PDA

View Full Version : Knock when coming into boost



ttls1
December 1st, 2013, 11:19 PM
Hi, A mate and I have been tuning my turbo LQ4 landcruiser and seem to be getting knock that we can not hear or get rid off. I have no doubt its definitely there as its in the log but would like a few ideas if possible on ways to correct it as neither of us are experts and were learning along the way. I have attached the tune and a couple of logs. It has been quiet hot here and the IAT's have been up there but I took it for a run at night when it was a bit cooler (log 7) and it still had it. Were running 98 fuel. Have tried pulling some timing out and giving it more fuel but its still there. Any info would be great. We had it in closed loop for the Log 7 run but it was OL for the others.

darcy
December 1st, 2013, 11:39 PM
You sure it's not just too lean?

Its 14:1+ just before it knocks

ttls1
December 2nd, 2013, 09:35 PM
Well we took 5 degrees out of the whole table and added lots more fuel. As you can see in the log it still has knock. Any ideas on where to go from here? The fuel is 98

wesam
December 2nd, 2013, 10:30 PM
You have to set B3613 to 55 kpa you have it at 120 kpa which make delay before entering PE
Also try to make B3643 1.5 so it will enter PE faster then try to make your AFR 11.4 with 12 degree of timing along the line
then try it

ddnspider
December 3rd, 2013, 08:26 AM
The first log looks like its too lean as stated above and too much timing. Then when you added fuel and pulled timing it looks like its very rich and not enough timing. Too rich/too little timing can cause knock too. I've experience that where I actually had to add timing to clean u the knock. Do what wesam said and make it enter PE faster, set the A/F to something reasonable, low 11:1's and timing safe as well like 10-12*. Your IAT's are getting high, but you are also running 98 octane so you have more cushion than 93.

barbermi
December 3rd, 2013, 08:53 AM
Thanks for the info we will try that. Should there be a transition to 11.4:1 or should it be there at 100kpa?

ddnspider
December 3rd, 2013, 11:16 AM
Thanks for the info we will try that. Should there be a transition to 11.4:1 or should it be there at 100kpa?

Pretty much you should always have a transition. My open loop fuel table transitions from stoic at low map to ~13:1@100kPA, but I have my PE table ramped from 14:1 to 11.2-11.4:1 over the rpm range. I.E. off idle and low RPM, you don't need to have it rich since you won't generate much if any boost. You just need to monitor and see at what rpm it will make boost and go from there.

barbermi
December 3rd, 2013, 12:05 PM
Thanks for the reply. We will adjust the boost VE so as it comes onto boost we see ~ 13:1. We will then transition it to 11:2 or so at 185kPa MAP. Once we have the AFR's sorted we will increase the spark table 5deg across the whole lot (we dropped it this much yesterday trying to eliminate knock). We are going to use the KR histogram to help us tune spark. Has there been much success using this histogram for tuning previously?

ddnspider
December 3rd, 2013, 12:22 PM
if the knock is real, yes it can help. I would start by setting the timing to something like 10* across the board in any g/cyl where you're in boost. Get your A/F sorted out first and then start worrying about adjusting the timing. Make sure that you get it rich enough once youre actually in boost. Anything over a couple psi and you'll want the A/F into the 11's with the richest spot at peak torque.

barbermi
December 3rd, 2013, 12:30 PM
Thanks again.
Should the boost VE and spark table be tuned while in open or closed loop?

ddnspider
December 3rd, 2013, 12:36 PM
all VE/boost spark tuning should be done in open loop. Then you can finish off the tune and go closed loop if you choose and do some logs and see how it looks and make adjustments as necessary.

barbermi
December 3rd, 2013, 12:51 PM
Cheers. We will have a go again this afternoon.

joecar
December 4th, 2013, 04:43 AM
+1 in OL.

( CL would be applying fueling corrections ahead of you while you're attempting to correct the VE table, i.e. your VE corrections would be frustrated ).

ttls1
December 4th, 2013, 07:56 AM
Ok cheers, we did some more tuning in OL and have it very close. One thing we have noticed is the PE does not work in OL so we have fudged the fuelling with th ve to get our afrs correct. What will happen when we go back to CL? Will the fuel trims try to correct to commanded afr and throw it all out? The commanded afr will be about 12.7:1

ddnspider
December 4th, 2013, 08:43 AM
PE tuning most definitely works in OL. Can you post your latest tune? I didn't see what kind of car/truck this was, but some tunes have crazy PE delays based on rpm or map.

ttls1
December 4th, 2013, 10:10 AM
That is interesting!

I have attached the last log we did along with the tune file.

Cheers.

16197
16198

ddnspider
December 4th, 2013, 10:54 AM
You have table B3618 set to 14.49 across the RPM range. You need to update this table to transition your A/F down to ~11.2:1 or thereabouts. That is why it appears that its not kicking in in open loop.

ddnspider
December 4th, 2013, 10:57 AM
Also this is a boosted setup and you have the MAF turned off, but are only running a 1 map VE table. This means the tune will NEVER see boost. You either need to go to a 2 bar map sensor and a custom operating system, or turn the MAF back on.

ttls1
December 4th, 2013, 12:26 PM
B3618 had been set to the figures below but it still didn't move into PE mode or have any effect when in open loop yet will change commanded AFR when in closed loop. We changed them all to 0.99 to rule out any changes which might effect our VE calibration. Im running COS5 and was under the impression this is 2BAR as I have a 2Bar MAP installed. The Boost VE is table A0009.

1.000000
1.000000
1.000000
1.000000
1.000000
1.000000
1.000000
1.000000
0.949907
0.930064
0.910222
0.889661
0.870008
0.849793
0.849793
0.849793
0.849793
0.849793
0.849793

ddnspider
December 4th, 2013, 01:42 PM
Ah, missed the COS5 part, that makes me feel better :)
1)I would also set your B3616 to something more reasonable like a stock F body table. At light part throttle its trying to kick PE on. But that is a separate issue.
2)I would also set B0120 to 0 since your speed density.
3)Are you using the A0011 valet mode pin as a 2 step?
4)I looked through the PE tables and nothing jumped out at me except what you have the PE vs. RPM table set to. Can you try setting it to 11.0:1 across the RPM range and make a quick pull?


B3618 had been set to the figures below but it still didn't move into PE mode or have any effect when in open loop yet will change commanded AFR when in closed loop. We changed them all to 0.99 to rule out any changes which might effect our VE calibration. Im running COS5 and was under the impression this is 2BAR as I have a 2Bar MAP installed. The Boost VE is table A0009.

1.000000
1.000000
1.000000
1.000000
1.000000
1.000000
1.000000
1.000000
0.949907
0.930064
0.910222
0.889661
0.870008
0.849793
0.849793
0.849793
0.849793
0.849793
0.849793

joecar
December 4th, 2013, 01:56 PM
B3618 had been set to the figures below but it still didn't move into PE mode or have any effect when in open loop yet will change commanded AFR when in closed loop. We changed them all to 0.99 to rule out any changes which might effect our VE calibration. Im running COS5 and was under the impression this is 2BAR as I have a 2Bar MAP installed. The Boost VE is table A0009.

1.000000
1.000000
1.000000
1.000000
1.000000
1.000000
1.000000
1.000000
0.949907
0.930064
0.910222
0.889661
0.870008
0.849793
0.849793
0.849793
0.849793
0.849793
0.849793Anything higher than lambda 0.87 is too lean for NA, anything higher than lambda 0.78 is too lean for boost.

joecar
December 4th, 2013, 01:58 PM
Did you edit the MAP scaler for the 2-bar MAP sensor...?

Boost
December 7th, 2013, 04:03 AM
Anything higher than lambda 0.87 is too lean for NA, anything higher than lambda 0.78 is too lean for boost.

Does this apply only to his setup or my directed injected "knocker" as well? Also, is this under all conditions? Thanks!

joecar
December 7th, 2013, 06:17 AM
Does this apply only to his setup or my directed injected "knocker" as well? Also, is this under all conditions? Thanks!Yes, applies to you too :)

Rich Z
December 7th, 2013, 08:31 AM
Anything higher than lambda 0.87 is too lean for NA, anything higher than lambda 0.78 is too lean for boost.

What will a B3618 table look like on a typical boosted car? Sloped downwards sharply when boost kicks in and then flat lined out to redline with 0.78?

barbermi
December 7th, 2013, 08:56 AM
Hi Joecar. The MAP scaler is done too. Out of curiosity we logged current gear when doing a power run. We noticed knock at the exact time the gearbox downshifted from 2nd to 1st when first applying power. The next few runs we held first then applied power with no knock. Whats the odds the shift change is giving us false knock?

ddnspider
December 7th, 2013, 09:45 AM
Hi Joecar. The MAP scaler is done too. Out of curiosity we logged current gear when doing a power run. We noticed knock at the exact time the gearbox downshifted from 2nd to 1st when first applying power. The next few runs we held first then applied power with no knock. Whats the odds the shift change is giving us false knock?Ive seen several cars knock right at the shift or on the downshift, it is more than likely the sensors picking up the mechanical sound of the transmission shifting that is in the same frequency range as the sensors are monitoring.

ddnspider
December 7th, 2013, 09:47 AM
What will a B3618 table look like on a typical boosted car? Sloped downwards sharply when boost kicks in and then flat lined out to redline with 0.78?
You need to know what RPM the car will make boost at first. If you don't generate hardly any boost at all below 3k rpm's theres no point in dropping to .78 at 2k.

Rich Z
December 7th, 2013, 10:10 AM
You need to know what RPM the car will make boost at first. If you don't generate hardly any boost at all below 3k rpm's theres no point in dropping to .78 at 2k.

Well, the last time I did a run that went into boost, I simply put it in fourth gear and just ran up the RPMs. At 69% throttle at 1751 rpm I was already at 105kPa. I was at 182kPa at 3700 rpm when the wastegates said "enough!". My log file is on this page -> http://forum.efilive.com/showthread.php?22840-Sanity-check-please/page14, post #135. Boost starts around frame 11920.

So I'm hoping I'm OK in that department.

barbermi
December 7th, 2013, 01:44 PM
Thanks again spider.

So far we have completed the main VE and the BEN factor is great. The boost VE we have adjusted so that our AFR is where we want it under boost. We still cant get the PE working unless it is in closed loop. So the BEN factor when in boost is way off but our end result is good. If we put the tune back to closed loop and disable the PE how will the AFR react when in boost. Seeing the engine is a lot richer than the commanded AFR will the PCM lean out our tune under boost? If so how much could we expect the tune to lean out during a power run?

ddnspider
December 7th, 2013, 02:20 PM
Well, the last time I did a run that went into boost, I simply put it in fourth gear and just ran up the RPMs. At 69% throttle at 1751 rpm I was already at 105kPa. I was at 182kPa at 3700 rpm when the wastegates said "enough!". My log file is on this page -> http://forum.efilive.com/showthread.php?22840-Sanity-check-please/page14, post #135. Boost starts around frame 11920.

So I'm hoping I'm OK in that department.

Looks like your boost threshold is pretty low, makes the car fun on the street :good: So then yes, you would likely have a sharper slope for your PE vs. RPM table. You still don't need to go to max rich right away, there can still be a transition.

ddnspider
December 7th, 2013, 02:26 PM
Thanks again spider.

So far we have completed the main VE and the BEN factor is great. The boost VE we have adjusted so that our AFR is where we want it under boost. We still cant get the PE working unless it is in closed loop. So the BEN factor when in boost is way off but our end result is good. If we put the tune back to closed loop and disable the PE how will the AFR react when in boost. Seeing the engine is a lot richer than the commanded AFR will the PCM lean out our tune under boost? If so how much could we expect the tune to lean out during a power run?


I guess I'm still stuck on that a bit though, how can your boost VE table table be what youre commanding if you cant command your AFR :) When in OL, your commanded AFR will be the richer of B3618 (PE vs RPM) and B3605 (commanded fuel when in open loop). Are you saying you updated B3605 to low 11:1 in the 100kPa column and its using that for all boosted regions?

barbermi
December 7th, 2013, 02:28 PM
We have fudged the boost VE based on measured AFR with a wideband. The commanded AFR is commanding a leaner mixture.

ddnspider
December 7th, 2013, 02:34 PM
We have fudged the boost VE based on measured AFR with a wideband. The commanded AFR is commanding a leaner mixture.

exactly what I was afraid of. Can you log commanded air/fuel? I think its GM.AFR or GM.EQUIVRATIO. That will tell you what the car is trying to get to, and then you can track that back to what table in the tune is being used.

barbermi
December 7th, 2013, 02:35 PM
We have logged commanded AFR in the logs posted previously but i cant for the life of me work out where we are getting the wrong command.

ddnspider
December 7th, 2013, 03:19 PM
We have logged commanded AFR in the logs posted previously but i cant for the life of me work out where we are getting the wrong command.

In the last log I saw you post on page 2, GM.AFR was 12.76.......and your 100kPa column in B3647, commanded fuel when in open loop, is 12.73. You PE table was set to 14.49, meaning that B3647 was the richer of the 2 tables, so it commanded 12.7.

darcy
December 7th, 2013, 03:48 PM
Try with with your Commanded OL Fuel & PE, something like this:
16205

NA fuelling ramps down to 12.5 as load increases, then once you're seeing a couple of psi, PE drops it further to 11.3

johnv
December 7th, 2013, 04:16 PM
Thats how I do it too darcy, makes for a better transition of fueling into boost.
Post up your latest tune file so we can take a peek.
Once the fueling is better dialed in, if your still seeing knock in the logs, you can try some race gas or octane booster, if knock is still present it is likely false knock from other sources and can be ignored.
Your main timing table needs some work too big a veriation from cell too cell.


What is your turbo setup ?

If single turbo don't bother trying to put it back into closed loop when tuned, and for that matter even with twins unless its a road car and your concerned about fuel economy I would leave it perminantly open loop.

joecar
December 7th, 2013, 04:48 PM
Hi Joecar. The MAP scaler is done too. Out of curiosity we logged current gear when doing a power run. We noticed knock at the exact time the gearbox downshifted from 2nd to 1st when first applying power. The next few runs we held first then applied power with no knock. Whats the odds the shift change is giving us false knock?Post a log file.

joecar
December 7th, 2013, 04:49 PM
We have fudged the boost VE based on measured AFR with a wideband. The commanded AFR is commanding a leaner mixture.This means the air modelling (VE and MAF tables) are incorrect.

joecar
December 7th, 2013, 04:52 PM
Maybe this will help (requires careful reading to see the interactions among tables), see post #4 here: Summary-Notes (http://forum.efilive.com/showthread.php?14188-Summary-Notes)

joecar
December 7th, 2013, 04:54 PM
I would log GM.EQIVRATIO rather than GM.AFR... I've have seen GM.AFR miss a few squiggles compared to GM.EQIVRATIO.

barbermi
December 7th, 2013, 05:00 PM
Thats how I do it too darcy, makes for a better transition of fueling into boost.
Post up your latest tune file so we can take a peek.
Once the fueling is better dialed in, if your still seeing knock in the logs, you can try some race gas or octane booster, if knock is still present it is likely false knock from other sources and can be ignored.
Your main timing table needs some work too big a veriation from cell too cell.


What is your turbo setup ?

If single turbo don't bother trying to put it back into closed loop when tuned, and for that matter even with twins unless its a road car and your concerned about fuel economy I would leave it perminantly open loop.

Its not actually my car its ttls1's car. Its a twin turbo lq4

Rich Z
December 8th, 2013, 06:10 AM
Maybe this will help (requires careful reading to see the interactions among tables), see post #4 here: Summary-Notes (http://forum.efilive.com/showthread.php?14188-Summary-Notes)

Wow! Lots of GOOD stuff in that referenced thread!

ttls1
December 8th, 2013, 12:35 PM
That's a good read cheers joecar.

I also found this post from you in September post #2

If I planned on running OL all of the time would this setup be fine? Provided we can get PE enabled in OL.


http://forum.efilive.com/showthread.php?23157-Power-Enrichment-and-Boost-VE&highlight=power+enrichment

joecar
December 8th, 2013, 12:56 PM
That's a good read cheers joecar.

I also found this post from you in September post #2

If I planned on running OL all of the time would this setup be fine? Provided we can get PE enabled in OL.


http://forum.efilive.com/showthread.php?23157-Power-Enrichment-and-Boost-VE&highlight=power+enrichmentYes, this is what some tuners have been doing, it gives easier conrol of fueling throughout the MAP range

( setting OLFA B3647 to provide sub-boost fueling upto 105kPa, and PE B3618 to provide boost fueling by setting B3613 to 110kPa (and making sure PE enables ).

barbermi
December 8th, 2013, 01:59 PM
Seems like the easiest way to set it up for sure. Thanks joecar.

barbermi
January 3rd, 2014, 10:04 PM
Bit of feedback for you all.
The knock when coming on boost was due to gear shifts. Bogus knock!
We have got the PE working correctly thanks to joecar. Until you mentioned we werent aware the tune will always run the richer AFR. Now all RPM's are set to 11.3 in the PE modifier and we have got the BEN factor nearly spot on when in boost.
We have set the commanded fuel back to 1 in the cells that were .99's and the STFT's are working. LTFT's are yet to learn and function. DFCO works a treat! Next step is start adding spark and see how it goes down the 1/4 mile!

Thanks for everyones input.

Boost
January 4th, 2014, 01:10 AM
Thanks for the update!