NTC Sensor Conundrum surrounding a Modified 3.2 JTS Engine.

[Ascari32]

Currently the 3.2 JTS engine is running very well with a 1k5ohm resistor in parallel with the NTC Sensor. Only when the weather was severe; minus 8 – 10deg.C., did the car fail to start first time in the morning. Disconnecting the 1k5 ohm resistor and the engine sprang to life immediately, confirming that the NTC is very much involved, not just in maintaining Manifold Catalytic Converter temperature, but also "Start – up" enrichment. Removing the 1k5 ohm resistor, it mimicked the use of a Choke as with conventional Ported Fuel Injection (PFI) engines.

Refitting the 1k5 ohm resistor and the engine has continued to perform well, albeit the weather is generally much warmer than hitherto. However, it is noticeable that temperature indicated by the dash gauge sits at around 90deg.C., when driving on local roads between Onehouse and Bury St. Edmunds.

Using these minor road, the engine is turning at speeds of 1100 – 1600 RPM, corresponding to 30 – 40 mph and the "Gas Burner Roar" from the Supersprint catalytic converter is less obvious. I take this to mean the engine is running nearer to normal operating temperature and thus both "ECU Dosing and Timing Advance/Retard" is close to normal, as if the original manifold cats were still in place.

At these low revs, I believe there are two issues to consider: -

1.) The inlet tract of the 3.2 JTS is taken from the front of the car, where it performs an element of forced induction. This seems logical, given how restrictive the standard camshaft timing is: 2.5deg. NVO; Static Timing, between 9deg. ATDC and 11.5deg. ATDC.

However, despite the CB camshafts, 23.5deg. PVO; Static Timing; which exists from 0.5deg. BTDC to 23deg. ATDC, the velocity of air intake, at such low engine speeds, is still not not sufficient to generate substantial upper cylinder cooling, thus engine temperature is that much higher as the cooling effect of the PVO is having negligible effect.

2.) The engine thermostat is known to be faulty; permanently open, so, although water is circulating much more freely, again speed is not sufficient for the radiator to take heat away from the engine.

However, returning from Bury St. Edmunds by way of the A14, speeds are sufficiently high enough to increase both the ram – air effect of the inlet tract and also force air through the radiator and thus the engine temperature falls by some 20 - 30deg.C. This drop in temperature provokes the ECU to resume "Dosing fuel and modifying timing advance/retard. Thus the "Gas Burner Phenomenon" of the Sports Cat resumes.

A further consideration is, although the thermostat has failed its operating value is given as 105deg.C., which is too high for the engine, given the man – cats no longer exist and therefore, there is no need for excess temperature to maintain "Light – Off". Using a thermostat, with an operate point of 82deg.C, to 95deg. C., fully open seems more appropriate. This is the value not only of the Holden V8's N/A engines, but also that of both the SAAB and Vauxhall 2.8 V6 turbo engines which are derived from the same Holden engine as the Alfa 3.2 JTS. This author believes, 105deg.C. is too high and has negative consequences for the metallic Supersprint sports cat.

The standard 3.2 JTS engine is noted for its low emissions, a function of the science surrounding the use of Man – cats, secondary cats, NTC Sensor, valve timing and the sophistication of the ECU software, although this sophistication does create considerable fragility in the way the engine performs. It is overcomplicated, temperamental and causes many secondary failures which can be difficult to pin point – burnt oil, lacquered cylinder heads, carbon deposits and failing coil packs to name but a few. Other vehicle manufacturers have achieved low emissions by using much less complicated techniques.

Lowering the engines operating temperature and fitting a good quality metallic main catalytic converter is advantageous to both the reliability of the engine and the recovery of losses due to the restrictive nature of the standard exhaust system, of which the man – cat are a part. Replacing the man – cats with straight through headers can recover 8 – 10 bhp alone, particularly if done in conjunction with improved valve timing, which further enhances engine performance – not least throttle response.

However, the position of the NTC Sensor, in the head, close to number two cylinder exhaust port is problematic. Investigating the issue surrounding this component, I have found that in some designs, the NTC is connected directly to the cylinder head, the purpose being to provide protection in the instance of coolant loss. Some too, seem to be fitted only just protruding into the water jacket a little way, which seems to be the case with the Alfa 3.2 JTS.

It is my belief; in Alfa's case, it is done deliberately to enable a very fast response time for the NTC to sense a change in manifold cat temperature as the cats need to be kept in the "Light – Off Zone" to ensure catalytic reduction when the engine operating speed changes with driving conditions.

This particular function becomes even more problematic when used in conjunction with Autodelta headers, Colombo Bariani Camshafts and a free flow exhaust system such as the Supersprints. Because of the very large positive valve overlap, the C.B. cams create and the fast flowing action of the Autodeltas and Supersprint exhaust, the NTC responds very, very quickly to change in temperature around cylinder two's exhaust port – the same with all cylinders, although it is only cylinder two exhaust port that is monitored. This results is a continual modification of both the ECU dosing of fuel into the cylinders and modification of ignition timing, advance/retard. This phenomenon is entirely due to the effectiveness of upper cylinder cooling promoted by 23.5deg. PVO and the free flow exhaust system.

A more effective way of monitoring the mean temperature of the engine needs to be found, other than by way of the fast changing value of the existing NTC Sensor. As all modifications to the engine have been undertaken with the express purpose of avoiding software modification, it is becoming increasingly difficult to find ways of gaining more performance without recourse to remapping.

However, a replacement thermostat and housing have just been acquired – after destroying a spare during bench testing. This, as previously stated, has an opening value of 82deg.C. On inspecting the housing, it seems an NTC Sensor would serve a better purpose if it was fitted into the Thermostat housing – directly in the flow path of coolant to the radiator. So once the thermostat opens, the NTC will sense the "Mean Temperature" of water circulating around the engine as a whole.

So a new TS10253 NTC Sensor will be fitted into the blanking plug of the Alloy Thermostat Housing, which is adjacent to the outlet to the radiator and just after the thermostatic element. I believe the regulating effect of the 82deg. thermostat will smooth fluctuations in NTC Sensor value which are seen at the position of the original in the cylinder head, close to cylinder two's exhaust port.

Secondly, the mean regulated temperature will be lower than standard as the thermostat operates at 82deg. and not 105deg. Thus my original criteria of lowering the engines working temperature is met.

I accept this in itself will provoke the ECU to inject extra fuel and modify the advance retard because the NTC value would indicate a "below optimum temperature" and the ECU will respond as if it were trying to return the manifold cats to "Light – off". Thus, a fixed value resistor will still need to be fitted across the TS10253 Sensor, the value of which needs to be determined to "Fake" the error and correct the indication on the dash temperature gauge.
     
P.S. It is not uncommon for an NTC Sensor to be fitted within the thermostat housing of cars and indeed believe Alfa themselves have done it before.

[Ascari32]

Still Waiting for my components to arrive, but decided to do a little more experimentation, between the rain, wind and bitter cold.

In this instance, I fitted a 270ohm resistor across the engine NTC Sensor. The car would not start, "off - throttle", from cold and it was only when I used the throttle I managed to coax some life into the engine. However, on my local test circuit, the engine was much quieter. Acceleration was good and on return it seems the exhaust tips were much cleaner - lacking any sooty deposits. On start - up however, I saw the dash temperature gauge was immediately up to 90deg., despite the car having stood overnight. Clearly, the 270ohms across the NTC sensor was too low and it was noticeable that on return the radiator cooling fans had cut in.

No matter, at least it confirmed the connection between the NTC Sensor and the radiator cooling fans. Again, this was always going to be an exercise in trial and error as there is absolutely no information on any site, to my knowledge to help give guidance.

On returning home, a little more research was done and the resistance value changed to 490ohms; 270+220ohms. This value, when in parallel with the NTC and the engine at 80deg.C. would equate to 177ohms,  an impedance, very close to representing 95deg.C, the NTC would be naturally. Essentially, the NTC Sensor, with 490ohms in parallel is "Fooling the ECU into believing the temperature is 95deg.C."

I have yet to see how well the engine starts after standing overnight and see at what point the radiator fans cut in, if indeed they do. However, the temperature gauge was sitting just below 90deg. and the coolant fans were no longer running after I made these changes yesterday, whilst the engine was still hot, but parked outside the house. I can only describe the the exhaust as "Lean - sounding" when ticking over, having lost its baritone character.

Obviously, it is still early days as I have yet to reposition the new NTC Sensor in the Thermostat Housing. I expect there to be difference in its mean value as it will be directly in the flow of the engine coolant. However, I remain convinced it may provide a better solution for me than where it is currently mounted.

https://amwei.com/ntc-2200-ohm-beta-3935k-radial-lead-epoxy-coated/

Clearly, the attached data sheet is not the same as the one fitted to the 3.2 JTS as both Bosch and Alfa are reluctant to disclose it. So there may be slight variations, but not enough to prevent a very close approximation being achieved. It was always going to be difficult in any case, as to my knowledge, no one else seems to tackled this problem, so limited information is available to me.         

[Ascari32]

Still the components for the thermostat modifications have not arrived. So I continue to explore the real function of the NTC sensor.

I have now tried three further experimental resistor values across the existing NTC Sensor, 270ohm, 490ohm and 760ohm. In every instance, from cold, the engine was difficult to start.

The reduction in noise level with the 1k5ohm fitted across the NTC, I felt may too be improved upon, hence the experiments with the three values given above. The most difficulty in starting the engine was with the lowest value; 270ohm, progressively getting easier; 760ohm, but still difficult, as the values increased: 1k5ohms, no problems at all. However, the noise level, although very much lower than with the 1k5, also started to increased, indicating "Dosing" was beginning to reoccur.

With any one of these values fitted, there is temptation to believe power had been lost as it seemed the "Gas - burner roar" deceived the ear into thinking the engine was more powerful. It is not! It was just the absence of roar made the exhaust sound appear more "Hollow". Without getting the car on the Dyno, between changes, there is no absolute way of proving this, but using my "Test Track", there seemed to be little difference.

I think I could repeat the same exercise many times over and come to the same conclusion, "Beyond Start - up, the NTC Sensor function is primarily to dose the cylinders with extra fuel when driving conditions change - coming off high speed motorways, into urban traffic. I also conclude, the advance/retard function, associated with the dosing function are inextricably linked and if there is no need to dose the cylinders, there equally is no need to modify timing advance/retard.

Given I have now dispensed with the manifold cats, there is no requirement for "Fuel Dosing", which hitherto maintained "Light - Off" temperatures in them. The Supersprint Cat is metallic and designed to function within emissions regulations, reducing pollutants to within legal limits. By the same token, the advance/retard function becomes redundant too.

So all that remains is to have a device, or a method of ensuring the ECU does not introduce these functions to the engine and also a method of getting something like the correct engine temperature displayed on the dash.

It all appears so much simpler now and the car is almost back to the noise levels of a standard 3.2 JTS with the full Alfa Romeo exhaust system, albeit, on acceleration it sounds much more exotic.

The only thorn that remains is the "Choke function", the NTC Sensor performs when the engine is cold. It is not a discrete function, more a bi-product of the action of the NTC Sensor when first getting the catalytic converters up to "Light - Off" temperature.

So, whilst one can be increasingly clever by designing a system to initially utilize the NTC Sensor as normal, then switching in a fixed value resistor automatically across it when the engine gets above a certain temperature - yet to be determined - it also introduces and further element of potential failure.

Therefore, I intend to test my new found knowledge by simply fitting a fixed value resistor across the NTC Sensor,  with a switch in series with it. On cold start - ups, the switch; which I shall mount in the cabin, will be "Open" and thus the normal enrichment function will allow fast star - up, regardless of the weather. Once the engine has settled down to normal "idle", the switch will be closed, placing the fixed value resistor in parallel with the sensor. Once the engine has warmed a little, even with the lowest value resistor 270ohms fitted across the sensor, the engine is very easy to start in any case.

The value of resistor must not be so low as to cause the radiator fans to cut in unnecessarily during normal usage however.

Leading on from this, it seems, there is no reason to move the NTC Sensor across to within the thermostat housing as this approach now seems pretty much academic. It may however be worthwhile as the reading of the dash fitted temperature gauge will always be "Fake". Fitting an extra NTC Sensor into the thermostat housing however, would enable me to fit a separate gauge into the cabin indicating the true value.

And that folks is about all to say - well from my perspective. It is clear, despite Alfa's Software, this engine "Is" capable of much improvement, without recourse to ECU Remaps. Autodelta headers, or for that matter any similar manifolds can be made to work well, thus improving performance greatly. In my case the exercise started out to be one of getting the engine temperature down, which ultimately spilled over into increased performance.

The power gains are interesting but it is mostly the increased torque that is remarkable. I have yet to get the car back onto the Dynamometer after I modified the MAF body and am hopeful of further gains as a result.

     

[Ascari32]

On the "Home Straight" - I hope.

Called Scholar Engineering today and they said my modified Thermostat housing was ready for collection.

I had just booked the car in for its MOT test tomorrow, so called back down to change the timing so the housing could be fitted at the same time. Adam suggested, "We get you legal with the MOT first and rebook for the Stat Housing later". - I agreed.

This gave me an opportunity to post the attached pictures and also start some prototype tests to control the "Dosing" element of the NTC/ECU routine.

I am not sure what the new temperature figure is going to be, compared with the NTC Sensor in its current position, hopefully a little higher, but whatever it is, I expect it to be more stable and not suffer the vagaries of constantly changing due to the new valve timing and the lack of manifold cats.

Time will tell and importantly I have the instrumentation now to read off the exact temperature at both positions.

There was a time when it would have been a huge design exercise to achieve what these two circuit boards can do and all for less than £10. Life is so much easier nowadays. In truth, I am probably going to have to pay more for some ultra - low impedance Source/Drain FET's than these little beauties, as I do not want to have any relay contacts involved.