Ion knock detection sensitivity

[BVHOG]

Is there a difference between calibrations regarding the knock sensitivity? If so could this potentially be adjusted by a reference table?

[wurk_truk]

Pretty cool question.  The ION sensing itself is built into the coil... no change there possible.  But, it should be all software driven from that point on HOW and WHEN it polls the sensor and also what amount is the ECM supposed to disregard.  I would like it...  maybe... at 100% at all times, but I don't know if there is software in there to disregard known bad reads?  And... ARE there known bad reads of this?

[cts1950]

I have been chasing knock for several years. On both of my bikes a 103 and a 95 The ecm shows fast retard events yet I can not hear them. Even with knock retard turned off or even set to .25 degree retard just to see if the events are recorded. I have paid attention to keeping stock wires and spark plugs in them both. I spoke to Springer once on the phone and we concluded I was chasing phantoms. He said that they see a lot of Fat and Retarded bikes come into his shop for tuning. Most people enrich the fuel mix and retard the timing to try to get rid of the perceived knock.  I wish there was a way to confirm if it really worked. When I was running the DTT ecm knock was easy to hear.

[Don D]

It is an interesting process. I would think our modified motors would have a different feedback profile to the ECU therefore need to be modified through the software. That could mean changing the expected values the ECU sees and or the time the samples are taken. I would like to see the ION sense circuit divorced from the coil and then we could add performance coils to our beloved motors. When modified in high compression states and under the demand of higher RPMs I can only think that more avalable voltage to the spark plug tip would allow us to open rather than close the gaps and power would rise, not a windfall but a small amount anyway. Would make sense also that if the spark plug gap was not set to stock value the ION feedback to the ECU would be altered. It is a bit like long and short arc welding.

[lonewolf]

Here's part of a PM from FBBR on ion sensing when I questioned him on "false knock retard". Pretty smart guy.

"What most folks fail to realize is ION sensing is really "model" based!! A long time ago I had a sticky(maybe in the OLD section when my "name was fatbyrr" ) about how ION sensing works. But here goes!

It is not a sensor like the piezo sensor. (even those knock sensor have to calibrated for the engine they are on, to "see" knock.)
But the “ION” sensing element that is in the coil does not really "sense" ionization at the plug. What it senses is resistance to the spark event that is sent across the gap. And as most folks know, the requirements increase as Ionization happens at the plug. But the important thing is "at what sensor value" is IONIZATION ( or the beginning of detonation) occurring? The way we know, is when we calibrate the system, it is with real COMBUSTION equipment on the engine. Combustion equipment along with a very fast and LARGE memory computer, can predict many combustion events. Some of the things Combustion (chamber installed very sensitive pressure transducers) equipment can measure are, MASS fraction burn for every cylinder event. (That is the percent of fresh charge of fuel burned for EVERY degree of crank rotation) Location of Peak pressure
( good for finding best spark at every speed and load) as well as pressure vs. crank angle degrees! So we get a real PV (pressure volume diagram) for EACH firing event for EVERY cylinder as the engine is running. ( a lot of computer memory!). As you know at the onset of detonation, it is obvious from the rise in cylinder pressures at the END of combustion that detonation is either occurring or just the very beginning of detonation has started! It is that pressure spike we induce by adding spark while the engine is running to find the start of detonation.
That is known from the pressure rise recorded by the combustion equipment! It is then only a matter of logging the output from the coil sensor and correlating the pressure rise (from start of detonation) to the value at the sensor. That is done at in a table! As the values change with speed and load! So the table for ION detection is calibrated using that equipment and as such is really model based. When you change the engine size or combustion chamber or “camshaft timing” those values might in some areas be close enough to function, but it doesn’t surprise me that you would either get “false” retard, or failure to detect real detonation under some speed and loads!

As I said even when a real “piezio” knock sensor is used, we also calibrate the signal
( from the sensor) vs. engine noise for a given application. And sometimes there are areas where we simply “turn off” the knock retard ( at very specific speed and loads) because the sensor cannot differentiate between the signal and “normal” engine vibration noise!
An ION sensing system is much better for OVERALL ability, and can sense the ONSET of detonation much faster ( at lower levels than the old sensors), so the ION sensing is still much better than the old “real knock” sensor system! Especially for a HARLEY!"

[cts1950]

If the the knock retard is in a look up table, is this something that a ECM programming tool such as TTS be able to access if the code were written to get to it? It seems to be a common problem on modified engines. We can adjust out how much timing the retard function takes out but not the sensitivity of the system. I would think it possible to create a module to go between the coil ion sense lead and ECM to modify the signal but beyond my skill level.

[Don D]

That is a question for Steve Cole
In the past we have seen pretty strong pushback when asking about such things but really we are handing TTS golden goose eggs to keep their product frontline and competitive, IMHO. That said many of the issues we bring up end up with fixes, albeit titled things that are not necessarily exact but they are nothing more than software switches and adjustments and they correct the issues so who cares.
Here is some information that details the true potential of ION sensing. The HD only scratches the surface with a "stop gap" rather then adaptive aproach, once again IMHO.

http://www.max-boost.co.uk/max-boost/internet_articles/Spark-Advance%20Control%20by%20Ion-Sensing%20and%20Interpretation.htm

[cts1950]

[http://www.max-boost.co.uk/max-boost/internet_articles/Spark-Advance%20Control%20by%20Ion-Sensing%20and%20Interpretation.htm
[/quote]

That was a good read thanks for the link. There is a lot more to this than meets the eye. When we change a 88 inch motor to a 95 or a 103 that must throw the look up chart entirely off with the increase in compression ratio.  Changing the breathing with low restriction air cleaners and pipes also must throw a wrench into equation.

[wurk_truk]

The actual problem lies in that the ION sensing DOES trigger adaptive strategies.  The Delphi uses two distinct sections to deal with ION events.  This has been discussed quite a bit here on HTT, too.   There is adaptive... that once an event is seen, the timing stays pulled.  This decays after a bit unless ION senses another event, and one can SEE the timing staying altered even when no ION events are present (doing data runs).   And there is also fast retard.  Fast retard is what is seen in TTS data graphs as ION events

Some of us feel, from playing around, of which DON says to NEVER do.........  if we zero out the adaptive tables, then all we are left with is Fast Retard.  Tone down the Fast retard table and the timing events will still have a little help, but won't pull massive amounts of timing anymore and also WILL NOT REMEMBER doing so.  I wouldn't recommend killing all the pull out of fast retard.  Bad gas would mean one needs a bit of protection.  But THAT can be set like 2-4 degrees instead of the 8* now.  But, ultimately that is a work around and not a fix.

Something to alter sensitivity would be nice.  But I cannot help but feel that the base cals take this into consideration somewhat.  A ST1  96" cal will be more sensitive than a St1 110 cal would be, will be more sensitive than a 113 built hipo cal.  One CAN see SOMETHING of that nature by simply taking a cal that triggers ION pulling timing a bunch with another cal... and then minimum ION events are seen.  Yes... that can be from fueling strategies are different, but still....

[HogMike]

Interesting question. Far too much for me to worry about, though.
It was suggested to me to leave the adaptive knock retard to 0 on the last 3 tunes I have had my tuner do for me. He likes to let the fast retard do the trick. We set max retard to 5.
I am playing around with my timing just to see how far I can go without knock issues creeping in.
On this trip, I am using adaptive knock retard, but, scaled back on how much. I find anything from 86 to 92 octane, just can't always get what you want!
I'm just about in my "happy place" as far as MY combination.
I'm STILL not sure how the adaptive will "reset" after a key off. From what I've read, your "learned" values will reset to base on a key off/on. This is to put timing back in on, say, a "good" load of gas in your next tankfull.
JMO, so, NO bashing!

[Don D]

I love to play around but try to stay away from sink holes. 

I have not become so forgetful to not remember how nice a carbed early Twin cam ran. They did and still do. They do not have adaptive anything and the only real electronic control is the MAP controlled spark. When set right (mechanical parameters) with proper fuel and spark timing they run every day and do not ping at all, uneven AFR and all. We technologists need to keep a look back sometimes and not loose track of the goal post ahead. Unless this bike of ours is just a masochistic play toy. Footnote: I AM NOT ANTI EFI

The Delphi ION sensing in it's current state and with the tuning tools offered is not an effective feature, simple as that. Why psychoanalyze it so deeply. Until we have the tools to make it work better then it remains an excellent feature for a stock or stage 1 motorcycle and with the limited controls all we can do is dumb it down not effectively use it nor can we trust the feedback in the logs. Too many false hits. Then add another layer the "PE" and it's timing scheme.
This is just a rerun of the "EGR" stuff. Steve, to his credit, fixed that

[wurk_truk]

Way upper left.  You don't see what we see in the flatlands of the middle.  Corn as far as the eyes can see.  Farmers no longer raise feed... they raise gasohol.  Thats the truth, actually... 40% corn is for fuel arounds these parts.

Except for Shell gas... corn is here to stay.  And ION could/can be a help with that.  For right now, without any tools to speak of... turn off adaptive and pick a tune that senses falsies the best.  No matters WHOSE base maps, they were all ultimately tuned in by the MOCO.  There ARE some base cals better than others on this issue.

Gasoline quality sucks here.  Tunes being done at 89 here to cover it up a bit.  Mega compression on the streets is dead.  1970's all over again.  Mid 180s ccp is tops.  and then a GOOD tune is needed.  Stoich in the Midwest for pump fuel is 14.2AFR. 

[Don D]

Other than a personal attack or two and a rank or two on the EMS nothing you said addresses the accuracy of the Delphi to proportion timing correctly.

[wurk_truk]

Touche on MY post then, huh Don... offer 'solutions' that have already been dealt with and call out personal attacks on me...  I will NOT suffer that... even from the GOD of all things Harley.  I KNOW I am a dumbass Don, but NOT on everything.

As far as the ION thing...  if too many hits are being seen, the thing those of us HERE see as a work around is to kill the adaptive knock control (nobody, in the 'know' has stated if the adaptive clears at each on-off of the key) and also try different base cals.  Some show WAY less activity than others do.  A sensitivity thing would be nice, but HIGHLY doubt it is in the works anytime soon.  Drinking Kool Aid HAS its advantages.  The Delphi does NOT proportion timing correctly, once totally stock config is left in the dust..  There IS no fix except for work arounds at present.

Swapping around base cals seems to work.  Especially when trying out cals that are for higher comp/larger CI engines.  Dial the CI in the constants to match build.  False 'ping' sure makes bike run crappy.

Actually in 100% agreement on two things... old bikes dealt with it ok (universally better gas then), and 2.. most will NOT invest the time to v-tune multiple base cals.  Tweakers will, tho

And me?  at the suggestion to kill it out completely?  Like an older bike?  Ummmmm... not me, anyways.  Fuel in Ohio and WVa lends me to LIKE a degree or two being pulled at times.  8*, no way.  I also do NOT want the bike to 'adapt' to ping or false ping... then the timing is pulled for a very long time.  I will tune at 89 AND  I will let fast retard do the heavy lifting after that.  But, THAT is the DUMB HILLBILLY in me, though.

[HogMike]

Touche on MY post then, huh Don... offer 'solutions' that have already been dealt with and call out personal attacks on me...  I will NOT suffer that... even from the GOD of all things Harley.  I KNOW I am a dumbass Don, but NOT on everything.

As far as the ION thing...  if too many hits are being seen, the thing those of us HERE see as a work around is to kill the adaptive knock control (nobody, in the 'know' has stated if the adaptive clears at each on-off of the key)

"The Delphi ECM utilizes Adaptive Spark Control based on information received from the knock detection system. This system learns a retard value to apply when knock is detected, and this value is retained between key-on cycles. However, at each key-on the remembered values will be reduced towards zero. This gradually clears out the learned knock adapt value and
serves to adapt to a change in conditions, such as a tank of low octane fuel being replaced with better fuel.
The Adaptive Knock Retard table determines the maximum amount of learned spark retard that can be applied, based on RPM and MAP. When the spark tables are being developed, this table should be set to zero degrees to prevent influencing the values. After tuning is complete, restore the table to the factory values or set the table to the values appropriate for your engine combination."
Figure 3-
?

and also try different base cals.  Some show WAY less activity than others do.  A sensitivity thing would be nice, but HIGHLY doubt it is in the works anytime soon.  Drinking Kool Aid HAS its advantages.  The Delphi does NOT proportion timing correctly, once totally stock config is left in the dust..  There IS no fix except for work arounds at present.

Swapping around base cals seems to work.  Especially when trying out cals that are for higher comp/larger CI engines.  Dial the CI in the constants to match build.  False 'ping' sure makes bike run crappy.

Actually in 100% agreement on two things... old bikes dealt with it ok (universally better gas then), and 2.. most will NOT invest the time to v-tune multiple base cals.  Tweakers will, tho

And me?  at the suggestion to kill it out completely?  Like an older bike?  Ummmmm... not me, anyways.  Fuel in Ohio and WVa lends me to LIKE a degree or two being pulled at times.  8*, no way.  I also do NOT want the bike to 'adapt' to ping or false ping... then the timing is pulled for a very long time.  I will tune at 89 AND  I will let fast retard do the heavy lifting after that.  But, THAT is the DUMB HILLBILLY in me, though.

THATS the clarification I'm looking for!
DOES is decay to zero? How soon? Key off?

[wurk_truk]

Mike, a couple of threads ago, we asked the same thing and didn't get a really satisfactory answer, for the adaptive decay rate.  This is the reason, simply for me only, to give a try to using only fast retard.  I do NOT know if, as a strategy, this will work, but I WILL give it a go.  I CAN see both sides of using adaptive on this.  If adaptive is being used, maybe only... say 3* of timing is pulled, but for quite a distance, while using fast retard only... THAT could pull MORE timing, but in bursts.  I surely know what that 'tink' sounds like, so MY thought is fast retard and only have available a few degrees of timing being pulled.  If I, do in fact hear a tink or two, I will data that area and go looking.

I full heartedly believe whats posted here about false events.

[HogMike]

I would think, reading the explanation of this, that you would see back to zero as quickly as you could fill your tank.
It could be me,, but on this trip it "seems" that my bike will run nicer, after a change from crap gas to good stuff, all other conditions remaining the same. I HAVE noticed more instances of ping when this feature is disabled (set to zero).
Just my limited experience in 2 different touring bikes.

[wurk_truk]

I HAVE noticed more instances of ping when this feature is disabled (set to zero).
Just my limited experience in 2 different touring bikes.

Good info  Did you zero out all the settings in Adaptive Knock control, or go to constants and turn off knock?  What I am thinking is leave knock turned on in constants, but zero out all the values in adaptive knock control.  As you know, I asked around before.  The Wizard, a GOOD member in standing here, not like me at all., told me this:

You may be REAL old but you have a good memory.

Back, about a century ago, Steve George (Fullsac.com) and I were having a discussion as to why he left his Adaptive Knock values at ZERO.... for those of you reading this that don't understand what this table does it's pretty simple...

The Adaptive Knock Retard Table lets you preset what the ECM will REMEMBER and STORE when the ECM sees an ION sensitive motor knock. Your value in the table is the number it will remember to reduce your timing so this motor knock will not happen again during THIS RIDE.

The Adaptive Knock Retard is not a permanent record such as the Adaptive Learning process for the VE tables as when you do a simple muffler modification and the ECM will make an adjustment for this modification over time.

The Adaptive Knock Retard will decrease or try to lose its values recorded with each key on/off cycle. Why? Let's say you get the mysterious bad tank of gas. Your AKR will adjust your timing to remove the knock but the next time you fill up there will be no knock with good gas, so this is why the AKR table will decrease its learned values over time.

Getting back to you Richard (Me... Mr Dick Head... Richard HAHA!!).... Steve said it was clear on his dyno runs working with his new X pipe, cams, head work and so on that the AKR would devalue his peak numbers or dip his curve. He could see it stand out like a sore thumb. Since he sells the TTS he will also give you a starter calibration if you also purchase his Stage 1 or Stage II kits. In this starter calibration he always leaves the AKR cells set to ZERO. Steve George knows what he is talking about but as it is with Dyno Operators or Home Brew Guys we all have different opinions and ways to verify.

Further... This is not how I work my calibrations. This is not how TTS recommends the use of the AKR, per Steve Cole. After VTuning it is recommended to restore the original values of the AKR tables as a safety measure. Steve George's argument is the Constants table knock retard is all you need.

Speaking of the "Knock Retard" located in the Tuning Constants Table. I never turn this off but I have been known to reduce the MAX KNOCK RETARD value to 5.

Hope this is clear.

What Wiz is talking about is in the constants, there are three constants to the left.  Size, injector and max retard... that max retard is what he adjusts to 5.  Me??  I am going to try 3.  To kill AKR, one needs to go to THAT table and zero it out.  See?  NOT the one in the constants table.  I bring this up, Mike, because I would think fast retard (the one in constants table) would have killed any actual audible ping even if the AKR was zeroed out.

Mike, in either event, what fuel did you v-tune at?  Maybe you may want to consider tuning using 89 octane instead of 93.  THEN when you are working the data runs to look at timing, if you pull timing where you noticed a ping. this would all wurk better for you, in regards to AKR.

[Steve Cole]

Adaptive Knock Retard is what is Learned by the ECM by watching where it has to pull timing. Sorry, but its stored values so when you shut the key off it is still there on the next startup. As time goes by if it does not knock it will take the learn values back to zero but this does not happen each time you shut the bike off. It does get set back to zero when the ECM gets programmed. The reason I have you limit it to zero when tuning is to keep it from changing your results. With it limited to zero the system does not learn any spark adjustments and when your trying to find the proper settings it's nice to know the ECM isn't changing things around on you.

My feeling is that it should be returned to our base settings to allow for that bad tank of fuel. If you get a bad tank are you going to run it out or dump it? Most are going to keep on riding and that's why I feel it should be there. Now to the Max retard table or Constant as it is done both with a table or a constant depending on which base software is used. IF you limit it there you limit the entire system. There is also another constant that allow you to switch the whole system off. All I can say is play at your own risk.

[Mr. Wizard]

  The Wizard, a GOOD member in standing here, not like me at all., told me this:

You see... right there was your mistake. Everything was going soooo well too.

[Mr. Wizard]

Adaptive Knock Retard is what is Learned by the ECM by watching where it has to pull timing. Sorry, but its stored values so when you shut the key off it is still there on the next startup. As time goes by if it does not knock it will take the learn values back to zero but this does not happen each time you shut the bike off. It does get set back to zero when the ECM gets programmed. The reason I have you limit it to zero when tuning is to keep it from changing your results. With it limited to zero the system does not learn any spark adjustments and when your trying to find the proper settings it's nice to know the ECM isn't changing things around on you.

My feeling is that it should be returned to our base settings to allow for that bad tank of fuel. If you get a bad tank are you going to run it out or dump it? Most are going to keep on riding and that's why I feel it should be there. Now to the Max retard table or Constant as it is done both with a table or a constant depending on which base software is used. IF you limit it there you limit the entire system. There is also another constant that allow you to switch the whole system off. All I can say is play at your own risk.

Hummmmmm.... must have been drinking when you told me this? Steve... you need to change to 2%, that whole milk will kill ya.

Pretty sure it was you that told me the above key on, key off routine dropping the AKR learned values by small increments over time.

Maybe I should switch to skim milk. Getting too old to handle the root beer now.

Then fill in this gap for me... over what period of time will the AKR devalue if the knock sensed is now gone or not seen anymore? Is this based in miles or clock?

[strokerjlk]

Hogmike. It could be as simple as a change in temp spark would fix your slight ping problem. Anyway you can keep it from starting to pull timing,the. Better off you are.
FWIW relying on ION sensing to control detonation is asking for trouble. Not that you are mike,but it is a practice some use.
truk.
after doing some 89 tuning I am still not sold,that it is a cure all. Case by case only.

I remember once in texas,I got some 93 that was just simly bad gas.
there wasent a chance of ION sensing,spark retard curing the detonation. Only thing to do was ride in a manner that could safely burn a couple gallon out,and get something better in.

[Steve Cole]

Pretty sure it was you that told me the above key on, key off routine dropping the AKR learned values by small increments over time.

It does reduce but does not reset it to zero, which is how I understand what has been posted. There is a big difference between dropping a small amount and resetting to zero in my book. The time and adjustment amount of adding spark back will vary based on what it has learned and for how long so there is no straight answer that fits.

[Mr. Wizard]

The Adaptive Knock Retard will decrease or try to lose its values recorded with each key on/off cycle.

I believe we are saying the same thing here... will decrease and when you and I discussed this you used Key on, Key off as an example over time. I don't think my stolen quote by WT said it would clear to zero.   

[wurk_truk]

Timing be damned.... Stolen?