Another MS conversion, Progress, at last. Options

[JamesM]

Progress!!

In the last week or so Mark and I have made some changes: put in a new fuel pump and filter and straightened out a couple of fuel lines, a new ECU, and new trailing arm bushings.

I think the fuel issue and the ECU have cleared out the background issues and left the real problem clearly in view. Between 2000 and 3000 rpm it bucks. The PW and MAP bounce around a lot. I'm not sure which is the cause and which is the effect.

I have done a lot of reading and made many changes to the tune without ever changing the bucking. Here is the most recent tune and a short log showing the bucking and system changes.

Does anyone have any idea what might be causing this?


https://www.dropbox.com/sh/p593bw6iyebgwjn/bANp3X25ic

Thanks for looking.

Sweet! logs! ill take a look. Is that the MSQ that was in use when the log was generated?

Yes. Thanks.

What wideband setup are you using and how is it set? LC1 0-5v?

[JimN73]

Progress!!

In the last week or so Mark and I have made some changes: put in a new fuel pump and filter and straightened out a couple of fuel lines, a new ECU, and new trailing arm bushings.

I think the fuel issue and the ECU have cleared out the background issues and left the real problem clearly in view. Between 2000 and 3000 rpm it bucks. The PW and MAP bounce around a lot. I'm not sure which is the cause and which is the effect.

I have done a lot of reading and made many changes to the tune without ever changing the bucking. Here is the most recent tune and a short log showing the bucking and system changes.

Does anyone have any idea what might be causing this?


https://www.dropbox.com/sh/p593bw6iyebgwjn/bANp3X25ic

Thanks for looking.

Sweet! logs! ill take a look. Is that the MSQ that was in use when the log was generated?

Yes. Thanks.

What wideband setup are you using and how is it set? LC1 0-5v?

Mark is the best one to answer that. Hopefully, he'll see this and let us know.

[McMark]

AEM UEGO 0-5v

[JimN73]

thanks, Mark

[JamesM]

Provided I have the settings for your wideband correct in my log viewer (AEM Linear) I would say you are lean pretty much across the board, and to the point of misfiring in the 2000-3000 RPM, 75-100KPA range, also looks like this lean condition may be masked partially by your current acceleration enrichment settings.

Your AFR is inversely mirroring your pulse width, which means your car is doing what the MS system is telling it to do, the problem is it is just not telling it to squirt the right amount of fuel so, tune, tune, tune.

The log file is pretty short, its hard to tell a lot from only 30 seconds of data and I would like to see more to really get a good idea but what I see is

1. Lean, like really LEAN

2. The lean condition is being compensated for by the accell enrichment which is probably going to be to much once the Ve table is set properly. Accell enrichment should be the last thing you tune and right now is just going to give confusing results, so you would be doing yourself a favor to just zero it out for now.

3. From the short bit at the end of the log that looks like you were idling, your idle vacuum is very poor (high 75-80KPA value) however you are also very LEAN there as well (15-17:1 AFR) Richen that up to the low 13's and I would not be surprised to see your vacuum increase at idle quite a bit, down to the 50kpa range at least I would guess (maybe more but I am not familiar with your cam to know). This should increase your throttle response and drivability a TON.

4. Your Ve values are seeming pretty high for how lean you are running. Are you sure you calculated your req fuel constant properly? You might want to think about scaling that before you do anything else.

5. Other than that, in the 30 seconds of data everything looks to be responding about as I would expect it to, MAP value responding properly to throttle position, pulse width being affected by Ve table and accell enrichment, and AFR then reflecting pulse width, low vacuum due to lean condition, everything seems to add up to me. It did not look like your throttle position sensor ever registered full open, but that may just be because you didnt hit the gas fully, looks like you never went over 50%.

I could send you a tweaked MSQ built from the log files that should fix that bit of the 2000-3000 range, but as the data was limited and you have a LOT of other changes that are going to happen to get to where you need to be, it might be better for you to adjust it yourself (also I don’t want to be liable should your engine explode)


-James

[JamesM]

Just for fun I am going to give a play by play of what I see (or think I see) in a portion of this log.

Starting at around the 19127 sec area you see the RPM decline to the 2500ish problem area, engine looks to be in a state of overrun, throttle closed high manifold vacuum. Around 127.8, throttle is opened back up to 20%, this is where the hold at 2000-2500 rpm range starts, MAP jumps and holds in line with the throttle position as expected. Small lean spike to 15 AFR from opening the throttle (looks to be a bit of delay from the AFR reading) pulse width jumps due to accell enrichment and hitting a higher value on the Ve table and we look to be back at an acceptable AFR range around 128.6. Here is the important part though, @ 128.9 throttle is being increased slowly but apparently not fast enough to trigger accell enrichment so you see the accell enrichment decaying and with it the pulse width decreasing, this causes the AFR to skyrocket to lean misfire (19:1)territory. If you were just looking at the RPM, MAP and VE table this swinging AFR would not make sense, but looking at all the data in the log it does. @130.8 TP change is enough to trigger accell enrichment again, PW goes up, AFR swings back down to the 13's, again all despite being in pretty much the same part of the Ve table. @132.1 Accell enrichment decays again, AFR shoots back up, and here is where we get some more good data. RPM, throttle position and MAP now remain pretty constant until about 143sec with NO accell enrichment to get in the way so we can see how well the Ve table alone is fueling the engine. What we see now is the AFR jumping between lean(~17:1) and leaner(~18:1) as it bounces between the 80 and 88KPA rows on the Ve table, which have a pretty large value change for a somewhat small MAP change. Probably lean missfiring as well. Jump to 144, throttle closed, engine in overrun back down to idle speeds @147. Idle AFR and MAP both way to high, lowering one should fix the other, but as they are tied the table will need to be adjusted all the way down.

So again to summarize what has turned into a huge post, turn off your accell enrichment and tune the Ve table from there with it off. You are way to lean but with the amount of accell enrichment you are shooting currently, attempting to richen up the VE table would result in a rich misfire anytime you get on the gas. Think about it, the amount of gas your accell enrichment is adding is enough to swing your AFRs down 4-5 points, so with your current settings, if you map was tuned properly for a mid 13:1ish cruise AFR and you punched the gas you would be in the 8:1 range and fouling your plugs. Accell enrichment is the last part of the tuning process and should be disabled completely until you get to that point. In my experience, with how rich these engines like to run, and the fact we have increased the spark advance in the lower RPM range over stock, I find that you don’t need a whole lot of accell enrichment for decent performance.

Get a good tune on the Ve table first, eliminate all other variables until that is done otherwise you are just chasing your tail.

[JimN73]

Thanks, James. I appreciate your help, but I didn't mean for you to stay up all night.

Jim

[DBCooper]

Very cool. Anytime Megasquirt is discussed the phrase "you'll end up learning how engines really run" is mentioned, but I don't think people give that much consideration. James is obviously way up the curve on this, but it's really worthwhile for anyone to understand the interrelationships of what's going on with an engine. There are parameters depending on other parametiers with layers of conditionals thrown in. Getting a feel for how it all fits together will help you tune carburetors and ignition in addition to what you learn about injection. An education and all good experience... especially after it's all together and running. Good one James. Thanks.

[McMark]

Great post James and DB! (IMG:style_emoticons/default/first.gif)

[JamesM]

Thanks, James. I appreciate your help, but I didn't mean for you to stay up all night.

Jim

No biggie I am sort of a night person, I typed that out while watching a movie and this sort of thing is fun to me so i guess you could say i was relaxing. (IMG:style_emoticons/default/biggrin.gif)

Really intrested to see what your impressions of the motor are once it is dialed in

[VaccaRabite]

Wow. Sounds like you are on the right path for sure.
When you get this tuned please send me your files! As our engines are so similar I may just be able to plug in your data and go.

I have zero doubt you will be able to tune the bucking out.
Zach

[JamesM]

Very cool. Anytime Megasquirt is discussed the phrase "you'll end up learning how engines really run" is mentioned, but I don't think people give that much consideration. James is obviously way up the curve on this, but it's really worthwhile for anyone to understand the interrelationships of what's going on with an engine. There are parameters depending on other parametiers with layers of conditionals thrown in. Getting a feel for how it all fits together will help you tune carburetors and ignition in addition to what you learn about injection. An education and all good experience... especially after it's all together and running. Good one James. Thanks.

(IMG:style_emoticons/default/agree.gif) Probably the most educational thing I have ever done on a car. Also its why I didn't just want to tweak a tune and say "here try this", its a pretty layered process and you really need to understand how everything relates to move through it.

Its a fun process though!

[JimN73]

Thanks, James. You asked the right question: Req_fuel OK?. No!

Fixed it. At Mark's suggestion I took out all of the acceleration enrichment and drove and tuned. You'll like the new VE table, 124 cells changed, average change 10, max change 24. (IMG:style_emoticons/default/driving.gif) (IMG:style_emoticons/default/driving.gif)

Did data logging until the laptop battery forced a shutdown. Fortunately, the log file was saved. It's attached. The log AFR is much lower than the AFR table for every AFR/MAP bin on the AFR table. I'm not sure what that means or what I should do about it. (IMG:style_emoticons/default/icon14.gif)

There is still some hesitation. I'm hoping/guessing that some acceleration enrichment will help that. When the battery recharges, I will do that and report back. (IMG:style_emoticons/default/type.gif)



https://www.dropbox.com/sh/p593bw6iyebgwjn/bANp3X25ic

[VaccaRabite]

(IMG:style_emoticons/default/piratenanner.gif) (IMG:style_emoticons/default/driving.gif) (IMG:style_emoticons/default/first.gif) (IMG:style_emoticons/default/smilie_pokal.gif)

Kick ass seabass! Looking forward to what your enrichment tables do when you add them back in.

Zach

[JimN73]

Zach, I started a while ago with minimal PW additions over rates that Mark set up some time ago and small changes to TPSdot threshold%. The mild bucking at a little over 2000 rpm (cruise and light acceleration) persisted, so I increased the req_fuel a bit. It seems better now, but I don't have any more time today to fuss with it. With moderate to aggressive throttle, it seems to go well.

There's an autocross tomorrow, so I'll fiddle more on Sunday.

[JamesM]

I have to do some work tomorrow morning where I will be in front of a computer for a long stretch of time, so I probably won't look at your new logs till then (and will probably have another huge post for you mentioning some other aspects that I left out of my last one to not complicate things.

but for now some things to think about...

Begin brain dump....

In theory, if you knew all the variables and operating characteristics of an engine you would be able to create a Megasquirt tune for it by hand without ever starting the engine. The reason we can't do this entirely on theory is because we do not know the volumetric efficiency of your (or any) specific engine. That is what the "fuel map" in megasquirt actually is, it is not really specifying an amount of fuel to inject but rather represents the volumetric efficiency of your engine at all its operating points. (this is why it is called the Ve map) What megasquirt then does is take Ve map of your engine and calculates it with the req_fuel value(and corrects for temp sensors) to determine how much fuel to inject.

req_fuel itself is a constant that represents the injector time in milliseconds required to get a stoich 14.7:1 burn at 100kPa manifold pressure 70deg F for your given cylinder displacement IF your engine pumped at 100% efficiency. As no naturally aspirated engine does pump with 100% efficiency (best guess for our engines is probably somewhere between 75-85%) what this means in terms of your Ve map is that IF you were tuning to achieve a 14.7:1 mixture at WOT the values in the upper portion of your Ve map would be in that same range of 75-85. We are more likely tuning to achieve around 12.5:1 at WOT so I would expect values in the upper portion of your map to be closer to the high 90's, maybe low 100's if everything else was set correctly. The fact that you were in the 17:1 range with Ve values already close to or over 100 is what prompted me to suggest adjusting your req_fuel to a properly CALCULATED value.

That all being said, the best way to initially set your req_fuel value is to calculate it, as unlike your Ve map, all the variables to determine req_fuel can be known ahead of time, you just need to know your engine displacement and injector flow rates at the pressure you are running them.

This is not to say that you can't make things work with an incorrect req_fuel value, it just means the numbers you are looking at won't make any sense from a theory standpoint. This is also why most people's Ve maps turn out to be completely useless to other people, because they were not setup using proper theory. A Ve map that has been setup using PROPER settings can be swapped to another persons car with the same engine configuration without re-tuning provided their settings are also correct for their car. It also allows for pretty simple injector swaps provided all correct values for both sets of injectors are known. This is more than just theory, I have done both.

So, back to your car, what you did when you changed the req_fuel was essentially richen your mixture across the board, where as you could have just adjusted your Ve map to richen the specific areas that were lean. Both ways work just fine, neither one is technically correct. IF you want to be correct in your theory, you basically need to go back to square 1 and set your req_fuel to a CALCULATED value and then tune your Ve map properly to that. The problem with that is you need to know your injector flow rate at the pressure you run them, and most published specs are at 3bar where as the 914 pressure regulator is usually set around 2 bar. Again, unless you are running them at published spec, the only way to be 100% correct would be to flow bench your setup as the flow characteristics of an injector might not be linear in relation to pressure.

This is all just food for thought/educational info and I will help you work through the tune no matter which way you want to do it, its just good to know this stuff so you can know what to expect.

What are your injector part numbers and what pressure are you running at? Maybe I can make a best guess on the req_fuel if I can find the specs.

...End brain dump

and if you were able to digest all that in one sitting then (IMG:style_emoticons/default/beerchug.gif) (IMG:style_emoticons/default/beerchug.gif) (IMG:style_emoticons/default/beerchug.gif) to you! Still just looking at the tip of the iceberg here.

[JimN73]

thanks for the brain dump, James. It makes sense, and it's nice to get it all in one in one place, not in bits and pieces as is often found in manuals.

If Zach hasn't installed his injectors yet, maybe he can pull a number off of them. Mark will know more about them, too.

The injectors are 23 gph injectors at 36 psi. I'm running at 30 psi and expect to boost req_fuel and the enrichments to compensate. As long as I don't have pulsewidths that get too long, that should work. We're using an AFR of 13 yielding a req_fuel of 16.1..

Did an autocross and some freeway driving today. Now that I know a bit more, it's obvious that the engine is still running lean. Tomorrow.

[McMark]

The injectors are custom. There isn't a part number. The specs Jim posted are the ones from the supplier.

16.1 Req Fuel would be at the supplier spec of 23 lb/hr at 36.5psi. But since we're running at 30 psi, I speculate something around 20 lb/hr which IIRC comes out somewhere in the 28 range for Req Fuel. But check that math.

[JimN73]

thanks, Mark. I don't know what the time window for the pulse widths is, but 28 seem to be a pretty big number, and with other enrichments, might get close to the limit.

Wouldn't it take that question out of the discussion if I increased the fuel pressure?

[JamesM]

thanks, Mark. I don't know what the time window for the pulse widths is, but 28 seem to be a pretty big number, and with other enrichments, might get close to the limit.

Wouldn't it take that question out of the discussion if I increased the fuel pressure?

The req_fuel calculation should be made with a 14.7 AFR if you are burning gasoline, the actual calculator on the megasquirt manual page has this built in as a constant. If the injectors are indeed pushing 20lbs/hr then 16.5 would be the correct req_fuel. If flow were directly related to pressure then the math says they would be at 18.9lbs/hr which would put req_fuel at 17.4. At their full 23lb/hr req_fuel would be 14.3. This at least gives us a rough range of where the setting should be for these injectors, somewhere between 14.3 and 17.4 roughly.

I would say probably go with Marks assumption they are running around 20lbs and start out with a req_fuel of 16.5.