Tom,

I used DSMLink in my DSM and I loved it. I am seriously considering putting a DSMLink in my EVO, but since the DSMLink is not a 'native' EVO implementation, there are a few trade-offs/bugs that I don't want to deal with. And seeing that the DSMLink guys are still a ways out from releasing their EVO version, I'm shopping for some sort of EMS.

I really don't like piggy-backs, but I have to give you kudos on everything that you have input into your product. It really looks like a nice tool.

My question deals with your knock features:
1. By looking over documentation, I see that you can log the knock sensor voltage. Can you monitor each cylinder individually or is this just an average of the four or some sort of total?
2. Do you have any way to warn/alert the driver when knock occurs over some value? i.e. flash check engine light, etc.
3. Corresponding with #2 above, if that is possible, is this value just a static voltage number, or is this variable with RPM, following the natural background noise of the engine. The background noise of the engine increases with RPM, so defining a ceratin voltage number will miss knock in the lower RPM range.

Please detail more on how the ECU+ monitors knock and what user settings are available to monitor, adjust, etc.


Thanks,
Eric

 

Eric,

1. The knock read-out is an average, not per-cylinder. The software inside the box can knows what it is, per-cylinder, but the datalogging isn't fast enough that you could distinguish per-cylinder anyway. I guess it's conceivable that I could have a user setting that picks what cylinder you'd like to monitor, if say you knew that one particular cylinder was the most knock prone.

2. The unit has a not-yet-used shift light driver, which I suspect will be useful for what you want. When I implement it, it seems like I'd want to make it trigger on some user-specified condition - certainly knock would be one of the avaliable conditions.

3. The knock thing is just starting to get some people looking at it. I saw one log of knock where yea, the voltage increases with RPM, but when there was real engine knock, the signal spiked up to like 4 volts (it usually hovers in the 1-2 volts range as I recall). In that case, a static number would actually work even though you'd think it wouldn't be optimal. But you know, it's all "just software" - if the current setup isn't working right, I'll change it.

Tom

 

So are you saying that 5 (FIVE) Volts reading on the knock monitor is bad no matter what RPM it is at?
*uhoh* :}

 

Just for reference I've been logging my knock sensor voltage with my LMA-2 from Innovative. I think every engine as different background noises but after a few thousand miles my car has about .04 - .05 volts at idle, .4v - .5v during normal driving and as high as 1.0v in the upper RPM at the track.

As fas as I can tell I've only logged knock one time and it was while drag racing. The reading jumped up to around 2.0v and instead of it being a spike like I usually see it looked like a hump on the graph ( __/ \__ vs. __/\__ ) if you know what I'm talking about. The "hump" was followed by a decrease in my timing as recored by my Pocketlogger.

 

It's pretty much 100% certain that the ECU+ doesn't monitor the knock sensor the way the Innovative unit does. The ECU+ filters (in frequency) and windows (in time) the information from the knock sensor, just like the stock ECU does. Almost all of the aftermarket knock readout units are just simple voltage amplitude monitors, which are less accurate than the way I'm doing it (at least in theory). If a knock readout device doesn't integrate itself *inside* of an engine management system, it's definitely doing things the cheap and easy way. Anyway... the point is that you can't really compare what the ECU+ reads out to what other units read out.

Tom

 

Tom,

Another quick question for you:

I know the ECU+ is a piggyback and isn't reading the stock ECU, but do you think it is possible for you to write some sort of code that actual logs the knock retard (degrees of ignition retard) that the stock ECU is doing?

That's really the whole goal with being able to monitor knock. I'm not sure how your knock routines are written for the ECU+, but I agree with you in your above posts that a simple voltage reading will not suffice.

But, that being said, what I (and I think most everyone else) is after is a way to read when the stock ECU says there is knock. The stock ECU already does a fine job of detecting knock and pulling the appropriate amount of timing, etc. So, if you could somehow read the timing retard from the ECU (whether it be reading the timing and comparing to the point of the timing map that you should be on or some other logic), that would be perfect.

That's just what I am used to with DSMLink. I'm really thinking about using your product, but I need to know a little bit more information first.

Thanks again,
Eric

 

Eric,

I don't think the timing retard is available in any way from the stock ECU. I think the engine timing is available through the OBD-II port, but even then that's not the amount of timing retard - its the actual timing. What I've seen, though, is that you really don't need that exact number. Instead, two things:

1. It's usually pretty obvious when there's knock, as the timing curve, instead of increasing linearly with RPM, does a nose-dive.

2. You can overlay multiple datalogs on top of each other, so if you have a conservative tune log (there'll be no knock there) on top of a current log, each showing the timing, and the current log shows a dip at around the same point you see the knock voltage go up, then you know that knock was causing the timing retard.

Tom

 

I personally have never looked into the code of the stock ECU, but I do know that it is possible to log the timing retard, as this is how DSMLink does it. But, I'm sure the guys over there have coded up special routines to extract the information and DSMLink deals with the ECU directly and also OBD-II, so I know we are talking about a different animal.

But, to add comments to your comments:

1. I agree that it is easy to see dramatic knock in a log of timing. It is pretty obvious. But, unless you know the exact timing map that you are on and have a reference to the stock timing tables (which most people don't have...some do), it is much more difficult to tell the areas of less knock, which in my opinion, are still very important. I, personally, tune my car to a 'knock-free' tune.

2. This is a great method and I have even suggested something similar to another user that was using a piggyback, but there is one big caveat in this. This method is only valid if your two runs were using the same exact load column in the timing table. Chances are the two runs won't be, because of the conservative tune versus the more aggressive tune, putting you on a different load column, which would in turn make it look like timing was being pulled. This method would only be valid on upgraded turbos where even a conservative tune would land you on the highest load map in the stock timing table.

It's sort of hard to explain here without writing a novel, but I appreciate your replies and understanding in this matter. I'm just trying to offer suggestions or possibilities for future enhancements. Your methods are definitely valid and useable, but you must be very-well trained and have access to some 'hard-to-get' data, that everyone may not have access to in order to definitely say that there is or there isn't any knock in any particular run. And, using these methods are 'after the fact', where I would like to know during the run, so I could let off the gas before doing more damage.

Tom, thanks again for your eagerness to reply. I truly appreaciate someone who stands by their product and are willing to make it better.

Thanks,
Eric

 

Tom,

I thought I would share the stock timing map for the Evo, thanks to Thomas Dorris over at DSMLink. This helps illustrate my point of the timing curves and the difference with being on different load columns. I have also attached the corresponding data.

Just being one load column over could drop the timing 3 degrees and not be knock. Overlaying two logs may cause some people to think that there is knock, when there isn't. Also, it could also go the other way, where people thing they are just moving around in the map, where they really have knock.

Eric

 

you by far seem to have more knowledge than me on this subject... but a simple question none the less: if you did identical WOT pulls (gear, load, temp, stretch of road) should'nt you be in the same load "column"? so in the ecu+ you use your 'conservative' WOT load tune and compare it to another WOT load tune.

 

Well, in a perfect world, yes, but under stock boost level, probably not, especially if your 'conservative' tune is using a piggyback.

Load is roughly proportional to airflow/rev, which is roughly proportional to boost. Temperature and many other variables come into play, but let's go with this for sake of explanation.

If you use a piggyback to do 'conservative' WOT pulls, what is making it conservative? I would guess richer fueling right? Well, what a piggyback does is 'lie' to the stock ECU about airflow and the resulting calculation by the stock ECU about airflow/rev will be off. So, if you then do a more aggressive WOT pull, agressive meaning leaning out via the piggyback, then you are passively increasing timing, so your timing curves will never match.

The one and only time they will match in the above scenario is when even your aggressive tune is well above the last load map in the stock ECU. But since using a piggyback essentiial moves you further into the conservative ECU load zones, your aggressive tune will never be above the last load map. I would venture to say that you would need to be pushing well over 60lbs/min of airflow (600+hp) before that would be the case.

I hope this made some sense. It's hard to explain things sometimes.

Eric

 

wow.... so your implying that by using a piggyback (by its very nature) you are decreasing the load category the stock ecu puts you in, in any given situation? Does this also imply that after some degree of tuning w/ a piggyback, it is possible to never attain the max load "column" that is programed into the stock ecu? If so... what difference will this make in outcome? Only knock calculation?

 

Yes, exactly. That't the whole basis behind piggybacks. By it's very nature it fools the ECU into seeing less airflow, which puts you on a lower load column in the stock ECU. That, in turn, will decrease fueling and increase timing, both of which you want to do for more power.

But, all the knock algorithms, etc, will still work in the stock ECU. I'm just saying that it is much more difficult to know what timing value you should be at for any given RPM. For example, looking at the stock Evo timing map that I posted earlier, when using a piggyback, there is no way to definitively tell what load column you are on, so in turn you can't really tell what you timing should be, so you really can't tell if the stock ECU is pulling a little timing here and there from knock.

That is my whole point behind needing to knock the knock retard from the stock ECU.

Hope this helps,
Eric

 

also... since w/ ecu+ the stock ecu is still free to make knock adjustments, and since you say tuning w/ the piggyback will push you 'down' in load 'columns' (toward idle, away from WOT), does this mean that the stock knock adjustments will become increasingly inadiquate, seeing as it will be less aggressive/prone/whatever in it's adjustments when it thinks that you are 'lower' in load 'columns'?

did that make sense? summary: does simply using a piggyback make the stock ecu's knock control less "conservative" (thus less usefull) ?

edit: you posted before I finished this post, thank you for that info. Tom, man where you at? What's your take?

 

The amount that you're pushed down on the curve really is a function of how much lower you have to push the airflow. Generally, you push it more with progressively large injectors, whereas if you're using stock injectors it isn't a problem. So if you're concerned about it, the best approach is to get a reflash for your stock ECU to make the internal calibration match your injectors. (Al, in one of the other threads, offered this up for $75 along with some other features.) With this done, you can use the ECU+ for fine tuning and get the best of all worlds.

With that said, lots of people successfully use piggybacks every day in some very fast cars and it works fine. It's kinda funny how people obsess over things like this, when some of those same people will install standalone ECUs and end up with some god-awful hand-tuned knock control strategy.

Tom