knock sensor and power/mileage

[mikesail]

I have been concerned that my bike has a minor problem with its motor. I have had it since new, and 21K miles of riding have consistently returned about 31 mpg, regardless of how I ride. It appears from casual discussions that its power is a little low, as with stock gearing it just barely will do a power wheelie in first and in second gear absolutely will not loft the wheel regardless of how hard I rev and dump the clutch.

My belief is that a bird normally gets mileage in the high 30's, many have reported over 40 when cruising. It appears that the bike should wheelie in second if pushed hard.

I have checked most of the engine sensors, but do not have the knowledge to properly assess the knock sensor. It seems that a false knock indication would force timing to maximum retard and give symptoms such as I am having, i.e. low mileage and slight loss of power.

Believeing that the knock sensor might explain this, I tried first disconnecting it. No good, the FI light gives the code for bad knock sensor, and guess what, the bike runs exactly the same. I believe this indicates that my bike has been running in the same limp home mode that it would with a bad sensor.

I next looked at the sensor with an oscilloscope on my bench, and gave it a tap. I have no comparison to make with another good sensor, but I have to believe that the 200 mV or so that came from a modest tap is excessive output. Also the waveform was not a pure tone as expected from a resonant piezo element.

Since the bike needs the sensor connected, and I now believe that the sensor is way too sensitive, I reconnected the sensor electrically to the bike, but mechanically isolated it with a shop rag wrapped around it. The bike still seemed about the same, and then after about 50 miles the FI light came on again. Every time I restarted the bike it would be off for a few minutes then come on again. So it appears that the ECU wants to see some minimal level of activity or it flags a fault.

The next experiment was to install the sensor back in the block, but I put a fat O-ring on the 12mm mounting stud it has. the sensor is then just hand tightened. I was hoping that this softer mount would decrease the level of vibration getting to the sensor.

I have now gone thru three tanks of gas with this setup. The FI light has not come on. The bike will wheelie much easier in first gear, and now can get a slight wheelie in second by dumping the clutch. Clearly the torque is greater now. My last tankfill returned 35 mpg, sometning that I have never remotely approached before. There is no doubt that this has changed things for the better.

I don't believe that my "fix' is at all adequate, and a new sensor should improve matters further. What this does suggest to me is that some of the variability that other have reported in their mileage might well be associated with a similar issue. I cannot think of a good way to tell what the timing is doing while riding, but a dyno and timing light could show what is happening.

After my experience others who have interest might try a similar trick to dampen the sensor and see if there is any change.

I would be very interested if someone would be willing to disconnect their sensor for one tank and report on the change in power and mileage, that would pretty much nail this issue down.

Next item is to either build an attenuator for the sensor connection or just get a replacement.

[beondwacko]

Interesting Idea putting a resistor inline with the knock sensor to get more total advance from the system. One thing you never mentioned was the octane of fuel that you regularly use. One of the first things that I would look into would be a dyno session so that you have a baseline HP number. I also admire your creativity of doing the O-ring as an isolator. Pretty cool. I believe that the factory ECU follows a pre programmed advance curve at different throttle positions and other inputs from temp' and baro sensors. When it detects a ping as we know it will pull out a degree or two of total timing untill it no longer hears the pinging. Standard fare. Another thing you didn't mention is your location and altitude. If your running up in the Mtns that would explain the lack of power. Also, stock exhaust ? Ait filter ? Power commander? Any other changes from stock ?

[Canadian Bird]

Why do you want to "wheely" your Bird? :icon_duh:

[beondwacko]

Why do you want to "wheely" your Bird? :icon_duh:

I know I do ,,,, but not at the drag strip of course.

[mikesail]

Why do you want to "wheely" your Bird? :icon_duh:

I know I do ,,,, but not at the drag strip of course.

Why? because it can

guess I haven't grown up enough

to the other serious questions:

I don't feel like paying for a dyno, especially if I am not able to run my own experiments. Thus my mention of the ability to wheelie, as this precisely indicates a level of torque related to my weight and body position. Sort of a poor man's dyno for me.

I did just add a pc3 last month. Made a minor improvement in mileage, about 1 better.

Otherwise the bike is stock, I am at sea level in so cal. I normally run regular, but the few times I ran 91 octane there was no change.

I suspect the attenuator would have to be a shunt load and capacitively coupled. A series resistor would not affect the signal since a piezo transducer is high impedence.

[JohnA]

This is my list of actions, moving down if the problem persists:

1. untighten and retighten the sensor to the correct torque (it does affect the output)

2. try adding a washer to decouple the sensor a bit

3. change sensor for a new one

4. Carefully inspect the wiring until the ECU

5. Ultrasonically clean injectors

6. fill tank with race fuel --- if it still persists, then swap ECU for a know good one

[mikesail]

This is my list of actions, moving down if the problem persists:

1. untighten and retighten the sensor to the correct torque (it does affect the output)

2. try adding a washer to decouple the sensor a bit

3. change sensor for a new one

4. Carefully inspect the wiring until the ECU

5. Ultrasonically clean injectors

6. fill tank with race fuel --- if it still persists, then swap ECU for a know good one

John, nice job on your website. Found some good references there as well.

I did use an O-ring to decouple the sensor, that is my major point of the post.

My real interest here is in quantifying what is going on. I don't want to just change things until it works. In particular I would like to calibrate knock sensor levels. I just have no idea of what the waveforms are like and the amplitudes involved.

I could put a scope to the bike while parked, but cannot collect data while riding with anything i own.

[JohnA]

John, nice job on your website. Found some good references there as well.

Thanks

Unfortunately I haven't got enough time to spend on it, so it is kinda stale at the moment.

. In particular I would like to calibrate knock sensor levels. I just have no idea of what the waveforms are like and the amplitudes involved.

You have to understand that the sensor doesn't pick up 'knock', despite the name.

It's just a microphone, picking up noise, vibration etc.

It is the filtering software that decides if that shape of noise at those revs and such load can be regarded as possible knock.

Every block is different, the resonance frequencies are different, and the ECU filtering code reflects that. It's not an exact science either.

Sometimes interference or dodgy connections lead the ECU into thinking that it should go into 'safe' mode. Nothing to do with the sensor itself.

I'd use some race fuel (50/50 is more than enough) and see if the problems persist. If they do, then you know that there is not really knock, just an electrical fault (you never know, maybe you really do have knock!)

[mikesail]

John, nice job on your website. Found some good references there as well.

Thanks

Unfortunately I haven't got enough time to spend on it, so it is kinda stale at the moment.

. In particular I would like to calibrate knock sensor levels. I just have no idea of what the waveforms are like and the amplitudes involved.

You have to understand that the sensor doesn't pick up 'knock', despite the name.

It's just a microphone, picking up noise, vibration etc.

It is the filtering software that decides if that shape of noise at those revs and such load can be regarded as possible knock.

Every block is different, the resonance frequencies are different, and the ECU filtering code reflects that. It's not an exact science either.

Sometimes interference or dodgy connections lead the ECU into thinking that it should go into 'safe' mode. Nothing to do with the sensor itself.

I'd use some race fuel (50/50 is more than enough) and see if the problems persist. If they do, then you know that there is not really knock, just an electrical fault (you never know, maybe you really do have knock!)

john, thanks for helping

It is hard to believe that the bike is knocking. But I could be wrong :icon_doh:

The plugs look good, and the bike has performed consistantly since new. It doesn't care a whit about what grade of gas I use

I understand exactly about the ecu filtering. My profession is circuit designer, although of the analog sort. I believe the wiring is sound, as the unit will detect the absence of the sensor immediately.

My understanding is that the sensor is a piezoelectric crystal which is set to be resonant at the natural ringing frequency of the block or possibly the piston. If it is not resonant then of course the ecu would have to do all filtering. but by making the sensor resonant one gets substantial prefiltering which makes life easier for the electronics. If it is not resonant then they would be generic parts like an o2 sensor, right?

The reason I mentioned tapping on the sensor while observing with a scope was to allow someone to tell me what they believe the relative sensitivity of such a unit is. my guess is that my unit is improperly potted and thus bangs around internally, causing extraneous output. But until I either get a new unit to compare or someone helps me with a comparison of their unit I cannot tell what I have going on with certainty.

Haven't managed to find any scientific papers on this topic and I am not a member of the SAE. I do enjoy conversing and sharing knowledge with others who share these interests

[JohnA]

I understand exactly about the ecu filtering. My profession is circuit designer, although of the analog sort.

There you go

If it is not resonant then they would be generic parts like an o2 sensor, right?

I haven't looked at the Honda sensor from upclose, but normally they *are* generic.

Haven't managed to find any scientific papers on this topic and I am not a member of the SAE. I do enjoy conversing and sharing knowledge with others who share these interests

I occasionaly buy SAE papers if I find them interesting.

For more advanced engineering discussions, you may like this forum:

http://www.eng-tips.com/threadminder.cfm?pid=71