MJLJ V4 + Knock sensor = possible good idea

[Salamander]

Hello
I thinking of making a sensor fusing like this.



The signal emitted by the Knock Sensor to the buzzer, you can take and use the input to the MJLJ V4, to lower grades of progress? or not?



is very good idea, but not much the topic of electronic order to implement and can work to improve the system.
someone you think of a better idea?
sorry my english

[Supercharged Nat]

thats looks like it might work! as its been said before, determining what knock actually is for your particular engine is the main problem- it might be happening but the sensor you have off say some other engine might not detect it. to get aorund this you could get one of the audio aplifier kits from maplins. that way you can hear knco your self during testing and you can then see if your detector circuit is picking it up too. if it is then all is good. then all you have to do is get megajolt to use that signal- hopefully Brent will post up the mods needed to the V3 some time soon.

[david jenkins]

As Nat suggests, the best/cheapest knock detector is mounted between the driver's ears!

There are a few suggestions here.

I fitted a knock sensor salvaged from a scrapyard, connected it to an amplifier, and then drive around wearing a set of ear protection headphones. It's amazing how the brain can process the appalling noises that a healthy engine can produce and pick out those that shouldn't be there!

Trouble is - it's hard to connect the brain to a Megajolt unit.

David

[Supercharged Nat]

Ive also had a go at the headphone/ amplifier but havent got around to using it yet. i did here a bit of knock the other day buts its the first time i have. my map/ fueling cant be far off but obviously under some conditions it can be generated. i think that is where the new megjolt will come in very handy as for those odd times it may happen, you can be safe in the knowledge itl be taken care of!

i plan to do some proper test driving listening for the knock, tweak my map for the ultimate performance (while staying just shy of knock) and then run a knock sensor etc as backup- hopefully anyway.

the diagram above looks like it will work, im just waiting for the V3 mods to be published and then im going to sling it all together and see if it works.

[Salamander]

Hi

check this and comments.

if the auxiliary input support 0-5v, this circuit can work with MJLJ?

http://www.msextra.com/manuals/MS_Extra ... .htm#knock

http://www.viatrack.ca/

[Alexander_Monday]

Since the firmware is not open source, you need some method of continuing to raising the AUX input as long as knock is detected, and lowering it when not, both at a controlled rate. Otherwise, say you set the first bin of the AUX input to retard 2 degrees. Now say you have knock that activates that bin. The ignition will retard 2 degrees, but 2 different things can happen after that. 1st, if the 2 degrees is enough to prevent the knock, the bin will no longer be active and the retard will go away and the knock return, ad infinitum. The 2nd is what if 2 degrees was not enough to make the knock go away? As long as the knock does not get worse you will stay in that bin. Minor knock not so bad? What if it happens higher in the bins with more substantial knock? You could of course over retard on purpose and that would make a good safeguard, but not a very useful control to stay at the maximum edge of power. Also, if you did get the perfect amount of retard in the bins what happens when you refuel and the gas is a little worse, or the humidity, temp, and bp change?

Here is what I propose, and please run this idea through the fire because I am only human and have been known to come up with stupid ideas.

Use the led output of the standard knocksense ("It contains an active filter that amplifies the knock signal and a comparator circuit that drives an LED which flashes on every knock event. The flash intensity is related to the knock intensity.") through a diode ( to prevent the knocksense from draining the charge off the cap that is ahead ) through variable resistor ( to adjust the charge rate ) to an electrolytic capacitor. The cap has a variable resistor to ground ( to adjust the charge drain off time ). The cap is connected to the input of a single ended supply +5V op amp set up as a buffer / follower that feeds the AUX input on a V4. As long as the knocksense is detecting knock the AUX voltage will continue to rise up to a maximum level. When the knock sense quits detecting knock the AUX voltage will fall. The rise / fall rate will be dependent on the adjustments of the 2 variable resistors ( not very eloquent because the rise fall will interact with each other, but simple ). Load the bins up with increasing retard as the AUX input gets higher. Now, when the first bin takes out 2 degrees, if the knock does not go away the input will continue to rise even if the knock has not got worse and the next bin will be activated with more retard ( say 4 degrees ) and on up the scale of bins.

[miniman82]

If I may, I think you're thinking about this the wrong way.
I'm not sure if the ATMEGA chip (Ii think that's what it is) can do anything with analog inputs, but even if it could a digital circuit would be the best way to go about it. In other words High for knock detected, Low for no knocking (or vice versa). I know you're thinking that by doing it the way you propose would make it easier to 'tune' to any particular engine, but I think it would only complicate matters.

I believe the correct way to do this is the same way MS does it: http://www.msextra.com/manuals/MS_Extra ... .htm#knock


Every engine creates knock signatures at a specific frequency in the audio spectrum. For this reason, it's impossible to design a circuit that's a 'one size fits all' type of thing. My idea would be to have MJ simply retard timing by X degrees everytime one of the inputs goes High/Low, and at the same time have a timout which seeks to return the timing to normal after Y amount of time expires. The external amp/filter circuit should be where all the tailoring to the engine in question should be happening. You could have notch filters, bandpass filters, amplifiers, or whatever you need to be able to get false-free triggers external to MJ. This way MJ is not predisposed to working with one type of circuit, and the end user has the freedom to choose how to set it up.

These are just suggestions, but suffice it to say I'm very interested in getting knock retard to work with MJ. If we can get it to work, MJ then becomes very powerful indeed. It would place it in league with the Safguard system, which costs hundreds of dollars more than MJ does. I know for a fact there are Corvair people who don't want to shell out $500 for a Safeguard, so something like this is what those people need. I personally think it would be great, because you never know when you'll get that sh!t gas at the pump. I know for a fact there have been times when my car hasn't run the way it should, and I've had to load a more conservative map to compensate. This would cure most of that.

[Alexander_Monday]

At least for the V4 its a Motorola 68H series, and yes it has ADC's. This is how it gets the MAP or TPS info, as well as the AUX input. These pins can be several things, however Brent has them set up as ADC inputs. If you check out the configurator, under Tools, Auxiliary Input Channel Options, you will see where the AUX input voltage inputs can be scaled to the value you want. You would have to mod the board by cutting a trace to R6, as it is an internal 5V pull up resistor, and I have not checked out what the time constant of the C15 R5 C16 filter is, it could be changed or deleted if necessary.

While I agree that a firmware change would be the best answer, I do not have the source code or know what compiler was used to be able to modify it. This is why I suggested a sort of modified AGC circuit to be able to crudely do what the firmware could do better.

Also, I still think the big impasse to our HP flat 4's (can you believe we get 3 to 4 times the HP out N/A, it just astounds me when I think about it) is the valve train noise after warm up due to not being able to use aluminum push rods with the high spring pressures.

[Alexander_Monday]

Oh, and the knocksense http://www.viatrack.ca/ I referred to has the filters and they tune it to your engine parameters (bore is the main determiner of the resonant frequency of the knock ).

[miniman82]

Also, I still think the big impasse to our HP flat 4's (can you believe we get 3 to 4 times the HP out N/A, it just astounds me when I think about it) is the valve train noise after warm up due to not being able to use aluminum push rods with the high spring pressures.

Ahh, but I also have a Corvair which has a hydraulic valvetrain, and I'll soon have a Type 4 on hydraulics as well. It's not noisy at all. If more people would use hydraulic lifters in VW's, it's wouldn't matter what pushrod you used but that's a debate for another forum.


Hey, here's an interestingly cheesy way to get retard- use the knock sensing module to trigger a relay instead of a light. The relay could switch between 2 VR sensors, which had an offset equal to the desired amount of retard. In this way when a knock was sensed, the relay could click over for a time period and give the VR signal from a pickup which lags the primary one by a few degrees. I'm sure that if it were only a degree or 2, there wouldn't be any issues with the EDIS module getting thrown off by seeing teeth at different intervals....

[Alexander_Monday]

If you have V4, another option for your relay route would be to program the second ignition configuration with retarded bins and switch to the second ignition configuration using the option switch input. I did read somewhere that there is a slight delay in switching on the fly, and it is not the best way to do knock control, but better than nothing.

[miniman82]

If you have V4....

Unfortunately, I have an earlier unit.
Didn't see a point to upgrade then but now...


So no one has deconstrusted the code yet?
Remind me why it's not open?

[brentp]

If you have V4, another option for your relay route would be to program the second ignition configuration with retarded bins and switch to the second ignition configuration using the option switch input. I did read somewhere that there is a slight delay in switching on the fly, and it is not the best way to do knock control, but better than nothing.

This technique works with the V3 as well, as it too supports the option switch. I think you're confusing the AUX input feature...

With regards to using the AUX input for correcting ignition advance, V3 will have similar capabilities very soon.

[brentp]

To frame the conversation, I think it would be good to describe the current capabilities of the Aux input feature, since it's fairly new.

The AUX input feature, combined with advance correction and the 'peak hold' capabilities provides a basic capability for enabling some type of knock compensation using an external device.

Presuming an external device offers a variable output voltage indicating the 'level' of knock, the advance correction table could be set up to react to the knock condition and pull back timing as programmed into the advance correction table.

The peak-hold feature holds the measured value for a configured number of ignition events. Once the peak hold expires, the value drops down to the current level.

Given the current design, to be effective at combating knock the ignition retard value programmed in the advance correction table must be high enough to reduce knock in order to achieve the desired feedback loop behavior.

As Alexander alludes, there currently isn't an automatic 'ramping up' of the measured value if the correction value does not sufficiently reduce knock.

The current feature was designed to be generic- to be useful for things like adapting to warm-up conditions, using a temperature sensor, and to act as a basic form of knock correction. Using the existing mechanism is dramatically better than coarsely flipping between two maps, but I agree for knock sensing applications it can be refined.

I'd like to improve it in the firmware, so I look forward to hearing your feedback on how it would work best while fitting nicely into the current design.

Thanks!

[brentp]

If you have V4....

Unfortunately, I have an earlier unit.
Didn't see a point to upgrade then but now...


So no one has deconstrusted the code yet?
Remind me why it's not open?

We have a history of publishing firmware source code, as demonstrated for the V3 hardware. We haven't packaged up the V4 firmware for publishing yet, but it is on the to-do list.