Rough idle when cold, but only for about 10 seconds

I'm trying to figure out a problem with my 02 Targa. When I start the engine in the morning it idles normally for about 10 seconds and then starts idling very rough for about 10 seconds and then behaves normally. The car is otherwise running extremely well without any other sign of problems When the problem is occurring: the RPM drops from about 1100 to ~700 the car shakes and seems like it is misfiring heavily it almost stalls The problem only occurs after sitting for a long period (either overnight or after a day at work). Given that the problem doesn't occur when warm I assume it is due to the engine's cold start values or possibly the SAI. The problem only started this spring after I had performed quite a bit of maintenance on the car so my current assumption is that I broke something. This winter I had replaced my water pump, ignition coils and plugs, and oil-fill tube (which required removing the alternator and SAI pump). So far I have confirmed the following the SAI pump is running at startup in these conditions and I assume it is blowing air I actually forgot/lost the clamp for the SAI hose but have replaced it with a regular hose clamp using a generic scan tool I have verified that the engine coolant temperature and intake temperatures are reasonable. Although after sitting overnight in my garage the intake temperature always seems to be about 5 degrees F above the coolant temperature mass airflow sensor is fairly stable at 1.5 to 1.8 lb/minute and it increases when I hit the throttle the dash shows a drop in voltage from ~14 to ~12 (not quite sure how low) when the problem occurs. My guess was this was the SAI kicking in. But the voltage measured from the scan tool looks constant From the scan tool history (attached) I have noticed the following the problem starts about 10 seconds in at (19 seconds in the CSV) the throttle position increases from 5-7 to 14 at this time even though I haven't touched the accelerator MAF values seem stable until I touch the throttle at about 44 seconds fuel rate increases from ~1.1 to almost 1.5 when the problem occurs My guess at this point is that one of the vacuum hoses near the alternator were broken since I had to move them when removing the alternator. None of them have any visible damage that I can see. I have attached a data log if anyone is willing and able to take a look. Hopefully someone else has debugged this problem in the past although I couldn't find anything after a lot of searching. Any idea what to try next? My next option is probably to bring it in to a professional but that would be admitting defeat after all of the DIY I performed this winter Unfortunately it takes a while to debug since I basically get one chance per day to try to diagnose.

Hoses can have a crack/split without showing any real signs of damage. Older hoses do not take kindly to being moved about. It was SOP when I worked on an older engine to replace the vacuum hoses with new ones. I kept on hand in the garage a supply of various vacuum hose material to have handy. Likewise I kept a supply of gas line hose for use when I did any work on the fuel system of a car. My advice would be to consider just replacing the vacuum hoses with new hose. If the vacuum hoses are special and not off the shelf you might want to run the hoses looking for damage. You can do this with the engine cold and off with a bright light and a mirror on a stick. You can also do this -- if you are very very careful -- while the engine is running. You pick the hose up, move it about a bit -- nothing violent -- to see if any movement causes the engine to react. However, if the hoses are old enough this can damage them so you might just want to replace them. Before you even touch the engine though consider performing a recalibration of the e-Gas. The steps to do this should be in the owners manual. For my 02 Boxster turn the ignition key off. Then turn it on and leave it on for 60 seconds. Do not turn it far enough to start the engine. Do not touch the gas pedal. After 60 seconds, turn the key off and leave off for 10 seconds then start the engine. Oh, I take it no CEL?

Too early to admit defeat One thing jumped out on me is "the dash shows a drop in voltage from ~14 to ~12 (not quite sure how low) when the problem occurs. My guess was this was the SAI kicking in. But the voltage measured from the scan tool looks constant" Almost sounds extra load not accounted for on the engine causing the stumbling so the throttle opens up to compensate for that to maintance idle speed. Sounds like the SAI pump is loading up the alternator somehow (but not sure why not at the beginning when the pump starts running but rather at 10s). My suggestion is to unplug the SAI pump in the morning before you start the car and see if you have the same problem. You may get a CEL due to SAI disconnect but that can be reset easily with the scan tool later.

In looking over the data I have a couple of questions: 1) Where are all the rest of the engine coolant temp readings? 2) Where are all the rest of the air intake temp readings? 3) Short Term Fuel % Trim - Bank 2 - I would not expect to see 0% readings. 4) Long Term Fuel % Trim - Bank 2 - I would not expect to see such dramatic swings. Based on my observation of these in various cars the readings almost looked swapped, the headings reversed. 5) The same applies to the Short and Long Fuel % Trim values for Bank 1. 6) I see no spike in load or fueling rate at the time of the RPM drop. 7) The change in absolute throttle position could be interpreted as e-Gas attempting to vary the throttle opening to maintain idle speed. 8) The only time I've seen a voltage drop or level below approx. 13.5 to 14 volts in my 996 Turbo is when I start the engine after a car wash or after the car has sat out in the rain. The belt slips and sometimes is quiet other times chirps a bit and slips until the water on the belt and pulleys dries or is slung away. If the battery is low the engine may not run right as marginal battery voltage is not generating for instance a strong spark. If you tell me the engine is dry then I might suspect the belt is bad or there is a belt tensioner problem and it is not keeping the belt tight enough. Might be worth checking the belt's condition at least.

Quotes source post: 11304037

Calculated load jumped from 4.7 to 6.3 when rpm changed from 1040 to 739, at 17.396s and 19.779s

Just because you can "hear" the SAI, doesn't mean that it's good. Some actually make more noise as they are failing. Have you taken off the hose and when it kicks on and confirm that it's blowing air out? The extra load could be from the SAI bringing it down as it may be grinding or on it's way to eventually sticking, thus placing more load on your electricals.

Wow. Thanks for all of the feedback. I definitely learned something new. I only have a few minutes to reply, but with any luck the recalibration of the e-gas was the problem. I wasn't aware of the need for this and I still can't find any mention of it in the user manual. But I just tried it and could hear the throttle intake adjusting for about 30 seconds. I started the car and the problem didn't appear. I don't think that it is conclusive yet though since I had started it already about 4 hours ago and the temperature outside has gone up. I'm not sure what the cold start temperatures are and at which temperature the SAI is required. I pulled out and greased my SAI last year (and blew out the dust) since it was getting noisy at startup so it isn't unreasonable to think that it is possible the motor is starting to seize. But I can feel the motor spinning when cold so I assume it is moving air. As for replacing hoses, it isn't an easy task since most would require pulling out the alternator again. But I have verified that the obvious ones are holding a vacuum simply by unplugging them after the engine was running. For the missing temperature values in the CSV...I'm not sure. That was the output from the scan tool and I'm not sure why they stopped including temperatures. But from the initial data I assume the sensors are working. With any luck the e-gas calibration will have solved the issue. Unfortunately I can't tell until tomorrow morning. After that I'll try disconnecting the SAI. Thanks :-)

Just an update for today. I started the car this morning with the scan tool attached. It is definitely better than yesterday but not completely solved. Although it may be back to what it was before the winter and my DIY repairs. The cold idle wasn't quite perfect before this winter but it wasn't localized to such a short time period and I assumed it was possibly due to installing 200 cell cats. The behaviour today had the same characteristics as yesterday except the drop in idle wasn't nearly as significant (from ~1050 to ~900 instead of from ~1050 to ~700) and it recovered faster (3-4 seconds instead of 5-7 seconds). There was still a drop in voltage visible from the dashboard gauge but not from the scantool output. I watched the gauge closer this time and the drop is definitely not down as low as to 12 volts. Maybe from ~14v to ~13v. The drop occurs in line 37 and 38 in the CSV. This time I didn't remove all of the time columns. The scan software (OBDwiz) records the time in the CSV for every metric. The things I noticed in this scan RPM dropped from 1058 to 906 O2 voltage (Bank 1 sensor 1) appears to have dropped from .245 to .22 O2 voltage (Bank 2 sensor 1) appears to have dropped from .28 to to .18 MAF increased from 1.5 lb/min to 1.7 lb/min fuel rate increased from 1.0 gal/hr to just over 1.12 gal/hr From the feedback above my guess is that drop occurs when the SAI kicks in and the computer doesn't have quite the right settings to adjust to the fuel flow when that happens. The e-gas calibration seems to have helped, but it might take a few more days to know for sure. Thanks Macster! Is it normal for the RPM to drop by ~150 when the SAI kicks in? It seems reasonable to me. Hopefully the e-gas calibration is the only fix required here. I'll post again after my next cold start.

Quotes source post: 11304123

I do not have the data spreadsheet in front of me but the load is as you correctly stated calculated. The throttle I suspect contributed to that reading.

Without reading ALL of the specifics in depth, I would suspect a leaking injector partially filling a cylinder with fuel while sitting. This would cause a misfire that occurs only on cold start and would in turn clear up as the excess fuel was burned off. If you can see a specific cylinder showing more misfire counts than the others, this may give you a clue as well as the fuel trim numbers for the bank that contains the problematic injector. If I'm missing something, feel free to clue me in.

Quotes source post: 11305559

For all models I can't say but for my Boxster the SAI is running as soon as the engine starts. That is as soon as the engine cranking noise subsides I can hear the SAI pump running in the background. The Turbo doesn't appear to have a SAI delay or lag either. The Turbo has a bit of a stumble at approx. 3 seconds after engine start. I sort of think this is when the switch over to low lift takes place. Since this is done with oil I think the DME waits a few seconds after engine start to ensure there is sufficient oil pressure and flow so the oil that is used for this doesn't result in starvation to the rest of the system. But once the switch over takes place -- it happens in an instant -- the engine is a bit rougher running but nothing scary. And as the engine idles and warms up the idle smooths out and when the engine RPMs drop the engine's idle is as smooth as it ever is. I detect no increase in RPMs when the SAI kicks off either. The SAI pump shut down is without any drama. Save for the noise going away I couldn't tell you by what the dash tells me, or my ears tell me, when the SAI pump shuts down.

Interesting. I'll try again tomorrow to see if the SAI starts immediately. There is definitely a brief drop in voltage and a change in RPMs about 7-10 seconds after the car is running. I've verified this with the A/C disabled just in case the ECE has code to enable the compressor or something similar at startup. My SAI is (now) pretty quiet so I can't actually hear it over the engine. I have been assuming it was part of the issue since it is an obvious component that would force the DME to adjust. To answer Ptech1, I doubt it is due to a leaky injector since the car runs fine for the first 8-10 seconds after startup. I would think any gas in the cylinder or exhaust would be burned off by then unless it impacts the O2 sensor somehow. However I have done 3 cold starts since the e-gas recalibration and I haven't seen the problem since. I still see the voltage drop at 8-10 seconds and the RPM still drops, but it now only drops from ~1100 to ~1000 or so. I think/hope the problem is solved although I'm still curious on how the e-gas calibration would have impacted this.

Sorry, missed the 10 seconds normal, then misfiring, etc... after glancing over everything else.

Just an update for anyone watching (or searching in the future). It has been over a week and the problem hasn't resurfaced. So I'm willing to conclude that the problem was just the e-gas recalibration. Thanks to Macster for mentioning it. I couldn't find any mention of that in my manual though which is a bit strange and it is also very odd that the problem only surfaces when the voltage drops at about 7-10 seconds after starting the car. I believe I have found the source of the voltage drop as well. I can hear the radiator fans starting up at the same time as the voltage drops on the meter on the dash. The interior lights dim as well so I'm sure the drop is real but I haven't started to debug yet. The fans always start about 7-10 seconds after the car has started independent of the temperature. Is it normal for the fans to draw enough current to visibly drop the the voltage on the dashboard guage? It not I assume the problem is one (or more) of the following Radiator fan motors have a partial short causing a significant increase in load (current) Poor connectivity between the battery and the fans (but before the voltage guage) Failing battery (it is about 4 years old but there are no other symptoms) Failing alternator (probably unlikely)

That's good progress. I think the more intriuging question is why the DME turns on the radiator fans. I checked your log again but those coolant reading at e missing. If the fans are on high, a little voltage drop does not sound abnormal. A failing fan can also draw more current than it should but still why the DME turns on the fans?

Thanks Ahsai. I checked it out this morning with the ODB scanner. Both the coolant and outside air temperature sensors were reporting reasonable temperatures (ie 60F) even when running. But I then realized that the AC would also trigger the front fans and sure enough the fans stopped after disabling the AC on the climate control. So my new theory is that the DME is starting the AC compressor and the front fans when cold (maybe to prevent windows from fogging or something) which is causing both a drop in voltage and an increase in load due to the A/C compressor. Somehow the increase in load must have triggered the drop in RPM before I recalibrated the e-gas. I'm still a little (pleasantly) surprised that the e-gas calibration resolved it. But assuming the theory is correct I think this is solved.

Understood now. You left out the detail that the ac was on before you start the car

Actually I didn't realize it was on. I just leave it in auto all the time. Disabling the AC on the climate control turns it 'off' auto and I think that disables the automatic adjustment of the fan. My guess is that enabling the A/C to dry out the air on a code start is probably a feature :-)