Yet Another E-Tune Review for Gernby
Originally Posted by sohc_mshue' timestamp='1361885201' post='22364277
Disabling the knock sensor is not the same as desensitizing it. Any modifications to the engine will affect the engine noise level and could trigger false knock. That is why running the engine with a known amount of advance that will not cause knock is important to fine tune the sensitivity so the knock sensor is actually useful. Have youe ever tuned a newer SI with a jrsc on flashpro? They trigger a ton of false knock at low rpms and the sensor must be desensitized in that area of the map or you will get unecessary knock retard.
There is a "knock ignition limit" that was determined by Honda testing for maximum spark advance before knock using the max likely octane fuel to be used on the street. This ignition limit if negative is retarded from the base timing tables. It does not show up as knock retard though when logging.
If you are running a super high octane race fuel and the knock retard goes away then that means you were in fact getting knock.
There is a "knock ignition limit" that was determined by Honda testing for maximum spark advance before knock using the max likely octane fuel to be used on the street. This ignition limit if negative is retarded from the base timing tables. It does not show up as knock retard though when logging.
If you are running a super high octane race fuel and the knock retard goes away then that means you were in fact getting knock.
The last sentence you wrote is utterly incorrect. The only time there should be no knock retard is when you are running super high octane race fuel. If you aren't running super high octane race fuel, then a good tune will apply knock retard (Octane Ignition Trim). That's a fact!
Optimal timing changes with octane, so you can't assume "base timing" is constant. Honda's knock control is designed to automatically adjust ignition timing UP and DOWN to compensate for the fuel it's running at the time. There isn't a "Knock Advance" value, so if you zero out the knock retard, the ECU can't ADVANCE timing when you run higher octane fuel.
If you truly want to setup the timing tables like honda intented then you will need to setup the base timing table using "super high octane" to find mbt and use the knock ignition limit table to reduce timing while on good pump fuel to avoid knock. Obviously most of your customers will not be going out to get 100 octane fuel to set this up so you are stuck setting the base timing table up using pump fuel while looking for mbt and monitoring knock.
If you reduce the timing 4+ degrees and are still seeing knock retard then likely the knock is false hence why i suggested desensitizing the knock sensor in those areas.
I tune for no or minimal knock retard. I can tell you from experience with the hot ass intake temps from the jrsc setups you will typically hit knock before mbt at WOT. I am not going to tune their base timing tables to MBT and let the knock sensor just do its thing when I know it will likely always be in knock retard under those conditions. Of coarse it will likely have some knock retard under varying road conditions/fuel conditions though. I would like to keep this to a minimal.
Yes I know that the ecu trims timing based on feedback from the knock sensor, but dont you think that if it is retarding 4 degrees then you should reduce some timing in the base table or tell the customer to use a top tier premium fuel (i.e. shell, bp, exxon, chevron, etc.) while you setup the base timing table for them?
You cannot be sure that this fuel has the best knock resistant properties out of all the pump fuels they will come across, but you will be playing things safe by assuming it is. Any lower octane fuel they come across will be compensated for by the knock control.
If you truly want to setup the timing tables like honda intented then you will need to setup the base timing table using "super high octane" to find mbt and use the knock ignition limit table to reduce timing while on good pump fuel to avoid knock. Obviously most of your customers will not be going out to get 100 octane fuel to set this up so you are stuck setting the base timing table up using pump fuel while looking for mbt and monitoring knock.
If you reduce the timing 4+ degrees and are still seeing knock retard then likely the knock is false hence why i suggested desensitizing the knock sensor in those areas.
The fact that you are still talking about "false knock" makes it clear that you just don't understand Honda's knock control. I really wish Hondata would have named that value something else, so it wouldn't be so confusing to people. Knock retard is not an indication of knock! You can actually subtract 4 degrees from your ignition maps, and knock retard will still be there! Knock retard is just a calculated adjustment based on long-term observed fuel quality. Please don't screw your customers out of the benefit of that feature!I tune for no or minimal knock retard. I can tell you from experience with the hot ass intake temps from the jrsc setups you will typically hit knock before mbt at WOT. I am not going to tune their base timing tables to MBT and let the knock sensor just do its thing when I know it will likely always be in knock retard under those conditions. Of coarse it will likely have some knock retard under varying road conditions/fuel conditions though. I would like to keep this to a minimal.
"Nope. I assume that the customer's gas is crap, and want to see ~3-4 degrees of knock retard on the low cam. I also know that retarding timing won't reduce the knock retard, since knock retard isn't caused by knock!"
I don't get a bunch of knock retard while tuning flashpro unless it has actualy experienced a knock count at some point. This is with knock sensitivity untouched on cars that are not prone to false knock. How else would the ecu know to adjust the k. control value (the ecu's estimated octane scalar on the knock retard table), but from knock sensor feedback? Also how can you get a consistant run to tune for mbt when your timing is constantly moving around due to non knock related retard.
Perhaps the knock retard you are experiencing is from some accumulated knock count that has raised the knock control value?
"What you consider safe is actually dangerous. Adequate knock retard is what makes it safe. Unless you spend many many iterations reducing knock sensitivity tables by 1% at a time until knock retard just gets to 0.01 degrees, then you don't really know how much worse their gas would need to be before it would actually start applying knock retard again. You also rob the customer of the benefit that could come with better fuel, since the ECU won't EVER advance timing beyond the base timing."
Ignition advance = minimum(MBT Ignition, MBT Ignition + knock ignition limit - (knock retard x knock control))
Taken from Hondata's help file. According to them timing will never be higher than your base timing table unless acted on by an ignition compensation. I have never seen this occur before (I'm not saying that Hondata has 100% figured out honda knock control, but from my experience this calculation for timing appears to be true).
"It would take some throwaway engines and very expensive equipment to recreate Honda's tuning process. That's why I studied the way Honda's tune works in stock form, then set my targets to tune for similar behaviors. My observation from stock '06+ S2000's is that Knock Retard is ALWAYS above 0, and typically as high as 7 degrees on the low cam. That's why I target 3-4 degrees of knock retard, so that the ECU will advance timing as much as 3-4 degrees when it has great fuel, but it can also retard fuel immediately with lower quality fuel."
If your knock retard lowers and you get 3-4 degrees more advance with great fuel then that means you were KNOCKING on the fuel you tuned on! This means that you are assuming your customer is running crap fuel while tuning. If they are in fact running good fuel then you will always have knock retard 3-4 degrees or greater whether they are running good or bad fuel. Sure this probably works out just fine and does not really degrade reliablity or drivability on the mostly stock vehicles that you are etuning, but i would hate to use this method on something actualy making big power.
" The fact that you are still talking about "false knock" makes it clear that you just don't understand Honda's knock control. I really wish Hondata would have named that value something else, so it wouldn't be so confusing to people. Knock retard is not an indication of knock! You can actually subtract 4 degrees from your ignition maps, and knock retard will still be there! Knock retard is just a calculated adjustment based on long-term observed fuel quality. Please don't screw your customers out of the benefit of that feature!"
How do you think the ECU determines fuel quality???? knock count...
"I guess I wouldn't consider your tunes to be any good then."
Thats ok. I don't consider yours to be top notch either
I don't get a bunch of knock retard while tuning flashpro unless it has actualy experienced a knock count at some point. This is with knock sensitivity untouched on cars that are not prone to false knock. How else would the ecu know to adjust the k. control value (the ecu's estimated octane scalar on the knock retard table), but from knock sensor feedback? Also how can you get a consistant run to tune for mbt when your timing is constantly moving around due to non knock related retard.
Perhaps the knock retard you are experiencing is from some accumulated knock count that has raised the knock control value?
"What you consider safe is actually dangerous. Adequate knock retard is what makes it safe. Unless you spend many many iterations reducing knock sensitivity tables by 1% at a time until knock retard just gets to 0.01 degrees, then you don't really know how much worse their gas would need to be before it would actually start applying knock retard again. You also rob the customer of the benefit that could come with better fuel, since the ECU won't EVER advance timing beyond the base timing."
Ignition advance = minimum(MBT Ignition, MBT Ignition + knock ignition limit - (knock retard x knock control))
Taken from Hondata's help file. According to them timing will never be higher than your base timing table unless acted on by an ignition compensation. I have never seen this occur before (I'm not saying that Hondata has 100% figured out honda knock control, but from my experience this calculation for timing appears to be true).
"It would take some throwaway engines and very expensive equipment to recreate Honda's tuning process. That's why I studied the way Honda's tune works in stock form, then set my targets to tune for similar behaviors. My observation from stock '06+ S2000's is that Knock Retard is ALWAYS above 0, and typically as high as 7 degrees on the low cam. That's why I target 3-4 degrees of knock retard, so that the ECU will advance timing as much as 3-4 degrees when it has great fuel, but it can also retard fuel immediately with lower quality fuel."
If your knock retard lowers and you get 3-4 degrees more advance with great fuel then that means you were KNOCKING on the fuel you tuned on! This means that you are assuming your customer is running crap fuel while tuning. If they are in fact running good fuel then you will always have knock retard 3-4 degrees or greater whether they are running good or bad fuel. Sure this probably works out just fine and does not really degrade reliablity or drivability on the mostly stock vehicles that you are etuning, but i would hate to use this method on something actualy making big power.
" The fact that you are still talking about "false knock" makes it clear that you just don't understand Honda's knock control. I really wish Hondata would have named that value something else, so it wouldn't be so confusing to people. Knock retard is not an indication of knock! You can actually subtract 4 degrees from your ignition maps, and knock retard will still be there! Knock retard is just a calculated adjustment based on long-term observed fuel quality. Please don't screw your customers out of the benefit of that feature!"
How do you think the ECU determines fuel quality???? knock count...
"I guess I wouldn't consider your tunes to be any good then."
Thats ok. I don't consider yours to be top notch either
Gernby, do you only tune Flash pro and Hondata stuff? I see a lot of the "E-Tuners" that only learn one specific software and what I find more times than I'd like to is that those "tuners" fail to understand the entirety of tuning. Im with MSHUE...no knock what so ever is the goal, I don't care what the hell you are tuning. There are times I'm sure you will see false knock because it does happen and like Matt said, the knock sensor will have to learn the baseline of what is normal and not. The whole point of tuning is to tune up to the point of MBT or below any knock level.
The statement of knock retard has nothing to do with actual knock makes no sense to me... please explain. The purpose of knock retard is to retard timing when knock is present
The statement of knock retard has nothing to do with actual knock makes no sense to me... please explain. The purpose of knock retard is to retard timing when knock is present
Diverse tuning experience is important. I've been tuning Honda and Toyota cars for 11 years. However, if your experience is significantly spread across multiple platforms, you will not master any of them. I have mastered FlashPro for NA applications, and don't believe anyone outside of Honda or Hondata understands more about how the DBW Honda ECUs work.
You guys seem to be mixing up knock control functionality from multiple ECUs / platforms, and don't fully understand knock control in the DBW Hondas. You seem to think "Knock Level", "Knock Retard", and "Knock Counts" are similar, but they are NOT. You assume that my allowance for "Knock Retard" equates to acceptance of Knock, but my tunes do not knock, and do not create "Knock Counts". It's due to my use of "Knock Retard" that I ensure zero "Knock Counts".
My own personal S2000, which I've tuned literally hundreds of times during product development, is tuned for 3-4 degrees "Knock Retard" on the low cam with zero "Knock Counts". Depending on the climate or the batch of fuel I'm running, my datalogs show more or less knock retard, but my knock counts are always zero, which is exactly how it should be! I want the ECU to adjust timing automatically for the current fuel I'm running! If you tuned my car, it would NOT do that.
The "Knock Sensor" is just a microphone. It isn't a "Knock Detector". It just lets the ECU monitor the noise level inside the engine. The noise level from that microphone was named "Knock Level" (misleading IMHO). If the noise level is 50% at 4000 RPMs, that does NOT necessarily mean that the engine is knocking! It only means that it is knocking IF the noise level exceeds a predefined value based on load and RPM. For example, if the noise level is 50% right after cylinder 3 fires, then jumps to 70% after cylinder 4 fires, you will get a "Knock Count" on cylinder 4. It is true that a "Knock Count" will increase the amount of "Knock Retard", but a good tune will result in "Knock Retard" that prevents "Knock Counts".
The ECU maintains a sort of long term noise level factor that is used with the Knock tables to determine "Knock Retard". If you get a tank of bad gas, the noise level from the microphone will increase, which will increase the long term noise level factor. If tuned right, the higher noise level factor will increase the amount of "Knock Retard". This doesn't mean that the engine is knocking, but actually means that the ECU is actively preventing knock. Even if you run super high octane fuel AND retard timing several degrees, the microphone will still pick up noise AND the ECU will still register a noise level greater than zero AND it should still apply some amount of "Knock Retard".
You guys seem to be mixing up knock control functionality from multiple ECUs / platforms, and don't fully understand knock control in the DBW Hondas. You seem to think "Knock Level", "Knock Retard", and "Knock Counts" are similar, but they are NOT. You assume that my allowance for "Knock Retard" equates to acceptance of Knock, but my tunes do not knock, and do not create "Knock Counts". It's due to my use of "Knock Retard" that I ensure zero "Knock Counts".
My own personal S2000, which I've tuned literally hundreds of times during product development, is tuned for 3-4 degrees "Knock Retard" on the low cam with zero "Knock Counts". Depending on the climate or the batch of fuel I'm running, my datalogs show more or less knock retard, but my knock counts are always zero, which is exactly how it should be! I want the ECU to adjust timing automatically for the current fuel I'm running! If you tuned my car, it would NOT do that.
The "Knock Sensor" is just a microphone. It isn't a "Knock Detector". It just lets the ECU monitor the noise level inside the engine. The noise level from that microphone was named "Knock Level" (misleading IMHO). If the noise level is 50% at 4000 RPMs, that does NOT necessarily mean that the engine is knocking! It only means that it is knocking IF the noise level exceeds a predefined value based on load and RPM. For example, if the noise level is 50% right after cylinder 3 fires, then jumps to 70% after cylinder 4 fires, you will get a "Knock Count" on cylinder 4. It is true that a "Knock Count" will increase the amount of "Knock Retard", but a good tune will result in "Knock Retard" that prevents "Knock Counts".
The ECU maintains a sort of long term noise level factor that is used with the Knock tables to determine "Knock Retard". If you get a tank of bad gas, the noise level from the microphone will increase, which will increase the long term noise level factor. If tuned right, the higher noise level factor will increase the amount of "Knock Retard". This doesn't mean that the engine is knocking, but actually means that the ECU is actively preventing knock. Even if you run super high octane fuel AND retard timing several degrees, the microphone will still pick up noise AND the ECU will still register a noise level greater than zero AND it should still apply some amount of "Knock Retard".
So you are saying that the ecu is already retarding timing before it even sees knock? I'm referring to timing not already set in the timing tables or the iat/motor temp compensation tables.
What the hell is the purpose of a knock sensor if it retards timing BEFORE knock is sensed?
What the hell is the purpose of a knock sensor if it retards timing BEFORE knock is sensed?
I wasn't going to post about this anymore, since I agree that knowledge is power, and I shouldn't give so much away for free. However, I don't want to leave one of my eTune reviews in this state, so I did some searching. Thankfully I found this post directly from Hondata, which says just about everything I've said. Enjoy the read! I've highlighted some points.
Honda ECU knock control confuses a lot of people, mainly because it does not work as they would think. You may get knock retard without any knock, and knock without any knock retard, so you would be excused thinking Honda knock control is broken.
From the K-Series ECU onwards, knock control basically means adjusting the ignition timing for different octane fuel.
In more depth, you need to know how Honda tunes their engines. The engine goes onto an engine dyno with a high octance spec fuel, and they are computer tuned to set the fuel to give a stoichiometric air fuel ratio (14.7, 14.68, 14.58, depending on how pedantic you are about the fuel here), and the ignition is tuned to MBT. MBT = mean best torque, which basically means the ignition is set to give maximum engine output at each load / rpm combination, irrespective of knock.
After the engine has been tuned to stoichiometric and MBT ignition, the WOT fuel compensation tables are used to enrich the fuel for high load / fuel throttle. For ignition, there is a variable amount of ignition retard from MBT to ensure the engine does not knock. The amount of retard varies by load and rpm, but also by the amount of knock sensor noise that the ECU sees. This knock control value is a fairly long term thing, where it may increase fairly quickly which knock sensor noise, increasing the knock retard, and then slowly decrease if they is not much noise, decreasing the knock retard.
The knock sensor processing is fairly smart. The ECU knows which cylinder is compressing, and knows that knock is most likely to occur at 18 degrees past TDC, so it can work out what cylinder is knocking. You can datalog this.
The thing that gets most people at this point is that the ECU doesn't retard / advance very quickly - it is applying a pre-programmed retard based on estimated fuel octane rather then retarding / advancing quickly when it gets a small amount of knock.
The next thing that confuses is that the ECU will retard, even without any knock. This is because it may thing that even with the highest octane fuel it can't reach MBT timing, so will retard anyhow.
Next, the ECU may detect knock, increase the knock control value, but if the knock sensitivty is low, it may not retard enough (or at all), so you get knock but no knock retard.
So how do you tune around this? In short, tune the ignition timing so you don't get any knock counts. Otherwise a small amount of knock in one place will cause knock retard everywhere. When this happens on the dyno you can get more power everywhere by retarding the ignition just where it is knocking.
Finally, allow the factory knock sensor to do its job by not giving it additional noise to worry about. Typically culprits are exhaust headers hitting something, worn engine mounts, noisy superchagers - all these can give 'false knock', which both decreases engine performance and masks real knock.
There's probably more, but that's it for now.
From the K-Series ECU onwards, knock control basically means adjusting the ignition timing for different octane fuel.
In more depth, you need to know how Honda tunes their engines. The engine goes onto an engine dyno with a high octance spec fuel, and they are computer tuned to set the fuel to give a stoichiometric air fuel ratio (14.7, 14.68, 14.58, depending on how pedantic you are about the fuel here), and the ignition is tuned to MBT. MBT = mean best torque, which basically means the ignition is set to give maximum engine output at each load / rpm combination, irrespective of knock.
After the engine has been tuned to stoichiometric and MBT ignition, the WOT fuel compensation tables are used to enrich the fuel for high load / fuel throttle. For ignition, there is a variable amount of ignition retard from MBT to ensure the engine does not knock. The amount of retard varies by load and rpm, but also by the amount of knock sensor noise that the ECU sees. This knock control value is a fairly long term thing, where it may increase fairly quickly which knock sensor noise, increasing the knock retard, and then slowly decrease if they is not much noise, decreasing the knock retard.
The knock sensor processing is fairly smart. The ECU knows which cylinder is compressing, and knows that knock is most likely to occur at 18 degrees past TDC, so it can work out what cylinder is knocking. You can datalog this.
The thing that gets most people at this point is that the ECU doesn't retard / advance very quickly - it is applying a pre-programmed retard based on estimated fuel octane rather then retarding / advancing quickly when it gets a small amount of knock.
The next thing that confuses is that the ECU will retard, even without any knock. This is because it may thing that even with the highest octane fuel it can't reach MBT timing, so will retard anyhow.
Next, the ECU may detect knock, increase the knock control value, but if the knock sensitivty is low, it may not retard enough (or at all), so you get knock but no knock retard.
So how do you tune around this? In short, tune the ignition timing so you don't get any knock counts. Otherwise a small amount of knock in one place will cause knock retard everywhere. When this happens on the dyno you can get more power everywhere by retarding the ignition just where it is knocking.
Finally, allow the factory knock sensor to do its job by not giving it additional noise to worry about. Typically culprits are exhaust headers hitting something, worn engine mounts, noisy superchagers - all these can give 'false knock', which both decreases engine performance and masks real knock.
There's probably more, but that's it for now.


