E46 330i Automatic - M54B30 Turbo

[Braymond141]

What does that have to do with strength during normal operation? Studs ripping out due to high in cylinder pressure and studs ripping out because the material properties have changed due to highly elevated temperatures are completely different failure modes and cannot be compared.

The first hand issue I saw was not related to the studs pulling out, but the motors grenading when tracked. I sold an S54 to a guy in AZ who had at least 5 3.0 M54s on stands that were all blown and he was done dealing with the aluminum block. We didn't go into too much detail, but he said they have horrible harmonics at high rpm. Although I'm sure this is where the ATI dampener helps a lot.

[PEI330Ci]

Do you find it troubling they rated a driveline component in HP capability and not torque?

No, not really. I know what they did to the internals, and I'm happy with it.

With an electronically controlled automatic transmission, the software is just as critical as the hardware for longevity. This is a BIG reason why I went with a stand-alone transmission controller; so that I can control the clutch overlap and duration, as well as the converter lock-up. The OEM software for example will lock up the converter on the gear change from 2nd to 3rd which gives a really bad shock to the drivetrain. I plan to keep the converter unlocked.....

People just repeat this, but there is never any proof. It mostly comes from the oversimplification that aluminum=weak, steel=strong.

Pro-Mod and Top fuel engines are aluminum blocks with liners...

Or because they regularly pull the threads out of the block when overheated...

I'm not certain it's only overheating.

I think in some scenarios, the combustion chamber temp "walks" the peak pressure during the combustion event closer to TDC. My theory is that there is a maximum peak cylinder pressure that the block/head can handle, and above that the weak point is the strength of the thread engagement with the block. This threshold I suspect is below what people would consider a high coolant temp, and the peak cylinder pressure event doesn't trigger knock sensors because it ISN'T an "explosion".

The first hand issue I saw was not related to the studs pulling out, but the motors grenading when tracked. I sold an S54 to a guy in AZ who had at least 5 3.0 M54s on stands that were all blown and he was done dealing with the aluminum block. We didn't go into too much detail, but he said they have horrible harmonics at high rpm. Although I'm sure this is where the ATI dampener helps a lot.

Oil pump shaft failures; even the VAC "upgrade" won't help.

As a counterpoint, I know of M54 engines that lived above 8000 RPM for sustained periods, but they were 2.5L and dry-sumped.

ATI damper + GSR Oil pump is a great combination to keep M54B30s alive.

- - - Updated - - -

[jakermac]

ATI damper + GSR Oil pump is a great combination to keep M54B30s alive.

More info on GSR oil pump please?

Also as an added point, my motor is NA, and a single Vanos (based on 97 Z3 2.8 aluminum block) as opposed to double Vanos. It has lived a decent life as a daily driver and autocross missile for the last 14,000 miles. It's a built internals motor with a lightened (40lbs) S52 crank, forged rods and pistons and an ATI damper. Rev limiter in 1st and 2nd is set to 7,800 and it has been there many many times. Blackstone Labs oil analysis says everything is in order over the last 4 oil tests.

[Greg S]

More info on GSR oil pump please?

http://store.bimmerworld.com/gsr-m52...aft-p2210.aspx

[5mall5nail5]

Will that not work with M50/S50/S52 pumps?

[NOTORIOUS VR]

Here's a much cheaper solution, I'm running their S52 pump... Love it and the price is right. Plus Achilles are great guys to deal with, very helpful.

M54 version: http://www.achillesmotorsports.com/A...-uopsk-m54.htm

M/S5x version: http://www.achillesmotorsports.com/A...1-am-uopsk.htm

[PEI330Ci]

Here's a much cheaper solution, I'm running their S52 pump... Love it and the price is right. Plus Achilles are great guys to deal with, very helpful.

M54 version: http://www.achillesmotorsports.com/A...-uopsk-m54.htm

M/S5x version: http://www.achillesmotorsports.com/A...1-am-uopsk.htm

Those look identical to the VAC parts that failed on my M54 engine, and many other M54 engines that I know of. In my case, I was running an ATI damper and the nut still loosened....

The GSR solution involves a LOT more machine work, so of course it will cost more.

[NOTORIOUS VR]

Those look identical to the VAC parts that failed on my M54 engine, and many other M54 engines that I know of. In my case, I was running an ATI damper and the nut still loosened....

The GSR solution involves a LOT more machine work, so of course it will cost more.

It's certainly NOT identical at all. You should probably revisit the site and have a look at how the shafts are keyed. It's a pretty obvious and large difference.

I can see how VAC's design isn't that great. the GSR setup is quite nice though I have to admit, it distributes the load over a much broader range with 3 pins and 4 bolts.

[PEI330Ci]

It's certainly NOT identical at all. You should probably revisit the site and have a look at how the shafts are keyed. It's a pretty obvious and large difference/

I missed the difference in the shape of the mating surface. OEM uses splines, the VAC piece uses a double lobe, the Achilles piece uses a square shape.

Since I have not tried the Achilles piece, it's fair to say that I have not experienced failure with it. (Nor have I heard of anyone with failures of that exact piece)

That said, I get a LOT of emails from people asking for advice on M54 engines, and I will continue to recommend the GSR piece because I like the design, and I know that it works. Everyone will have a personal preference, and this is mine.

Well....my preference is actually a 4 stage dry-sump...but that's not very practical on a street car. :P

[NOTORIOUS VR]

I missed the difference in the shape of the mating surface. OEM uses splines, the VAC piece uses a double lobe, the Achilles piece uses a square shape.

Since I have not tried the Achilles piece, it's fair to say that I have not experienced failure with it. (Nor have I heard of anyone with failures of that exact piece)

That said, I get a LOT of emails from people asking for advice on M54 engines, and I will continue to recommend the GSR piece because I like the design, and I know that it works. Everyone will have a personal preference, and this is mine.

Well....my preference is actually a 4 stage dry-sump...but that's not very practical on a street car. :P

Certainly the GSR version is very nice, practically eliminates the failure point completely would be my guess.

[wazzu70]

So can anyone link to actual failures of the M54 without there being an install error (overtorque) or overheating? They may be out there but I have never seen one and its certainly not a common mode of failure. Additionally studs load the joint in a much better way and any high performance engine should use these IMO.

For the oil pumps, the main reason they fail is torsionsl vibration from the crankshaft. It would have been incredibly nice if BMW had used a bedplate architecture for the engine block like they did with the next N5x generations, but what can you do. The aluminum blocks have reinforcments from the main caps back to the block, but the cylinders are not linked together.

There is actually a motorsport oil pump for the M54 engine the cars used in the Koni challenge class. Its quite expensive though as you can imagine so not a viable option for mere mortals. The GSR pump and ATI damper is the "affordable" ticket as Adam mentioned.

Until I actually see a head stud pull out from combustion pressure Ill still believe its caused by operator error.

[PEI330Ci]

So can anyone link to actual failures of the M54 without there being an install error (overtorque) or overheating?

I know of a couple of cases where the engine assembly was very carefully controlled, and the ECU (Motec) didn't have overheating or knock show up on the logs. As they repeatedly experienced failures, they became increasingly careful with both engine prep, and the tune.

I just shot one of the teams an email to follow up on where their at. Not sure how much I'll be able to share though...

- - - Updated - - -

There is actually a motorsport oil pump for the M54 engine the cars used in the Koni challenge class. Its quite expensive though as you can imagine so not a viable option for mere mortals. The GSR pump and ATI damper is the "affordable" ticket as Adam mentioned.

TMS sells the kit, they are $3k.

[wazzu70]

I would be very interested to hear what the team has to say. Even if they cannot share the details, just a confirmation there has been issues with studs pulling out due to something other than overheating or assembly error would be useful!

Yeah, the kit on Turner was what I was referring to but I didn't want people to crap their pants at the price tag!

Your build I think will help show the M54 platform is fully capable of being turbocharged for a street application. Thanks for documenting things so well and being so open about your approach!

[PEI330Ci]

I would be very interested to hear what the team has to say. Even if they cannot share the details, just a confirmation there has been issues with studs pulling out due to something other than overheating or assembly error would be useful!

It would appear that overheating was the culprit. The cooling system was not "normal", so there were other factors in how temperature was read. Once they figured out the cooling issue, the blocks lasted a LONG time in the racing environment. The team has since changed over to another powertrain due to political reasons, so that's the end of that data source.

For anyone that's scratching their head on the above, the issue was related to how the system was pressurized. If you change the position of the coolant reservoir, and the location and type of the coolant pump, you can add a false pressure to system that causes coolant over-flow to occur at lower engine temps. Once you lose pressure in the coolant system, the coolant flashes to vapor much sooner, and you lose the ability to transfer heat from super-hot surfaces. In this scenario, you can have coolant circulating, but it doesn't pick up heat, so the coolant temps won't increase drastically until the head itself overheats. By that time, the block is usually "done".

The trick in NASCAR is to run super high coolant base pressure in the system (They use an accumulator bladder to stabilize the pressure) to keep the coolant from boiling while "drafting". That...and coolant pumps that circulate 3 -4 times as much volume as an OEM system. (I'm told upwards of 200 GPM)

[autowa]

I would be very interested to hear what the team has to say. Even if they cannot share the details, just a confirmation there has been issues with studs pulling out due to something other than overheating or assembly error would be useful!

Yeah, the kit on Turner was what I was referring to but I didn't want people to crap their pants at the price tag!

Your build I think will help show the M54 platform is fully capable of being turbocharged for a street application. Thanks for documenting things so well and being so open about your approach!

Hit up scotcH (not butters) on the track subforum he has experience with track
N/A. M54 info

Sent from my SGH-I337M POS

[iRodD]

It would appear that overheating was the culprit. The cooling system was not "normal", so there were other factors in how temperature was read. Once they figured out the cooling issue, the blocks lasted a LONG time in the racing environment. The team has since changed over to another powertrain due to political reasons, so that's the end of that data source.

For anyone that's scratching their head on the above, the issue was related to how the system was pressurized. If you change the position of the coolant reservoir, and the location and type of the coolant pump, you can add a false pressure to system that causes coolant over-flow to occur at lower engine temps. Once you lose pressure in the coolant system, the coolant flashes to vapor much sooner, and you lose the ability to transfer heat from super-hot surfaces. In this scenario, you can have coolant circulating, but it doesn't pick up heat, so the coolant temps won't increase drastically until the head itself overheats. By that time, the block is usually "done".

The trick in NASCAR is to run super high coolant base pressure in the system (They use an accumulator bladder to stabilize the pressure) to keep the coolant from boiling while "drafting". That...and coolant pumps that circulate 3 -4 times as much volume as an OEM system. (I'm told upwards of 200 GPM)

Good info. Thank you

[wazzu70]

Hit up scotcH (not butters) on the track subforum he has experience with track
N/A. M54 info

Sent from my SGH-I337M POS

Good call. Also the Spec E46 race class has started and I know a number of people with cars in the class locally so I will probably start getting some more feedback once things get off the ground more.

I recommended requiring an ATI damper and an oil pump upgrade, but these were declined to be allowed due to cost reasons. We shall see if the cars hold up fine or if these components are deemed necessary.

http://spece46.com/

[PEI330Ci]

Hit up scotcH (not butters) on the track subforum he has experience with track
N/A. M54 info

Arek is one of the people I've kept in touch with...

Good call. Also the Spec E46 race class has started and I know a number of people with cars in the class locally so I will probably start getting some more feedback once things get off the ground more.

I recommended requiring an ATI damper and an oil pump upgrade, but these were declined to be allowed due to cost reasons. We shall see if the cars hold up fine or if these components are deemed necessary.

http://spece46.com/

I was consulted when they were drawing up the rules for this class; my opinion at the time was that the GSR pump was the only thing the M54 needed to be raced.

Hard Motorsports just experienced an engine failure recently racing SpecE46, that would be a good place to check what didn't work....

[PEI330Ci]

While writing the control modules for Meth control into the Motec M1 ECU, I included pressure sensor inputs at various points to help monitor system health, and also adapt PWM based on pressure. The most obvious scenario would be where both pre-turbo and manifold injection was being used, and the plumbing resulted in different pressures at each injection point. Because my control methods are based on mass flow vs pressure, it was only natural to include pressure sensors.

This brought up an interesting problem.

Most fluid pressure sensors are available in 75 PSI, 150 PSI, and 500 PSI ranges. Most pressure pumps for water/meth systems are designed to run at 160-180 PSI. This means you are forced to use a 500 PSI sensor, with the resultant loss of resolution.

I discussed the issue with Joel @ Racespec, and he came up with a custom sensor solution for my application:





So now we have pressure data.

The next problem is flow. Aquamist has stopped offering flow meters as free-standing devices....I guess I need to contact Howerton to see what he can do for me!

[5mall5nail5]

Thats cool to have custom sensors made but do you not think 10 - 12 bit (or more) ADC is not enough for 500 psi sensors? Its nice to be able to quickly replace a failed sensor with an off the shelf part. Cool none the less.

[PEI330Ci]

Thats cool to have custom sensors made but do you not think 10 - 12 bit (or more) ADC is not enough for 500 psi sensors? Its nice to be able to quickly replace a failed sensor with an off the shelf part. Cool none the less.

Valid points.

Who said I don't have a spare sensor? :P

[wazzu70]

Never get in the way of a man's data resolution! What speed are you sampling at?

[PEI330Ci]

Never get in the way of a man's data resolution! What speed are you sampling at?

Anything up to 1 kHz...although you can apply averaging filters to any of the sample rates, at whatever averaging period you want.

I should clarify that you can set the above up for each individual sensor....some ECUs only let you set the rate for batches, or all logging channels and there is no access to filters.

[wazzu70]

Dang, 1000hz is fast! I am monitoring at 100hz (10ms) right now for cylinder to cylinder accel for misfire diagnostics which is incredibly fast. Generally pressures/temps for intake and exhaust are logged at 20hz (50ms) which is also pretty moving

Sounds like a nice setup and being able to set different rates for different sensors is very nice. You can log most stuff at 1hz and the transient info at a faster rate. Also setting the filter for each channel is nice too.

I need to see what my standalone can log at now... Its been so long I dont remember!!

[PEI330Ci]

Dang, 1000hz is fast! I am monitoring at 100hz (10ms) right now for cylinder to cylinder accel for misfire diagnostics which is incredibly fast. Generally pressures/temps for intake and exhaust are logged at 20hz (50ms) which is also pretty moving

Sounds like a nice setup and being able to set different rates for different sensors is very nice. You can log most stuff at 1hz and the transient info at a faster rate. Also setting the filter for each channel is nice too.

I need to see what my standalone can log at now... Its been so long I dont remember!!

I think high sample rates are misleading in conversations, unless you have practical experience with how much "noise" is present in signals. Most people would flip out if they saw a 1 kHz trace of fuel pressure. It's all over the place....even on really good systems. The guys chasing really big power, and using individual cylinder pressure sensors, will be looking to make the pressure as seen by the injection nozzle as steady as possible. (Big power is all about consistency in mixture preparation and cylinder burn) So that data is useful. For the rest of us, the highest meaningful resolution is usually based on how fast the data from the lambda sensors is acquired. (In most cases, that's roughly at 20 hz) So then the questions is whether the sensor output is acquired at 20 times a second, or averaged to 20 samples a second. The latter being more useful, but most people don't have a choice.

In M1 Build, you can set the voltage input sample rate to a completely different value than the data refresh rate, and on top of that, you can set the rate at which calculations are made using that data. An example would be Fuel pressure that is acquired at 100 hz, read into the control functions at 20 hz, and a visual warning function is calculated at 1 hz. You also have the ability to write hysteresis functionality in the time or voltage domain.