ECM Hope?

[MidLife]

Not sure why your comment "that seems to have only happened to bikes coming your way" is directed towards me.

Because of this......

I've seen first hand, many melted knock sensor switches, wire harnesses, and melted body panels from excessive heat from riding with the light on and ignoring it.

I've otherwise never heard or read from a poster about his bike starting to flash 25/26 and then melting away.

With your help, I was able to snatch one of the very few new ECUs still available (Thank you very much!), but experienced nothing melting down at all while riding with my 25/26 MILluminated 1300 prior to the swap (only the standard 10% or so MPG drop reported by many).


One thing I can see though....is an overheating bike melting a sensor and triggering a "justified" (and fixable) 25/26........which is not the same as a 25/26 code melting the bike away.

 

[Andrew Shadow]

And most of the claimed fixed involved simply disabling the idiot light.
Which sounds to me like what cutting or disabling this wire would do.

I don't ignore warning lights, and sure don't tell folks to ignore them and keep operating their bikes until they sort it out.

I agree completely with your two statements above Larry. This is treating the symptom, not the cause.

I just don't think bypassing a warning light is a fix, and more of a band-aid than anything else.

I agree again.
However, if I understood and remember the video correctly, there was another scenario offered as a solution as well. The other option was adding an additional wire to completely separate the ground path circuit for the knock sensors from the ground path circuit of several other accessories that Honda lumped all together. Doing this very thing has been very successful in resolving many automotive problems attributable to what is often referred to as electrical noise, especially with audio systems. There is always the possibility that a weak ground circuit design is the cause of this problem as well. I have not had the time to look at the 2007 and earlier wiring diagrams vs the 2008 and later to see if Honda changed this ground path in any way, and if so what did they change.

I can not see how adding this additional and separate ground path for the knock sensor circuit can cause any damage as it simply shifts the ground path from one place to another. If this is done using proper electrical practices, it actually could provide a better ground path for the knock sensor circuit. While it may not cause any harm, whether or not it does any good is another question all together that needs to be vetted.

 

[W0QNX]

I can not see how adding this additional and separate ground path for the knock sensor circuit can cause any damage as it simply shifts the ground path from one place to another. If this is done using proper electrical practices, it actually could provide a better ground path for the knock sensor circuit. While it may not cause any harm, whether or not it does any good is another question all together that needs to be vetted.

I agree. That is the main solution of his fix on the first of the GL1800's. I've been reading and ciphering his research results as things allow.

I have looked over the before 07 ST wiring diagrams and the after 07 ST diagrams and we may be very similar in "bad design".

The main fix for the wings was changing where the shield wire on the knocker sensor wires are grounded. The ST circuits do show a difference in the knock sensor shield wire termination points (grounded connected at point on the bike) but i haven't seen where these actually are on the bike. The shields on both sensors connect to the G/P (green pink) ground wire but the G/P ground wire terminates at as different location on the diagram and bike. I don't know where these are yet physically on the ST.

His main finding is the knock sensors send a small voltage (under .25 volts) voltage created via a peizo electric vibration device (which is the knock sensor) to the ECM where an amplifier increase the voltage to large "spike" level so the ECM can see it easier. He also says the ground points being like and ST, all lumped into one causes stray RF to by in the ground circuits.

The main thing of his fix is he found that ECM's where "locking up" due to too much miscellaneous RF found in the ground circuits of the per 05 wings. He's changed where the ECM Logic ground terminate by a cut splice and new singular ground onto the engine head. This he believes reduces the stray RF found in the multi connect ground circuits.

The entire goldwing THREADS (more than one) and VIDEOS (3 so far) give up to 5 failure end points in the bad ECM's from "good but tripped" to "junk". He's getting some to reset with the new ground point.

I think this may be a possibility for the ST too since many of know the ST has a couple large multi connected ground points.

I'm tired of typing on I don't know the answer yet, this needs more research and thought so we can try this on the ST. It may be as simple as changing the knock sensor shield wire ground to a different location will recover some ECM's or lengthen the life of many with less RF noise in the circuits being amplified in the ECM circuits through a better ground point.

The different G/P ground wire points (images below) shown on the 03to07 diagram and the 07after diagram. I'm not even sure if this is where we need to concentrate but as of today I think it is the first thing that should be tried.

The G/P end point on 03to07:



07after diagram:

 

[Paul]

The green/ pink wire grounding point is under the rear of the main fuel tank on both series of bikes.

 

[Andrew Shadow]

His main finding is the knock sensors send a small voltage (under .25 volts) voltage created via a peizo electric vibration device (which is the knock sensor) to the ECM where an amplifier increase the voltage to large "spike" level so the ECM can see it easier.

Another good point.
The lower the voltage and amperage the more important the connections need to be with most electrical circuits. When the voltages are this low any increased resistance has an amplified effect. This leads to the possibility that adding an additional separate ground wire fixes the problem (if that is indeed the case) not because it separates the knock sensor from the rest of the grounds, but because it re-introduces an extremely good low resistance ground connection.

We already know that there is a poorly designed ground system in a couple of areas of the ST1300 wiring design. Maybe we should all go remove, clean, and reinstall our ground connections for the knock sensor circuit, if not all of them.

 

[SMSW]

Maybe we should all go remove, clean, and reinstall our ground connections for the knock sensor circuit, if not all of them.

I never understood the practice of grounding different pieces of equipment on a bike to the most convenient (read closest) part of the frame or engine. Yes, it might be cheaper, but over time some of them fail and tracking down the problem can take a while. It has always made more sense to me to run them back to one common point - like the ground bar in a breaker panel.

Yes, for the problem at hand with the ECU, all the grounds should be clean and tight.

 

[DannoXYZ]

Dark Sun pointed out multiple areas that may be problematic, starting with measuring resistance to chassis-ground of ground-wires at ECM connector. They were far from zero-ohms! I suspect Honda also threw in knock-sensor as add-on feature later and took some short-cuts. They broke some basic design rules by using chassis-ground for knock sensors.


First, unpowered sensors that generate their own signal have extremely weak output that's extremely sensitive to interference. One practice that helps them is to have their own dedicated "floating ground" at ECM, also known as "sensor ground". Not to be mixed-up or confused with "chassis ground" or "power ground". Honda did this properly with pretty much all sensors: O2-sensors, IAT, ECT, MAP & TPS sensors have their own dedicated sensor-ground at ECM (grn/org). This is most likely filtred before connecting to separate chassis-ground wire at ECM (grn/pnk). This filtering prevents spurious signals being dumped to chassis-ground elsewhere from creeping back into dedicated sensor-ground line.

Another example of floating common sensor ground (brn/E2) from Toyota Supra.

Knock-sensor is even more critical because its signal is lots lower than O2-sensor. Here's example of what knock-sensor sends to ECM:

It picks up LOTS of noise from engine: combustion, valve-train, etc. There typically is some serious filtering done on this signal to extract specific frequencies and amplitudes that's signature of detonation events. Specific frequency vary based upon bore-diameter of engine. It's so critical that Porsche has an entire KLR computer dedicated just to monitor knock-sensor and adjust timing. Notice that knock-sensor has its own dedicated signal, ground wires and shielding braid not shared with anything else! That's level of isolation needed to extract cleanest signal without interference.

Honda didn't do good job here with grounding knock-sensor to block that's shared with lots of other devices without any filtering. AND shielding-braid also shares a power-ground with other devices, not to dedicated floating-ground used by other sensors. So it's no wonder it gives up and throws permanent code.

 

[DannoXYZ]

I'm getting some ECUs in with code 25/26 issue. Here's my plan starting with my perfectly-working '06 with no code issues:

1. scope knock-sensor signals at ECM connector at idle and 4500rpm for "before" control measurement.

2. measure resistance to chassis-ground from ground-wires at my ECM connector. See if they line up with bad readings shown in GL1800 thread
2a. IF they're not close to zero-ohms, clean ground-wires at other end where they connect to bundles and chassis
2b. IF still not zero-ohms, add supplemental silicone tinned wires from ECM connector grounds to chassis-ground.
2c. move shielding-braid ground for knock-sensors to floating sensor common-ground of other sensors.
2d. verify if these mods changed or not changed bike's behavior in any way.

3. scope knock-sensor signals at ECM connector at idle and 4500rpm for "after" measurement to see if mods changed signal in any way.

4. test bad ECUs when they arrive on my bike. First, reset ECUs and clear codes:
4a. IF ECU stays clear with no codes above 4000rpms, then there very well may be wiring-differences between my bike and one ECU came from.
4b, IF ECU still throws code 25/26, it's obviously not wiring problem and is something internal to ECU
4c. sniff exhaust with wideband datalogging below and above 4000rpm failsafe trigger to detect any changes in AFR


We've had recent reports where it's wiring that caused this. Sub-harness between sensor and main-harness had rubbed something hot and got frayed and melted. Shorting out signal wire to shielding. In that case, replacing sub-harness took care of issue.

 

[MidLife]

Maybe we should all go remove, clean, and reinstall our ground connections for the knock sensor circuit....

I'd rather change it per Dark Sun, if doable on the ST?

 

[Mellow]

Let's not turn this personal. This should be a thread with constructive comments.

If you don't agree with what's being said, just ignore it or simply watch it. You know me, I have a way to deal with non-constructive comments and if it persists I will ban someone from any comments on it.

This is the first time we've see ANY positive comments on a possible approach to this issue that may lead to some type of solution. Maybe you like that solution, maybe you don't, but we'll let those brave enough and smart enough to dive into specifics give it a shot and hopefully come out with some new information.

 

[Chris09]

Let's not turn this personal. This should be a thread with constructive comments.

If you don't agree with what's being said, just ignore it or simply watch it. You know me, I have a way to deal with non-constructive comments and if it persists I will ban someone from any comments on it.

This is the first time we've see ANY positive comments on a possible approach to this issue that may lead to some type of solution. Maybe you like that solution, maybe you don't, but we'll let those brave enough and smart enough to dive into specifics give it a shot and hopefully come out with some new information.

point taken, my apologies.

 

[rdmnt2002]

I'm getting some ECUs in with code 25/26 issue. Here's my plan starting with my perfectly-working '06 with no code issues:

1. scope knock-sensor signals at ECM connector at idle and 4500rpm for "before" control measurement.

2. measure resistance to chassis-ground from ground-wires at my ECM connector. See if they line up with bad readings shown in GL1800 thread
2a. IF they're not close to zero-ohms, clean ground-wires at other end where they connect to bundles and chassis
2b. IF still not zero-ohms, add supplemental silicone tinned wires from ECM connector grounds to chassis-ground.
2c. move shielding-braid ground for knock-sensors to floating sensor common-ground of other sensors.
2d. verify if these mods changed or not changed bike's behavior in any way.

3. scope knock-sensor signals at ECM connector at idle and 4500rpm for "after" measurement to see if mods changed signal in any way.

4. test bad ECUs when they arrive on my bike. First, reset ECUs and clear codes:
4a. IF ECU stays clear with no codes above 4000rpms, then there very well may be wiring-differences between my bike and one ECU came from.
4b, IF ECU still throws code 25/26, it's obviously not wiring problem and is something internal to ECU
4c. sniff exhaust with wideband datalogging below and above 4000rpm failsafe trigger to detect any changes in AFR


We've had recent reports where it's wiring that caused this. Sub-harness between sensor and main-harness had rubbed something hot and got frayed and melted. Shorting out signal wire to shielding. In that case, replacing sub-harness took care of issue.

Anxiously standing by. Thanks in advance for your efforts DannoXYZ!!!

 

[RichKat]

Happy New Year guys
Am I correct that Honda changed the ECM in '08?
What did they change? I have a '11 are they affected by this issue?

 

[SMSW]

Happy New Year guys
Am I correct that Honda changed the ECM in '08?
What did they change? I have a '11 are they affected by this issue?

I have no idea what was changed internally, but the connectors for post '08 are larger and I think they have more terminals/connections. IIRC, from pictures in other threads, the post ECU's are physically larger than the early units. Failures in the new units ('08+) are relatively rare (based on few if any threads on that subject here).

 

[Igofar]

It’s not just pre 07 years, as I’ve seen several on new late model unit escort bikes.
But to be fair, these bikes always have a lot of jury rigged wiring stuff added on.
Let’s keep our fingers crossed.

 

[Andrew Shadow]

Am I correct that Honda changed the ECM in '08?
What did they change?

No one is really sure as Honda does not share information about their ECM's and their programming, which is why none of the ECM repair shops have come up with a repair for this. @jfheath has reported that he experiences less ignition knock with his 2009 vs his previous 2006(?). He has also reported having more torque available at very low RPM's. Both of these suggest different programming as a minimum.

I have a '11 are they affected by this issue?

The reported incidents of undiagnose-able code 25/26 failures is much higher with the 2007 and earlier model years. Whether that has anything to do with changes made by Honda in 2008 or not I don't think anyone knows this either. There have been 2008 and later model year failures reported by Larry, so if this is simply an age related failure that will eventually affect the 2008 and later model years to the same degree seems as yet to be determined.

 

[jfheath]

@jfheath has reported that he experiences less ignition knock with his 2009 vs his previous 2006(?). He has also reported having more torque available at very low RPM's. Both of these suggest different programming as a minimum.

To put some meat onto that. I had my 2006 from new, and I had a 2013 built A9 model from new.

The instant I got on the A9, it sounded 'raunchier'. A bit throaty, more of a noisy grunt. Not much more, but noticeable.

The A6 was much worse at climbing steep hairpins than my previous ST1100. I had to get my hand on the clutch lever before taking a very tight left hander. Forget counter steering, I'm talking about bends that you can only just get round. There was not enough power to drive the bike up the gradient and round the corner without slipping the clutch to keep the revs high enough to stop it from stalling.

By contrast, the A9 pulls up the same steep hill bends without having to worry about the clutch. It picks up the power from low down revs and just drives out of the bends. First gear stuff but the comparison is valid - a new ST1300A9 compared to an 70,000 mile 7 years old ST1300A6 - but the A6 had always behaved like that, and the engine was better at the end of my time with it than it was at the start - partly because I had balanced the starter valves, partly because the A6 pinked throughout its life when hot and when under heavy acceleration. Just a feint tinkling. The A9 has never pinked and there has never been any signe that the mixture is too lean.

In Summer, in Spain on high mountain passes, the engine would get too hot - it would reach a point where it sounded different as if the motor was tightening up and struggling - I thought with lubrication - as well as the pinking. I wondered if the oil was too thin at those temperatures. So we'd find a view and let it rest for a while. Then it was ok. Again, I was thinking 'too lean'

The A6 occasionally displayed an odd phenomenon. Usually not long after filling up - but not always. After cruising along without much throttle activity, a few times, I would aenter a bend or roundabout where I had slowed down, gone through the gears and relied on the power to lift me out of my lean out of the bend. It wasn't there. It wouldn't accelerate. I'd have to pull the clutch, blip the throttle a couple of times and then it seemed to wake up. A friend - who was high up in Honda UK and had lots of friends in the technical side - asked them about it. They told him that thye ECU changes engine modes accordding to how it is being used, and if you are coasting it takes a couple of throttle blips to make it wake up and switch into more responsive mode. Whether that is correct or not, I don't know, but I always blipped the throttle during slodown if I was wanting to pull away again.

The A9 has never done anything like that.

If I was to buy another, I wouldn't be buying one earlier than an 08 - mainly for those reasons.

 

[MidLife]

Here's example of what knock-sensor sends to ECM:

And that's even without any knocking, right?

Honda didn't do a good job here with grounding knock-sensor to block that's shared with lots of other devices.....

May also explain why @Igofar is observing more 25/26 codes on hyper-farkled bikes with even more interfering grounded devices.

Do pre and post 08 bikes have different Knock Sensing grounding?

 

[DannoXYZ]

And that's even without any knocking, right?

May also explain why @Igofar is observing more 25/26 codes on hyper-farkled bikes with even more interfering grounded devices.

Do pre and post 08 bikes have different Knock Sensing grounding?

Looking at wiring-diagram, ‘08+ has exactly same knock-sensor configuration. ECU has different pin-outs with 2 additional lines, but I haven’t looked too closely at what’s different. So yes, we may be seeing more reports of code 25/26 problems in coming years from ‘08+ bikes.

Yup, those are raw waveforms from knock-sensor. Filtering and ADC has to occur before ECU gets possible candidates of knock events. Here’s some filtre circuits used with AEM and Microsquirt programmable EFI systems.

You can see it filtres out high/low and has bandpass on specific knock signature. On Porsche Turbo’s KLR knock-computer, it extracts signals centred around 6.52khz based upon bore of cylinders (errors in docs below saying 5.7khz). Here’s what final filtred knock-sensor signal looks like, lowest teal trace below:

With single knock-sensor, it actually monitors knock events and retards ignition on per-cylinder basis. In addition to extra ignition retard on each engine-rev, it’ll also lower boost to reduce BMEP. Then throw MIL on dash wih 2-2 code when knock is detected along with going HIGH on knock-output=yes/no terminal for external knock countres. There’s also other codes for sensor, wiring, boost-controller failures as well as issues internal to computer itself.

Info on Porsche’s hardware here: KLR - knock detection hardware
And step-through of algorithm. KLR - knock detection software

In factory trim, I was never been able to trigger knock-protection using 87-PON/AKI petrol, system was programmed with wide safety margin. Finally got some knocking by advancing ignition 5-degrees under WOT. Using 91-PON then cleared that up. Almost doubling factory boost levels from 10psi to 18psi required using 50/50 mix with E85 for ~100-PON. Got +100hp bump from just software changes alone!

This knock-detection & mitigation algorithm was actually copied and released as standalone J&S Safeguard a decade later.

 

[DannoXYZ]

NOTE: Detonation causes knock/pinging sounds and always occur after spark-ignition near TDC. Increasing heat and pressure as combustion progresses causes formation of radical akyl groups. These are extremely unstable compounds and will spontaneously ignite/detonate on their own, forming 2nd wave-front opposing initial one. Their collision generates intense pressure and hot-spots which creates knocking/pinging sounds. It’s literally tapping on piston with tiny hammer!

Higher-octane petrol has higher percentage of more stable aromatic (benzene ring) hydrocarbons that resists radical akyl formation under intense heat and pressure of combustion better than straight-chain HC.

Pre-ignition is completely different phenomenon and occurs near BDC when mixtures are easiest to ignite. This causes combustion to push down on piston that’s being moved up on compression stroke by crank. Engines typically don’t last more than couple revolutions after pre-ignition event.

"Damage from pre-ignition is much more severe and instantaneous than that from detonation. Typically, with pre-ignition, you will see holes melted in pistons, spark plugs melted away, and engine failure happens pretty much immediately. "

pre-detonation = no such thing

Good article: https://blog.edgeautosport.com/detonation-vs-pre-ignition