For any discussion not specifically related to your project
By Black Knight
#48676
It has been a long time ago I was here.
Many good things done with my speeduino project. I have afr tuning with in .2 afr with auto tune and have very little hand tuning needed as for afr. I know from old school how to get very good ignition timing and have it. I am at the point where I need to find the best ignition timing with the ecu so I decided to set up a knock sensor to work with speeduino.

It is now working through my controller and shows up on a volt meter and a recording software on a laptop. I have a frequency filter for both volt meter and the recording software that filters most of the engine noise out and amplifies the knock signal. I can see the sensor spikes very clearly and hear them with play back on the sound recorder.
This was tested with a knock sound I created that sound very similar to engine knock. My filter may be different for real knock but that is an easy change to my controller.
I will be setting it up to log on TS next as I have the controller set up to give 0v to 5v also.

I am getting ready to use it to tune timing.
I am old school and know what to do the old school way but this will be better.
ECU tuning is soooooo much easier.

What I need is some input as to how to use this signal. My thought is to slowly advance timing until I see the first signs of knock signal and back off until I see no knock signal.

Is this correct?
I may find slight knock and add some fuel to cool things down and see if knock clears up.

Good thing I have spare engines and pistons. HAHAHA

Anyone worked with timing and Knock sensors?

Any help here to get it right the first time would be great.

I will set up the controller with three out puts.
One for volt meter shown in volts.
One for sound recording.
One for Speeduino/TS logging.

Almost done.

If you can, please educate me on how to use this set up.

Very cool sensor.

Thanks in advance.

Black Knight
User avatar
By PSIG
#48679
Many things we could discuss on this topic ;) but first question is what fuel you are using? Second is why you have knock (short valve timing, excessive compression, poor cooling, boost, nitrous, etc)?

I prefer to work from the other direction. I first use good fuel that will allow full correct timing, and tune to find it. Then I switch to the fuel type Mr Cheap will use and repeat timing tuning with slowly added load until I detect detonation. Assuming the fuel quality is below what is needed for proper timing, I first try to cool everything, such as the engine coolant temperature, incoming charge, pistons, etc. Heat is what causes detonation, so I get rid of as much as I can. If that is not enough, I adjust the timing to retard (delay) the cylinder pressure peak in order to eliminate the knock before it happens, as it takes more retard to stop it when it starts than to avoid it first.

By starting with good fuel, I also find the stable cylinder and exhaust temperatures. This is important, as retarding for crappy fuel adds heat, and again heat causes detonation. You can get into chasing your tail with this. Adding fuel can also retard the burn, but costs more power than timing (to a point). So I generally tune the fuel to peak-torque Lambda and adjust timing as-necessary, either to peak power or to the det threshold if det-limited. Specific engines or power-adders may need specific plans. I always test with pulls longer than what the customer might do, as heat accumulates, and a long run may put it into det that didn't show during tuning.

I prefer using an audio amplifier with piezo sensors, as I have yet to find a "smart" controller that can sense knock onset among the high-speed engine noises better than the human ear. Some of the "window" controllers are getting pretty good, but most controllers are not designed for high-performance uses. If you don't have an amp or controller, no problem, you can detect knock the old-school ways without a knock sensor. I prefer to not need to detect det at all, with the engine designed for the fuel to be used, or the right fuel for the engine you have; but not all customers are willing to build the engine right or spend an extra $5 on a tank of fuel for their expensive engine. :roll: The secret to not damaging anything is to use the right fuel, otherwise be patient and work-up to it. Detect it at lower loads before it can do damage. That's me. Do your thing. Have fun!
By Black Knight
#48697
I am having fun and knock sensor should make it more fun.

Subaru 2,2 with 11.2 to 1 comp, CC'd and polished combustion chambers with no sharp points at all, valve faces and backs polished, 91 octane pump gas.

With this work done to the heads there is almost no aggressive knock. So a sensor should help me find the knock threshold.

I can see it now.

I know how to tune the way you stated. I have been doing that for years the old school way and have now done it with speeduino. This way of tuning works great and I have a tune that has good torque and good throttle response with NO KNOCK from the way you stated.

But there is more to be gained from more accurate timing, not just very good and close.
That is what knock sensors are for. I know other ways of finding best timing but it is more complicated and is more work.

I would not want to work with a generic knock sensor and try to set it up to work when the factory has it done right for 30 years. I just got the factory proven sensor to work and show up on speeduino/TS.

Now the question is. HOW TO USE IT.
Not why am I using it or if it works.

Thanks for your input. It will help others to know there is more than one way to fine best timing but I am going to have fun using the knock sensor and see if I can do what factory has done for a long time that works great.

So, the question is. Is my way of using it correct?

Thanks to all.
Black Knight
By Black Knight
#48707
LPG2CV wrote: Mon Mar 01, 2021 9:26 pm Hey BK :)

What firmware version are you using?
Alway good to hear from you. Read some of your latest postings. You are getting good.

firmware 202002.

Why do you ask?

Very interested in why you ask.

Very happy with how it is all working. Just good fun now.

Chat later

Black Knight

Ify ideas and a few beers.
User avatar
By PSIG
#48708
I think you missed my suggestions on how to use it. ;) You can also compile data into Speeduino logs in order to see relationships in MegaLogViewer with factors such as CLT, MAP, rpm, advance, Lambda, etc. In this way, you may find the most efficient way to de-tune the engine to the point it will run det-free with the best performance.

BTW - you may have misinterpreted my use of terms "best" or "proper" timing (correct timing for most efficient conversion of fuel to power) to adjusted or altered timing (manipulated timing in order to avoid detonation). Running altered timing with det-limited fuel is not correct timing, but rather manipulated timing in order to compensate for other shortcomings, such as fuel octane or AKI. With lower-quality fuels, you cannot run accurate best timing, as the variable fuel quality means you have to run more conservative from tank-to-tank.

By using better quality fuel, your timing may be tuned more accurately with less variation due to fuel quality issues. This is why I suggested that first, in order to establish true and accurate best timing, knowing exactly what that is. Then re-fuel with the pump stuff and easily find your det limit, knowing the timing is correct at every point, only limited by detection of det. Experiment (or use data) to find the best balance of fuel and timing for best power with a wider safety margin.

Nothing wrong with that set of compromises, but they are compromises, and becomes a challenge to guess how conservative to tune it based on how bad the next load of fuel might be. The conservative tune costs power and efficiency, so there are limits of how far you can go. That's where the inaccurate timing comes-in. That's your judgment call as the tuner, to balance a better tune against unknown fuel factors, with some safety for the customer's and fuel vendor's unpredictable choices.

Likewise, if you run programming that retards when it senses det, then det has already started and will require a larger amount of retard to stop it, then creep back towards the det limit. This can cost a large amount of performance, running 80% of the time excessively retarded to stop the det before re-advancing. With OEM cars the owners have no clue, being happy to pump whatever they like into the tank; but we know. ;)
User avatar
By Chris Wolfson
#48712
High quality fuel has the advantage to be of consistent quality, that makes it so important for tuning the spark.

After tuning for best performance with good fuel, you can make some test runs with lower octane quality. That will give you a quick idea where the problems surface first, in a comfortable load /rpm window. Reduce advance in that region to fit, than you can expand the change manually to the rest of the tune. That way you get a "not harm the engine" tune. You may do that as "switchable maps". How you add more fuel is not simple to answer, as too much enrichment may have adverse effects, too.
There is no value in getting a perfect ignition map with a tank of some crappy fuel, as the next load may have very different characteristics. Also, to demand maximum performance from an engine running on cheap fuel is somehow illogical. If you want max power, give it the best you can get...

The "automatic, perfect knock control" is identical to a self tuning ignition and will stay a dream, however you call or describe it. A knock sensor integrated into the ECU is only a last emergency exit to save the engine. It works well in some regions, but unreliable in the high rpm/ load range. Factory engines take care of this problem and retard in the upper region, so no knock adjustment should be needed

Sometimes even that does not work, when unhappy conditions accumulate. Like Bad gas, cheap engine oil, high oil consumption, old lambda sensor, hot weather and a dirty radiator meet with high load. Just to name a few.

Engine power does not increase linear and then, suddenly, start to detonate. There is a region where more advance does not increase power. A good, save tune should retard just to the point where power starts to drop.
You can find that point in data logs made under identical conditions or on a rolling road. Wind, temperature and air pressure should be taken care of if you can´t do it in one session.
By Black Knight
#48713
Yes, good fuel and using more than one sensor and more than one data is all very true.
I am doing things very similar as all of you have said. Very true. That is how I got this good tune now.

Just maybe knock sensor can be one more bit of data to help.

I have a very good tune with out it.

Now I am finding ways to see and hear the very first onset of knock with as little as 2mv at 5000hz.
problem knock is closer to 25mv to 35 mv on this motor critical knock looks closer to 50mv.

I don't know how much it will help but I intend to find out.

Thanks for the input.

I have been using many of the sensors and data to tune much as you have said.

Very good and thanks.


Black Knight.
By Black Knight
#48717
JHolland wrote: Tue Mar 02, 2021 9:50 am What knock sensor are you using?
The stock one that came from factory.

I think I should trust the research of the factory.

They spent the time to to make it work as well as possible and find the location and angle that works best.

Subaru through hole one wire.

So far it looks like they did a very good job.

Black Knight
Vr Conditioner Noise when cranking

So, i tried a true arduino and same result. I also[…]

blitzbox

I've finally figured out why MAP and Lambda weren'[…]

Hello, I bring news!! Let me tell you that after […]

Honda J30 running

Okay, we now need to talk about violence and backg[…]

Still can't find what you're looking for?