Mad Scientists in here

[aminga]

So I bought one of the new ISAC controllers and in planning for the installation I had an idea. It was such a rare occurrence I thought I'd let everyone in one it

So I witnessed a case of sudden onset head gasket failure Saturday and got to thinking I'm sometimes guilty of not watching the Scanmaster going down the track the knock alarms and knock gauges are nice but I always found my gauge a little touchy about reporting knock. So on the ISAC I believe it's possible to trigger "Launch boost" mode by grounding one of the inputs to the controller. If I could ground this input at a certain KR level I could add another failsafe to the mix.

Now the question is how.

 

[dr_frankenstein]

could use a latching SCR.

when the input recieves enough voltage it will trigger and latch.

thus grounding your ISAC to return you to low boost mode.

the hard part would be determining at what output voltages of the sensor you would want to trigger the SCR. depending on the variables of the SCR and the type of Knock sensor, some juggling with resistance values should find the voltage range for the particular SCR.

Adding a Pot could offer some adjustability on the fly if you wanted to control the level of knock required to trigger the SCR.

A.j.

 

[rmar]

The old silicon control rectifier in the hat trick......

 

[JayC]

You really dont need to do that. You can use one of the existing failsafes in place of what youre trying to do. Im pretty sure the Alc Low Flow/Pressure failsafe is just looking for a ground.

Getting the knock signal back in there, youre on your own there. The only ways I can think of to do it are kind of expensive.

 

[Razor]

The problem I see is false knock. If we cant get guys to figure out if they have real or false, how can we get a circuit to do that. The other thing is the knock circuit works by applying a ground. The longer the ground is applied, the higher the knock.

Couple things can pop a headgasket. One is detonation, the other is excessive cylinder pressure. Assumming :redface: engine was built properly.

The detonation can be from a few different angles. Lean and octane are the two cuplrits. Lean can be handled through the WB sensor, octane is alcohol amount in terms of an alky system. Alcohol amount can simply be done as pressure, then set a low limit based on boost run.. if it goes below that limit, trip your gate. You need to establish a baseline watching your knock sensor and engine performance to say it needs X amount of pressure to support Y amount of boost.

Sometimes **** happens when racing. Plenty of alcohol, plenty of fuel, proper timing and air fuel.. and the gasket ****s the bed. After all cars on C16 blow stuff up.

 

[aminga]

Jay, I was wondering if you guys had thought of that. Apparently so.

Razor, yea the part of this little project that is hard is false knock. If you limit it to Knock in third gear you might get some of it.

It my not be dooable but it's been a little mental diversion for me.

I'd love to see someone do a "recognizing false knock" in the How To section.

 

[JayC]

Jay, I was wondering if you guys had thought of that. Apparently so.

Well.. there's a lot of reasons to not do it that way. Like Julio said, false knock would be a real issue.

Not to mention your ECM already has a built in provision to deal with knock.

Doing it with drops in boost being the method of control would be a chore within itself. You'd have to figure out what attack rate you wanted to use on the boost, you'd have to figure out how the attack rate effected the boost, etc.

Really the way the ISAC's failsafe work is much better. AFR leanout will save your engine long before knock detection would. With the ISAC, once your A/F ratio goes half a number above your target, it drops boost back as far as possible. You're going to get to the trigger A/F ratio long before you get to knock unless you're running your system on the very edge of detonation.

 

[aminga]

Well.. there's a lot of reasons to not do it that way. Like Julio said, false knock would be a real issue.

Not to mention your ECM already has a built in provision to deal with knock.

Doing it with drops in boost being the method of control would be a chore within itself. You'd have to figure out what attack rate you wanted to use on the boost, you'd have to figure out how the attack rate effected the boost, etc.

Really the way the ISAC's failsafe work is much better. AFR leanout will save your engine long before knock detection would. With the ISAC, once your A/F ratio goes half a number above your target, it drops boost back as far as possible. You're going to get to the trigger A/F ratio long before you get to knock unless you're running your system on the very edge of detonation.

Here's what got me spooked. This was from my last trip to the track same tune as I've run for the past 3 outings No change in boost. But I stopped at a different gas station to fill up.

The first and second gear could be false but on the first run there was knock in third gear out of nowhere.

I need to fix a few things in my combo. On a stock turbo I'm not moving enough air for the TT correction to be stable, it's too fast for the turbo I have. Eric made it better but the real solution is on my work bench right now (a TA49)

Like I say it was a thought experiment that kept me occupied.

 

[JayC]

Thats one of the reasons wideband correction is a tool and not a crutch. Thats why I recommend using the wideband to get your tuneup close and then disabling it. Sounds like he has the gain too quick in the chip, which makes the wideband really erratic.

 

[WSLN 6]

You could add the knock detector into the mix as a safegaurd by using a third gear switch as the "on" for a relay that is triggered by the knock sensor. The relay could then trigger the low alky on the ISAC. Maybe use a diode inline to avoid backfeeding?

 

[Razor]

The 2004r has a 3rd gear switch built in that could be used.

But I have seen cars blow headgaskets in 1st and 2nd gear.

I have also seen headgaskets go cuasing the knock sensor to go off after the fact.

There is no failsafe for racing. I agree with Jay, wideband is the first issue to watch. I run a FAST Dash logger and setup my alarms based on input values I want/dont want to see.

 

[JayC]

You could add the knock detector into the mix as a safegaurd by using a third gear switch as the "on" for a relay that is triggered by the knock sensor. The relay could then trigger the low alky on the ISAC. Maybe use a diode inline to avoid backfeeding?

Well, again, I just dont feel like using knock is the best course of action. With the stock ECM, you can set a quick retard attack, set up a big timing pull for knock and I mean, thats pretty safe. Trying to do it with boost just presents some issues, even if you can get the signal back in.

Air/Fuel is the way to go. Plus you have overboost protection with the ISAC as well as low pressure/low flow warnings you can plumb in. If you want to utilize knock for protection, Id do it from a timing standpoint rather than trying to integrate it into controlling boost.

 

[salvageV6]

Use a 4017 counter chip and a CMOS flip flop to select how many of the knock sensor pulses to ignore (if any).

I made some updown and countdown timers to work the knock counts on the fly all with early 80's rocket science technology.

Won't count down when counting up ie. pulses still on the line, counts down when knock clears.

Then you can use that output to ramp up your pump or when you reach a high count number kill the boost solenoids and back off the boost.

All you do is shift your MAP sensor voltage to trick the pump into thinking there's more or less boost present, just an offset.

A small computer PIC microcontroller chip with basic could probably do it easier but I like CMOS as there's no inherent software or timing delays excepting a few nanoseconds.

It's a bit cumbersome to decide the speed and counts to use but 3rd gear safety works the best, and you should try to eliminate false knock with most setups anyways to keep timing from being pulled.

80's rock steady CMOS technology or you can cut open a Caspers knock gauge and use the last green light or first red light to do something.

The ISAC failsafes are nice and simple to use and work well for sure.

I use the ISAC now and removed most of my electronics.

If you want to get into KR control you first need to build a counter circuit to hang on the blk/yel. stripe wire to watch how the bursts occur and how many counts you get relative to what you see on the scanmaster, timing pulling via the ECM, and what if anything you hear.

Then you can decide what counts to take your action(s) upon.

A test circuit I built has a 2 digit counter as well as BCD 4 digits LED counters with a couple of chips.

Also used a turn on timer to ignore cranking KR, simple reset button to clear the counts.

I haven't done up any schematics yet I just work with chips I have from whatever junk I can scavenge from and have saved up over the years.

It's not hard to do if you have any CMOS chip skills some basic circuits are in the Radio Shack books you could buy as well.