Knock events counter on F9T DME

[Mscromer]

Been driving my newly rebuilt engine around and I’m getting a lot of knock events on the counter in the F9T software. My AFM gauge is fluctuating right around 14.7 so I wouldn’t think I should be getting any knock events. Is this normal? I haven’t messed with the FQS on the DME because like I said my AFM is right at 14.7. Don’t know the position of the switch, I just left it where it was when it came from F9T. Engine seems to be running great once it’s warmed up.

[Tom]

I see more than I expected too on my F9 DME, especially compared to my Curtis counter. I suspect (but don't know) that some of it is just driveline/engine noise triggering the system. But also I think the F9 is proving to be an education as to how active the KLR really is. For years many of us relied on the knock indicators driven off pin 15 of the KLR. Based on what I'm seeing, that pin 15 output does not seem to report knock every time the KLR is actually pulling timing. I'm wondering if the KLR has some threshold over which it sends a blip to pin 15, and under which it just happily goes about its business making minor timing adjustments? With apologies if I already asked that same question, but wondering if @johnb has any intel on that?

Also, pretty sure the f9 datalog will show your FQS setting... (unless I'm mixing up data loggers...)

[Mscromer]

Well that makes me feel better that your seeing it also. When I’m looking at the live data while driving I’m seeing the KLR advance and retard the timing constantly like you think it should. I’m hoping John will see this and give a little more info. If he doesn’t I’ll email him. I wish he had detailed info on what we should see and what different values mean on these data logs. Like is it normal to see a lot of knock counts. And you are correct the FQS is on the data log. I didn’t notice that. Mine is set at 0. So if my AFM is bouncing around 14.7 there would be no reason to adjust the FQS correct? Thanks for your input.

[johnb]

@Tom No there is no special threshold for pin 15, it just gets pulsed if the KLR detects knock.

That said, the KLR is always pulling ignition timing, 100% of the time. It's a side effect of how the code is written and has nothing to do with knock. It's a kind of quantization error - the ignition signal output from the KLR is always fired by a timer interrupt, but the timing of the input signal from the DME is random. This delay is needed to synchronize the phase of the internal timer to the ignition signal.

From my investigation into this years ago, I wrote that it was around:

...260us below 1500rpm, and around 170us above that (it needs to be longer below 1500rpm because the timer period is longer - see the speed measurement section if that doesn’t make sense).

[1]

The reason that there are 2 different delays is that there are 2 different timer periods used in the code, depending on engine speed. This is an unfortunate complication brought on by the fact that the puny 8048 only has an 8-bit timer. It's used for measuring engine speed, but with only 8 bits it lacks the dynamic range needed to measure the full range of engine speeds. At low rpm, the rpm counter would overflow, so the timer has to be slowed down to prevent that.

Anyway, this means that you have a fixed delay in the ignition of 260us up to 1500, then 170us for the rest of the rpm range. Obviously this corresponds to a varying angle depending on engine speed - the higher the rpm, the bigger the angle.

At 3000rpm, there's 50 revs per second, or 20ms per rev. That means around 55us per degree, so 170us is over 3 degrees.

I wonder if this is what people are seeing? I don't have any idea how the FT9 DME counts knock events - does it have an input from KLR pin 15? If not then I don't see how else it could be doing it except by the actual ignition signal delay, but then there's no way to tell if it's really knock of if it's just the normal delay.

[1] https://jhnbyrn.github.io/951-KLR-PAGES ... ignal.html

[Tom]

@Tom No there is no special threshold for pin 15, it just gets pulsed if the KLR detects knock.

That said, the KLR is always pulling ignition timing, 100% of the time. It's a side effect of how the code is written and has nothing to do with knock. It's a kind of quantization error - the ignition signal output from the KLR is always fired by a timer interrupt, but the timing of the input signal from the DME is random. This delay is needed to synchronize the phase of the internal timer to the ignition signal.

From my investigation into this years ago, I wrote that it was around:

...260us below 1500rpm, and around 170us above that (it needs to be longer below 1500rpm because the timer period is longer - see the speed measurement section if that doesn’t make sense).

[1]

The reason that there are 2 different delays is that there are 2 different timer periods used in the code, depending on engine speed. This is an unfortunate complication brought on by the fact that the puny 8048 only has an 8-bit timer. It's used for measuring engine speed, but with only 8 bits it lacks the dynamic range needed to measure the full range of engine speeds. At low rpm, the rpm counter would overflow, so the timer has to be slowed down to prevent that.

Anyway, this means that you have a fixed delay in the ignition of 260us up to 1500, then 170us for the rest of the rpm range. Obviously this corresponds to a varying angle depending on engine speed - the higher the rpm, the bigger the angle.

At 3000rpm, there's 50 revs per second, or 20ms per rev. That means around 55us per degree, so 170us is over 3 degrees.

I wonder if this is what people are seeing? I don't have any idea how the FT9 DME counts knock events - does it have an input from KLR pin 15? If not then I don't see how else it could be doing it except by the actual ignition signal delay, but then there's no way to tell if it's really knock of if it's just the normal delay.

[1] https://jhnbyrn.github.io/951-KLR-PAGES ... ignal.html

Ah yes, I remember having a similar conversation with you about the timing latency in the KLR when i was making my ignition timing gauge. I was able to detect curious latency through the KLR, which you confirmed and explained to me back then. Seem to recall a long conversation about the ref sensor pin TDC offset too. I was picking up the ref sensor signal after the S100 and calculating advance based on when the ignition was actually firing -- trying to confirm timing in general and detect when the KLR was pulling timing. I do recall there was a slight delay in timing pre and post klr no matter what, which you explained.



Interesting that there is no conditioning of Pin 15. That's good news in a way, as Pin 15 doesn't show nearly the activity that the F9 logs do. I've wired pin 15 directly to the interrupt pins of various microprocessors -- which will catch virtually any intentional pulse (and even transient noise) -- and those pulses are few and far between compared to the f9 logs. Purely a suspicion, but I'm inclined to think the F9 log is using some logic/algorithm to calculate and report when knock occurs, and that it is over-reporting as a result. I recall similar reporting issues as he sorted out the throttle positions states (and am still unsure if the ISV duty cycle reports correctly). None of that is intended to undermine the F9 DME, which seems rock solid, or the effort in general -- but it's possible there is still a little fine-tuning needed in the logging software. Or I'm completely wrong -- also a real possibility.

[Mscromer]

So after everything yall both said to which I have no idea what all that was. Lol. Sounds like as long as I have 14.7 AFM, having a high knock count isnt actual knocks that I should worry about?

[Tom]

So after everything yall both said to which I have no idea what all that was. Lol. Sounds like as long as I have 14.7 AFM, having a high knock count isnt actual knocks that I should worry about?

Sorry about that. On a more practical level, your real risk is at wide open throttle, not when cruising at 14.7:1. What is your AFR when you floor it and are on boost? For a typical tuned 951, a good target is 11.8:1, with anything much higher than 12 to 12.5 adding unnecessary risk on typical pump gas... If your car is bone stock, running stock boost curves, then the factory chip/maps are fine of course... I wouldn't stress too much about the f9 knock counts -- that seems to be typical for that DME, whether accurate or not.

And BTW, the closed loop system can bring a car back to 14.7:1 even if the base tune is fairly far off. So, don't take too much comfort in seeing 14.7:1 when cruising. The closed loop system can mask issues. That closed loop system is not active at full throttle, so any shortcomings in the base maps will reveal themselves when your motor is at its most vulnerable... That's why I say it's much more important to monitor AFR when under full boost.

If you want another perspective on whether your car has knock, people have been using these knock counters for years....

[Mscromer]

So after everything yall both said to which I have no idea what all that was. Lol. Sounds like as long as I have 14.7 AFM, having a high knock count isnt actual knocks that I should worry about?

Sorry about that. On a more practical level, your real risk is at wide open throttle, not when cruising at 14.7:1. What is your AFR when you floor it and are on boost? For a typical tuned 951, a good target is 11.8:1, with anything much higher than 12 to 12.5 adding unnecessary risk on typical pump gas... If your car is bone stock, running stock boost curves, then the factory chip/maps are fine of course... I wouldn't stress too much about the f9 knock counts -- that seems to be typical for that DME, whether accurate or not.

And BTW, the closed loop system can bring a car back to 14.7:1 even if the base tune is fairly far off. So, don't take too much comfort in seeing 14.7:1 when cruising. The closed loop system can mask issues. That closed loop system is not active at full throttle, so any shortcomings in the base maps will reveal themselves when your motor is at its most vulnerable... That's why I say it's much more important to monitor AFR when under full boost.

If you want another perspective on whether your car has knock, people have been using these knock counters for years....

Thank you for the translation. Lol. I’ll keep all that in mind. I do have a little bit bigger cam 480 lift 233 duration. I’ve only got 20 miles on the engine so far so I’m a ways from increasing the boost.

[walfreyydo]

Knock (as in detonation or ping) is really related to four things: AFR, ignition timing, load and fuel octane. All play a role in whether you get knock, so its not just AFR's. Under normal cruising conditions, ie light load, you should really never see knock, even when extremely lean, unless your timing is extremely advanced. You would typically see a misfire or bucking/surging if its overly lean at light loads. Under load (such as going up hill in a low gear, flooring the accelerator at low RPM or under boost), is when you normally see knock/detonation and this is where all four factors can play a role.

Since you are running an aftermarket cam you may want to ensure you reinstalled the distributor rotor and have it clocked correctly and verify timing with a timing light and compare to what the F9 DME total timing is reading. That could be one scenario where your timing is too far advanced even though you havent modified the ignition timing tables, which as far as I know, is not possible with the F9Tech DMEs (unless chipped or reflashed).

[johnb]

Knock (as in detonation or ping) is really related to four things: AFR, ignition timing, load and fuel octane. All play a role in whether you get knock, so its not just AFR's. Under normal cruising conditions, ie light load, you should really never see knock, even when extremely lean, unless your timing is extremely advanced. You would typically see a misfire or bucking/surging if its overly lean at light loads. Under load (such as going up hill in a low gear, flooring the accelerator at low RPM or under boost), is when you normally see knock/detonation and this is where all four factors can play a role.

Since you are running an aftermarket cam you may want to ensure you reinstalled the distributor rotor and have it clocked correctly and verify timing with a timing light and compare to what the F9 DME total timing is reading. That could be one scenario where your timing is too far advanced even though you havent modified the ignition timing tables, which as far as I know, is not possible with the F9Tech DMEs (unless chipped or reflashed).

You can't change timing by clocking the distributor - it's controlled entirely by the speed and ref sensors. The rotor is just timed so that the contacts are closed for a wide enough window to cover the full range required in the timing maps. Clocking it will either have no effect on timing, or else result in the spark just not firing if you go too far.