Issues with exhaust backfires and hesitation with purge solenoid function.

[jigga009]

Hello,

I have a purge solenoid wired into my G4+ Xtreme for my Subaru application. I have the ECU controlling the solenoid via PWM in order to be able to get more granular control over just how much fumes are coming into the intake manifold when the ECU commands the purge solenoid on. From what I can see so far, it seems to work, but there are some issues that I have not been able to solve.:

Issue 1  - My purge solenoid (a Bosch 0280142353) is being controlled by my G4+ Xtreme with PWM, but I am finding that very low duty cycle numbers (<10%) seem to provoke some rather large AFR swings of up to 40% in CLL when the purge solenoid is activated. For example, duty cycles of perhaps 3-5% can cause an AFR swing that taps out the CLL correction at 40%. Not sure if it's an issue with the settings I have for frequency and the like; it may or may not be a big issue given that you can only input values of 0.5 multiples into the PWM table, but it is something I have noticed, nonetheless, and thought I might mention. I spoke to Simon in tech support about it, and he could not figure out why such small duty cycle figures were causing such a large AFR swing for the ECU to compensate for. Could this be an issue with the set frequency I am using to operate the purge solenoid?

Issue 2 - The real issue - On occasions when the purge solenoid is activated (and then CLL is busy pulling fuel out to compensate), and then I take my foot off the throttle, going into overrun, and then back into the throttle when the purge solenoid is immediately activated again, the car seems to go transiently lean causing quite a large "bang" out of the exhaust. It's such that it can be detected on my pre-turbo backpressure sensors.

My initial suspicion was that the issue was due to the closed loop "reactivation timeout" since in other logs it coincided with when the timeout was active and CLL was "asleep". However, in the latest log I have, overrun fuel cut was not activating for other reasons (so no reactivation timeout after overrun fuel cut), and yet the re-opening of the purge solenoid seems to cause a transient lean event that occurs too quickly for the CLL to catch, causing a brief period of choppy/stuttering power, followed by a loud bang out of the exhaust if I keep my foot steady at the same throttle %.  

 I have consulted tech support, and I've been told that the "reactivation timeout" is not user-adjustable unless I pay up for a G4X. I was also told that I cannot see the actual duty cycle that the ECU is commanding unless I pony up for a G4X.

My current suspicion is that it has something to do with my Closed Loop Lambda Gain or Rate settings, but I'd like some smarter people to look at this first before I go on a fishing expedition.

Simon from tech support even came up with the slick idea of having the purge solenoid do its purging only during overrun in order to try to avoid the issue, but what happens there instead are loud explosions again in the exhaust during purge events as the vapor runs through the engine and hits the hot exhaust manifold. So in other words, it simply moved the issue elsewhere.

Some notes regarding my setup:

- Closed loop lambda is averaged of 4 Lambda sensors (Lambda 1,2,3,4) as I have sensors on all 4 cylinders.

- There are 2 backpressure sensors - one for each scroll - AN8 and AN9, so overlay both up on top of each other, and you can see when the vapor is lit up in the exhaust manifold where it shows as a small brief intense cluster of exhaust pressure spikes with no actual boost increase or throttle input at certain time indexes in the log.

- Purge solenoid is on Aux 6, and is PWM controlled. I have a Virtual Aux 1 also set to Purge Solenoid, so to know when purge solenoid is actually venting into the manifold, you would need to overlay both AUX 6 and Virtual Aux 1 and watch where they both are "ON". I have it set up this way primarily to keep the purge solenoid from purging when in 1st gear.

- You might notice that overrun fuel cut is not working in the log - I'm pretty sure its due to my lockout TPS setting of 1%. A setting of 1% is typically fine in cooler weather on my car, but I took this log on one of the hotter days we have had so far this year with ambient temps at around 30C, and I noticed that the throttle cable runs over the hot engine and seems to stretch a little with the heat under the engine bay especially in low speed where there is little airflow into the engine bay (its that or the OEM throttle return spring might be feeling the effects of the under hood heat), so I need to adjust the lockout to perhaps around 1.3% to give some room so that the function continues to work properly when things heat up especially on hot summer days. As a plus though, not having the overrun fuel cut working during the log conveniently helped rule out the closed loop reactivation lockout as the cause of the issue (which I originally suspected).

Screenshot of my datalog to show spikes in EMAP

You can see what I am referring to in the image above occurring at time indexes 1:02, 1:06, and again at 1:18.04. The "bang" can be "seen" as the jump in both EMAP traces at those time indexes. You can also see how the "lean" condition that occurs as soon as the purge solenoid opens when throttle is reintroduced following throttle lift off preceded by a purge event, as well as what the CLL is doing prior and after each event. Again, you can see on all 3 occasions that the purge solenoid was active, and then I let off the throttle briefly to coast, and then went back into the throttle, and then experienced a slight stuttering/hesitation as car ran lean and then BANG (out of the exhaust), and then the car continues as if nothing happened. 

If I turn off the purge solenoid completely via the ECU, the car drives normally, no matter what I'm doing with the throttle. I'd rather keep the purge on though, as I am not a fan of the smell of fuel venting from my gas tank. I can literally smell it inside the car if I roll the windows down and drive slowly. I can also detect other drivers keeping their distance when the purge solenoid is deactivated, no doubt because they can smell raw fuel emanating from the car.

ECU map and log

Any assistance from our wise members would be much appreciated. 

Thanks!

[jigga009]

OK an update - I was able to solve this issue by putting the purge solenoid on a delay timer. 

Seems that purge solenoid activation can catch the closed loop fuel correction out briefly if a second purge event is triggered within close temporal succession to the end of the first purge event. More specifically, if the second purge event occurs while the ECU is still busy pulling fuel in response of the first purge event. 

So by putting the purge solenoid on a 3 second timer delay that starts when the conditions are met for the purge solenoid to be activated (timer starts counting when ECU determines that it is time to purge, and actually activates the purge solenoid 3s later), and resets at the end of the purge event, you now force the ECU to have a minimum delay of 3s between purge events, regardless of whether the purge conditions are met sooner. This allows the ECU a minimum of 3s after the solenoid closes to continue stabilizing fueling before a subsequent purge event can be triggered (at which time closed loop fueling starts pulling the appropriate amount of fuel again). 

[EvocentriK]

Appreciate the detailed explanation and corrective action. I’m about to tackle this myself and you probably saved me some headaches!!