Open Loop EE Advice

[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.

[spfautsch]

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.

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...



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

[Terminal_Crazy]

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

[spfautsch]

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.

[steveo]

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...

[spfautsch]

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"?

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?

you could lug it up hills.

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

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. :-(

[steveo]

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)

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.

[spfautsch]

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.

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.

[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:

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.