Programmable digital knock filter?
-
- Posts: 2886
- Joined: Sun Aug 02, 2009 9:16 pm
- Location: Bayside, Melbourne, Victoria
- Contact:
Re: Programmable digital knock filter?
There a GREAT thing & can't talk highly enough of it not only for making an 808 upmarket with knock sensing but just the knock LED or LOG triggering!
GREAT WORK FESTY!!!!
GREAT WORK FESTY!!!!
Re: Programmable digital knock filter?
Here's a quick log of me mucking around with the filter module's settings.
The engine is idling, and by adjusting the module's parameters the knock being reported to the ECU varies from intense, constant knock to no knock being reported at all. A LED connected to the output helps get the settings in the right ballpark to start with, a quick knock isn't always visible in the logged knock count like here - the knock retard is increased but the knock count isn't obvious:
The engine is idling, and by adjusting the module's parameters the knock being reported to the ECU varies from intense, constant knock to no knock being reported at all. A LED connected to the output helps get the settings in the right ballpark to start with, a quick knock isn't always visible in the logged knock count like here - the knock retard is increased but the knock count isn't obvious:
- Attachments
-
- knock_settings.xdl
- 12P log of knock settings changes
- (309.45 KiB) Downloaded 466 times
-
- Posts: 17
- Joined: Sat Dec 26, 2009 2:19 am
- cars: 95 Monte Carlo LS
90 Grand Prix LE
Re: Programmable digital knock filter?
i believe that would be due to only the MSB of the knock counter register being transmitted over ALDL.
Re: Programmable digital knock filter?
It could also be related to the slow old laptop I was using to log, it's so gutless that it can't update the monitor window and log packets at the same time - I'll try it again with a faster machine and see if the log looks any better.
Things like this don't make much sense to me - the knock flag is clear and count is 0, but in the second line the knock retard increases so my guess is that knock was probably detected between sample 2126 and 2127, but was gone by the time packet 2127 was sent?
The knock filter would have processed heaps of samples between those two messages, so maybe a message6 would show more granular detail... not that it really matters because the KR value shows what's going on anyway.
Things like this don't make much sense to me - the knock flag is clear and count is 0, but in the second line the knock retard increases so my guess is that knock was probably detected between sample 2126 and 2127, but was gone by the time packet 2127 was sent?
Code: Select all
Sample Time Knock Count KR
flag
2126 234.417 Clear 0 7.4
2127 234.524 Clear 0 8.1
2128 234.629 Clear 0 8.1
-
- Posts: 2886
- Joined: Sun Aug 02, 2009 9:16 pm
- Location: Bayside, Melbourne, Victoria
- Contact:
Re: Programmable digital knock filter?
Had the ECU been diss connected before logging?
If it goes into areas it's learnt knock before it could just be from ram memory?
If it goes into areas it's learnt knock before it could just be from ram memory?
Re: Programmable digital knock filter?
Yep, I'm in the habit of isolating the battery whenever it's not being run.
And the engine was holding a pretty steady idle, so shouldn't have been moving in and out of learned areas?
I think it's just similar to what you were seeing with your logs - sampling at 10Hz is going to miss some fast events.
And the engine was holding a pretty steady idle, so shouldn't have been moving in and out of learned areas?
I think it's just similar to what you were seeing with your logs - sampling at 10Hz is going to miss some fast events.
- VL400
- Posts: 4991
- Joined: Sun Mar 01, 2009 2:54 pm
- cars: VL Calais and Toyota Landcruiser. Plus some toys :)
- Location: Perth, WA
- Contact:
Re: Programmable digital knock filter?
12P does not have learned knock, its purely reactive on knock at that instant. Its just that the knock detected flag is set and cleared every 12.5ms that you can miss it while logging.
Re: Programmable digital knock filter?
Thanks, the log makes perfect sense in that case
Is the knock count cleared at the same rate?
Is the knock count cleared at the same rate?
- VL400
- Posts: 4991
- Joined: Sun Mar 01, 2009 2:54 pm
- cars: VL Calais and Toyota Landcruiser. Plus some toys :)
- Location: Perth, WA
- Contact:
Re: Programmable digital knock filter?
Yeah knock counts are the same 12.5ms, the flag is just a zero check of the knock counts to clear/set it.
Am sure you have worked it out from testing, but 16 knock counts = (very close to) 1mS of knock duration sampled every 12.5ms.
Am sure you have worked it out from testing, but 16 knock counts = (very close to) 1mS of knock duration sampled every 12.5ms.
Re: Programmable digital knock filter?
I hadn't done the maths, but that certainly matches up with the max knock count being around 200 in the logs.VL400 wrote: Am sure you have worked it out from testing, but 16 knock counts = (very close to) 1mS of knock duration sampled every 12.5ms.
I was mucking around with a sig gen the other day and found that with very high integration time/low gain set, the detected knock bandwidth got as narrow as 250Hz
I don't think those extreme settings would work too well in the real world though...