Problems starting engine when hot - any troubleshooting tips?

[Art]

Murphy's Law - The sensor I forgot to mention...

Now you've noted DTC 23 in the stored (historic) DTC memory clear it and see if it reappears. If it doesn't reappear its safe to assume the oxygen sensor is good, don't rule the stored DTC may be left over from an old fault that has been fixed without the stored DTC being erased. It could also be an intermittent fault, only time will tell.
 
You don't need the Honda test harness to complete the remaining tests outlined in the workshop manual. The test harness is only used to read and clear DTC's which you now understand can be done using a jumper cable or paper clip. If you want to complete the tests use your multimeter to check the resistance between the oxygen sensor connector, ECM and ground:
oxygen sensor connector black/green cable to ECM (grey connector) pin 2, expect closed loop;
oxygen sensor connector black/green cable to ground, expect open loop.

Bottom line if you clear the DTC and it doesn't re-appear I'd assume the oxygen sensor is good.

Originally Posted by AKxx70

Tried to start engine - and it started just fine despite the engine being hot. Not sure if this was a coincidence.

If I was a gambling man I'd wager on a dirty connection, in disconnecting and reconnecting the oxygen sensor you've made a better connection between the terminals.

[AKxx70]

Quite possibly. The code does not re-appear and the engine seems to start fine cold and warm.
I started it cold and it was fine. Waited until it warmed up (fan kicked in) and turned it off. Went for a cuppa. Came back and the bike started fine again. I stopped and started it a few times - no issues (touch wood).

I have not taken the bike out yet as it is raining, cold and generally miserable today. Will test it properly later during the week.

And, by the way, thanks a lot, Art, for pointing me in the right direction!

[Art]

Happy days

[AKxx70]

The trouble with hot engine start has come back    But no error codes or anything. So I started to read about how to test O2 sensors and found some interesting videos.

There are two fairly crude methods to test operation of O2 sensors without having test computers - one is when the sensor is removed and the other with the sensor in-place.

Both methods require the sensor cable to be disconnected and external "12V" is supplied to the heater (two white wires). A DC voltmeter is used to monitor voltage generated by the sensor element itself (the other two wires). The voltage should change when the amount of oxygen around the sensor changes. The sensor should be hot and its voltage stabilised prior to starting the test (takes about 2-3 min)

Update: Just reading about Narrowband vs Wideband types of sensors and wondering if the above methods are applicable to wideband sensors or not?


Method 1: Sensor has been taken out from the exhaust. Use a burning candle in a jar covered with a lid. Wait until the candle burns out all oxygen in the jar and can no longer burn, then quickly put the sensor in the jar - the voltage should jump up due to the lack of oxygen in the jar. Take the sensor out of the jar and the voltage should drop back to the initial value in a few seconds.

Method 2:
When testing the sensor in-situ just twist the throttle and see if the sensor reacts with a change of its output voltage.

My test results were:
- Method 1: the sensor was generating about 0.95V and this output has not changed at all when the sensor was inserted in the jar.
- Method 2: the sensor was generating less voltage (due to heat dissipation, perhaps..) but still no reaction to throttle movements.

(Observation: when the "12V" is disconnected from the heater the sensor voltage slowly goes back to zero in less than a minute. The DC through the heater element was about 0.7A so the resistance of the heater element was about 12.5V / 0.7A = 18 Ohm approx.)

The sensor is fairly new (12K miles on the odo) and it did not look dirty. I sprayed carbs cleaner into the sensor's little holes and also left it to soak in the carbs cleaner overnight. Dried and re-tested in the morning and there was no difference whatsoever. Looks like it is a faulty sensor after all.

I fitted the O2 sensor eliminator plug as a test (which is just a 320 Ohm resistor simulating the sensor heater and an open circuit for the sensor element) and I saw no difference in engine operation, idle or when riding. Not sure what to think about the 320 Ohm resistor compared to the 18 Ohm resistance of the real heater.

The behaviour is now quite consistent: the engine starts almost instantly when cold. And it almost never starts straight away when hot, but usually requires one or two "dry" start attempts (with throttle fully open and fuel supply cut) before starting normally again.

I read somewhere that the ECU would ignore the O2 sensor readings during the engine start and would initially run in the open loop mode when idling until the engine gets hot. But what if the engine is already hot? Would the faulty O2 sensor (or the lack of it) make the mixture too rich and cause engine flooding?

And another observation: I read that the heater should heat the sensor up to 600F (300C) or more for narrowband or 1000F (500C) or more for wideband types. Mine did not feel that hot when heated because I could touch the sensor body near the cable end with my bare hand.

[Art]

Remote diagnosis is difficult and only as reliable as the information given, you appear to be over thinking this, the PGM-FI system checks the O2 sensor (connection, circuit & heater) so we don't have to. Having cleared the DTC's the absence of any returning current or stored DTC's together with the continued difficulty experienced in starting a warmed up engine would be good enough for me to assume the fault is nothing to do with the O2 sensor or any other PGM-FI sensor. Of course it could be an intermittent fault in which case it may return and can be dealt with at some time in the future. It could also be a missed fault, for example, it's all too easy to see an oxygen sensor DTC, replace the sensor and consider the fault to be fixed without any consideration given to the failure being premature, oxygen sensors should be good for 50 to 100,000 miles or more if they fail prematurely it could be due to an underlying fault which, obviously, would cause the replacement to fail prematurely too.

Time to look elsewhere and I'd be looking towards something such as an air intake issue which wouldn't, on its own, trigger the MIL. My prime suspects would include: failed intake duct valve; air duct leak; air box leak; clogged or dirty air filter, air hose leak, vacuum pipe leak and then some, none of which on their own would trigger the MIL.

Another suspect, as mentioned much earlier, could be the spark plugs. You're perfectly correct that 12,000 miles is well within NGK's stated service limit of 100,000 miles, however at 10+ years they should at least be given a visual inspection. The dilemma here, and exception to the 'if it's not broke don't fix rule' is having gone to the time and effort to take the spark plugs out then why not replace them?

or maybe...

Originally Posted by AKxx70

I replaced the original exhaust on my Mk2 with an aftermarket one, removed the servo motor and used a Healtech servo eliminator plug (Exhaust Servo Eliminator (ESE) - ESE-H02). Worked a treat.

or did it?

[AKxx70]

Thanks for your ideas Art. Yes, I agree that the root cause of fuel mix richness might not relate to the fuel injection system electronics but simply to the lack of air coming in for whatever reason. Or the spark plugs. Or as some suggest even the valve gaps.

Before taking this course of investigation I would still attempt one more test. When I had the O2 sensor disconnected and no eliminator fitted the MIL light was on. Yet the engine was starting fine when cold. And when hot too (as far as I remember, but this is what I need to double-check). If this is the case then there might be an explanation as I think there is one important difference between running the engine without the O2 sensor connected (MIL lit) and running it with the eliminator.

My logic is as following: if the sensor is not connected then the ECU knows about it and does not try to use the readings (0V) from the missing sensor - it will just use its own maps. But if the eliminator was connected it would trick the ECU into thinking the sensor was there and it will use the 0V reading as an indication of mixture being too lean (remember, the more oxygen - the less voltage) and would try to compensate by adding more fuel to the mix making it permanently too rich for the "closed loop".

I am going to run the engine without sensor and without the eliminator and see if the hot start improves or not. It is just a theory (because the ECU might consider 0V as an out-of-range value and ignore it if it has been programmed to do so) but it is worth checking.

As for the change of the exhaust - the issue was there before with the old exhaust and servo fitted. I did not notice any change in engine operation after fitting an aftermarket exhaust. The only change I noticed was the sound and the looks  And maybe a bit of weight saving. The MPG is the same too - approx. 11 miles per litre (Honda, why did you want us to practice x4.5 multiplication every time we ride the bike?!!!)

Will keep you posted.

[AKxx70]

Ok, it was not the O2 sensor at fault after all.

I was about to start opening the bike up to get to the plugs and check the sparks but wanted to familiarise myself with locations of all sensors first. I do not remember why but I tapped on the TP sensor with my fingers. Then the bike would not start when cold no matter what. It would not even start with my spare car battery. Spinned ok though.

So I removed the TPS to check the resistance and the way it changed when rotating - it looked ok. Put it back on the bike trying to make sure that it was at a similar angle. The engine started just fine. I ran the engine through a few hot/cold cycles - it started fine every time.

It was all fine for a couple of days. Happy days, I though. Then the same hot start problem came back.

I made a simple TPS voltage checking device using 9V battery and L7805 voltage regulator (gives a stable 4.98V output) and tested the sensor. Voltage changes were ok, no sudden jumps. I re-installed the TPS back and re-calibrated the idle output voltage to 0.5 V. The engine starts hot and cold ok in the garage so far.

I will be riding the bike for the next few days and see if this TPS is slowly failing.

(Ah, I forgot to mention that after fiddling with the TPS position I performed the "hearsay" procedure to "reset" learned TPS data in the ECU: disconnect battery for 10 min; reconnect and turn ignition on; slowly twist throttle all the way up and then back down again; turn ignition off. Start normally.)

[Art]

As far as I know if the ECM receives an out of range voltage from the oxygen sensor it will ignore it, similar to during a cold start and warm up except the MIL will come on. The engine will run more or less normally with the ECM using a pre-defined fuelling table and data received from the TP, ECT and CKP sensors to determine the correct fuelling etc.

If the ECM detects a DTC other than the oxygen sensor it may run as above on the pre-defined fuelling table and data received from the TP, ECT and CKP sensors or it may go into fail safe (limp home) mode and run on reduced RPM's and power or, if the DTC is critical the ECM will shut down (crank no start) not to be confused with a low battery voltage.

Therefore if there are no DTC's and the battery has tested good (>9.5v when cranking) the fault must be elsewhere, the assumption being the fault is not with the after market exhaust and servo eliminator.