Now here's a question..

If i can run 19.5psi and my o2 is 0.92..
Can i just keep turning up the boost till it gets too lean? What reading is too lean? What should i stay above?

Also.. i'm at 0.92 BUT with the Mike J Maf mod i still have lots of room to richen it up.

So ideally, would i make it as rich as possible, then turn the boost up until it runs too lean (just above that of course)?
And if that's the case.. i could do the fuel pump rewire/upgrade and that would even make it richer and allow me to run more boost right?

What else should i be monitoring/be weary of.

BTW: I'm running a 16g, 3" turbo-back, stock (yes stock) fuel pump.

------------------
13.904 @ 99.878mph - still need to learn how to drive

You still have to be careful of detonation. A lean mixture isn't the only thing that will cause it. There comes a point where, for whatever octane rating of gas you're using, the pressure and heat will ignite the fuel no matter how rich you're running.

I don't think I'd reccomend more that 19 - 20 on street gas.

If you want to run race gas all the time though, that's a different story..

Jeremiah, don't trust only the o2 sensor's output, you also need an EGT to give you the other perspective on your picture. My mechanic told me (can anybody confirm this?) that when the o2 is exposed to sufficient heat, that can cause it to send a "rich" voltage signal, even though it could be running lean.

A better indication of when you start to run lean is that your EGTs will suddenly start rising quickly, whereas before they were rising at a steady rate. I observed this myself, watching how the the EGTs suddenly went up (the injectors were maxxed out), but my o2 was still showing full rich.

Please be careful.

------------------
Marta
dsmgrrrl@hotmail.com
90 Eclipse GSX
92 Talon TSi AWD

I'm not sure I believe that information regarding the oxygen sensor. I don't know for sure, though.

One thing I do believe is that the oxygen sensors were never designed to do anything but switch at a 14.7:1 A/F ratio. In other words, they were not designed to be consistent nor accurate at one extreme of the measuring range. There was no incentive for the sensor manufacturer to test the sensors for linear or consistent output above 0.75V.

This means (IMHO) that you really cannot trust the oxygen sensor nearly as much as people say. While checking the O2 sensor output is much better than nothing, it is also much worse than using an EGT gauge to monitor internal engine temperatures.

Jeremiah, were you to datalog your car, you would likely find that your timing is being reduced by your ECU because of knock. It seems that most people who datalog their cars (or who use TMO gauges) find, much to their surprise, that their knock sums are pegged and their timing is way back despite high oxygen sensor readings and relatively low EGTs. You might find better performance by lowering your boost some, so the ECU can advance your timing more.

I also not sure exactly what you are doing to run 19.5 psi on a 16G without getting fuel cut, but that will probably make a difference too.

The only other difficulty that you may run into depends on your car. 1990-1992 cars had a weak OEM intake manifold gasket that can be damaged by high boost pressures. Later cars, or cars with newer gaskets, have a metal gasket installed that will take the pressure.



------------------
S.
1000 AAQ: 1000q.dsm.org
ECU Primer: members.home.net/costall/ECUprimer/index.html

So it looks like i need an EGT then eh?
I really would like to datalog my car.. but i have no means possible
No fuel cut.. thanx to Mike J for his MAF mod.
And i have a 93, so i have the metal gasket.

So you're saying, if the timing is retarded that will make me go slower than if i reduced boost pressure?
And..is there to monitor the timing? How would a knock LED help?

------------------
13.904 @ 99.878mph - still need to learn how to drive

~~~~~~~~~~~~~~~~~
So you're saying, if the timing is retarded that will make me go slower than if i reduced boost pressure?
And..is there to monitor the timing? How would a knock LED help?
~~~~~~~~~~~~~~~~~

Yes 20 retard of timing is much more of a hit then 2 psi of boost.

Monitor timing... I'm been playing with that idea.... but haven't got it yet. The best bet is the datalogger to watch your knock sum.

A knock LED is helpfull to tell you when your knock sum has maxed out but not good for telling if your timing is being retarded and to what extent. Here's how it works:

If there is very little knock the nock sum will be low... if the knock sum is less then 3 then the ECU will start advancing timing. This will eventually increase you knock and thus your knock sum. Once your knock sum goes over 3 but less then 7 your ECU will no longer advance your timing... but it won't retard it either. But once the knock sum goes over 7 the ECU starts retarding your timing... more retard for higher knock sum. Once your knock sum goes up to 43 you ECU has already retarded the timing to it's maximum and now decides to shut your boost down through the BCS. This is where you would see the knock LED come on. So it helps to tell you when things are really bad... but say your knock sum was 40... you'd have huge retard and not see the knock LED.


Clear?


------------------
Mike J

-Quadcylla-
92 Laser RS-T, FWD

Vortex Racing
http://www3.sympatico.ca/robin.jackson2/HomePage.htm
Alt Email: vortexracing@hotmail.com

Kinda clear..

So the LED only comes on when the knock sum is maxed out right? So you can't tell when the sum is 3 or 7 or whatever? So basically it tells you when it's too late then, because by then, it's retarded your timing and reduced your boost right?

Okay.. i've got an EVC so i don't have a BCS, does that mean the most it can do is just retard the timing? Or will it go to the ecu to reduce boost also?

So how did you come up with that number or timing retard (20) to boost? (2psi) Is there a scale or ratio (like... knock sum of 10 = retard of 20 > 2psi?)

So figured how to monitor it already!
Your MAF mod allowed me to run huge boost, now help me max the car out so the only variable would be my driving skills (or lack thereof)

------------------
13.904 @ 99.878mph - still need to learn how to drive

This may be a bit of a newbie question, but how can we tell when the engine is beginng to be retarded? I guess the LED mod would only tell you when your fully retarded, is there anyway to tell when the ECU starts to pull it back (without datalogging software)? Is 15psi on stock turbo with nearly full rich ok? Just sitting here late at night and this post got me wondering.

Ya it will only tell you when things are really bad and by then it's too late for you performance... but it lets you know that you've got a dangerous situation in you engine and you'd better get off the gas. The only I know of so far to monitor that stuff is the datalogger. Give me more time to look into other ways.

If your BCS is not hooked up your ECu can't touch your boost level. So it is up to you to reduce it at that point... unless you're feeling lucky

Anyone out there have a full schematic for the MPI system that I can look at to help me with this project?



------------------
Mike J

-Quadcylla-
92 Laser RS-T, FWD

Vortex Racing
http://www3.sympatico.ca/robin.jackson2/HomePage.htm
Alt Email: vortexracing@hotmail.com

Mike,

How does the Knock sensor work to tell the ECU at which point to retard timing? Is there a voltage output, like the 02 sensor that the ECU gets it's readings or is there an internal calculation done?

Is it possible to tap directly off the 02 sensor or is the BCS the only option. If I knew how the system actually came up with those values (0-3-7-43) then I might be able to help you design a way to monitor the system.

Jazz..

Okay now you're getting close There is a detonation sensor in our cars that just listens for knock. I assume every time it hears something it sends a voltage spike to the ECU thru pin 9. I also assume the louder the knock the higher the spike. The ECU has some circuitry that sits there listening to the detonation sensor and keeps track the total sum of knocks over a given time frame and keeps refreshing that value. Let's say it's over 10 seconds... the ECU keeps track of how many spikes there were over the last 10 seconds... this is knock sum. As knock increases in frequency (happens more often) the knock sum will increase.

So one idea is to tap into pin 9 and run a LED off it to show you every voltage spike. This does not give you a usable value but I assume you'll be able to see whether there is more knock or not by the frequency of flashes.

So can you tap in anywhere and read the actual knock sum? Well there is only one way since this value never leaves the confines of the ECU... and that's to ask the ECU for that information thru the diagnostic port. Well then you need to know the protocol to ask for it and interpret what you recieve. This is what the datalogger does (among other things).

So can we emulate the knock sum value? Hook up a listening circuit to do the same calculation the ECU is doing to arrive at it's knock sum value? This sounds cool but you'd have to find out over what period of time the knock sum is looking over. You'd also have to figure out how much of a voltage spike would result in a recognised knock by the circuitry. I don't know if every spike is counted or to avoid every noise counting it takes a certain amount of voltage spike to count. There are so many unknowns here and to get it all right in the end you'd have to have a datalogger to confirm whether you are close or not.

I think you might have to be careful too with adding LED circuits or listening cicuits cuz you don't want them to draw so much power away from the signal that what should be triggering as knock doesn't.

I've been pluggin the digest with questions about this stuff and have been getting no help so we're in no man's land here I'm still willing to work on this if there are any ideas out there. I will prolly do the LED set up myself... hooked to a switch cuz it will flash a lot I'm sure.


------------------
Mike J

-Quadcylla-
92 Laser RS-T, FWD

Vortex Racing
http://www3.sympatico.ca/robin.jackson2/HomePage.htm
Alt Email: vortexracing@hotmail.com

[This message has been edited by Mike Jackson (edited May 02, 2000).]

If the ECU actually does do it's measurements that way then why not biuld a counter instead. It'll save you from having to count each blip of the LED. You could also add a timer to reset the counter every 10 seconds.
I think you should ask Todd Day about this one.

Jazz..

I did, but he didn't have anything to help . I also guessed at the 10 seconds... no one seems to know the actual time span. Plus it doesn't reset it's self every 10 seconds it's constantly reflecting the knock over that time span at any given time. The way you are suggesting would have 0 knock sum every 10 seconds and then it will build until the next reset. See what I mean?

------------------
Mike J

-Quadcylla-
92 Laser RS-T, FWD

Vortex Racing
http://www3.sympatico.ca/robin.jackson2/HomePage.htm
Alt Email: vortexracing@hotmail.com

Jeremiah: an EGT will help, yes.

Another problem is that no one is really sure what type of signal is output from the knock sensor to the ECU. It seems that the ECU has some significant amount of processing circuitry and code which are dedicated to the knock sensor.

It would likely be a lot easier to build a monitor on the ignition timing, since all of the signals are externally accessible.



------------------
S.
1000 AAQ: 1000q.dsm.org
ECU Primer: members.home.net/costall/ECUprimer/index.html

YES!!!
monitoring the timing would be great. I know that Mike J was looking into that.

I didn't get any replies to my post to the digest about monitoring timing ... Sean if you know anything that would help please feel free

------------------
Mike J

-Quadcylla-
92 Laser RS-T, FWD

Vortex Racing
http://www3.sympatico.ca/robin.jackson2/HomePage.htm
Alt Email: vortexracing@hotmail.com

Oops. See the other thread about monitoring ignition timing.



------------------
S.
1000 AAQ: 1000q.dsm.org
ECU Primer: members.home.net/costall/ECUprimer/index.html