Knock control on the MS2
#1
Elite Member
Thread Starter
iTrader: (10)
Join Date: Jun 2006
Location: Athens, Greece
Posts: 5,979
Total Cats: 356
Knock control on the MS2
Finally, after many obstacles, the latest version of my Enhanced MS2 has the foundation ready for proper knock control.
Input from the knock sensor on the processor pin. It is already filtered through the DSP stage:
What the MS2 sees:
Now, for the firmware changes...
Input from the knock sensor on the processor pin. It is already filtered through the DSP stage:
What the MS2 sees:
Now, for the firmware changes...
#3
Elite Member
Thread Starter
iTrader: (10)
Join Date: Jun 2006
Location: Athens, Greece
Posts: 5,979
Total Cats: 356
Raw stock knock sensor output (pink) Vs the filtered output going to the MS2 (yellow) - you can see that the stock sensor is of the resonant type.
Sudden knock:
Reaction time of the DSP stage (~1.7ms):
Sudden knock:
Reaction time of the DSP stage (~1.7ms):
#6
Elite Member
Thread Starter
iTrader: (10)
Join Date: Jun 2006
Location: Athens, Greece
Posts: 5,979
Total Cats: 356
It is on my expansion board. I can make it work for the MS3 (or do a separate "knock-only" board) however I don't want to have a "competing" product with James. So wait for his board again.
#9
2 Props,3 Dildos,& 1 Cat
iTrader: (8)
Join Date: Jun 2005
Location: Fake Virginia
Posts: 19,338
Total Cats: 573
the latest MS3 beta firmware (yes, I said beta) does a pretty decent job of utilizing the knock signal. it's a little hard to sift through all the noise in the logs (actually beta 19 seems to forget to log the raw per cylinder values) but the peak-decay combined value is logged.
also the windowing is pretty nice because you can see exactly what cylinder is making the most noise and retard or enrich on a per-cyl basis.
also the windowing is pretty nice because you can see exactly what cylinder is making the most noise and retard or enrich on a per-cyl basis.
#10
Subscribed.
Reverant, just to be clear:
- Will current owners need our boards' hardware updated?
- Are any other wiring/hardware changes on the car needed?
- Will the severity of the knock be detected to allow for gradual mitigation (timing/fuel adjustments in steps)?
I really appreciate your persistent work - I know this has been on the way for a long time.
Cheers,
- Tom
Reverant, just to be clear:
- Will current owners need our boards' hardware updated?
- Are any other wiring/hardware changes on the car needed?
- Will the severity of the knock be detected to allow for gradual mitigation (timing/fuel adjustments in steps)?
I really appreciate your persistent work - I know this has been on the way for a long time.
Cheers,
- Tom
#11
Elite Member
Thread Starter
iTrader: (10)
Join Date: Jun 2006
Location: Athens, Greece
Posts: 5,979
Total Cats: 356
Subscribed.
Reverant, just to be clear:
- Will current owners need our boards' hardware updated?
- Are any other wiring/hardware changes on the car needed?
- Will the severity of the knock be detected to allow for gradual mitigation (timing/fuel adjustments in steps)?
I really appreciate your persistent work - I know this has been on the way for a long time.
Cheers,
- Tom
Reverant, just to be clear:
- Will current owners need our boards' hardware updated?
- Are any other wiring/hardware changes on the car needed?
- Will the severity of the knock be detected to allow for gradual mitigation (timing/fuel adjustments in steps)?
I really appreciate your persistent work - I know this has been on the way for a long time.
Cheers,
- Tom
2) No other changes needed on the car.
3) The severity of the knock is detected as you can see, the signal fed into the MS2 is analog. My plan is to give a proportional retard option - ie retard 4* if the knock is 40%, retard 7* if the knock is 70*.
#13
Great much-needed work!
Will you share with us how you discriminate knock from background noise? Do you:
- Compare noise levels during the valid knock crank-angle window against outside-the-knock-window?
- Just compare noise levels during the valid knock window against a background-noise-vs-RPM plot?
- Don't do any crank angle windowing?
- listen for the telltale "descending bell" note?
Will you share with us how you discriminate knock from background noise? Do you:
- Compare noise levels during the valid knock crank-angle window against outside-the-knock-window?
- Just compare noise levels during the valid knock window against a background-noise-vs-RPM plot?
- Don't do any crank angle windowing?
- listen for the telltale "descending bell" note?
#14
Elite Member
Thread Starter
iTrader: (10)
Join Date: Jun 2006
Location: Athens, Greece
Posts: 5,979
Total Cats: 356
Right now there is no knock windowing, although I have provisions on my current board for it. If my free time permits, I will see if I can actually make it work.
The method to be used will be simple:
- Only look for knock inside the knock window
- Noise outside the knock window will not be used at all (this is a limitation of the MS2, I would use it if I could)
- Display the knock percentage. User decides the knock percentage threshold which triggers the knock reaction system. The threshold is the noise profile of the engine at various RPMs. Probably an 8-point table.
- May add a sensitivity adjustment (factor) based on MAP, IAT (higher MAP or IAT = more likely to knock, so make it more sensitive)
- Once the knock threshold has been crossed either remove a given amount of timing until no knock is there, then gradually restore the timing back (current MS2 strategy), or, given that we have an analog signal in now, reduce timing proportionally to the knock intensity, and gradually restore it.
The method to be used will be simple:
- Only look for knock inside the knock window
- Noise outside the knock window will not be used at all (this is a limitation of the MS2, I would use it if I could)
- Display the knock percentage. User decides the knock percentage threshold which triggers the knock reaction system. The threshold is the noise profile of the engine at various RPMs. Probably an 8-point table.
- May add a sensitivity adjustment (factor) based on MAP, IAT (higher MAP or IAT = more likely to knock, so make it more sensitive)
- Once the knock threshold has been crossed either remove a given amount of timing until no knock is there, then gradually restore the timing back (current MS2 strategy), or, given that we have an analog signal in now, reduce timing proportionally to the knock intensity, and gradually restore it.
#16
Elite Member
Thread Starter
iTrader: (10)
Join Date: Jun 2006
Location: Athens, Greece
Posts: 5,979
Total Cats: 356
This is for the Enhanced MS2 that I build only. It is an integrated solution, there is an expansion board inside the case, much like the MS3X board, that has the knock solution on it. For the MS3, you need to get Jame's (jsmcortina on msextra.com) knock board. I would get this and install it for you, however he does not have any at the moment to sell.
#20
This is for the Enhanced MS2 that I build only. It is an integrated solution, there is an expansion board inside the case, much like the MS3X board, that has the knock solution on it. For the MS3, you need to get Jame's (jsmcortina on msextra.com) knock board. I would get this and install it for you, however he does not have any at the moment to sell.