SpecE30 and Megasquirt


#81

The same guy that at CMP in Apr, pulled my DME, opened it and looked at my chip. The difference is that one inspection is rigorous and the other isn’t.[/quote]

And what’s to stop me from flashing the stock tune and then re-flashing right after inspection with the cheater tune? its even easier than swapping a chip. if this is solely for diags, it seems a bit overkill. esp since not all of us are even interested in looking at MS log outputs. swapping parts is perfectly acceptable to me and if that fails i will take the car to someone who can diagnose it better. I know my limitations both in skill set and time. I’m not trying to be a BMW Master Tech at this point in my life. Those who want diags could install a passive logger and be done with it, the rest of us will just keep things the way they are now.[/quote]

This was a standard post race inspection when they looked at the chips of the top 3 and one pulled at random. I was the random.

No one is going to be interested in logs until they’re tearing their hair out over an intermittant problem that seems to be defying a solution. I don’t care who you take the car to, if the mechanic can’t replicate the problem he’s not going to figure out the cause. And when the problem only occurs in 4th geat >5krpm, the mechanic is going to be hard pressed to replicate that unless he takes the car to a dyno.

BMW Master tech’s aren’t some kind of miracle worker. When he can’t replicate the problem and everything tests fine in the shop, what does he do? How much would it cost for the BMW Master tech to spend the day at a dyno with a pile of spare parts…thousands of dollars in his time, dyno time, and parts? You guys that have not gone thru this seriously underestimate just how difficult this kind of situation can be.

The passive logger idea sounds interesting, but no one has yet presented a finished product that can do the job. And the possibility has to be considered that the passive logger could create it’s own gremlins by affecting the DME’s inputs. Few ways of gathering info are truly “passive”.


#82

Ranger -

You are proposing back surgery for a guy who has trouble because his shoes are worn out!!!

I honestly do appreciate what it is like to chase an intermittent problem. I honestly CAN’T appreciate swapping out several used parts trying to chase a problem. Start with new parts. Their failure rate will be much lower than the junkyard parts bin.

It’s OK to swap used parts if that’s the way you roll, but it isn’t reasonable to suggest imposing a lot of cost on all of us.

Data acquisition is open under the current rules. Log data all you want. While you are at it, figure out a plug-and-play way to REALLY see if the OEM DMEs have been modified.

The number of people suffering from intermittent problems that can be solved with data logging is miniscule.

Sorry if my jab in the other thread (where you say you don’t want to replace plug wires that shock you because they are only a year old) was not taken in the mostly-lighthearted fashion it was given. But damn, if you aren’t willing to replace known failing parts…

If you assume everyone is having intermittent problems and everyone wants to tinker, the MS idea has merit.


#83

[quote=“Steve D” post=57949]Ranger -

You are proposing back surgery for a guy who has trouble because his shoes are worn out!!!

I honestly do appreciate what it is like to chase an intermittent problem. I honestly CAN’T appreciate swapping out several used parts trying to chase a problem. Start with new parts. Their failure rate will be much lower than the junkyard parts bin.

It’s OK to swap used parts if that’s the way you roll, but it isn’t reasonable to suggest imposing a lot of cost on all of us.

Data acquisition is open under the current rules. Log data all you want. While you are at it, figure out a plug-and-play way to REALLY see if the OEM DMEs have been modified.

The number of people suffering from intermittent problems that can be solved with data logging is miniscule.

Sorry if my jab in the other thread (where you say you don’t want to replace plug wires that shock you because they are only a year old) was not taken in the mostly-lighthearted fashion it was given. But damn, if you aren’t willing to replace known failing parts…

If you assume everyone is having intermittent problems and everyone wants to tinker, the MS idea has merit.[/quote]
This isn’t about tinkering. Nothing in my scheme would require or encourage tinkering.

I bet that anyone that spends a decade in the series will at some time be bedeviled by symptoms such as Jim and I have had.

Almost all parts can be tested. Every used part I have is “known good”. The problem is that sometimes “tested” and “known good” turn out to be inexplicably not so good. Like a known good DME and AFM that seem to hate each other for no reason that makes sense. Or a part that tests good in the garage but fails when it gets warm or experiences vibration. Or the part that tests “within spec” yet because the spec is so loose the part is responsible for engine management that is far from optimum.

It’s all to easy too end up with race car full of parts that have tested good over and over again, yet has intermittant engine management problems. This shit happens and will make a person crazy.

Re. aren’t willing to replace known failing parts. Jim’s a smart guy and he knows a lot, but just because he says that my plug wires are bad because one of them shocked me, doesn’t make it so. When pliers squeeze on the rubber boot to pull a plug wire, the pliers reduce the thickness of the rubber insulation. At some point the rubber insulation under the plier jaws will become thin enough that the >20kV can arc thru. Maybe the wires are bad and maybe they aren’t. But getting shocked thru pliers isn’t a conclusive test.

In any event, I’m weary of this whole discussion. If someone wants to start a thread on passive monitoring, I’d read that with interest. I don’t like to fight, and that’s what this seems to be turning into. I withdraw the idea.


#84

[quote=“vmwerks” post=57902][quote=“dgorman” post=57900]My vote:

Drop another 50lbs (free)
Add more camber to front (free)
Allow for further suspension/setup changes
Allow for lightweight flywheel

Skip motor/ECU mods completely.[/quote]

Makes sense to me - IIRC Ranger was quoted as saying that dropping 100lbs from the car cannot be felt… he can keep the weight in if he wants to… :slight_smile:
[/quote]
Find the quote or apologize for the cheap shot.


#85

Ranger -

There are some techniques you can use with passive monitoring to limit current draw (to not affect the system as a whole). You can then use maths (either in-line or post-process) to derive other metrics if needed. In some cases it’s not cheap, in some cases it is (roll your own approach).


#86

[quote=“dgorman” post=57953]Ranger -

There are some techniques you can use with passive monitoring to limit current draw (to not affect the system as a whole). You can then use maths (either in-line or post-process) to derive other metrics if needed. In some cases it’s not cheap, in some cases it is (roll your own approach).[/quote]
I hear you but that’s too abstract to be useful. Either someone can point us toward a plugnplay (or nearly so) passive logging system with extremely high input impedances for a reasonable price, or they can’t.


#87

Stop pulling plug wire boots with pliers and they likely will not fail on you! I pull mine by hand…with the engine running…crazy, I know…

Mark


#88

Ranger,
AIM MXL dash provides 1MOhm input impedances for example :slight_smile:


#89

[quote=“dgorman” post=57957]Ranger,
AIM MXL dash provides 1MOhm input impedances for example :)[/quote]
What is the max sample rate?


#90

[quote=“jlevie” post=57961][quote=“dgorman” post=57957]Ranger,
AIM MXL dash provides 1MOhm input impedances for example :)[/quote]
What is the max sample rate?[/quote]

Evo4: Total max sampling frequency 5000 (5KHz) samples per second.
Pro05: Total max sampling frequency 4000 (4KHz) samples per second.
Pista: Total max sampling frequency 2000 (2KHZ) samples per second.

Solo DL: Not sure yet but much less expensive than the above.

Carve it up how you like. :slight_smile:


#91

[quote=“dgorman” post=57968][quote=“jlevie” post=57961][quote=“dgorman” post=57957]Ranger,
AIM MXL dash provides 1MOhm input impedances for example :)[/quote]
What is the max sample rate?[/quote]

Evo4: Total max sampling frequency 5000 (5KHz) samples per second.
Pro05: Total max sampling frequency 4000 (4KHz) samples per second.
Pista: Total max sampling frequency 2000 (2KHZ) samples per second.

Solo DL: Not sure yet but much less expensive than the above.[/quote]
That would work for everything but the CPS signal, and it is critical for diagnosing a miss or cutout. You need at least 10 times higher rate for the CPS to tell what if it is valid.


#92

[quote=“jlevie” post=57969]You need at least 10 times higher rate for the CPS to tell what if it is valid.[/quote]Interesting. Can you dumb it down for me? The crank rotates at, say, 100 revs per second. 2000 Hz sampling would give a sample every 18 degrees. Every 20th sample there should be a “hit” from the CPS, right?

Arguably if CPS is the only thing the AIM dash won’t diagnose, if there are no other issues in the logs there’s your problem?

Does the DME throw a code if the CPS is acting funny?


#93

[quote=“jlevie” post=57969][quote=“dgorman” post=57968][quote=“jlevie” post=57961][quote=“dgorman” post=57957]Ranger,
AIM MXL dash provides 1MOhm input impedances for example :)[/quote]
What is the max sample rate?[/quote]

Evo4: Total max sampling frequency 5000 (5KHz) samples per second.
Pro05: Total max sampling frequency 4000 (4KHz) samples per second.
Pista: Total max sampling frequency 2000 (2KHZ) samples per second.

Solo DL: Not sure yet but much less expensive than the above.[/quote]
That would work for everything but the CPS signal, and it is critical for diagnosing a miss or cutout. You need at least 10 times higher rate for the CPS to tell what if it is valid.[/quote]

Jim,

I only see MS datalog sample rate of 333Hz. Do you know if it’s higher?

Having lived with a programmable ECU for 7 years, I would say this. A built-in datalogging capability rarely helps you in chasing down an intermittent problem. To be useful, an independent separate datalogging capability is needed.


#94

[quote=“Steve D” post=57971][quote=“jlevie” post=57969]You need at least 10 times higher rate for the CPS to tell what if it is valid.[/quote]Interesting. Can you dumb it down for me? The crank rotates at, say, 100 revs per second. 2000 Hz sampling would give a sample every 18 degrees. Every 20th sample there should be a “hit” from the CPS, right?

Arguably if CPS is the only thing the AIM dash won’t diagnose, if there are no other issues in the logs there’s your problem?

Does the DME throw a code if the CPS is acting funny?[/quote]

There’s some Analog to Digital rule of thumb re. required oversampling rates in order to be able to reproduce the orig signal, but I don’t know what it is off of the top of my head. But imagine a sine wave of one full cycle. In order to reproduce the right shape, how many times would you have to take a sample of the single full cycle in order to have enough information that you could reproduce the single full cycle? In 90deg the signal goes from 0 amplitude to full amplitude. In order to accurately spot 0 amplitude and full amplitude you’d probably want to sample it no less than every 10deg and then interpolate. Every 10deg of 360 is 36 samples.

How many teeth are there in the harmonic balancer? Lets say 60 to make the math easy. I’ve no idea what the real # is.

(100revs/sec)(60teeth/rev)(36samples/tooth)= a sampling rate of 216Khz. That’s not do-able. That’s a sampling rate on the order of HAM radio.

I pulled that all out of my butt so don’t be shocked if it’s way off.


#95

Having said all that, my theory is that the CPS isn’t so much a “crank position” sensor but a “top dead center” sensor. I have no proof of this, but I think that the critical piece is the gap in the teeth at TDC not the teeth themselves. Therefore, so my theory goes, the DME is looking at a rapid sine wave and then a gap in the sine wave. That would mean that a noisy sine wave wouldn’t matter, as long as the gap in the sine wave was clear.

My reasoning is this…the DME doesn’t need a sensor to tell it the position of the crank. As long as the harmonic balancer is spinning the DME can interpolate the crank’s position from the timing of the gap in the teeth. That would be easier from a design perspective then trying to keep track of the exact shape of the sine wave from the rapidly moving teeth.

Stating the obvious, the gap is there for a reason and it’s not to make it easier to get to a couple bolts.

I got this idea when I stumbled across a (non-M20B25) BMW reference to a TDC sensor that seemed really similar to our CPS.


#96

[quote=“jlevie” post=57969][quote=“dgorman” post=57968][quote=“jlevie” post=57961][quote=“dgorman” post=57957]Ranger,
AIM MXL dash provides 1MOhm input impedances for example :)[/quote]
What is the max sample rate?[/quote]

Evo4: Total max sampling frequency 5000 (5KHz) samples per second.
Pro05: Total max sampling frequency 4000 (4KHz) samples per second.
Pista: Total max sampling frequency 2000 (2KHZ) samples per second.

Solo DL: Not sure yet but much less expensive than the above.[/quote]
That would work for everything but the CPS signal, and it is critical for diagnosing a miss or cutout. You need at least 10 times higher rate for the CPS to tell what if it is valid.[/quote]

I haven’t done the math on the CPS (looks like Ranger was taking a stab at it), however, then you could integrate your own solution (we have a board that we programmed) and there are many out there. There is some IP here so I’m being generic on purpose. While it’s not a “plug-and-play” unit and requires people to know basic coding skills and basic EE, it is a solution. We can get up to 30KHz by adjusting PWM logging. Output an analog signal back to the dash.


#97

[quote=“Ranger” post=57973][quote=“Steve D” post=57971][quote=“jlevie” post=57969]You need at least 10 times higher rate for the CPS to tell what if it is valid.[/quote]Interesting. Can you dumb it down for me? The crank rotates at, say, 100 revs per second. 2000 Hz sampling would give a sample every 18 degrees. Every 20th sample there should be a “hit” from the CPS, right?

Arguably if CPS is the only thing the AIM dash won’t diagnose, if there are no other issues in the logs there’s your problem?

Does the DME throw a code if the CPS is acting funny?[/quote]

There’s some Analog to Digital rule of thumb re. required oversampling rates in order to be able to reproduce the orig signal, but I don’t know what it is off of the top of my head. But imagine a sine wave of one full cycle. In order to reproduce the right shape, how many times would you have to take a sample of the single full cycle in order to have enough information that you could reproduce the single full cycle? In 90deg the signal goes from 0 amplitude to full amplitude. In order to accurately spot 0 amplitude and full amplitude you’d probably want to sample it no less than every 10deg and then interpolate. Every 10deg of 360 is 36 samples.

How many teeth are there in the harmonic balancer? Lets say 60 to make the math easy. I’ve no idea what the real # is.

(100revs/sec)(60teeth/rev)(36samples/tooth)= a sampling rate of 216Khz. That’s not do-able. That’s a sampling rate on the order of HAM radio.

I pulled that all out of my butt so don’t be shocked if it’s way off.[/quote]


#98

At redline the engine is turning at 6500rpm. The tooth spacing on the harmonic balancer is 60 teeth/rev (good guess Scott). That means 6500cps. The sensor is a magnetic reluctance type which generates something very close to a sign wave, so to see the signal with enough resolution to tell if it is valid you need about 10 samples per cycle, or a sampling rate of 65khz.

But when using a MegaSquirt it suffices, for trouble shooting, to simply look at the rpm log data. If it is dropping out the cause will be loss of timing data from the CPS. With a stock DME you can’t be sure that a drop in the tach output really means a loss of timing data. The DME could be locking up, loosing power, or the tach output circuit could be flaky. At one point or another I’ve seen all three cases.

Steve earlier opined that the sort of problem that a MegaSquirt could help solve was minuscule (sic). I beg to differ. At the last event 2 cars out of 20 had problems that could have been instantly diagnosed with an MS. A good educated guess did fix them (this time), but that is still 10% of the field present at RA. Memory is a tricky thing, but my recollection is that I wind up working on one or more cars at most events trying to solve some sort of performance problem. I’d hardly cause that minuscule.


#99

[quote=“Ranger” post=57975]Having said all that, my theory is that the CPS isn’t so much a “crank position” sensor but a “top dead center” sensor. I have no proof of this, but I think that the critical piece is the gap in the teeth at TDC not the teeth themselves. Therefore, so my theory goes, the DME is looking at a rapid sine wave and then a gap in the sine wave. That would mean that a noisy sine wave wouldn’t matter, as long as the gap in the sine wave was clear.

My reasoning is this…the DME doesn’t need a sensor to tell it the position of the crank. As long as the harmonic balancer is spinning the DME can interpolate the crank’s position from the timing of the gap in the teeth. That would be easier from a design perspective then trying to keep track of the exact shape of the sine wave from the rapidly moving teeth.

Stating the obvious, the gap is there for a reason and it’s not to make it easier to get to a couple bolts.

I got this idea when I stumbled across a (non-M20B25) BMW reference to a TDC sensor that seemed really similar to our CPS.[/quote]

The gap is there for cylinder identification, all right, but there are a couple types of noise that can give the ECU problems. A piece of noise right where the missing tooth goes can make it lose track of the missing tooth and lose its cylinder identification, while noise in the sine wave between pulses could look like an extra tooth and get your timing off slightly if the ECU fails to filter out pulses. Here’s an example of someone capturing the first type of noise issue at around 4,000 RPM with a MegaSquirt set to log the CPS (and no other signals - many MegaSquirt code variants have special tools for troubleshooting CPS issues, beyond just watching the RPM signal).

This is for a Ford, not a BMW, but the idea is the same. You’ve got 36 teeth and one missing, instead of 60 teeth with 2 missing. What you’re looking at is a bar graph where the height of the bar represents the time between teeth, and there is one missing tooth so you get a double-height bar when you see a missing tooth. However, on one revolution, a noise pulse gets through the filter and makes the missing tooth disappear. (He fixed this with a change to hardware filtering, in case you were wondering.)


#100

I would be all for it if I had the money, but if we use the ms we need to use it to it’s full potential and not just for data logging. Throw out the flapper box, install a real throttle position sensor, and a wideband O2 sensor. MS has a autotune program that fine tunes the maps once you enter in ballpark running numbers. Autotune has to work on a running laptop while you drive and I think you need your laptop plugged in to log data also. You don’t need to worry about people cheating if you let everyone use whatever tune they want.

If everyone is worried about people chipping we should just make it legal. They cost 20$ on ebay. I have one sitting in a box in the garage from my autox days.