Regarding calculation of airmass,

Would it be better to take the temperature from the stock sensor location, or to take it at the intake tube? I feel like I could argue it either way.

I have a couple extra sensors. I'm thinking about installing one in the tube and logging it, if for no other reason than to get an idea of the delta-T between the two locations. Since the stock one would be retained I wouldn't have to change anything within the tune, I just need the data for some external calcualtions.

Discuss.

 

I like it the nearest possible of the head because it give a more accurate reading of the ''real air temp'' that enter the cylinder.

In fact at WOT the reading would probably be really similar in the 2 locations but I never dataloged it.

 

the tube would be a better location, the intake manifold holds a lot of heat.

 

On the manifold? Heatsoak. Not good.

 

There are compensation tables in Honda's PGM-FI systems for engines that had the IAT placed in the manifolds. The tuning engineers had to make multiple tables to compensate for heat soak. So if you go and move it to the cold side without changing the compensation tables..... your probably not going to gain much.

If you have a fully tunable EMS then running the IAT on the cold side would show a more realistic temperature reading. It has been shown that carefully removing the cover off the IAT and revealing the thermistor will help response time as well.

Here is a stock F/H series IAT with the hard material removed from the thermistor.



The thermistor, as seen here on the tip, is encased in a clear epoxy. Probably best to leave that as is!

Doing this does not change how the temperature is read. It just gives the OEM IAT a faster response time.

OEM = 11.5s
Modified OEM = 4.4s

 

nice tip ghost, will definitely keep this in mind!

 

Sorry, I haven't had much time to type a thoughtful response on my post. There is some good info on here already though, and this was what I was looking for. A good discussion on the why's and how's of the IAT sensor.

I'm more concerned with a general theory of how the ecu calculates airmass, and how I can calculate it externally for building a good VE model. I feel that I can see it both ways. What we (and the ecu ultimately) are concerned with is calculating the mass of air that the engine is consuming at any given point in time. The temperature of the air is needed to determine the density in order to convert the volume to mass. My question is whether the temp before or after the throttle plate gives the best picture of what is getting in to the engine.

On one hand, it seems that the temp before the manifold has had a chance to do work on the air would be more accurate. On the other, it seems that the temp inside the manifold paints a better picture of what is going into the cylinders. Calculating the airmass before the throttle plate could skew the results, since theoretically not all of that air gets in. But also, the temperature doesn't begin to change (mostly) until the air is in the manifold.

Either way, the ecu has been calibrated from the factory for whichever arrangement the motor came with, so saying one is "better" than the other on a stock setup is irrelevant. If the hot air is not a better representation, there is a transfer function somewhere that correlates it to the actual temperature, and it has been used to compensate.

As far as the stock IAT sensors and heat soak... I have a couple of theories on that as well. It looks like there are two layers of plastic around the thermistor. One around the sensing tip, and a second ring that contacts the manifold itself (albeit loosely). There is most likely an air gap between the two, so there are four resistances before the manifold can start heating up the thermistor. It's also protruding into the airstream enough that the effects of heat soak should be reduced. I can't confirm the air gap though, since I can't inspect for it without destroying the sensor.

As far as the response time goes, it's totally possible that the slow response is on purpose. A slow response will keep the temp signal from changing too fast, in turn keeping the temperature curve more level and giving the ecu a better idea of the average intake temperature. A temp signal that is changing constantly and quickly would tend to cause the iat compensation tables to jump all over the place, feasibly causing the ecu to never be able to settle on an AFR. We are, after all, dealing mostly with 20-25 year old ecu technology with our typical projects around here.

As far as this:

While those numbers are interesting, they are meaningless to describe the performance of the iat sensor. Response time has to be correlated to the change in temperature in order to be normalized to any temperature change. The time constant, tau, is derived from the time that it takes the sensor to respond to a step temperature change. Industry standard is 63%. So while 4.4s would be pretty slow for a delta-T of 2*F, it would be pretty quick for a delta-T of 500*F.

I might hook up the DAQ this weekend and see if I can come up with a tau value for a couple of the sensors I have hiding around. You've piqued my curiosity.

I appreciate everyone's input.