Open Loop EE Advice

I've been struggling with knock on a freshly built LT1 for several weeks. It seems to be happening in taller gear ranges at cruising speeds, and only became blatantly evident when IATs started staying above 80f all day (would see little or no knock in the mornings when IAT was lower).

Engine Specs:
Stock crank, new scat rods, cast .030 over pistons, recip assy balanced within .5 grams
10207643 casting heads, LE1 port job, manley 1.94 / 1.50 ss valves, Lunati 73925 springs
LE ported stock intake
CompCams 07-467-8 - 230/236 @ 0.050", .576 / .570 lift, 113deg LSA installed 3 deg advanced for 109deg ICL
CompCams 1605-16 Ultra Pro Mag 1.6 RRs, TFS guideplates, ARP 7/16 studs
Remanned stock opti w/ OE spec pickup, new coil, ICM, 8mm wires
StainlessWorks 1-5/8" standard length headers w/ high flow cats into 2-1/2" duals w/ X pipe

Compression ratio should be more or less stock. Other than EGR delete / block-off all other emissions devices are functional (AIR, CCP).

I've tuned VE using narrowband data with the assumption my closed loop calibration was more or less in the ballpark of accurate. That said, a large amount of VE was removed below 2500RPM - in the neighborhood of 25%. I'm sure some of this is related to the cam, but as the IATs have been climbing I've noticed ever increasing knock in these cruising ranges especially in higher gears, to the point it's stumbling so bad I thought I had a wheel bearing about to lock up. Having pulled an enormous amount of timing out of this area with no result, I'm relatively sure it's going lean and needs more fuel due to the reduced effective displacement caused by valve overlap. I've chased down all the usual causes of false knock and feel relatively confident most of it's real. Plugs show no signs of severe detonation but are otherwise bone dry. Above 3000 RPM or in PE (still using stock PE AFRs) it runs like a top and no knock. So this morning I used eehack to disable BLMs and command AFR to 13.5. Knock isn't completely gone but greatly diminished, and when I turn BLMs back on the knock count starts rolling again.

Even though it idles great in CL, I think I want to try running OL with MAF and wondering if there are any pitfalls I should be aware of. Am I safe leaving the cats on?

I had to turn-off all the knock retard/detection tables. I was constantly getting knock retard even though it wasn't knocking. It seems that more lift and roller rockers mess with the knock sensor unless you have a LT4 knock module.

I've researched up and down that subject and though I'd be open to trying one out, LT4 knock modules have become as rare as hen's teeth. Without a wideband and better knowledge of what I'm doing I'm extremely reluctant to detune knock retard.

I'm sure some amount of what I'm logging is false, as I can get the knock count to increment at hot idle by just blipping the throttle. In addition to that, the exhaust system seems like it resonates well at the frequency the knock circuit is tuned to. I picked up huge amounts of knock count by whacking the collector with a small piece of brass rod. I also found I had cracked one of my exhaust tips with a backfire (killed my MAF cleaning it, severe flood condition) and it's "quacking" like a goose was picked up. I've also seen drivetrain noise set it off, hard shifting, acceleration over rough pavement.

But...

The really debilitating stuff is happening between 1300-2100 rpm at 60-75 mph. In cold air it's barely noticeable, but the warmer and thinner the air gets, the worse it becomes to the point it "feeds" itself and KR ends up pegged at 12 until you downshift. Switching from OL to CL with eehack, the counts start climbing as soon as BLMs are enabled so it seems pretty obvious it's going lean in CL.

Knock graph from 5-12

Current OL tune

My take on this as i've been through it, is most of what you have read about anything doesn't relate to any observations you will see.
Most of my changes have seemed counter intuiitive to anything i've read. YMMV

Steveo's Trimalyzer is now on version 1.3b
Have you viewed the knock counts in EEHack and in the analyzer view?

The LT1 doesn't have AE acceleration enrichment so any blip of the throtlle leans the motor until the PCM can figure out a new pulse width.
This is/can be burst knock which has it's own tables of retard.
You can get knock without counts increasing & increasing counts without knock.

OL runs with a lot less counts as i can run it richer. In CL i keep retarding timing and still get knock counts.
That image does look heavy though. So to cover some ideas, You have checked nothing is loose or catching underneath?
You have the right knock sensor for the PCM?

Also my current PCM has a fewer knock counts with an LT1 knock module than the previous PCM with an LT4 knock module, so don't get fixated by that notion either.

HTH some
Mitch

Quote:

---

Originally Posted by Terminal_Crazy

Steveo's Trimalyzer is now on version 1.3b

---

I'm aware. I've modified the build I'm running to ignore frames with corrupt data as these inevitably get picked up as knock events. See here. Some of my logs have as many as 300 trashed frames, but my commute is about 65 minutes each way so that's a lot of data. I think this might be a y-body quirk but haven't had the time to investigate further.

Quote:

---

Originally Posted by Terminal_Crazy

Have you viewed the knock counts in EEHack and in the analyzer view?

---

Always.

Quote:

---

Originally Posted by Terminal_Crazy

This is/can be burst knock which has it's own tables of retard.

---

It was my understanding that burst knock is pre-emptive timing retard that is added to compensate for the lack of a pump shot. I would have to review all the code again, but as I recall all of steve's knock analysis routines only plot an event if knock count increments [edit] regardless of if there's knock retard.

Quote:

---

Originally Posted by Terminal_Crazy

You can get knock without counts increasing & increasing counts without knock.

---

Did you possibly mean "increasing counts without retard" in the last sentence? If so I think what you're getting at is knock in MAP cells lower than 45kpa generates no KR. Got it. [edit] If I cover up the knock plot with my thumb so I can't see anything to the left of 45kpa I'm no less freaked out.

Quote:

---

Originally Posted by Terminal_Crazy

OL runs with a lot less counts as i can run it richer. In CL i keep retarding timing and still get knock counts.
That image does look heavy though. So to cover some ideas, You have checked nothing is loose or catching underneath?

---

I'm going to get it on a lift tomorrow and verify, but yes. I see nothing to indicate the exhaust is touching anything. I'm going to put a fuel pressure gauge on it tomorrow too - I installed a new racetronix 255l/hr pump before I started doing any serious tuning and pressures were good after that so I didn't think about it further.

Quote:

---

Originally Posted by Terminal_Crazy

You have the right knock sensor for the PCM?

---

Pretty sure. Delco 10456126 both replaced during build.

Logging tonight in OL wasn't as good as I'd hoped - the knock plot isn't great (456 points in ~60 miles). But I still feel like it's running leaner than I'd like b/c I filled up and calculated 22.3 mpg for today. Seems unlikely with the way I drive. But some logs from last week were in the neighborhood of 1100 events.

I feel like some of what's getting picked up as knock may be misfires / lugging that gets amplified by the drivetrain. It's a 22 year old car and has rattles and squeaks like any other with 160,000 mi on the odometer. Maybe I'm operating it in a load range that the cam won't allow except with dense air. :confused:

Thanks for bouncing the ideas around - always good to discuss with someone who's been there and done that!

assuming you have ruled out a dead fuel pump or large fueling error, just disable the knock sensor already. you really don't need it if it's causing you this much trouble. knock detection on modified engines is hit or miss, yours is a miss.

I just got back from my buddies exhaust shop where we went over the exhaust system looking for anything that might have been touching and he didn't see anything that bothered him. I'm going to go drive a bit with a fuel pressure gauge and then I think I'm ready to kick caution to the wind and take all the teeth away from knock detection.

Quote:

---

Originally Posted by steveo

assuming you have ruled out a dead fuel pump or large fueling error, just disable the knock sensor already. you really don't need it if it's causing you this much trouble. knock detection on modified engines is hit or miss, yours is a miss.

---

Already? You're telling me to do the equivalent of pulling the pin out of a live grenade and wedging it into a stongbox along with about $10k in cash. Already...

:thumbsup:

How the hell do the LS guys get away with stuff like this? Are the noise filters that much better on those PCMs?

Have you tried OL with the wideband rather than narrow band O2's ?

I wouldn't turn off the knock sensor yet until you've exhausted all paths.

I would check your VE tables, the fueling is very jaggy and looks like it needs smoothing a lot.

sent you a pm.


Mitch

I have no wideband(s) but seriously debating it now. I don't really trust the narrow bands unless someone comes up with some new tuning suggestions for closed loop w/ headers because I've literally tried everything. Of course I'm reading that widebands can lie also, so ???

The VE tables are like that from blind tuning w/ numerous tools. I noticed some of that last night as I was adding a few more percent to the range I've been seeing missing / whatever. I'll spend more time smoothing tonight. With the MAF I don't know how much that affects things, but VE will be difficult to tune without some type of AFR sensor.

This morning when I put my ear near the oil pan w/ engine running it reminded me that the headers are extremely loud. Because they're fairly close to the knock sensors I have to wonder if they aren't causing a lot of the problem. Whatever the case, I probably won't be driving the rest of the week so I'll have time to get the wheel bearing replaced and possibly fix the leaky clutch slave cylinder and see about making a custom hose. The only fitment problem I had with the headers is that there's no way to route the clutch hydraulic hose without it touching the left O2 sensor.

Quote:

---

Originally Posted by spfautsch

Already? You're telling me to do the equivalent of pulling the pin out of a live grenade and wedging it into a stongbox along with about $10k in cash. Already...

---

if you don't botch your timing table or run shitty unpredictable gas the knock sensor is 100% useless. think of all the points distributor engines that ran for 500,000+ miles without one (with occasional points/cap/rotor, of course)

i have the same thing to say about trying to get a perfect stoich ratio across the board. if a carb guy can come along and get his engine running better in 5 minutes, you're doing it wrong.

if i came over to your place and did a really good open loop tune on your car and chopped your knock sensor and o2 wires you would probably crap your pants when you found out how smooth, how cool, and how efficiently it ran across the board after a few runs of just tuning by feel.

just because the tools came with your stock vehicle doesn't mean you have to use them. your target needs to be a good running and reliable car, not good sensor feedback and factory-like reactions to events.

i ran my last trans am wideband tuned. before i sold it, its final tune aimed for 16.5:1 or sometimes leaner at low loads and nearly 60 degrees of timing advance for cruising. really lean for decel too, so DFCO transition was really nice. 13.5:1 and ~30 degrees of all in timing for heavy throttle.

i had pipe plugs where the o2 and knock sensors were. no cats. it smelled like a chemical fire going down the highway, water was pouring out of the exhaust, and the engine loved every minute of it.

the plugs read perfectly clean. no soot in the tailpipes. the fans almost never kicked in it ran so cold, you could lug it up hills, you could redline it down hills.

the kid that bought it off of me drives the piss out of it still and it wont blow up. i'll tell you what too, that kid will never have to worry about any sensor failing except his optispark...

Quote:

---

Originally Posted by steveo

if i came over to your place and did a really good open loop tune on your car and chopped your knock sensor and o2 wires you would probably crap your pants when you found out how smooth, how cool, and how efficiently it ran across the board after a few runs of just tuning by feel.

---

I think you're close to converting me. I just have a bit of a hang-up on not wanting to risk cracking a piston with detonation. If this past winter wasn't the most psychologically and financially difficult one I can recall in over 10 years I'd be responding with "what would it take in addition to buying you an airline ticket"?

Quote:

---

Originally Posted by steveo

i ran my last trans am wideband tuned. before i sold it, its final tune aimed for 16.5:1 or sometimes leaner at low loads and nearly 60 degrees of timing advance for cruising. really lean for decel too, so DFCO transition was really nice. 13.5:1 and ~30 degrees of all in timing for heavy throttle.

---

This being my first experience with open loop, I began to wonder if there's any point in continuing to use PE with an OL tune. Why not just tell the AFR table what you want, where you want it?

Quote:

---

Originally Posted by steveo

you could lug it up hills.

---

Now you're just screwing with me. I want that back!

Quote:

---

Originally Posted by steveo

i'll tell you what too, that kid will never have to worry about any sensor failing except his optispark...

---

The optispark is actually a quite durable piece of hardware that got a bit of a bad rap IMO b/c GM sandwiched it into an unfortunate location. The only problem I've had with mine was caused by a cheap sensor that wasn't suitable for use in the thermal environment the opti lives in. Evidently chrysler used a physically identical pickup in one of their distributor applications, and the cheap aftermarket optis that are all over that one auction site are all equipped with them. According to the previous owner that's where mine came from. When the engine got to temp the high-res pickup would just stop functioning. Unfortunately it looks like AIP has stopped making the OE spec optical pickups, so we're stuck paying through the nose for a complete assembly from MSD or Delco. :-(

Quote:

---

This being my first experience with open loop, I began to wonder if there's any point in continuing to use PE with an OL tune. Why not just tell the AFR table what you want, where you want it?

---

it's nice to have both. PE for really heavy TPS (when you stomp on it) and low vacuum regions of open loop for really heavy load (where you need more power and need to keep combustion temperature down, but you aren't necessarily trying to accelerate rapidly)

Quote:

---

I think you're close to converting me. I just have a bit of a hang-up on not wanting to risk cracking a piston with detonation.

---

the knock sensor isn't there to prevent you from cracking a piston by making horrible decisions with timing advance; it's to allow low quality fuel to be used while maximizing performance with higher quality fuels. it's not a very good tuning aid, and railing up timing until the knock point is a bad approach to maximize power, imo. just add timing until it stops making more power (or less vacuum, if we aren't at WOT) and call it a day. this is super easy with eehack's timing skew feature.

Quote:

---

Originally Posted by steveo

the knock sensor isn't there to prevent you from cracking a piston by making horrible decisions with timing advance; it's to allow low quality fuel to be used while maximizing performance with higher quality fuels. it's not a very good tuning aid, and railing up timing until the knock point is a bad approach to maximize power, imo.

---

That wasn't my intended methodology. I was seeing a staggering amount of knock so I removed 3+ degrees of timing and it made no difference so I started looking at fueling. I think I might have got into a situation where the more timing I removed the worse it got because it would fall on it's face sooner. Also, see my comment about VE below.

Quote:

---

Originally Posted by steveo

just add timing until it stops making more power (or less vacuum, if we aren't at WOT) and call it a day. this is super easy with eehack's timing skew feature.

---

That would be easy with someone to do the driving. I think I just had an idea to add to eehack...

Edit: thinking on this, the only way I can imagine it being "easy" is with a driver and a chassis dyno. You're going to be chasing a moving target (MAP). Or am I missing the point? Is this the intended purpose of auto spark?

I ran a bit with a fuel pressure gauge the other day and it looked good to me ~38psi @ idle (45kpa) and ~44psi at WOT. Nice and steady too. Impressive b/c it looks like the original pressure regulator.

After giving it some thought I'm going to go back to square 1 with my tune and start from scratch with the original VE tables. I just looked back at the second log I took before I killed my MAF (and subsequently started tuning SD) and the IAT was over 100F then. In hindsight the knock map wasn't bad at all. I had to remove a lot of VE up to around 3000 rpm, and I'm beginning to think it wouldn't be bad to leave that more or less alone except for 0-1000 rpm where it was so ridiculously rich. It was raining the day I started tuning idle without the MAF and I nearly passed out from the fumes in my garage waiting for it to switch to CL.

Quote:

---

Originally Posted by spfautsch

I think I might have got into a situation where the more timing I removed the worse it got because it would fall on it's face sooner.

---

A funny thing happened when I tamed the max spark retard table from essentially stock to this:

Code:

---

0.00
0.00
0.00
0.00
0.00
0.00
0.00
1.50
3.00
4.00
4.00
6.00
6.00
9.00
9.00
9.00
12.00

---

All the banding in the knock plot went away. So apparently the knock sensors were picking up a misfire or who knows what from lugging up a hill, and I'm guessing the resultant KR made it snowball by worsening the cause.

I'm back to closed loop for the moment. Other than testing the lower max retard settings I did some trial and error with integrator delays and found settings that have the worst BLM split over all cells around 2.

So I switched back to speed density last night in anticipation of doing some logging today and my idle went back to garbage. It hunts, and falls on it's face when feathering the clutch. I've tried bumping integrator / transport delay to ridiculous numbers and it still wants to die taking off. So it sounds like I'm definitely going to end up with an open loop SD tune, as enabling the MAF takes all the bark away after 4000 rpm when the cam really starts to wake up.

Makes me wonder if a hack to enable MAF at 0 tps would be possible, but doubtful I'll end up caring about closed loop after VE tables are baselined.

Steveo - do you think there would be any value in installing widebands in the narrowband bungs in the collectors? If not I would have to yank the cats.

Quote:

---

Originally Posted by spfautsch

So I switched back to speed density last night in anticipation of doing some logging today and my idle went back to garbage.

---

so, your VE table is out to lunch.

Quote:

---

as enabling the MAF takes all the bark away after 4000 rpm when the cam really starts to wake up.

---

at high rpm where airflow is much more linear, maf and speed density should be fairly equal. so your MAF table is out to lunch

Quote:

---

Steveo - do you think there would be any value in installing widebands in the narrowband bungs in the collectors? If not I would have to yank the cats.

---

yeah, that's a great place, go for it

honestly, the more i listen to you the more I think you're tuning by these feedback numbers instead of by feeling what the engine needs. if it's limp add more fuel. if that makes it worse take fuel away instead.

Quote:

---

Originally Posted by steveo

so, your VE table is out to lunch.

at high rpm where airflow is much more linear, maf and speed density should be fairly equal. so your MAF table is out to lunch

---

I guess I missed adding a qualifier to that sentence - the idle goes to $* when it switches to closed loop SD. Before it's up to temp in SD or with the MAF in OL or CL, idle is awesome. You can barely tell it's a 230 degree cam. Not sure if you caught the post where I stated I'd switched CL back on. Whatever the case, yes, VE needs to be addressed.

Drove it yesterday with almost all non-PE KR disabled. Found some interesting stuff out. See below.

Quote:

---

Originally Posted by steveo

honestly, the more i listen to you the more I think you're tuning by these feedback numbers instead of by feeling what the engine needs. if it's limp add more fuel. if that makes it worse take fuel away instead.

---

I agree, you've converted me. But I think I need to resolve item #1 (VE table out to lunch) first? My first attempt earlier this month at getting VE close wasn't good as I'm still learning, but I think this approach will be better. Here's what I started on yesterday.

Attachment 11802

This should be relatively close and a lot smoother than what I ended up with tuning the table directly (and a little haphazardly) with trimalyzer. I still need to smooth some of the edges manually - this was aiming at -4% (slighly rich). Unfortunately my clutch slave has gotten to leaking worse and I'm afraid that's going to have to get fixed before I can log with this change.

A few things I observed yesterday.

1) When I switched off the MAF the other night I forgot to increase the PE Enable TPS vs RPM table so I get a broader range of feedback for logging. When I did this for the drive home the mid-throttle power was noticeably much better. Much, as in accidentally smoking the tires at 2800 rpm where it's never really displayed that much torque with this cam. I'm wondering if the narrowbands aren't as inaccurate as I began to suspect, and my VE table is just way too rich.

2) Without KR at cruise I started to notice knock counts at the oddest place. I had assumed this was primarily happening while lugging up a hill in 6th @ 75mph, ~1700 rpm. But I noticed none of this yesterday. More often than not the knock counts seem to happen after cresting the hill when TPS starts to decrease. I'm kind of baffled with this - wonder if it could be going a little rich and popping when the unburned fuel hits the cats (it's not plainly audible).

3) Lugging up a hill in 6th in that load range is not generating much along the lines of knock counts, but the engine is definitely not smooth. Almost feels like a cylinder is dead or way out of balance, but when you rap it to 6800 it's as smooth as glass. Now that I'm sure the knock isn't real I'm hoping I can play with timing in this range and see if it can get better.

Quote:

---

Originally Posted by steveo

so, your VE table is out to lunch.

---

You had it pegged. Applying some changes that added fuel in the 400-1200 rpm ranges that were lean by 8% or so has smoothed idle down considerably.

I think I've realized what's been causing my problems - tuning with too little data. After gathering logs for several days and setting a target of -3% in trimalyzer I was just applying all changes. What helped immensely was setting the minimum record count to a fairly high number. For me 750 really helped weed out the cells that haven't gotten enough "law of averages" applied to them.

This has brought me to the realization that I'll be logging data all summer to get enough to correct the areas outside of normal cruising. Calibrating my MAF will have to wait until at least next summer. Will probably need new tires by then as well. :)

Hi
I've just pent the last few weeks using trimalyzer to get the ve tables all smoothed out.
After spending 2 evenings swapping the shifter (springs gone soft) for the hurst the wideband went to shit.
I'd trapped the 02 sensor cable.
In the process the ve tables are now screwed.
Went OL. It was idling at 22:1.(quite happily)

Ve tables were maxed so dropping injector size 2 points (5%) it was idling at about 12:1. Pe is about 9:1

F***me does the car like that. It howls now.

Obviously explains why o2's and wideband didn't quite match but was never that far out when using CL

It definitely does like the fuel.


After The run (a really good one I might add ��) Idle had dropped

Quote:

---

Originally Posted by Terminal_Crazy

I'd trapped the 02 sensor cable.

edited for brevity...

Obviously explains why o2's and wideband didn't quite match but was never that far out when using CL

---

I'm not sure I follow, but I'm intrigued. Care to explain better - I can be a bit slow on the uptake.

Idling at 12:1 must smell pretty raunchy - are you parking in a garage? Woof!

I just spent some time fine-tuning the A/C and did a bit of test driving. When I initially charged the system last sunday (before applying VE corrections at idle) it died twice on me. With the "better" VE corrections it actually idles a little more steadily with A/C on, likely from the additional load causing more IAC opening / airflow. I can't imagine running a cam like this with an aluminum flywheel. The way it's running currently I can back into my driveway without instantly killing it, but backing uphill into the garage is still a little dicey.