if I thought it was easier.) Thoughts and comments are welcome.
