Running lean on hot restart

[142 guy]

MS2 Extra on a 1971 Volvo B20E - stock engine still running the original Bosch Djet 036 injectors but with a 36 psi base rail pressure and pressure referenced fuel pressure regulator. Conversion from MS2 batch / fuel only to sequential fuel with spark control.

On a hot restart my AFRs will go really lean at idle (up to 17-19) for a couple of minutes. After a couple of minutes operation, the AFRs return to the more normal range around 14.7 - 15.1. I was initially thinking vapour in the fuel line or low fuel pressures; however, a fuel pressure check came out OK and the actual restart is quick, so I don't think I have a vapour issue. EGO control is not active. The MAT / CLT correction is zero.

On a cold restart, the AFRs are in the target range after the ASE and warm up enrichment are finished and certainly do not go radically lean like they do following a hot restart.

I thought I had a log of a complete hot restart showing the change from high to normal AFR; but, no. I will get a log of a cold and hot start for reference. However, I did pick off a couple of points on logs at idle with roughly the same operating conditions, one after a hot restart with moderately high AFR and one after the engine had stabilized.

Hot restart
AFR 16.5
RPM 988
MAP 58 kPa
Pulse width 2.07 ms
MAT 63 C

Normal Operation
AFR 14.5
RPM 987
MAP 57 kPa
Pulse width 2.08 ms
MAT 59 C


No warm up enrich either case. What interests me is that the fuel PW, RPM and MAP are almost identical and yet the AFRs are so different. Just based on the ideal gas law equation, the MAT temperature difference should have no where near that effect on AFR.

I have an LC1 wideband. Is it possible the wideband is doing something weird following a hot restart? However, the engine does seem to run a little rougher on a hot restart suggesting that the high AFR is correct.

Any suggestions as to what might be causing this?

[flak monkey]

Every motor I've installed Megasquirt on has done the same thing (5 motors and counting now)... I suspect that it's a IAT heat soak related issue, but I could well be wrong. I never really tried fixing it before, so I'd be interested in the answer to this too.

[142 guy]

When I was running batch fire / fuel only, I had the opposite issue (low AFR on hot restart) which I attributed to an inlet air temperature problem. The original B20E Djet fuel injection had an air temp sensor that was way up front / away from the manifold and would measure air temperatures that were much lower than the actual manifold temp, particularly at idle. The temp skew would become insignificant once the car was moving / engine under load. I installed a new sensor directly in the intake plenum (open element GM style) which seemed to minimize the issue with the rich condition on a hot restart. Now that I have switched to sequential on MS Extra, its the opposite, running very lean on a hot restart. In order to cause lean operation, the temp measured by the IAT would have to be higher than the actual air temperature.

[yzn]

What i did to my car to fix the hot restart issue, is first i relocate the sensor from the intake manifold to intake pipe just before the throttle, this helped a little bit but the problem still there. What i did then is tune the mat/clt correction by increasing the number at lower flow and tapered to zero (you have to retune the idle area after doing this). Another thing i did is increased the last two value form the ASE Taper table and changed the setting to ignore MAT correction during ASE to ON from the cranking/startup setting menu. Now i have no problem starting the car after a long drive in warm weather.

[Dennis930]

I did as yzn has done. My ASE taper is twice as long at high MAT temperatures as compared to the lower temperature range. And I factored in 25% clt/Mat correction at low rpm's and tapered to zero by 2200 rpm's. I will see how well it works when the daytime temperature are in the 90's. One of the problems I have is that the intercooler is above the engine (air cooled), which heat soaks from rising engine heat when shut down. After restart the MAT increases from the intercooler heating the air. My IAT sensor is located after the intercooler and before the throttlebody. It is a VW 1.8 turbo sensor which is totally plastic except the thermistor element (which looks like a small resistor), and the element is directly in the air path. So, the only part that can heat soak is the element itself which is a very small mass. The data logs I have taken do not show my IAT as being a heat soak problem. None of this makes any sense, you would think the AFR would be rich doing these corrections. But reported AFR is around 14.1 and then drops to my target 13.8 after the engine settles in. I believe that heated injectors and/or fuel has a role in this.

[142 guy]

I understand how extending the high temperature ASE would help drop the AFR during a hot restart. I was considering that as a possible 'patch' if I could not address the fundamental cause (about which I am still a bit in the dark). I am at a bit of a loss as to how the MAT / coolant correction will help. The description of the correction is that it blends some coolant temperature into the air temp to give a corrected air temperature. I am not sure what blending means. Does the algorithm take (1-% correction value)x(measured air temp) and add (% correction value)x( measured coolant temperature) to create a corrected air temperature? If so, once the coolant is hotter than the air, this would create a blended air temp that is hotter than the measured air temp. This would tell the fuel equation that there is less oxygen in the air charge and that the fuel needs to be reduced which would tend to cause the engine to run leaner. Not the direction that I want to go in. Or, am I misunderstanding something about how the MAT/CLT correction works?

I was doing a little more exploration and there are anecdotal claims (not supported by actual code evidence)that some OEMs and aftermarket ECU vendors apply an adjustment to the air temperature measurement at high(ish) manifold air temperatures to reduce the temperature used in the calculation. Effectively saying that at high temperatures, the measured value is incorrect, or needs some bias if you don't like the word incorrect.

I had tried a brief run at using the MAT/CLT correction to address the lean condition on hot restart. From my brief experimentation, it either had no effect or seemed to make things worse, at least in terms of the idle being less stable. I am inclined to try increasing the MAT air density corrections for temperatures above 45 C, perhaps making the correction line close to horizontal above 45 C. Or, has somebody already tried this strategy and determined that it is ineffective?

[flak monkey]

Talking of biasing the curve, MS3 has this MAT Air Desnsity curve - I can't remember if MS2 has the same? My air temp sensor is in the intake ducting about 4" infront of the throttle body. Previous engines have had the sensor in the intake manifold and other places, but always had the same issue.

I'm thinking about removing some of the correction at high temperatures to see if it helps. I rarely see any genuine MAT temperatures above 50*C when driving, but after a hot restart they could be at or close to engine temperature.

MAT air density.JPG

[yzn]

I am at a bit of a loss as to how the MAT / coolant correction will help. The description of the correction is that it blends some coolant temperature into the air temp to give a corrected air temperature. I am not sure what blending means. Does the algorithm take (1-% correction value)x(measured air temp) and add (% correction value)x( measured coolant temperature) to create a corrected air temperature? If so, once the coolant is hotter than the air, this would create a blended air temp that is hotter than the measured air temp. This would tell the fuel equation that there is less oxygen in the air charge and that the fuel needs to be reduced which would tend to cause the engine to run leaner. Not the direction that I want to go in. Or, am I misunderstanding something about how the MAT/CLT correction works?

Think about it in way how the CLT reading affecting the MAT reading. You have to readjust the VE table in the lower flow area after enabling the mat/clt correction if you didn't do that it will lean out.

Note: flow (load*rpm), if your idle between 600-1200 rpm and load 20-60 kPa the x axis for the mat/clt correction should be roughly between 10,000 (or 0) and 80,000, try with y value started at 25% (lowest air flow) tapered to 0 at above 80,000. Try that and see what happen.

[cactus]

Does your coolant temp get heat soaked and start higher on a hot restart after maybe 10 minutes being off? I have noticed this on my B20 but don't know if it is common on all cars. It goes up 10 to 20 degrees F but then drops over a little bit. Is this part of your situation at all?

I also struggle to understand how to handle the hot restart condition because I know the manifold gets really hot after a short rest before some airflow cools it back down somewhat. With the air temp sensor far away from the engine it doesn't see the soak but the air coming in has got to be different for a little while and I don't quite see how the available adjustments handle that. This situation has to be somewhat common?

John

[VcrMiata]

Just wanted to add a "me too" for increasing ASE taper on higher temps.

I reviewed hot start logs and found my lean condition on hot start always lasted about 25-30sec, so my ASE taper on hot starts is now 30sec.
I tried adjusting MAT correction (up to 100%) too but it had little/no effect. Also, I couldn't get a definite correlation between MAT temp and the leanness being reported. It was, however, pretty well correlated with time .

[grom_e30]

Just wanted to add a "me too" for increasing ASE taper on higher temps.

I reviewed hot start logs and found my lean condition on hot start always lasted about 25-30sec, so my ASE taper on hot starts is now 30sec.
I tried adjusting MAT correction (up to 100%) too but it had little/no effect. Also, I couldn't get a definite correlation between MAT temp and the leanness being reported. It was, however, pretty well correlated with time .

you can also set it to ignore mat correction during ASE as well

[VcrMiata]

you can also set it to ignore mat correction during ASE as well

Absolutely, that's how I figured the significant part of leanness on hot start didn't have much/if anything to do with MAT correction (in my case). Once that was proven I moved to using ASE taper to workaround the issue.

[142 guy]

I looked at my existing logs and things were quite messed up, primarily because I couldn't resist the urge to adjust while doing the logs. Since the data was not directly comparable between logs, I did two tests this morning. A start from cold and allow to run until it looked like the MAT had stabilized then stop and wait about 10 minutes for the manifold to heat soak and then do a restart. It was much cooler today and the starting ambient temperature was around 13 C.

Yesterday I sinned and adjusted my Ve values a bit, so my AFRs in the cold start log are lower than what I had been using previously. In these logs the EGO correction is off, the MAT / coolant correction is flat and the air density correction is the default (ideal gas law) curve. Nothing was changed between logs (other than it appears I might have been resting my foot on the gas pedal).

At the end of the cold start log, the MAT is up to around 41.6 C and the AFRs are pretty stable around 13 - 13.3 once warm up and ASE are finished. In the hot restart, the MAT has heat soaked to 62.7 C prestart and and starts dropping once the engine starts running. At the point ASE is off the MAT has dropped to around 56.2 C and the AFRs are running in the high 16s. As the MATs drop, the AFRs also drop. The MAT looked like it bottomed out around 50.4 C and the AFRs were in the 13.7 - 13.9 range. The MAT never dropped to a value as low as the end of the cold start run and looked like they were starting to climb so I terminated it.

It almost looks like my AFR problem emerges when the MAT get above the 40 C range. I am speculating that the approach might be to set the idle VE values to meet the AFR target on an engine following a cold start once it is up to operating temperature and then use the air density curve to introduce some correction when the MAT is 50 C or higher?

The logs are too large to attach, so here are links to the logs on Google drives

cold start
https://drive.google.com/open?id=0B4XVg ... TlsWDJaVTA

hot restart
https://drive.google.com/open?id=0B4XVg ... mJVN1FjaWc

[142 guy]

With respect to the comments about MAT correction during ASE and using ASE to correct the lean running condition. The first is that my lean period extends way beyond the ASE on a hot restart. Extending and increasing the ASE might provide a patch; but, in my case I do not think it would be effective. If you look at my hot restart log, I would have to extend the ASE taper out for multiple minutes before the MAT has dropped enough that the AFRs have recovered to the values that were comparable to the values on a warm engine following a cold start.

Yesterday I had done some testing with unusually high air temperatures for early May, in the order of 32 - 33 C. On a hot restart test that I did, as long as the engine was idling the MAT remained in the 50 C or higher range. It was not until I started driving the car that the MATs dropped and the AFRs recovered to more reasonable values and that initial period of driving was a pain because the AFRs were really lean and engine operation wasn't very smooth. If I were going to adopt the extend the ASE strategy, following a hot restart I would have to start driving the car before the ASE terminated or I would be into high AFR operation.

If the problem is related to the MAT (which is still an if) and is a high MAT anomaly, I would be more inclined to try applying some air density correction starting at 45 C. Last year I did some testing and once the car is up to 50 km/h or higher, the MAT drops to about 2 - 3 C higher than the cold air inlet temperature. Since the ambient temperature rarely reaches 40 C here, applying air density correction above 45 C would mean that the correction would have no effect under normal operating conditions and would only effect operation at low speed or hot idle. Unless somebody can spot something obvious in the logs, I thin that is where I am inclined to go.

As an interesting observation, I have an electric fan in pusher mode. With ambients in the 32 C range and the car just sitting idling, the fan is only cycling on about 10-20% of the time. While the engine temperature might be just fine, the under hood temperatures are probably getting pretty high which contributes to the MAT problem. There might have been an advantage to the original viscous coupling fan that the B20E came with in terms of managing under hood temperatures.

The other interesting factor that might be at play is that I know that the B20 head develops a temperature skew from front to back when idling. The front of the block where the megasquirt sensor is located remains very steady, fluctuating about 2 - 3 C as the fan does its thing. However, at idle the back of the head where the dash sensor gauge is located will creep up to about as high as 12 - 13 C higher than the front of block temperature during extended idling. This temp skew drops to about 2 C once the engine speed increases.

[slow_hemi6]

I live in Northern Australia. I know what you are going through. I only allow my MAT curve to decrease 2degrees C starting from 40C to 64C (96% to 94%). The rest of the curve is normal but this gives me normal hot restarts.

[142 guy]

I live in Northern Australia. I know what you are going through. I only allow my MAT curve to decrease 2degrees C starting from 40C to 64C (96% to 94%). The rest of the curve is normal but this gives me normal hot restarts.

Tentative results suggest that this approach seems to be working.

I did a couple of tests yesterday evening after the engine had been sitting for about 10 hours. After the engine reached its normal coolant temperature from a cold start I adjusted the Ve values in the idle area of the fuel map to give me my target AFRs. I then shut the engine down, let it heat soak for about 15 - 20 minutes and did a restart. Sure enough, after the ASE finished I was running with AFRs about 2.0 - 3.0 higher than previously. Starting around 45C, I gradually moved the air density correction curve up to return the AFRs into the target range. I probably still have more correction in the curve than your 2%. I will do some more cold followed by hot starts this weekend to confirm whether this approach works with the B20 (as in gives repeatable results) and if it does, perhaps test further to determine if some more refinement of the adjustments to the curve is required.

[VcrMiata]

My tuning also included MAT air density adjustments.  My MAT correction curve is much flatter above 30C than the original ideal gas law curve. However I did this as part of my fuel tuning, not hotstart tuning. My tuning was done trying to aquire the same target idle AFR using different but stable CLT and MAT temperatures (as much as  possible) with some extrapolation on the MAT curve.

After this I discovered that even when my start MAT was within the tuned MAT correction region, I was going lean for up to 30 sec after a hot start.  This was when I decided to extend my ASE taper.  After ASE completes, MAT correction is restored. (Now that I read this I'm wondering why I still have MAT correction turned off.)

I didn't mean to imply I did not adjust MAT air density correction curve, however re-reading it sure seems that way. In my case MAT correction was not done as part of hot start tuning, I did this a part of my fuel tuning. After fuel tuning, I had a hot start issue for a period of time after start that could not be attributed to MAT and I could not find any other way to address other than ASE taper.

[142 guy]

After this I discovered that even when my start MAT was within the tuned MAT correction region, I was going lean for up to 30 sec after a hot start.  This was when I decided to extend my ASE taper.  After ASE completes, MAT correction is restored. (Now that I read this I'm wondering why I still have MAT correction turned off.)

I could see extending ASE taper as a viable solution to addressing a very short period of high AFR following a hot restart. Perhaps after fiddling with the MAT correction, if I also experience an short period of high AFR after a restart, that might be a better fix than bending the MAT correction curve even further away from the theoretically correct value.

[142 guy]

For a couple of days now I have been running the engine with the MAT correction adjusted to give no change in air density once the MAT exceeds 35 C (curve essentially goes horizontal at 35 C). So far, this seems to be reasonably effective at reducing the high AFR condition that occurs on a hot restart. It doesn't get the AFRs exactly back to the target values; but, close enough that with about 5% or less EGO correction it is on target. Once the car has been driven for about 5 minutes the steady running MATs have dropped to 35 C or less, MAT the correction disappears and the EGO correction is just doing its normal corrections.

The last few days have been cooler so my MATs have been dropping to less than 35 C when driving. Driving during ambient high temperatures might result in my steady running MATs exceeding 35 C. If the steady running MATs get much above 35 C, the MAT correction will cause the fuel mix to run rich. It will be interesting to see if this strategy is still as effective with high ambient temperatures.

[142 guy]

Time for an update.

After playing around with the MAT/CLT correction and then finally moving my MAT sensor from the manifold into the air inlet, I still was not resolving my run lean on hot restart condition. During one of the test sessions that was logged with the sensor mounted in the air inlet, I saw my measured air inlet temperatures go up and the AFRs drop. That got me to thinking that the lean hot restart issue has nothing to do with MAT sensor heat soak or air temperature measurements in general. I ended up pulling the sensor out and hanging it in free air so that the temperature measurement remained constant at around 20 C during the a hot restart test. Sure enough, on a hot restart the AFRs go lean and then within a few minutes resolve back to where they should be at idle (no EGO during tests) all the while the MAT input into the MS2 remains constant at 20 -21 C.

I then started wondering about the fuel injectors. I attached a thermocouple to the base of one of my injectors. Running the engine up to temperature and then shutting down gave be a starting injector base temperature around 40 - 43 C. This climbed up after shut down reaching a max of around 69 - 70 C. I started the engine up and sure enough, high AFRs. However, as the injector temperature gradually started dropping, the AFRs also dropped. Once the injector temperature was in to the 45 C range the AFRs were back in their target range.

I repeated the test with a small change. After doing the run up to temperature and shut down, I did a restart before the injectors had a chance to heat soak (about 45 seconds) and the AFRs were spot on target. Wait about 5 - 10 minutes until the injector temperatures have peaked and I have really high AFRs.

I think the problem is some kind of injector heat soak issue. I discovered that some time in 1972 - 1973 Volvo started retrofitting my B20 engine with phenolic spacers under the injector holders to help with lousy operation following hot restarts, so I think that they were aware that there was a problem of some kind. The common wisdom was fuel vaporization in the injector body; but, if that was the case that should resolve itself very quickly as opposed to taking minutes to resolve if the engine is just idling. After a little research, I discovered that the JEEP inline engines also suffered from a hot restart problem and the solution was insulating blankets and thermal barriers for susceptible injectors. I think the JEEP is like the Volvo B20E in having the intake and exhaust on the same side of the head which provides a nice radiant source of energy to warm up the injectors.

While doing a little internet surfing trying to find documentation about the effect of heat on injector operation, I came across a FORD patent covering injector and fuel temperature and pressure measurements. This was in the context of a returnless fuel system. What was interesting was that in the section of the patent that is called State of the Art (or something like that) the patent referenced something called adhoc HICOMP systems (Hot Injector Compensation) primarily for the purpose of explaining that the patent proposal was not for a new type of HICOMP system. I draw from this the conclusion that the OEMs might be a little more aware of this issue with hot injectors than seems to be visible on the internet.

I have returned the MAT/CLT correction back to the default, added a heat shield between my injectors and the manifolds and have configured the EGO such that I allow quick start up of EGO when the engine is hot with up to 15 % of error correction. So far this seems to manage the most nasty part of the lean run condition and make the engine operable. Aside from better thermal insulation of the injectors, I am not sure that there is much else that can be done to resolve the problem

I am not saying that this is the cause of all lean hot restart conditions; but, if you are having an unresolvable problem, give some consideration to heat soak on the injectors.