Home built Hot Rod

[Mike D'Silva]

Looks just like my old Dynomax 17231 muffler system. That was very good around town, zero drone when cruising, but full throttle above 5000 it would fucking scream. Looking foward to getting your opinion of actual driving.

[MFX]

6 minutes ago, Mike D'Silva said:

Looks just like my old Dynomax 17231 muffler system. That was very good around town, zero drone when cruising, but full throttle above 5000 it would fucking scream. Looking foward to getting your opinion of actual driving.

I have gone out on a quick run and I am really surprised at how good it is. So simple and I really thought it would be super loud, and it isn't.

[Mike D'Silva]

1 minute ago, MFX said:

I have gone out on a quick run and I am really surprised at how good it is. So simple and I really thought it would be super loud, and it isn't.

Told ya!

What about initial performance.. any perceptible difference? I used a 2.5inch muffler BUT I still used 2.25 tubing... (step downs on the connectors to the muffler)

[MFX]

1 minute ago, Mike D'Silva said:

Told ya!

What about initial performance.. any perceptible difference? I used a 2.5inch muffler BUT I still used 2.25 tubing... (step downs on the connectors to the muffler)

I think I actually got it wrong in the video as the muffler is actually 2.25, but the headers end in 2" so I did the same stepping it up at the V bands.

[MFX]

As for the performance. I need to crunch the actual data, but the initial butt feeling is that I may have lost a tiny bit down low, but now it really wants to rev at the top end. Not a huge difference, but I think it is what it needed.

[Coastr]

The butt dyno never lies, right?

[MFX]

This week I go through some data and make a few tweaks to see what sort of difference the new exhaust on Harry 911 has made.

 

 

[Mike D'Silva]

my suspicion is that you have more top end, and less down low. 

You could split your logs so instead of calculating time to go from 2500 to 7000, calculate time from 2500-4500 and then do 4500-7000... do both logs.

Maybe your second run, is faster between 4500-7000 compared to the first run? And slower between 2500-4500 ... The cumulative effect being basically the same total time from 2500-7k as your first log.

 

I think you are running quite rich, (if your sensors can be relied upon)  between 4k and 6k... you dip down to 0.7ish  which is around 10.5 afr... really rich. I don't know what the optimum lamba target is for these motors but I would be comfortable with 12.2 -12.5 afr or 0.83-0.85 lambda.. maybe they tuned it a little rich to allow for variances between cylinders.. 

Can you post your lambda target table?

[MFX]

1 hour ago, Mike D'Silva said:

my suspicion is that you have more top end, and less down low. 

You could split your logs so instead of calculating time to go from 2500 to 7000, calculate time from 2500-4500 and then do 4500-7000... do both logs.

Maybe your second run, is faster between 4500-7000 compared to the first run? And slower between 2500-4500 ... The cumulative effect being basically the same total time from 2500-7k as your first log.

 

I think you are running quite rich, (if your sensors can be relied upon)  between 4k and 6k... you dip down to 0.7ish  which is around 10.5 afr... really rich. I don't know what the optimum lamba target is for these motors but I would be comfortable with 12.2 -12.5 afr or 0.83-0.85 lambda.. maybe they tuned it a little rich to allow for variances between cylinders.. 

Can you post your lambda target table?

I will dig up the lambda table when I get home. As for the logs, I might sit down with them a bit more on the week end and try those figures.

[Peter M]

Your target lambda looks sensible but the ECU is really struggling to control the actual lambda and is varying wildly at a steady state throttle opening and loading. It should be achieving an actual much closer to the target

The ECT looks really low to me assuming it is mounted on a cylinder head.  I would expect it to be about twice what is being shown.

It could also do with a bit more acceleration enrichment with the rapid opening of the throttle to limit the over lean condition.

I reckon you have more surprise and delight coming if you can solve these tune opportunities. 

[Peter M]

I think your logs should look more like this.  Looking at a similar run on a similar time scale shows:

Greater acceleration enrichment

Less variation in the actual lambda to target plus none of the wild cycling from rich to lean.  The target lambda below would be in the 0.85 to 0.81 range with 0.81 at WOT at 6,000rpm  

Not shown on this screen but another, the IAT on this ramp run was 27 degrees and the cylinder head temp went from around 130 to 152 degrees over the run.

 

[MFX]

10 minutes ago, Peter M said:

Your target lamba looks sensible but the ECU is really struggling to control the actual lamba and is varying wildly at a steady state throttle opening and loading. It should be achieving an actual much closer to the target

The ECT looks really low to me assuming it is mounted on a cylinder head.  I would expect it to be about twice what is being shown.

The ECU should also be able to better handle the rapid throttle opening to WOT without the swing to over lean.  There is certainly a blip in the injector duty cycle from the acceleration enrichment setting (Sorry, I don't know how Link manage this and what they call it) but it's obviously not enough to avoid the lean out.

I reckon you have more surprise and delight coming if you can solve these tune opportunities. 

The ECT is not that close to heat, as there were not many places I could mount it. It does look like my AFR's are pretty rich and could use some tweaking, so I will have a look into it.

8 hours ago, Peter M said:

I think your logs should look more like this.  Looking at a similar run on a similar time scale shows:

Greater acceleration enrichment

Less variation in the actual lamba to target plus none of the wild cycling from rich to lean.  The target lamba below would be in the 0.85 to 0.81 range with 0.81 at WOT at 6,000rpm  

Not shown on this screen but another, the IAT on this ramp run was 27 degrees and the cylinder head temp went from around 130 to 152 degrees over the run.

 

I agree my lambda table seems like it is all over the place, and could use some tweaking. Unfortunately I am not sure if it was better before the exhaust change as it obviously wasn't working.

[OZ930]

Did Neil specify an exhaust size for this engine ?

[Peter M]

11 hours ago, MFX said:

The ECT is not that close to heat, as there were not many places I could mount it. It does look like my AFR's are pretty rich and could use some tweaking, so I will have a look into it.

I agree my lambda table seems like it is all over the place, and could use some tweaking. Unfortunately I am not sure if it was better before the exhaust change as it obviously wasn't working.

Jeff,

Mounting the temp sensor on the head will give better resolution due to the wider temperature range measured and will model the actual AFR requirements better so will likely allow you to improve your cold start performance in particular if you set up an appropriate cranking enrichment  and immediate after start enrichment tables.  This also gives you the opportunity to increase fueling when extremely high engine/head temps are encountered. I notice that the pre 3.2 engine builders often use a clamp arrangement to fasten the sensor to the cylinder head.

Looking at the data log shown at around 10 minutes 42 seconds into your above video shows that your AFR's aren't consistently rich or consistently lean.  It's the cyclic variation between that is the concern.  A bit less than 0.8 lambda is not much of a problem, 1.0 lambda is a problem and you are losing power and risking detonation (or having to pull timing out which is lost hp!) when running WOT.

I don't know why this cyclic variation would occur unless you are so rich that it is causing the sensors to misread as lean once they are operated outside their acceptable lambda range?

I'd would suggest you run it on a dyno and compare your lambda reading to that of the dyno's tail pipe sniffer and see how they compare. Are the plugs clean or black with thick soot?

I don't think your new muffler is causing this issue.  Great work BTW.

Don't be discouraged, you are almost there and it will be well worth the head scratching and effort! 

[Pork Chops]

47 minutes ago, Peter M said:

Jeff,

Mounting the temp sensor on the head will give better resolution due to the wider temperature range measured and will model the actual AFR requirements better so will likely allow you to improve your cold start performance in particular if you set up an appropriate cranking enrichment  and immediate after start enrichment tables.  This also gives you the opportunity to increase fueling when extremely high engine/head temps are encountered. I notice that the pre 3.2 engine builders often use a clamp arrangement to fasten the sensor to the cylinder head.

Looking at the data log shown at around 10 minutes 42 seconds into your above video shows that your AFR's aren't consistently rich or consistently lean.  It's the cyclic variation between that is the concern.  A bit less than 0.8 lambda is not much of a problem, 1.0 lambda is a problem and you are losing power and risking detonation (or having to pull timing out which is lost hp!) when running WOT.

I don't know why this cyclic variation would occur unless you are so rich that it is causing the sensors to misread as lean once they are operated outside their acceptable lambda range?

I'd would suggest you run it on a dyno and compare your lambda reading to that of the dyno's tail pipe sniffer and see how they compare. Are the plugs clean or black with thick soot?

I don't think your new muffler is causing this issue.  Great work BTW.

Don't be discouraged, you are almost there and it will be well worth the head scratching and effort! 

Always enjoyed your contributions Peter.

[Peter M]

1 hour ago, Pork Chops said:

Always enjoyed your contributions Peter.

Oh shucks Mr Chops, thank you!

Who we really need is @Redracn who used to do this as a living and will give really insightful advice.

[MFX]

13 hours ago, Peter M said:

Jeff,

Mounting the temp sensor on the head will give better resolution due to the wider temperature range measured and will model the actual AFR requirements better so will likely allow you to improve your cold start performance in particular if you set up an appropriate cranking enrichment  and immediate after start enrichment tables.  This also gives you the opportunity to increase fueling when extremely high engine/head temps are encountered. I notice that the pre 3.2 engine builders often use a clamp arrangement to fasten the sensor to the cylinder head.

Looking at the data log shown at around 10 minutes 42 seconds into your above video shows that your AFR's aren't consistently rich or consistently lean.  It's the cyclic variation between that is the concern.  A bit less than 0.8 lambda is not much of a problem, 1.0 lambda is a problem and you are losing power and risking detonation (or having to pull timing out which is lost hp!) when running WOT.

I don't know why this cyclic variation would occur unless you are so rich that it is causing the sensors to misread as lean once they are operated outside their acceptable lambda range?

I'd would suggest you run it on a dyno and compare your lambda reading to that of the dyno's tail pipe sniffer and see how they compare. Are the plugs clean or black with thick soot?

I don't think your new muffler is causing this issue.  Great work BTW.

Don't be discouraged, you are almost there and it will be well worth the head scratching and effort! 

Thanks mate. I will look into mounting the sensor on a clamp. When I put it in I was struggling to find somewhere decent to mount it. 

I have driven the car to work today and I will have a go at playing with the tables a bit more on the way home tomorrow.

[Peter M]

CHT Clamp details:

Cylinder Head Temp Sensors - Pelican Parts Forums

 

[Peter M]

Jeff,

Just thinking about it a bit more in Coles this morning, you haven't made any changes to your closed loop Lambda gain table?  Assuming it isn't running mega rich causing the sensors to misread, it's like it's over correcting the fueling that is then causing that slow oscillation in the Lambda readings.  I think you can reduce the gain sensitivity at higher rpm to prevent the over-correction.

Sorry, I can't really help, we need an expert! 

 

[MFX]

1 hour ago, Peter M said:

Jeff,

Just thinking about it a bit more in Coles this morning, you haven't made any changes to your closed loop Lambda gain table?  Assuming it isn't running mega rich causing the sensors to misread, it's like it's over correcting the fueling that is then causing that slow oscillation in the Lambda readings.  I think you can reduce the gain sensitivity at higher rpm to prevent the over-correction.

Sorry, I can't really help, we need an expert! 

 

Thanks for that clamp sensor. It does look different to the one I have which just looks like an earth connector (like a washer with a wire attached to it) that I just bolted onto one of the engine cover bolts into the cam tower. I am sure I could make something up similar to what is in that thread though.

At the moment it is not using closed loop, although looking at the cold start set up it seems that even at full engine temp (which seems to be 40-50 degrees) it is still running a correction. The issue is, I really have no idea what I am doing. I need to find another tuner who is closer and and more available.

[Redracn]

I haven’t been following this but here I am thanks to Peter. I have only reviewed the last couple of pages so feel free to correct any assumptions I make.

The videos are great a lot of effort must go into each one and the result  looks fantastic a lot of quality work there and on the car.

Dyno Power: Not something I would be too worried about given the time between runs or at all unless you are competing.  If concerned about the engine check compression, do a leak down and if really concerned get the oil tested. 

A couple of notes.

The air/exhaust extraction on that chassis dyno looks a bit sus especially for an engine with its intake so close to the rear. Even the slightest contamination of the intake air with exhaust fumes is bad news for power production. Difference could just be the wind direction at the time.

Given the time between runs there is so much that can be different not only atmospheric conditions, tying down etc but work done on the engine etc. just removing and replacing the crank sensor and/or its bracket can easily alter the ignition timing. Rule IV)  Always check the ignition timing always. What seems like a small inconsequential change like exhaust connections can have a surprisingly large effect. Now if I only had a $ for every time someone told me nothing was changed. 
 

Head temp is good as long as the tables are modified to suit. Oil is way to slow to warm up. I don’t like the clamp on a fin or similar design. Good head temp is right on the head (not cooling fins) and thermally bonded using a thermal transfer compound. If a clamp must be used then it should be thermally bonded to the fin and preferably insulated from the cooling air flow. 
 

I just saw that closed loop lambda is off which is best while tuning the efficiency(Fuel) tables. It can be on if it is known that it is working well and tracking the required lambda but then you need to use the fueling offset it applied at any given point to modify the efficiency table. Some ECU can self learn the efficiency table but I don’t think the Link is one of them. It is best if the tables and various corrections get the lambda to within +-5% of what is required before letting closed loop lambda operate especially given the lambda sensor is the most unreliable sensor in the car.

Logging if it isn’t already should be your best friend and a must when tuning and looking for issues. There are many more I items I would add such as lambda error(difference between what you want and what it is) or the target lambda so the analysis can show it and the error calculated, lambda correction (if in closed loop). Unless you are chasing emissions compliance it is best to delay turning on the lambda sensor heater for 30 seconds or more after engine start to ensure no liquid water in the exhaust it’s a real sensor killer. 

Pro Tip: when displaying logged data turn off the channel auto scaling. Showing lambda from 0 to 10.0 is a complete waste. Manual scale it 0.70 to 1.2. Same for temperatures etc that way you will always see everything on the same scale (apples with apples) each time you load logged data and can quickly at a glance see any issues. 
 

Books can and have been written on this stuff and still not covered it all.

 

[MFX]

25 minutes ago, Redracn said:

I haven’t been following this but here I am thanks to Peter. I have only reviewed the last couple of pages so feel free to correct any assumptions I make.

The videos are great a lot of effort must go into each one and the result  looks fantastic a lot of quality work there and on the car.

Dyno Power: Not something I would be too worried about given the time between runs or at all unless you are competing.  If concerned about the engine check compression, do a leak down and if really concerned get the oil tested. 
Head temp is good as long as the tables are modified to suit. Oil is way to slow to warm up. I don’t like the clamp on a fin or similar design. Good head temp is right on the head (not cooling fins) and thermally bonded using a thermal transfer compound. If a clamp must be used then it should be thermally bonded to the fin and preferably insulated from the cooling air flow. 
 

I just saw that closed loop lambda is off which is best while tuning the efficiency(Fuel) tables. It can be on if it is known that it is working well and tracking the required lambda but then you need to use the fueling offset it applied at any given point to modify the efficiency table. Some ECU can self learn the efficiency table but I don’t think the Link is one of them. It is best if the tables and various corrections get the lambda to within +-5% of what is required before letting closed loop lambda operate especially given the lambda sensor is the most unreliable sensor in the car.

Logging if it isn’t already should be your best friend and a must when tuning and looking for issues. There are many more I items I would add such as lambda error(difference between what you want and what it is) or the target lambda so the analysis can show it and the error calculated, lambda correction (if in closed loop). Unless you are chasing emissions compliance it is best to delay turning on the lambda sensor heater for 30 seconds or more after engine start to ensure no liquid water in the exhaust it’s a real sensor killer. 

Pro Tip: when displaying logged data turn off the channel auto scaling. Showing lambda from 0 to 10.0 is a complete waste. Manual scale it 0.70 to 1.2. Same for temperatures etc that way you will always see everything on the same scale (apples with apples) each time you load logged data and can quickly at a glance see any issues. 
 

Books can and have been written on this stuff and still not covered it all.

Thanks for that awesome response. The only real change between first and second dyno runs was the exhaust was cut off and re welded on with V band clamps, but like you say, it could have been a whole heap of things. 

I like your idea of turning the heating off for 30 seconds. Not sure if the Link will let me do that, but I will give it a go. I suppose at the moment I should turn them off unless I am doing a log, as they are not actually being used by the ECU. 

I am also concerned about the current lambda fluctuation, I am just not sure about how to use the Link software and what to change in what map, as I really don't want to mess it up. I need to really sit down and learn it, then I can easily keep going out and doing logs and tweaking the figures until it is running to the target AFR. 

[Redracn]

1 hour ago, MFX said:

Thanks for that awesome response. The only real change between first and second dyno runs was the exhaust was cut off and re welded on with V band clamps, but like you say, it could have been a whole heap of things. 

I like your idea of turning the heating off for 30 seconds. Not sure if the Link will let me do that, but I will give it a go. I suppose at the moment I should turn them off unless I am doing a log, as they are not actually being used by the ECU. 

I am also concerned about the current lambda fluctuation, I am just not sure about how to use the Link software and what to change in what map, as I really don't want to mess it up. I need to really sit down and learn it, then I can easily keep going out and doing logs and tweaking the figures until it is running to the target AFR. 

Hi MFX

It is also not advisable to turn the lambda sensors off when driving. Best to remove them from the exhaust if not being powered and properly controlled by a sensor amplifier.  If you have turned off closed loop control then as long as the exhaust has no air leaks such as wrong direction slip joints (or v bands that aren’t perfectly sealed prior to the sensor and the engine is not misfiring then the reading should be good so any fluctuations would most likely be real. Could be a lot of reasons such as fuel rail resonance (dampers are available and fitted to a lot of cars) . Misfires make the lambda sensors read leaner than reality.
Yep best to learn the software and tuning fundamentals before getting into it as you can waste a lot of time and do damage going in blind. 

Welding in the Vband has changed the flow path particularly along the exhaust pipe walls and any change in diameter (cross section area) or disruption is generally bad for the flow. Not saying this IS an issue but it has to be on the list of possibilities if you are chasing peak power otherwise ignore and enjoy. 

[Peter M]

7 hours ago, MFX said:

At the moment it is not using closed loop, 

 

1 hour ago, MFX said:

I am also concerned about the current lambda fluctuation, I am just not sure about how to use the Link software and what to change in what map, as I really don't want to mess it up. I need to really sit down and learn it, then I can easily keep going out and doing logs and tweaking the figures until it is running to the target AFR. 

Jeff,

Sorry I assumed you were running closed loop so went down an unrelated rabbit hole of lambda correction, gain tables and possibly over rich conditions!

So the fueling beyond the starting table(s) is entirely dependent on your VE table then?  If that's the case you just need to adjust some figures in the table to correct the varying Lambda over the WOT speed range.  Looking at your data log again from around 10:23 of your video suggests to:

Increase VE in the cell nearest to MAP 94kPA and 3,300rpm by about 12%

Decrease VE in the cells nearest to MAP 93kPA: and 4850rpm by 8% and at 5900rpm by 5%

Increase VE in the cell nearest MAP 92kPA and 7000rpm by 5% 

Then adjust the VE cells around the above cells to blend it all in so it's a smooth transition both across and down at least 2 or 3 cells as I expect the adjustments above would be all in the top row of the table.

Give it another ramp run and see if that gives you a flatter Lambda curve at WOT across the rev range that is closer to your target of 0.82ish.  If it's still a bit lumpy, try some other minor VE table adjustments based on your data log until you are happy with it.

 

 

[MFX]

16 minutes ago, Peter M said:

 

Jeff,

Sorry I assumed you were running closed loop so went down an unrelated rabbit hole of lambda correction, gain tables and possibly over rich conditions!

So the fueling beyond the starting table(s) is entirely dependent on your VE table then?  If that's the case you just need to adjust some figures in the table to correct the varying Lambda over the WOT speed range.  Looking at your data log again from around 10:23 of your video suggests to:

Increase VE in the cell nearest to MAP 94kPA and 3,300rpm by about 12%

Decrease VE in the cells nearest to MAP 93kPA: and 4850rpm by 8% and at 5900rpm by 5%

Increase VE in the cell nearest MAP 92kPA and 7000rpm by 5% 

Then adjust the VE cells around the above cells to blend it all in so it's a smooth transition both across and down at least 2 or 3 cells as I expect the adjustments above would be all in the top row of the table.

Give it another ramp run and see if that gives you a flatter Lambda curve at WOT across the rev range that is closer to your target of 0.82ish.  If it's still a bit lumpy, try some other minor VE table adjustments based on your data log until you are happy with it.

I will have another look and do some minor tweaks to start with to make sure I don't mess anything up  

This is a copy of the current fuel table.