MAP sensor circuit

[Fred]

To be honest, I agree with the VEMS people, but some others might not... it certainly is an important sensor, but in practice manifold "noise" might outweigh electrical noise. I'd personally like mine both inside too. Then it's just a matter of getting the hose out of the box and to the right place. The lag from air movement certainly isn't an issue if you hook it up right.

Fred.

[jharvey]

I'm trying to find vendors for the MPXA4100AC6U I seem to recall that was the MAP we want to use. I could only find it at mouser, and with a 6 week factory lead time. Although it also noted an order of 25 ships immediately, so I'm not sure, perhaps the 6 week is for qty 1. Who knows.

Do we want to use this MAP, and if so are there other vendors?

[Fred]

http://octopart.com/search?q=MPXA4100AC6U

Seems OK. This is the preferred unit for those wanting to run without boost. It is also the preferred unit for those wanting atmospheric correction. However, the 4250 will do an OK job for either of those cases too. The MPX4115A is available more readily but is no good for engine use, only atmospheric reading inside the case. The higher in boost reading capability you go the worse the resolution gets in the normal 0 - 100kPa range. Hence stick to the lowest you can for your setup. Mine will need to be 6300 or 4250 models for most stuff. I doubt I'll ever need a 6400 unit. I would prefer to use the 4100 for my atmospheric pressure sensor though. They all have the same footprint, so you can safely design around it without a worry.

4 bar (3 boost) : http://octopart.com/search?q=MPXH6400AC6U
3 bar (2 boost) : http://octopart.com/search?q=MPXH6300AC6T1
2.5 bar (1.5 boost) : http://octopart.com/search?q=MPX4250A
1 bar (atmo only) : http://octopart.com/search?q=MPX4115A (no hose barb)
1 bar (no boost) : http://octopart.com/search?q=MPXA4100AC6U

Fred.

[GrowlingandBiffo]

I'm trying to find vendors for the MPXA4100AC6U I seem to recall that was the MAP we want to use. I could only find it at mouser, and with a 6 week factory lead time. Although it also noted an order of 25 ships immediately, so I'm not sure, perhaps the 6 week is for qty 1. Who knows.

Do we want to use this MAP, and if so are there other vendors?

http://search.digikey.com/scripts/DkSea ... 100AC6U-ND

[Delta]

http://octopart.com/search?q=MPXA4100AC6U

The higher in boost reading capability you go the worse the resolution gets in the normal 0 - 100kPa range. Hence stick to the lowest you can for your setup.
4 bar (3 boost) : http://octopart.com/search?q=MPXH6400AC6U
3 bar (2 boost) : http://octopart.com/search?q=MPXH6300AC6T1
2.5 bar (1.5 boost) : http://octopart.com/search?q=MPX4250A
1 bar (atmo only) : http://octopart.com/search?q=MPX4115A (no hose barb)
1 bar (no boost) : http://octopart.com/search?q=MPXA4100AC6U

Fred.

Not necessarily. Well in absolute terms yes always, but if your A/D converter is of high enough resolution then no problems at all with any of the higher BAR sensors. A good example is the EMS8860 - it comes with a 5 BAR (4 boost) sensor on board and is perfectly at home on an NA car - but the A/D converter for the MAP sensor is 16bit. TBH I'm wary of any product that has less then 12bit A/D converters for this sensor as it really does hurt the tractability of even moderate boost cars. 16bits means it doesn't matter what sensor you use, 12 bits is kind of the limit for 3bar boost in a noisy environment and 10bit is really only ok if your NA only.

Now - people are probably going to get shitty at me for saying that, but even good A/D converters generally have 1LSB error rate without ANY noise. So for 10bit that equates to 9 bits of useful info and hence 512 points over which to spread 100kpa - which is fine. 512 points over 200kpa is juuust ok. 512 over 300kpa is definitely not IMHO - especially once you add noise. 2048 points (12bit 1LSB error) covers most sensors damn well unless noise is a big problem. 16bits means you can exclude the last 4 LSB's and STILL cover 5 BAR easily.

[jharvey]

Appears that was my bad, I didn't catch the MPXA4100 vs the MPX4100.

[Fred]

Now - people are probably going to get [shizzle] at me for saying that

Why get angry when I can just set you straight? :-)

A good example is the EMS8860 - it comes with a 5 BAR (4 boost) sensor on board and is perfectly at home on an NA car

This is still not a good idea regardless. A large part of the noise is not relative to the scale of the sensor and therefore has WAY more impact if the scale is setup that way instead of minimally to the maximum pressure required.

TBH I'm wary of any product that has less then 12bit A/D converters for this sensor as it really does hurt the tractability of even moderate boost cars.

Where is the empirical evidence for this statement? There are zillions of ms1 cars out there with 8 bit sensors for MAP running just fine on high boost. "tractability" is not lost in the slightest. If you meant something other than what that word means in your statement, please elaborate.

16bits means it doesn't matter what sensor you use

As explained above, that is not a true statement.

12 bits is kind of the limit for 3bar boost in a noisy environment and 10bit is really only ok if your NA only.

What about all those people with MS2 and 30psi that idle around and have great bottom end and top end and and and? This is not a realistic statement either. In fact, experience shows us the opposite. The lower bit rate sensors MASK the noise and allow a smoother running engine because of less random changes in output PW. Yes, you could mask off the noise in software by ditching the LSB stuff but then what exactly do you have again? The same low bit count ADC you started with.

but even good A/D converters generally have 1LSB error rate without ANY noise.

Provided the error is consistent and repeatable it does not matter AT ALL :-) I've watched the ADC counts and can occasionally get them to read a constant value even with a noisy power supply setup and pot contacts that just won't sit still :-)

512 points over 200kpa is juuust ok. 512 over 300kpa is definitely not IMHO - especially once you add noise.

My experience and that of many others disagrees with this. Show me how much fueling difference a 0.8kPa change makes % wise? Not significant. Put the other bit back in and it becomes 0.4kPa which is off the scale of "I care".

When you stop and think that lots of systems don't interpolate between cells and ran as small as 8x8 tables and 8 bit ADCs and ran well enough you have to concede that it is not a practical problem and that you are being idealist (a good trait mostly) on this.

I generally agree that more bits = better, but I disagree that it is required. It just isn't real-world important.

Fred.

[Delta]

Most places around the world you need to pass an emissions test if you fit aftermarket management - while its possible to run a 2-3bar map sensor with 8 bit - I absolutely guarantee its not making the best power it could, and I'm also sure it wouldn't pass a modern emissions test.

The more bits the better - and also the more bits the better the resolution so you CAN run a larger range map sensor without problems. You can run a boosted car with Alpha N if you really want to - will it run well - no. Will it make good power - probably. Will it run ok at low throttle openings - yep probably. When you stand on the throttle - will it have good response and be able to pull in top gear from low revs - absolutely not.

The same can be said of cars that have small tables and 8 bit ADC's. They make the power, they run _ok_ but they do NOT make the max power they could everywhere, and quite often they have weird on off throttle characteristics that make them a pain on the track and funnily enough in car parks (of all places).

But fair enough - if you guys think its ok thats what we run with. As usual I'm in the "not quite normal" catagory hahha, I know lots of people with stupid sized cams and big turbos (me being one) and without good resolution (I have 3-4" vac at idle) my car is undrivable.

[Fred]

I absolutely guarantee its not making the best power it could, and I'm also sure it wouldn't pass a modern emissions test.

I can't agree with that either. Although I know why you say that, there are ways of getting around it. Basically you need to move your idle such that it sits in only one MAP location always and the minor variations in pressure don't cross a border into another MAP location. Then you tune as normal to get the desired AFR and therefore emissions. It can be done, it just takes more skill and patience and talent.

The more bits the better - and also the more bits the better the resolution so you CAN run a larger range map sensor without problems.

I agree that more = merrier :-)

You can run a boosted car with Alpha N if you really want to - will it run well - no. Will it make good power - probably. Will it run ok at low throttle openings - yep probably. When you stand on the throttle - will it have good response and be able to pull in top gear from low revs - absolutely not.

Agreed, and there are so many good reasons to not do that. We are discussing having less resolution in MAP, not no map at all :-)

The same can be said of cars that have small tables and 8 bit ADC's. They make the power, they run _ok_ but they do NOT make the max power they could everywhere, and quite often they have weird on off throttle characteristics that make them a pain on the track and funnily enough in car parks (of all places).

True enough, I wasn't advocating their use, just that if they can run and our readings are 4 times better and our tables have 4x as many cells then it's probably not going to affect too many people.

But fair enough - if you guys think its ok thats what we run with.

There is an option to use an external ADC, but that's more expensive and adds a lot of complexity that we really don't need up front. For the fussy amongst us there could be a patch/hack/mode for an external 16 bit ADC for this purpose in the future, but right now, the last thing we need are more headaches and complexity.

As usual I'm in the "not quite normal" catagory hahha, I know lots of people with stupid sized cams and big turbos (me being one) and without good resolution (I have 3-4" vac at idle) my car is undrivable.

I bet it could be made very drivable and reasonable with a much more coarse system than what you have though. I agree that the resolution in that range needs to be A1 if that is all you have between idle and WOT.

Fred.

[shameem]

Just a thing i discovered recently - I was poking around with a friend's ms2 setup to see why the sensor data was noisy. After trying various things - i finally decided to dig into the ms2 schematics and found out that the ADC reference is tied to the voltage regulator output with a LC filter. So i rigged up another regulator just for the Vref and threw some mighty caps and the spikes went away. So the moral of the story is - if you have noisy ADC reading - it may be due to transients in Vref.