Home built Hot Rod

[Pork Chops]

Now you’re just giving us more stuff to worry about! Mine are as far back from the primary collector as they can be, but clearly well within 1m given it’s a 911! What does one do??

Maybe an X crossover pipe and put a single sensor at the crossover point?

[MFX]

2 hours ago, Redracn said:

Some more hopefully useful info regarding Lambda measurements.

Back in the early 90’s when I was designing my first Lambda Meter we did extensive testing of the sensor location in the exhaust. What we discovered is that the exhaust gas flow from a cylinder after say a 4 to 1 would not fill the entire pipe but remain as a smaller diameter flow in the larger pipe. It could also spiral along the bigger pipe. This means that often the lambda sensor if placed within a meter of the collector will mainly read the lambda of a single cylinder and not necessarily the one you think based on any spiralling which can also change with rpm and load. 
So without seeing the installation it is highly likely that the so called “ Bank “ sensors are actually mainly  reading the lambda of individual cylinders. 

You can see me weld mine in here ( the video should start at the sensor install).

Like Nick said, on a 911 you are hard pressed getting them much further back, so I suppose you are just relying on all of the cylinders on that bank to run the same (even though they probably arent ).

 

[Redracn]

I didn’t mean to scare anybody. I was just pointing out some issues with Lambda measurements. Where you measure the mixture matters especially if all the cylinders(injectors including fuel rail resonance) are not identical.  
You could put in individual sensors for each cylinder but often they are close to the engine and the mixture is not fully burnt at this point so readings can be leaner than reality. Or put in multiple sensors around the pipe to get an average. But in the end unless you are racing all this is not necessary to have a reliable engine that performs well. That’s why fuel settings are usually set conservatively. Engines are surprisingly tolerant of mixture variations. 
Measurements after turbos tend to give a good average , measurements before turbos can overheat the Lambda sensors giving incorrect readings and shortening their life.
When tuning race engines on an engine dyno the mixture is adjusted for maximum torque and the resulting Lambda recorded so that the same mixture can be set when doing the final tuning at the track. 

Reality is you have 6 x 1 cylinder engines that are interconnected in various ways and interact with each other. While we would like them all to be identical that is never the case. The challenge is of course to get them as close as possible and then tune each one individually within the limitations of the tools available. 
 

58 minutes ago, MFX said:

You can see me weld mine in here ( the video should start at the sensor install).

Like Nick said, on a 911 you are hard pressed getting them much further back, so I suppose you are just relying on all of the cylinders on that bank to run the same (even though they probably arent ).

 

Glad to see you have the sensors angled from 2 to 4 (clock face) as that is the best angle.  With them right next to the vbands you need to ensure that the vbands absolutely do not leak and are perfectly sealed. Yes air can be drawn back up the exhaust.  When doing tailpipe measurements you want the pickup at least 500mm into the pipe. 
Yep close enough to have a bias to one of the cylinders. 

[st3ve]

I am not an EFI professional but many years ago I converted an air cooled MC to EFI using an Autronic SM4 ECU.  As a general comment, I found that a compromised initial sensor setup makes tuning considerably more difficult and the end result ragged and less tolerant of changes in operating conditions.  Also, input from someone intimately familiar with your chosen ECU is invaluable.  I got onto an SM4 guru and reconfigured some sensors, got him down to tune again from scratch and the difference was remarkable.  Basically the  same peak power but a big difference in driveability and how it behaved from cold to hot, and also altitude.

Like you I had the challenge of "coolant temp" which is one of the biggies for an ECU, any air cooled motor is going to challenge the ECU because even the best cylinder head temp setup is going to follow a somewhat different temperature curve than water temp the ECU is expecting.  My understanding was that when the engine is cold, vapourised fuel condenses on the inlet port walls - this is the primary requirement for enrichment.  So I aimed to get my sensor as close to the inlet port as possible and settled on a NTC thermistor which was very small and which I potted with thermally conductive epoxy into a drilled hole in one of my inlet flanges.

You can just see it (pink wire) on the centre cylinder here...

I wrote a table for the ECU readings based on the datasheet of the thermistor (cross checked against known temps) and this setup has worked really well from day one.  I get good correlation to ambient temp and a quick response up to operating temps around 90-110C from memory.  I recall seeing a max of about 130C on the dyno at one stage - was a good indicator that we needed to give it a break.

It would seem to me that in spite of the location of the engine, the 911 has excellent cooling with the fan, fine fins and all the shrouds.  I can see it would be easy for a sensor to get an unrealistically cool reading such as yours is delivering.  A sensor like mine might be of use if you can't sort out something else.  Unfortunately a major change to the temp sensor values will probably necessitate a proper re-tune so try to make any other sensor refinements all at once.

I made two other major changes to sensors which helped.  One was to move inlet air temp sensor from a general ambient reading near the inlets to a location inside one of the throttle bodies (I used one of the blanked off secondary butterfly holes of the GSXR TB) so it read air just above the butterfly.  The SM4 uses a table called 'charge temp estimate' and this helped that whole system to work a lot better.
The other was a separate MAP sensor connected to a plenum linked to all TB's below the butterfly.  Some experimentation with plenum volume delivered a useful MAP reading which is used with TPS to calculate a 'load' reading which is what drives the main fuel table.  That then allowed me to use the inbuilt SM4 pressure sensor for barometric pressure, which compensates on the fly.  Not so important in Oz with comparatively slow elevation changes (the ECU otherwise samples MAP before startup and uses that for baro) but all these changes together made for a very well behaved engine in all conditions.

O2 sensors, readings thereof, auto tune, closed loop etc was a whole other rabbit hole but best to get the base setup right first.

HTH

[MFX]

Here is a pic of my temp sensor (circled in yellow) which you can see is just bolted to the cam tower under the edge of the cover. I flipped the end of the cable around this morning so that the tail now heads out the other side under the cover so maybe get the end of it a bit closer to the head, but I am not very confident it will make any noticeable difference. The only places I can see I could mount it without pulling everything apart is either a clamp to one of the fins, or possibly connect it to the base of the ITB, similar to what @st3ve suggested (circled in purple), although I am not sure how much better that would be?

[Redracn]

20 minutes ago, MFX said:

Here is a pic of my temp sensor (circled in yellow) which you can see is just bolted to the cam tower under the edge of the cover. I flipped the end of the cable around this morning so that the tail now heads out the other side under the cover so maybe get the end of it a bit closer to the head, but I am not very confident it will make any noticeable difference. The only places I can see I could mount it without pulling everything apart is either a clamp to one of the fins, or possibly connect it to the base of the ITB, similar to what @st3ve suggested (circled in purple), although I am not sure how much better that would be?

First thing I would do is check the calibration of the sensor in the ECU and if any doubt a simple and easy field test is an ice slushy for 0C and boiling water for 100C.

Its hard to tell from the picture but it looks like the sensor is on top of the bracket? Is the sensor itself mounted further up the what looks like an electrical eyelet? 

I wouldn’t mount an ECT sensor on an ITB as the temp there is also a function of airflow especially if an insulation spacer is used and you also have the issue of the fuel standoff in the runner causing substantial cooling in the inlet tract. Runners and ITB can become quite cold when flat out and quite hot when stopped due to heat soak.. The temperature of the cooling fin is also a function of the rpm so while better than an ITB still far from ideal. If you can mount the sensor under the bracket and direct to the head and use an insulating washer between the sensor and the bracket would be a good improvement as would some thermal transfer paste between the sensor and the head. The idea is to get as much heat to the sensor while not having other items such as the bracket drain it away. A lot of cars have thermal insulation washers/spacers on the fuel rails. 

[st3ve]

Have you checked the sensor calibration?  Take it off and dunk in some boiled water, or bolt it to a plate and warm that up, see what the ECU reads and compare against a measured temp.  Use a thermocouple if possible, IR temp guns are really useful but sometimes way off on shiny metallic surfaces.  It seems unlikely to me that that point there is only 40-50C in operation??

[Redracn]

A reasonable solution might be to drill down the bolt from the top and put a small NTC thermistor a few mm from the tip and bond it in with a thermally conductive epoxy. You could also only drill the bolt with a smaller hole for the wires and mount the sensor past the tip and fill the cavity in the head with thermal paste. If you when for a thermocouple a 1.5mm drill would do. 

Perhaps something like this could double as the mount bolt. Seems to be lots of sources.

https://www.bosch-motorsport.com/content/downloads/Raceparts/en-GB/58273931119693579.html#/Tabs=58279435/

https://www.msel.co.nz/epages/motorsportelectronics.sf/en_AU/?ObjectPath=/Shops/motorsportelectronics/Products/SENTEMSSM6

[st3ve]

Snap Redracn!    That is a good point about runners cooling with large flow, same way carbs ice up in cold weather I guess.  In my install the thermistor is just a couple of mm from the head interface, has really good thermal contact and is on the head side of the rubber isolator.  But something to consider for sure.

As another aside - these similar motors are efi out of the factory.  What type and where did Porsche put their CHT sensor in for example a 993?  They must have measured temp somewhere/somehow.  It's always best to leverage prior art if at all possible, much development often went into it.

[Redracn]

3 minutes ago, st3ve said:

Snap Redracn!    That is a good point about runners cooling with large flow, same way carbs ice up in cold weather I guess.  In my install the thermistor is just a couple of mm from the head interface, has really good thermal contact and is on the head side of the rubber isolator.  But something to consider for sure.

As another aside - these similar motors are efi out of the factory.  What type and where did Porsche put their CHT sensor in for example a 993?  They must have measured temp somewhere/somehow.  It's always best to leverage prior art if at all possible, much development often went into it.

Yea I have fond memories of the carby on my rotax powered KTM600 dirt bike icing up after a long blast at WOT. Kill switch became my best friend. 
 

Yours is a good situation as you have tapped into the alloy rubber to head transition piece. With the O ring seal not hindering heat transfer from the head and the rubber isolating the ITB and runner from it it would be difficult to improve upon other than some thermal transfer paste on the adapter  flange. Amazing how much difference some thermal transfer paste makes when between two flat surfaces. 

[Pork Chops]

4 minutes ago, Redracn said:

Yea I have fond memories of the carby on my rotax powered KTM600 dirt bike icing up after a long blast at WOT. Kill switch became my best friend. 
 

Yours is a good situation as you have tapped into the alloy rubber to head transition piece. With the O ring seal not hindering heat transfer from the head and the rubber isolating the ITB and runner from it it would be difficult to improve upon other than some thermal transfer paste on the adapter  flange. Amazing how much difference some thermal transfer paste makes when between two flat surfaces. 

Sorry officer, I couldn’t slow down as my throttle blade was iced open at WOT!

[MFX]

1 hour ago, Redracn said:

First thing I would do is check the calibration of the sensor in the ECU and if any doubt a simple and easy field test is an ice slushy for 0C and boiling water for 100C.

Its hard to tell from the picture but it looks like the sensor is on top of the bracket? Is the sensor itself mounted further up the what looks like an electrical eyelet? 

I wouldn’t mount an ECT sensor on an ITB as the temp there is also a function of airflow especially if an insulation spacer is used and you also have the issue of the fuel standoff in the runner causing substantial cooling in the inlet tract. Runners and ITB can become quite cold when flat out and quite hot when stopped due to heat soak.. The temperature of the cooling fin is also a function of the rpm so while better than an ITB still far from ideal. If you can mount the sensor under the bracket and direct to the head and use an insulating washer between the sensor and the bracket would be a good improvement as would some thermal transfer paste between the sensor and the head. The idea is to get as much heat to the sensor while not having other items such as the bracket drain it away. A lot of cars have thermal insulation washers/spacers on the fuel rails. 

The sensor is basically an electrical eyelet. I have use the bolt for the fibreglass cover and mounted the sensor under the fibreglass straight onto the cam tower. Are you suggesting I get one of those M6 temp sensors and mount it in the sam bolt hole I am currently using?

[Redracn]

5 minutes ago, MFX said:

The sensor is basically an electrical eyelet. I have use the bolt for the fibreglass cover and mounted the sensor under the fibreglass straight onto the cam tower. Are you suggesting I get one of those M6 temp sensors and mount it in the sam bolt hole I am currently using?

Yep replace the bolt with one that has an integrated sensor.

[Zelrik911]

2 hours ago, st3ve said:

  What type and where did Porsche put their CHT sensor in for example a 993?  They must have measured temp somewhere/somehow.  

993 Buried deep in a rear head - pr1ck to get out!!!!!!!

[Mike D'Silva]

2 hours ago, MFX said:

The sensor is basically an electrical eyelet. I have use the bolt for the fibreglass cover and mounted the sensor under the fibreglass straight onto the cam tower. Are you suggesting I get one of those M6 temp sensors and mount it in the sam bolt hole I am currently using?

Jeff, that's how the kits from Rasantperformance mount them. I mounted mine under the shroud, directly on top of the cam tower.  There are plenty of other people here that know a lot more than me, but mine seemed to work fine. 

[MFX]

10 hours ago, Mike D'Silva said:

Jeff, that's how the kits from Rasantperformance mount them. I mounted mine under the shroud, directly on top of the cam tower.  There are plenty of other people here that know a lot more than me, but mine seemed to work fine. 

I am thinking the calibration of mine might be off which is the issue.

[Zelrik911]

1 hour ago, MFX said:

I am thinking the calibration of mine might be off which is the issue.

Even the Factory had problems with the HTS on the (3.6) 964 & 993. They originally had a one-wire Sensor and then moved on to a two-wire lead with an earth wire. It was a common update to move all one-wire Sensors to the later 2 wire units. 

I suppose this shows that the head temp was hard to measure accurately and that it is a key component for air-cooled management systems. 

[bumble]

Drill and tap into the banjo fitting on the oil feed on the left hand side cam tower.  Very consistent temp. measurement here...

 

You can even buy them like this...

[MFX]

4 hours ago, bumble said:

Drill and tap into the banjo fitting on the oil feed on the left hand side cam tower.  Very consistent temp. measurement here...

 

You can even buy them like this...

That looks interesting Mark. The only issue I could see is that it would be greatly influenced by the oil temp, but most tuners have a consensus that the oil takes too long to warm up to help with the 'warm up' tune on the ECU.

[bumble]

Yes, the thinking is that the oil temp is a more consistent proxy for the engine temperature.  The Bitz racing CIS to EFI conversion was rendered much more reliable using this method.

[Redracn]

The primary use of coolant temp is for engine warm up. Oil Temp is way to slow to rise and is a poor proxy for the engine temperature. Far better to use head temperature and do it reliably. Also with head temperature corrections can be applied to improve engine performance and protection that can not be done using oil temperature.

[bumble]

Obviously slower to react - but that was the point - we wanted it to remain rich until it got up to temp. The Bitz ECU is rather primitive with few inputs.

[Peter M]

9 hours ago, Zelrik911 said:

I suppose this shows that the head temp was hard to measure accurately and that it is a key component for air-cooled management systems. 

Zelrik,

I have a different take on the the reason they changed to a two wire system.

I think measuring head temp accurately is straight forward.  I just think they found that the dis-similar metals caused corrosion that effected the earthing of the sensor that was easily resolved by moving to a separate earth wire on later sensors. 

1 hour ago, bumble said:

Obviously slower to react - but that was the point - we wanted it to remain rich until it got up to temp. The Bitz ECU is rather primitive with few inputs.

Gee whiz Bumble, that doesn't sound any more sophisticated than a manually operated choke on a carburettor!

By comparison, a Motec M1 offers:

Enrichment table based on CHT - On mine it enriches 20% at 20 degrees and progressively reducing to nil at 105 degrees

Engine Crank Enrichment table based on CHT -  Enriches 600% at -20 degrees down to 137% at 120 degrees only when the engine is cranking over on the starter.

Post Start Enrichment table based on CHT - Enriches up to 200% at -20 degrees for up to 30 seconds.  There is an option to extend enrichment up to 60 seconds.  Mine does do an enrichment if CHT is greater that 120 degrees. This takes over from the Engine Crank Enrichment table when the engine fires and starts spinning at more than 300rpm.

None of this would be practical if the temperature was based on oil temperature as it doesn't model the actual fuel atomisation conditions well at all.  How's it work in reality? Starts beautifully hot or icy cold. 

 

PS: I'm looking forward catching up again.  Maybe when Jeff invites us down to drive Harry!😀

 

Maybe I was bit harsh about the choke quip, but surely even the WUR on CIS is more sophisticated! 

[Redracn]

15 hours ago, Peter M said:

Zelrik,

I have a different take on the the reason they changed to a two wire system.

I think measuring head temp accurately is straight forward.  I just think they found that the dis-similar metals caused corrosion that effected the earthing of the sensor that was easily resolved by moving to a separate earth wire on later sensors. 

Yep corrosion is one of the reasons along with the use of loctite. Surface oxidation of aluminium is a really good insulator as is loctite. Then there is also the technical issue of shared current paths of signal and high current devices including ignition coils and the radio frequency and DC from them being injected straight into the ground cct of the sensors for the ECU to read.  Single wire sensors were simply NEVER EVER going to work. 

Running excessively rich until it warms right up is very bad news. Cylinder wash down is real. 
I have seen race teams use no coolant sensor but they properly warmed the engine with external sources before starting. 
 

[MFX]

Working through the tune again today. I have tested the Head temp sensor and it seems pretty close to the range. Between 10 and 50 degrees it is pretty much spot on with my laser temp gun. It drops off a bit at the far ranges. It was reading 5 degrees in the 0 degree ice water and got to around 68 degrees with the recently boiled water my temp gun read at 75. I have looked to find a better place to put it, but I am not sure there is. I could make a little bracket similar to the ones shown on Pelican, but my fins seem much tighter than the ones on there. It would need a 2mm thick arm to clamp in the gap in the fins and I am not sure it could be strong enough to hold. 

I think I will change the parameters in the warm up to fade off to 0 by 40c so it is not messing with my fuelling when warm and see how it goes.

I am also going to recheck the balance of the ITB's as I suspect they could be out a bit.