Soo, maybe others are apprehensive about E6K lack of knock sensor input.

I discussed an idea I had w/ a technician at my work who is helpful to me and into cars and it *might* just work.

I theorized that one could use the factory knock sensor and computer and have its output translated in a "black box" and have that "blackbox" output spliced into one/several of the E6K sensor input's correction map that can affect timing and possibly fuel.

With no knock sensed everything would be as normal, but w/ knock you could have E6K think there was a change in one of its correctable maps and adjust accordingly.

This tech. theorized one could use a PIC chip that is around $100 and easy to program. He used one in a Battlebot he helped build and was surprised by the programming simplicity.
He said he would engineer it for free since it sounded interesting- He just needs to know 1.factory knock sensor output type 2. Haltech sensor output types to correction maps and what affect they have on map.

So questions- is factory knock equipment worth employing w/ E6K? Anyone else try something like this? Is anyone interested? How much would you pay for knock control for E6K?

Thanks, Ian.

 

The only ignition adjustment map I know of on the Haltech is the cold cranking one.&nbsp That makes it useless unless you trick the Haltech into thinking it's at cranking RPM.

Nice try...


-Ted

 

I was thinking more along the lines of converting 3-5 volt Mazda knock computer signal to the equivalent of some more PSI to the MAP sensor. Gotta have a decently higher PSI MAP sensor than boost you are running for this, so it might cost some more for people running near the limit of MAP sensor.
Yep, fooling the Haltech is what this operation would be all about, just like piggy back system on a factory ECU. Piggy back for stand alone...

 

If you alter the MAP (load) sensor, wouldn't that adjust the fuel delivery also?&nbsp Or was that your intention?


-Ted

 

Your e6k has 2 inputs. Go to the input/output page. you can set either trim control or spare input function to IGNITION . look at the wiring loom near main haltech plug. There will be 2 weatherpack connectors. the ceter wire (check haltech diagram) will be a 0 - 5 volt input. Using this input you can trim your total spark advance
-7 to +8 degrees. Get your technician to determine if 0 - 2.5 volts or 2.5 to 5 volts on the input causes your emgine to retard. once you have determined this it will be a fairly simple task to interface the mazda knock amplifier to this input. A number of things to bear in mind, are that

1. Mazda only allows knock retard from around 800 rpm to 4500 rpm. The actual rpm is in series 4/5 workshop manual.

2. Knock will generally only occur a number of degrees after a sparkplug has fired. you will need to consider this in your knock interface.

3. Mazda takes these variables into account to prevent false knock or general mechanical engine noise from triggering knock retard. Engine mechanical noise increases with rpm

4.Most major OEM's use a similar setup to mazda

5. "Tricking sensor inputs is no way to acheive you aim. It is 2002 not 1975. Surely we no longer have toaccept this compromise.

Chris

 

A couple more thoughts, you could either do a propotional spark retard scheme or a jump to max retard setup . Basically proportional would mean that short duration knock would give you a bit of retard and long duration knock would give max retard. As for the amount of time it would take to effect the retard it would depend on how often the processor accessed the input. As far as bringing the spark advance back to normal you could use a time constant such as an r/c circuit to bring the advance back in slowly. This helps to avoid a re-occurence of Knock. As far as the rate of retard/advance is concerned most ecu's retard/advance from 1000degrees/sec to 7000/degrees per second so this is really not an issue.

Chris

 

If you alter the MAP (load) sensor, wouldn't that adjust the fuel delivery also? Or was that your intention?

Yup Ted, I thought enrichening and retarding were both a good plan to halt detonation.

Autotronic- I hadn't thought of using the trim controls for this, like I said its new to me. I don't see how trimming this way is superior to adding to MAP signal; I would have to run ignition and fuel trim instead of just MAP input, right? Just less of a hack? I need to make the tech read my Haltech manual and come up w/ ideas. He's a wiz at programming PLCs and such.


1. Mazda only allows knock retard from around 800 rpm to 4500 rpm. The actual rpm is in series 4/5 workshop manual.

From what I've seen, the knock control computer only has input from knock sensor, it has no idea what rpm the engine is at. This must be the ECU ignoring the knock computer input at these rpms.

2. Knock will generally only occur a number of degrees after a sparkplug has fired. you will need to consider this in your knock interface.

Right, otherwise its pre-ignition, not detonation or "knock". OK, you have to spell out to me why this matters? I planned on setting up Haltech as normal and then using knock control as an extra safety margin. If any knock is detected I want instant and full counter measures and I will stop and find out why it happened. I know I can't get my foot out of it when I hear knock as fast as the computer can pull timming and add fuel. I don't want to set it up to run "on the edge" of knock- especially w/ 15yr old sensor and computer

3. Mazda takes these variables into account to prevent false knock or general mechanical engine noise from triggering knock retard. Engine mechanical noise increases with rpm

Yes, the knock computer seems to have components that would filter the signal and look for the right frequency of sound. The tech identified one timer in particular that he said was for very fine increments in time.

4.Most major OEM's use a similar setup to mazda

Yup, its just Nippon Densho stuff. In the '80s Japanese automakers agreed to consolidate componentry to keep costs down.

5. "Tricking sensor inputs is no way to acheive you aim. It is 2002 not 1975. Surely we no longer have toaccept this compromise.

He he, the piggy back industry is really taking off recently, but where was it in 1975? Seriously, this is a Haltech E6K that runs in DOS, not exactly up to the minute or even a Motec. But, it won't kill me to do it the "right way" so I'll look into trimming.

We hooked the knock computer up and put it on the scope. We could see a great signal from the sensor that really varied w/ how we tapped on it. When we put the knock computer output on the scope we weren't able to get anything no matter what we did. It doesn't seem that it should be too picky about what sound it is considering knock since the test for it is a plastic hammer on the engine lifting eye and that sounds nothing like what I heard when my engine knocked.

Anyone have more good ideas? Try another knock computer? Give up? Thanks for the replies, Ian

 

Your typical knock sensor is just a fancy piezo microphone.&nbsp The knock box should be doing the frequency filtering, since the knock sensor itself isn't sophisticated enough to do this by itself.&nbsp I believe the magic number was 5,000Hz or something very close to it; I found this number a while ago perusing a bunch of SAE papers that went into knock information on a rotary engine.&nbsp So the knock sensor picks up ALL engine noise information, and the "black box" does all the filtering - only the pertinant[sp?] magic frequency should trigger an output signal from the black box to the ECU...


-Ted

 

5K Hz, hmm. Thanks Ted, I will see if I can duplicate that frequency and get some output out of the knock box.

 

I did do all the setup and testing of this knock box for Jon (RX ENGINEERIG) as he was looking to have a box with a light and siren on it to tell him about engine knock. This was to be used while tuning. I could get an output from the box shown on a oscilliscope by simply tapping the sensor on desk lightly. Never did finish the setup.

Chris

 

Thanks for the info Chris! Sounds like I might have a bad knock box Ian.

 

I had heard a solution a while back about using an input set to NOS to signal a timing retard event using some sort of simple circuit.

You guys know this stuff in amazing detail... you need to come over to the piston side and kick it around for a bit so that us piston-based mortals can have some good insight into what the hell we're really doing.

__________________________
MY00 Subaru Impreza 2.5RS-T

 

I think you may be able to adjust the ignition by use of the "ignition trim" if I remeber correctly. That would use resistance to vary the timing.

 

Yeah.
if the stock unit sends a ground or 12V+ signal, you could hook up this unit to a relay and have the relay complete the circuit to the ignition trim whenever it sees a knock. You would have to preset the amount of retard. Just make sure this setting is not changed accidentally. It would be bad news getting more advance timing when your engine sees a knock.

 

Instead of trying to kludge the Haltech, why not so what some OEMs (SAAB, at least) do and use the knock sensor to trim boost pressure? Either use it to control the wastegate, or an electric/pressure-driven manifold popoff valve.

 

A relay would have to much time delay in its operation, but you are on the right track.

Chris

 

Chris,
Tell me more. I figure even at 8500 rpm an engine has 7ms between power strokes for any rotor. I don't know how much time a relay may take and not sure what other timings may need to be accounted for.

 

Quote: Instead of trying to kludge the Haltech, why not so what some OEMs (SAAB, at least) do and use the knock sensor to trim boost pressure? Either use it to control the wastegate, or an electric/pressure-driven manifold popoff valve.

I wanted to set this up as just a bit more security if my WG fails, pop-off valve fails or the Haltech tunning can't handle something (bad fuel, etc). Not to run knock control as part of normal operation. My opinion is that this doesn't work well w/ boosted rotaries as one knock incident can toast an engine.
Its on the "back burner" right now though- just gotta get it running :-). Thanks for posting your ideas everyone! I appreciate the input.

 

A relay would be too slow in the transition off-on and on-off you want to look at using a transistor.

Chris

 

Now that I am totally confused about how to get rid of knock...
Check out my mods and tell me what I need before I try to tune my engine (or even start it). I do have the Haltech trim ****...
I don't want to ruin my new setup. Let me know!