IAT sensor heat soak
#1
IAT sensor heat soak
I've got an NB2 N/A with an intake similar to https://www.cxracing.com/air-intake-...-MIATA9905-KIT
I have an additional heatshield between the front engine lift bolt and the back of the strut tower to block off heat from the header.
As I'm driving, my MAT shows ambient. When I come to a stop, MAT stays around ambient. But as soon as my FM fans kick on, my MAT spikes to 100F+. I built a small shield that goes down the side of the radiator to block flow rushing around the side of the rad into the engine bay, and the front of my airbox is pretty close to sitting up against the inboard side of the headlight.
I'm not sure where I else I would want to place my IAT sensor. Anywhere in the piping would put the sensor right in the way of the rad fans. My only other thought would be to somehow attach it into the 90* silicone elbow on the TB.
I have an additional heatshield between the front engine lift bolt and the back of the strut tower to block off heat from the header.
As I'm driving, my MAT shows ambient. When I come to a stop, MAT stays around ambient. But as soon as my FM fans kick on, my MAT spikes to 100F+. I built a small shield that goes down the side of the radiator to block flow rushing around the side of the rad into the engine bay, and the front of my airbox is pretty close to sitting up against the inboard side of the headlight.
I'm not sure where I else I would want to place my IAT sensor. Anywhere in the piping would put the sensor right in the way of the rad fans. My only other thought would be to somehow attach it into the 90* silicone elbow on the TB.
#2
You could run a turbo VW IAT sensor, weld a bung onto your intake. The thing is all plastic and the sensor element is a whole lot better than the GM crap. Get it out in the middle where the flow is, not the stale air on the wall of the pipe.
You can get a weld bung here...
https://www.ebay.com/itm/Intake-Air-...YAAOSw6rhdC6Jx
You can get a weld bung here...
https://www.ebay.com/itm/Intake-Air-...YAAOSw6rhdC6Jx
#3
You could run a turbo VW IAT sensor, weld a bung onto your intake. The thing is all plastic and the sensor element is a whole lot better than the GM crap. Get it out in the middle where the flow is, not the stale air on the wall of the pipe.
You can get a weld bung here...
https://www.ebay.com/itm/Intake-Air-...YAAOSw6rhdC6Jx
You can get a weld bung here...
https://www.ebay.com/itm/Intake-Air-...YAAOSw6rhdC6Jx
#6
I did some investigation yesterday to see if replacing the sensor itself would fix the issue.
The sensor is located behind a heatshield and is threaded into a plastic body that is attached to the metal heatshield. The sensor is also on the "cold" side of the filter.
I'd say the sensor is in the best possible spot within the engine bay to not get heat soaked.
I'm in agreement that the brass sensor body gets hot. I let the car idle till the fans kicked on. Let the fans cycle twice.
By this time my MAT went from ambient (60F) to 90F to 120F after each fan cycle. Oil temperature was at about 160F. (I don't really see higher than that
I checked the body of the sensor after the sensor itself was reading 120F, and the body was reading close to 140F using my IR thermometer.
Touching the body of the sensor, it felt warm, but not 140F warm. This was bizzare to me because a shiny brass surface would make the IR thermometer read lower than true temperature.
TS was reading 120F, the IR thermometer pointed at the sensor read 120F.
Now to check the sensor reading being influenced by the body - I blew on the sensor tip. This caused a decent drop in temperature. As soon as I stopped blowing, the sensor got warm again. The sensor was also in ambient air at this point. I was holding the sensor by its wires.
This is how I can conclude that the brass sensor body heats up the thermister.
So, obviously my sensor needs to be moved into a spot with more air movement?
I changed my sensor position into my middle airpipe (dead center in front of the valve cover). and repeated the test with letting the fans cycle. This time the heat soak was way way worse. The sensor was screwed in far enough where the sensor tip was within 5mm of the middle of the pipe.
This would in theory allow the sensor tip to always be in moving air, and thus cool it down.
So, this begs the question: Is my intake air really getting that hot?
I don't think it is because my AFRs were going way way lean. Idling happily at 14:7 to swinging to 16:1/17:1 Assuming that the MAT Air Density table in the tune is correct, this would mean that my engine is sucking in colder air than the sensor's reporting.
Based off my MAT Air density table, the values for 59F is 101.7% and 120F is 90%. This would meant that the air coming in is 12% less dense, and therefore (if the equation for VE is linear) our AFR would be 12% higher. From 14:7 to 16:5
Another way I tested if my intake stream was getting heated by the fans, was moving the sensor into the bumper. When the fans turned on, the sensor's reported temperature barely went above ambient. However the mixture did go rich by about .5 points.
This makes me think that my real air temperature that the engine is seeing is being heated, ever so slightly.
I'd say the KISS (keep it simple, stupid) is to just run the sensor in the bumper. That way I avoid all heatsoak, and I have a smaller change in idle AFR. I just miss out on true engine air temperature, and could miss out on safety functions like pulling timing when MAT is too hot.
There's no saying that the temperature difference now between bumper and engine will be the same when ambients raise to 90F.
I looked into using the CLT/MAT blended vs flow correction table. I'm not super sure how it works. There's multiple posts on the Megasquirt forums that I need to read to understand it.
I will be switched to a plastic body sensor. I'm just not sold on the fact that a plastic one won't heat soak either.
The sensor is located behind a heatshield and is threaded into a plastic body that is attached to the metal heatshield. The sensor is also on the "cold" side of the filter.
I'd say the sensor is in the best possible spot within the engine bay to not get heat soaked.
I'm in agreement that the brass sensor body gets hot. I let the car idle till the fans kicked on. Let the fans cycle twice.
By this time my MAT went from ambient (60F) to 90F to 120F after each fan cycle. Oil temperature was at about 160F. (I don't really see higher than that
I checked the body of the sensor after the sensor itself was reading 120F, and the body was reading close to 140F using my IR thermometer.
Touching the body of the sensor, it felt warm, but not 140F warm. This was bizzare to me because a shiny brass surface would make the IR thermometer read lower than true temperature.
TS was reading 120F, the IR thermometer pointed at the sensor read 120F.
Now to check the sensor reading being influenced by the body - I blew on the sensor tip. This caused a decent drop in temperature. As soon as I stopped blowing, the sensor got warm again. The sensor was also in ambient air at this point. I was holding the sensor by its wires.
This is how I can conclude that the brass sensor body heats up the thermister.
So, obviously my sensor needs to be moved into a spot with more air movement?
I changed my sensor position into my middle airpipe (dead center in front of the valve cover). and repeated the test with letting the fans cycle. This time the heat soak was way way worse. The sensor was screwed in far enough where the sensor tip was within 5mm of the middle of the pipe.
This would in theory allow the sensor tip to always be in moving air, and thus cool it down.
So, this begs the question: Is my intake air really getting that hot?
I don't think it is because my AFRs were going way way lean. Idling happily at 14:7 to swinging to 16:1/17:1 Assuming that the MAT Air Density table in the tune is correct, this would mean that my engine is sucking in colder air than the sensor's reporting.
Based off my MAT Air density table, the values for 59F is 101.7% and 120F is 90%. This would meant that the air coming in is 12% less dense, and therefore (if the equation for VE is linear) our AFR would be 12% higher. From 14:7 to 16:5
Another way I tested if my intake stream was getting heated by the fans, was moving the sensor into the bumper. When the fans turned on, the sensor's reported temperature barely went above ambient. However the mixture did go rich by about .5 points.
This makes me think that my real air temperature that the engine is seeing is being heated, ever so slightly.
I'd say the KISS (keep it simple, stupid) is to just run the sensor in the bumper. That way I avoid all heatsoak, and I have a smaller change in idle AFR. I just miss out on true engine air temperature, and could miss out on safety functions like pulling timing when MAT is too hot.
There's no saying that the temperature difference now between bumper and engine will be the same when ambients raise to 90F.
I looked into using the CLT/MAT blended vs flow correction table. I'm not super sure how it works. There's multiple posts on the Megasquirt forums that I need to read to understand it.
I will be switched to a plastic body sensor. I'm just not sold on the fact that a plastic one won't heat soak either.
#7
I also tried having the sensor live in other places. I put the sensor on the passenger side firewall, and the heat soak issues still showed up, and when driving there's very little airflow in that spot, so the sensor stayed hot.
Passenger side headlight seems promising, but packaging the intake tubing with the FM reroute/crossflow radiator is going to be a challenge. Under the coolant hose, and I'm running my intake right in front of the fan, over the coolant hose and I'm stressing the hose, and causing hood clearance issues.
I also had the thought of using a MS GPIO to control a STDP relay to connect either an engine bay sensor, or outside bay sensor depending on idle.
Passenger side headlight seems promising, but packaging the intake tubing with the FM reroute/crossflow radiator is going to be a challenge. Under the coolant hose, and I'm running my intake right in front of the fan, over the coolant hose and I'm stressing the hose, and causing hood clearance issues.
I also had the thought of using a MS GPIO to control a STDP relay to connect either an engine bay sensor, or outside bay sensor depending on idle.
#8
MAT/CLT comp basically tapers final MAT between CLT and MAT. If at idle you see that you are going rich because the air actually entering the engine is warmer than the sensor reads you can multiply idle RPM and KPA to make "flow", then set a percentage of CLT to add to MAT.
Megasquirt does straight retarded stuff when you disconnect a thermistor. I think you can build a custom calibration for the sensor that won't, but by default using the calibration generator the first couple and the final ADC values are all 180C/F or something stupid like that. While the relay is switching your MAT would spike crazy.
Would be nice if megasquirt devs would allow you to map something other than the IAT ADC to IAT. You could make a can connected box that blends two sensors based on conditions. They are too busy adding features nobody asks for, then wonder why nobody is available to test. ******* morons.
http://www.msextra.com/forums/viewto...=72538#p554226
Megasquirt does straight retarded stuff when you disconnect a thermistor. I think you can build a custom calibration for the sensor that won't, but by default using the calibration generator the first couple and the final ADC values are all 180C/F or something stupid like that. While the relay is switching your MAT would spike crazy.
Would be nice if megasquirt devs would allow you to map something other than the IAT ADC to IAT. You could make a can connected box that blends two sensors based on conditions. They are too busy adding features nobody asks for, then wonder why nobody is available to test. ******* morons.
http://www.msextra.com/forums/viewto...=72538#p554226
#9
Elite Member
iTrader: (10)
Join Date: Jun 2006
Location: Athens, Greece
Posts: 5,978
Total Cats: 356
The issue is not IAT heatsoaking. Set the IAT correction curve to 100% all across the board and you will still lean out when the car heatsoaks. I believe the infamous heatsoak at idle issue has a different root cause, which the stock ECU doesn't have because they know something we don't. One thing is for sure, the stock correction curve exacerbates the issue.
#10
I would love love a way to add a thermistor to the fuel injector body and then have a way to shift reqfuel accordingly. I have a feeling it would fix so many problems, hot starts and drifting idle especially with large injectors. The factory ECU probably counts time at certain conditions and shifts fueling based on that.
Or, it's just MAF.
Or, it's just MAF.
#11
MAT/CLT comp basically tapers final MAT between CLT and MAT. If at idle you see that you are going rich because the air actually entering the engine is warmer than the sensor reads you can multiply idle RPM and KPA to make "flow", then set a percentage of CLT to add to MAT.
Megasquirt does straight retarded stuff when you disconnect a thermistor. I think you can build a custom calibration for the sensor that won't, but by default using the calibration generator the first couple and the final ADC values are all 180C/F or something stupid like that. While the relay is switching your MAT would spike crazy.
Would be nice if megasquirt devs would allow you to map something other than the IAT ADC to IAT. You could make a can connected box that blends two sensors based on conditions. They are too busy adding features nobody asks for, then wonder why nobody is available to test. ******* morons.
Megasquirt Support Forum (MSEXTRA) ? (Feature Request) Can we please have input sharing! (View topic)
Megasquirt does straight retarded stuff when you disconnect a thermistor. I think you can build a custom calibration for the sensor that won't, but by default using the calibration generator the first couple and the final ADC values are all 180C/F or something stupid like that. While the relay is switching your MAT would spike crazy.
Would be nice if megasquirt devs would allow you to map something other than the IAT ADC to IAT. You could make a can connected box that blends two sensors based on conditions. They are too busy adding features nobody asks for, then wonder why nobody is available to test. ******* morons.
Megasquirt Support Forum (MSEXTRA) ? (Feature Request) Can we please have input sharing! (View topic)
That would make sense to me.
The issue is not IAT heatsoaking. Set the IAT correction curve to 100% all across the board and you will still lean out when the car heatsoaks. I believe the infamous heatsoak at idle issue has a different root cause, which the stock ECU doesn't have because they know something we don't. One thing is for sure, the stock correction curve exacerbates the issue.
I understand what you're saying, that the density table makes the issue worse by bringing the IAT sensor into the mix. I do believe that if I had a magical sensor that gave me perfect charge air temperature, the table could be tuned to account for actual charge heating properly.
The table is making the issue worse because its getting bad data. So, the root there isn't the table, its the hardware. I have a couple of different plastic body sensors that I'll try the same tests with.
Also, running the fans takes a significant amount of power, and thus the idle gets wonky. There's something to think about there.
The MS3 Firmware allows the fan to command an idle up. I honestly think this isn't playing super nicely with the CL idle algorithm. I'll try disabling the % idle up, and only command 100 extra RPMs, and then command no extra RPMs. The idea being that if I'm on a hot summer night, with the lights on, A/C on, radio, blower fan and then the rad fans, I don't think that the CL idle can compensate without the rad fans calling for more idle.
#12
I would love love a way to add a thermistor to the fuel injector body and then have a way to shift reqfuel accordingly. I have a feeling it would fix so many problems, hot starts and drifting idle especially with large injectors. The factory ECU probably counts time at certain conditions and shifts fueling based on that.
Or, it's just MAF.
Or, it's just MAF.
Speed-density has its issues because of its assumptions, and these assumptions also allow us to use it very easily.
I've wondered what the process would be to stick a MAF from a modern car (with similar airflow characteristic) in my Miata. I'm assuming I'd need to map it, but then I also understand that MAF position matters for getting the right readings.
lol, and then do we hit the issue of MAF sensor heatsoak? Then the incoming air can't cool down the sensor at the right level to get good g/s readings.
Again though, this idle-lean issue I think can be partly healed by fixing my IAT sensor heatsoak. Its not like I can go anywhere anyways nowadays!
#13
Retired Mech Design Engr
iTrader: (3)
Join Date: Jan 2013
Location: Seneca, SC
Posts: 5,009
Total Cats: 857
I would love love a way to add a thermistor to the fuel injector body and then have a way to shift reqfuel accordingly. I have a feeling it would fix so many problems, hot starts and drifting idle especially with large injectors. The factory ECU probably counts time at certain conditions and shifts fueling based on that.
Or, it's just MAF.
Or, it's just MAF.
ASE Massive for CLT 180F and higher: Like 57%
Lots of EGO authority down low: Like 17 - 20 %
Those will make the car drivable until fuel and air cool down the injectors. The EGO also helps idle on hot days.
MAF cannot account for long dead times due to hot inj coils.
#14
That is more or less what I currently do, except I've turned it up to around 190 degrees. Much cooler and I can still manage a very lean idle with no noticeable dip getting on the throttle so I just let it run lean. My car seems to love to hold heat.
I had wrote a long post about MAF and some other ****, but was a total thread derail. I do have a MAF and I log it while running speed density to back figure the curve, I can tell in high speed logs that it would be beneficial over map simply because it reacts a hell of a lot faster. I also run the full IAT correction table with it's built in IAT, yesterday was a hard cold snap and I was averaging ~98% EGO correction while my table was tuned mostly around 60-70 degrees. I hardly see the hot end of the correction curve, but haven't seen summer with my setup yet...
Still, using big honkin' injectors with a wandering deadtime is a mean feat. I might heat shrink a glass bead NTC to one of my injectors, it would be very interesting just to plot it against average EGO. I could easily make an addon CAN box that calculates a deadtime on the fly and updates the megasquirt. I am pretty sure updating deadtime requires restarting the controller though. Regardless, it should be stupid simple to implement directly on the megasquirt. I don't need 7 generic PWM outputs...
My car runs fine, just spikes lean and runs retarded rich occasionally at low loads and speeds due to lame fixes. I'm just chasing the last 5% I guess.
Fans running might load the alternator a little bit and make you need more idle valve to still keep an idle, but electrical load will also mess with deadtime. Run the windows against the stops if you have power ones and watch AFR to see if it changes. Check megasquirt to see if it sees voltage sag. If it doesn't nothing you can do, if it does sag then you can play with deadtime voltage correction. If you go lean then your deadtime gets too short as the voltage drops. If you go rich then deadtime is too high when voltage sags. If you are swinging load and RPM wildly while testing then you really can't trust the AFRs unless you know it's generally tuned flat across that range.
As for idle, I think the fan duty adder just stacks on top of the loopup table and dashpot. It will not artificially lower the minimum idle duty cycle that the closed loop, just gives it some extra throttle to keep it from stalling. If you can really load it down and stall it out you can again check the voltage in megasquirt and have it add PWM idle duty to compensate. Again, this the last 5% but I want what I want!
I had wrote a long post about MAF and some other ****, but was a total thread derail. I do have a MAF and I log it while running speed density to back figure the curve, I can tell in high speed logs that it would be beneficial over map simply because it reacts a hell of a lot faster. I also run the full IAT correction table with it's built in IAT, yesterday was a hard cold snap and I was averaging ~98% EGO correction while my table was tuned mostly around 60-70 degrees. I hardly see the hot end of the correction curve, but haven't seen summer with my setup yet...
Still, using big honkin' injectors with a wandering deadtime is a mean feat. I might heat shrink a glass bead NTC to one of my injectors, it would be very interesting just to plot it against average EGO. I could easily make an addon CAN box that calculates a deadtime on the fly and updates the megasquirt. I am pretty sure updating deadtime requires restarting the controller though. Regardless, it should be stupid simple to implement directly on the megasquirt. I don't need 7 generic PWM outputs...
My car runs fine, just spikes lean and runs retarded rich occasionally at low loads and speeds due to lame fixes. I'm just chasing the last 5% I guess.
Fans running might load the alternator a little bit and make you need more idle valve to still keep an idle, but electrical load will also mess with deadtime. Run the windows against the stops if you have power ones and watch AFR to see if it changes. Check megasquirt to see if it sees voltage sag. If it doesn't nothing you can do, if it does sag then you can play with deadtime voltage correction. If you go lean then your deadtime gets too short as the voltage drops. If you go rich then deadtime is too high when voltage sags. If you are swinging load and RPM wildly while testing then you really can't trust the AFRs unless you know it's generally tuned flat across that range.
As for idle, I think the fan duty adder just stacks on top of the loopup table and dashpot. It will not artificially lower the minimum idle duty cycle that the closed loop, just gives it some extra throttle to keep it from stalling. If you can really load it down and stall it out you can again check the voltage in megasquirt and have it add PWM idle duty to compensate. Again, this the last 5% but I want what I want!
#16
Boost Czar
iTrader: (62)
Join Date: May 2005
Location: Chantilly, VA
Posts: 79,499
Total Cats: 4,080
The issue is not IAT heatsoaking. Set the IAT correction curve to 100% all across the board and you will still lean out when the car heatsoaks. I believe the infamous heatsoak at idle issue has a different root cause, which the stock ECU doesn't have because they know something we don't. One thing is for sure, the stock correction curve exacerbates the issue.
#18
I thought the NA8 vac dump was only for starting? ASE is the fix for that and not really all that jank, considering the vac dump is a jank fix already. Is it used around idle, pwm or something? Would totally kill me, wish I could shrink my FF970's (1120 @ 60psi) around idle and low loads, not make them bigger.
#19
Has anyone here tried to control the PRC valve with MS and tie it into the startup sequence, similar to the OEM controller?
I agree, increased ASE and EGO authority seems to work fairly well to combat hot starts, but it would be interesting to test.
As a crude test, you could also just unplug the pressure regulator and see how it affects hot starts..
I agree, increased ASE and EGO authority seems to work fairly well to combat hot starts, but it would be interesting to test.
As a crude test, you could also just unplug the pressure regulator and see how it affects hot starts..
#20
I thought about the PRC valve. I wonder if the same could be accomplished with the cranking pulse and ase which is basically what I'm currently doing. The only issue I have is that the level of heat soak seems to be affected by the off duration. So if I shut off the engine and re-start after 10mins it is much worse then a shut off and restart. My current tune is not ideal for either it is lean after a longer period and rich after a shorter but the drive ability is good. I'm only running 9.5psi so I decided to enlarge the plug gap from .024" to .030" which did help with the lean condition after starting which is in the 15-16:1 range.