10 second 190,000 mile low boost 3.0 build

[Force Fed Mopar]

You need to check my 2.6/3.0 interchange thread

[Ondonti]

http://autospeed.com/cms/article.html?&A=0353

anyone think I can find some sort of voltage output on this device that correlates to volume or something? Still might try this. would be nice when tuning someone else's car etc.

And after doing some more reading, I am going to probably pick up a knock sensor buffer, which ups the low current from the knock sensor into a useful 0-5v current.
Vishnu sells one with the Exed engine management setup. 100 dollars though. Anyone have a cheaper buffer idea?

[Aries_Turbo]

the problem with trying to derive a 0-5v signal from the knock sensor is that the knock event is so fast that youll easily miss it. also youll have more background engine noise as the revs go up so the signal that means knock at 1500rpms is going to be drastically different at 7000rpms.

listening with your own ears is the best bet short of a microprocessor based system.

Brian

[Shadow24]

http://www.viatrack.ca/

Interfaces with the MS real easy and has an LED indicator. I have mine but haven't gotten mine installed yet its only $60 without a sensor though

[Ondonti]

the problem with trying to derive a 0-5v signal from the knock sensor is that the knock event is so fast that youll easily miss it. also youll have more background engine noise as the revs go up so the signal that means knock at 1500rpms is going to be drastically different at 7000rpms.

listening with your own ears is the best bet short of a microprocessor based system.

Brian

Well the idea is not to be looking for something, but just to use some of the abilities I have. Thats why I would like to have both systems up and running. This is more of a datalogging thing.

Just because you hear knock doesnt mean you know what RPM it happened at. Logs would allow going back and pinpointing when it might have happened.
MS has its own threshold system for reading knock voltage (based on rpm) so I think the knocksense is sorta silly because you will end upalways having high rpm knock if you wish to see low rpm knock (basically the knock sensor will go off the moment your rpms get around 5k even if you are not knocking).

Now if this device is accomplishing the exact same thing as the vishnu buffer then it is cheaper. My issue is that i don't want it guessing what is knock, I just want it increasing the current so I can datalog. Then let me decided.

I also found out that the 2nd wire on my knock sensor is shielding wire to help knock out interference from the alternator etc.

[Aries_Turbo]

oh really? i didnt realize that MS had its own RPM based knock table. i may have to look into that. I can easily whip up the interface circuit.

EDIT!!!!!! I just looked at the MS2, MS2 Extra and the MS3 documentation. i dont see anything about a 2d RPM referenced threshold table. i still dont believe its implemented like that yet.

brian

[Shadow24]

do tell. AFAICT you had to feed it the signal somehow (either the GM conditioner or the KnocksenseMS unit) THEN it could "see" knock

[Aries_Turbo]

it (the current MS scheme) doesnt have the capabilities like i want. thats why i need to finish my knock interface box (been too busy on other stuff). it has a 2d rpm referenced threshold table to determine at a given rpm, what the knock threshold level should be.

the only aftermarket knock solution that i know of that claims to do this, is the turboxs unit. http://www.turboxs.com/more_info.php?ID=212 that unit wants you to rev the engine so that it can determine overall noise levels but it seems that they want you to do it at no load sitting there but the ringing of the block changes the overall noise level and that depends on load and boost. a block rings more at 15psi than it does at no load and it rings even more at 20+psi.

the factory turbo ecu's do too but that doesnt help 3.0L guys.

Brian

[Ondonti]

This is what I am refering too
A simple knock sensor buffer would do the trick. If the knocksense can do it for cheaper then great, I just don't want knock counts, I want straight voltage and let me decide what is knock based on listening device etc.
so if the knocksense can be used without it saying "this is knock" then that would be nice.
The vishnu just ups the tiny current from the knock sensor instead of trying to interpret anything. The problem is that it costs 100 dollars. The knocksense seems like a bad idea when it comes to integrating it with Ms.

Brian, if you could make something cheaper, I think you could sell units to just about anyone with LC-1 or zeitronix since all they would need is the buffer and a datalogging system (and a brain that can help them interpret results).

Ive read before about the knock sensor settings. Aggressive and safe modes etc. I would probably use safe mode because I would want to actually remove that timing rather then rely on active knock control.

[Aries_Turbo]

what version of the megasquirt software has that table? ms2?

ive never seen that table before. that is great. its basically the same as the factory ecu stuff minus the individual cyl timing control.

can you take a pic of your mitsu knock sensor that you want to use? i assume that its a narrow range one that is tuned to a specific frequency and doesnt require any additional tuning.

also, what is your bore size? if you find other engines with the same bore size that the knock sensor is a tuned narrow range sensor, you can use it on your motor.

Brian

[Aries_Turbo]

man.... i just looked. i never saw the description of that table before. that makes it so much easier now. after we determine what type of knock sensor you have, i should be able to whip up a circuit pretty quick.

Brian

[Shadow24]

hmmm, i'd be interested in trying your circuit to see how it compares to my KnocksenseMS. what is different about your circuit?

i remember that table, i just was thinking there was a table table (like VE or spark) for knock that i didn't know existed

[Aries_Turbo]

with the knocksense, it determines when you have knock.

when doing it my way, you let the table in the MS determine when you have knock based on the rpm values which is more accurate.

you have to use a narrow range knock sensor thats specific to your bore size though rather than the bosch one that the knocksense uses. a sensor similar to the stock turbo chrysler one.

Brian

[Ondonti]

Uh well I guess I can dig around for a picture. how am i actually going to help you determine anything with a picture?

[Shadow24]

lucky me has a stock mitsu sensor that came with the motor from the factory

[Ondonti]

The knock sensor I have is OEM 3.0 6g72 12 valve knock sensor, probably the same one that is on the DOHC's. I am using the mitsu block so it has the knock plate in the block valley with sensor installed.

[Aries_Turbo]

i can pretty much tell if the knock sensor is a wide range or a knock sensor tuned to a frequency by its physical construction by looking.

here is the factory chrysler knock interface circuit with component values.



for the "output microprocessor" output to clear the capacitor you need to make a circuit that applies power to the base of the capacitor at the exact moment the coil fires. the megasquirt rpm coil pickup circuit would work for this. you would have to take out the coil and spin the distributor till you see the coil firing pulse (use a meter) and make sure that the circuit provides +voltage to the base of the capacitor clearing transistor.

this excerpt from some factory documentation should help that make sense.

1985, '86 & '87 Detonation Sensor and Circuit

Anytime the engine is detonating (above .8 volts at 2,000 RPM or above 4 volts at 4,000 RPM), the detonation sensor outputs an a.c. voltage on the BK/LG wire. This voltage goes into the logic module to the base circuit of Transistor 1. See Fig.30. The resistor and capacitor in parallel with the base circuit of Transistor 1 stabilizes the base circuit voltage. Transistor 1 acts as diode. That is, any positive voltage on the base of Transistor I is allowed to go to the collector and charge up Capacitor 1. See Fig. 30. Since the voltage output of the detonation sensor is a.c. voltage, the voltage impressed at the base of Transistor I will go from a positive voltage to a negative voltage. The positive voltage that has charged up Capacitor 1 now "wants" to go back to the detonation sensor. This voltage is blocked at the collector of Transistor I. In this way Transistor 1 acts as a diode, allowing current flow in one direction (charging up Capacitor 1), and blocking it in the other direction.

Once a capacitor is charged up with voltage, it will keep that charge for a very long time or until it is discharged to ground. The voltage stored in Capacitor 1 is impressed at the input to the A/D converter. See Fig. 30. The logic module monitors the A/D connector every time the coil is fired. Once the engine has detonated, Capacitor 1 voltage level increases. The logic module now takes away the maximum allowed advance on just the cylinder that is detonating (as determined by the sync signal) by 4.2 degrees up to a maximum of 2.8 degrees when there is part throttle and a no boost condition. When the engine is in boost, 2.1 degrees up to a maximum of 15 degrees is taken away from total timing. After the coil has been fired the logic module monitors the A/D converter again to "see" if the voltage has been lowered at Capacitor 1. Remember it was said earlier that a capacitor will hold its charge until allowed to go to ground

here is the MS coil circuit. IRQ1 goes to the base of the capacitor clearing transistor. youll have to refer to the MS 2.2 board documentation to see what to do with the XG1 and XG2 jumper.



im sure ill have to clean this info up a little more but here it is for now.

Brian

[Ondonti]

I follow the diode talk cause I ended up using bridge rectifiers on my fuel pump voltage drop kit but I don't understand the point of the coil stuff.

Here is the sensor. Looks the same as the evo 8 sensor in this pic. I also have the connector and pigtail on the other side of the connector with 2 wires coming out that instantly turn into weird shielded wire again. I think the shielding wire is just grounded eventually or something...?

I am guessing they are not exactly the same because you can get the evo sensor for cheaper then the 6g72 sensor

here a 3000gt one looks exactly the same.


From what i read on an EVO site (and didn't really understand) the knock filters were program'd in dsm's to be a pitch higher then what the actual sensor/motor was...something like that.

Anyways I know a knock buffer works because people have used the vishnu device on evo's.

[Ondonti]

Here is real oem 3000gt parts






torque to 17 ft/lb.
I did just read that 1g dsm vs 2g dsm have different knock sensor values/frequencies. 2gdsm and Evo seem to be similar but use different filters in the ROM (ecu) and it might be possible that certain part numbers only differentiate wire length and connector style etc while the sensor might be the same.

[Aries_Turbo]

looks like a tuned range sensor. should work with this circuit.

the coil stuff is because the knock circuit has a capacitor that charges up and holds the charge to give the ECU time to take a sample. once charged up, a capacitor doesnt lose its charge (unless shorted to ground) so you would always get false knock because the voltage would stay in there, sending that signal to the ecu. the coil circuit is to discharge the capacitor upon each ignition event so that you start fresh (ie ~0v) after the ignition fires and listen to that cylinders knock.

when that coil interface circuit is connected to the negative side of the coil, when the coil fires, a positive pulse will be generated because the ground side is disconnected at the ecu and you now see the voltage from the positive side of the coil. this pulse powers the capacitor clearing transistor on the knock circuit. i need to do a little more digging. i forget how long the coil circuit is off for. if its too long, the knock circuit wont have enough time to charge up the capacitor for the ecu to take a knock reading and you wont get high enough voltages.

i may have to add another device to make sure that:

1. the pulse that is sent to the transistor is a positive one

and

2. its short enough to ensure proper operation.

Brian