P J Moran

Using my scan tool, I am reading some knock and retard under moderate loading. I can't hear it, but I understand that the ECM's supposed to hear it before I do. How can I tell if the detected knock is real or false? No knock is detected at cruise - only when accelerating as if to pass someone (but only at half throttle, or so - NOT WOT).

I've backed off the timing quite a bit, but am still detecting knock. I can keep adjusting my spark table, but am beginning to wonder if I'm chasing a ghost, here.

------------------
'87 IROC 5.7l TPI - original owner!

ORION

Although I may not be of much help here, but I had a similar problem and when I simply replaced the knock sensor it went away (though nothing showed any symptoms). The car constantly retarded etc. and sometimes acted REAL FUNNY under load.

ede

why don't you post this on the prom board and see if you get any help, i'd think someone there would be more likely to help you out. i'd move it but there's a chance you'll find your answer here

------------------
ICON Motorsports

1st & 3rd

P J Moran

It's not really a PROM question. It's just that by burning several chips that I discovered this problem.

Thus far, I have cut the timing by as much as 12* in some areas, but the ECM is still detecting knock. I have ultimately retarded the timing far beyond that the computer ever did (9* was the most retard the ECM ever applied, and we know that's usually more than actually needed). But knock is still detected, and the spark is still retarded, the same amount at the same "place".

This sure looks like a case of false knock. These are not full throttle passes from a stop. These are runs from a 60mph cruise to 90+ mph as if passing someone on a two lane road. Mash the gas (but not WOT), auto downshift, accelerate past, then back off the gas. Every time, knock is detected just after downshift! The retard goes away in just a few seconds (based on the "decay" rate?).

I'm tempted to bypass the knock sensor and start over with my "programming"; adjusting timing based on audible knock. I know this isn't very good, because by the time it's audible, damage is being done! Besides, in a noisy car, you can't always hear detonation until it's really bad. I would like to retain the ESC capabilities, but...

In case it matters, I'm running Edelbrock aluminum heads, calculated 10.0:1 CR, Magnum roller tip rocker arms, and a mild Speed Pro cam (209/214/471/490/112).

I would like confirmation that this is really a case of false knock, and find out why it occurs under these very specific conditions. Also, I would like some guidance on how to get the knock sensor to report only real detonation. Understand that I have been running with a "desensitizing circuit" that I got off some website, but it apparently hasn't solved the problem. Thanks in advance.

------------------
'87 IROC 5.7l TPI - original owner!

Grim Reaper

Be darn careful with the assumption that "inaudible" knock is a "false" knock. In fact, it most likely REAL knock but your eprom did its job. The source of your "ping" can be a variety of things and you may NEVER totally eliminate it. Most people do not have access to a scan tool so they never really see this. So guys with scan tools do; and they freak out, spendiing a ton of money and time to eliminate it, only to still have it.

I say this, because I too have been down your path. So I ultimately took another route. I started to reduce the effects of the knock retard by altering the "attack/recovery" rates until I virtually eliminated it (and also turned inaudible knock to audible knock on a few occasions too). But this has given me the greatest success.

The problem with the standard eprom is that when the "slightest" ping is detected it pulls a ton of timing out and it lingers for a long time...robbing you of power. What you have to do is reduce the "attack rates" (I run no higher than .25-.27) and increase the "recovery rate" (86-90%) at the highest spark advance I can run.

This way, I generally get no more than a couple of degrees pulled out at higher rpms; and I get my highest "EFFECTIVE" spark advance. No scan tool displays your effective spark advance because it is played with so much. But I generally look at "relative + my actual base" to calculate my effective spark advance.

There is also spark retard added by the "octane" routine. But that's another story again. It's a "long term" retard. I have just been playing with that so I can "recover" that retard as quickly as possible. Unaltered, you will loose 4* off your total spark advance and will not recover it until HOURS of running or by turning off the engine.

P J Moran

Thanks for the response, Glenn. But since I have taken out far more timing than the computer ever did, how could this be real spark knock? I think that something else is tickling the knock sensor when the tranny downshifts and the ECM mistakenly retards the timing (and then gradually runs it back up). I have made over a dozen runs and a pattern is clear. It'll suddenly take out 5 to 9 degrees and gradually put them back in as though the knock incident has passed. It does this regardless of the timing in the table. If the ECM takes out as much as 9* and is satisfied with that, and as a result I (eventually) cut the timing by 12*, the ECM shouldn't even see knock the next time, right? I did the same thing (only more so) the ESC did, didn't I?

I would like to play with the attack and recovery rates, but my editor doesn't give me access to those variables. At any rate, I still want to get my spark table such that no detonation is ever detected.

The saga continues...



------------------
'87 IROC 5.7l TPI - original owner!

Grim Reaper

Won't happen. I know there is code in the 8D eprom to deal with the locking of the transmission and I suspect earlier eproms have similar code. But trust me, if you "cap" the spark retard, that "inaudible" knock can become "audible". Been there, done that.

The trick I am finding is to set your total desired timing to around 34* (+/- 2*) and then control the amount of retard put into it. Of course, this does not address the "octane spark retard"; which also must be controlled.

Many of these parameters are not defined in the TDF if you are using TunerCat. You must get your hand on "a hack" and create a TDF to handle many of these. That is how I am controlling the "octane spark retard".

I will say that when you get "control" of the spark retard and can get an effective spark advance in the 30s, it is amazing how well the car performs. Far more responsive and pulls much better. You even get a higher top end with TPI. 8-9* of retard is costing you 20+ HP.

But forget about trying to reduce your spark ... you STILL will get retard occurring and UNTIL you adjust the "attack/recovery" rates, it will be high (lowering your effective spark advance) and linger a long time.

You are not the first to notice this and attempt to fix it; nor will you be the last. Just trying to save you a lot of frustration and steer you in the right direction.

CanadianBeast

Glenn what tables are you refering to when you talk about the attack and recovery rates.

P J Moran

Thanks. But again, you seem to be trying to control how the ECM controls the spark retard, which is good, and something I would like to eventually do. Whereas I am trying to keep the spark knock from ever happening in the first place. Are you saying that is physically impossible? There will always be a ping or two under certain circumstances, and that that can't be prevented by altering the timing curves?



------------------
'87 IROC 5.7l TPI - original owner!

Grim Reaper

Yup. I tried many things (short of a new engine)...still have some knock detection occurring @ WOT, YET I can lean out my engine to 17.6:1, advance my spark to 46.9* and climb a 6-7% grade with 70-75 kpa and not encounter knock. That is of course with a functioning EGR.

When the EGR starts to kick out at 80 kpa, THEN knock detection will occur. Just shows you how well the EGR functions and why it is not a mechanical problem; but an actual "ping" occurring and the knock sensor is doing its job.

That is when I decided to attack it from another route; controlling the level of retard induced by the knock sensor...and that has given me the best results.

And YES, lowering my spark advance STILL detects knock @ WOT (along with richening the mixture). Bottom line, there will always be some knock detected. I too thought like you "then this must be a false kncok", so I virtually disabled the knock sensor and that is when my "inaudible" knock became "audible'. So it is very real.

The controlling of the knock sensor's "attack/recovery" rates have produced the best results of attaining a high effective spark advance without inducing audible knock.


[This message has been edited by Glenn91L98GTA (edited March 08, 2001).]

FastBroker

I would bet on:

conn-rod knock or

piston slap or

a bad knock sensor or

the sensor is mounted in a non-stock location or

there is something (accessory bracket, brace, fan shroud, etc) hitting your engine/accessories somewhere...