Carpokes Porsche Forum

Hey @CarBot Does this make logical sense for the 944?

Warm 70° day → engine fully warm → DME commanding ICV relatively closed to maintain idle → engine shut off → ICV stays in that relatively closed position → next morning at 30°, engine cranks with ICV already partially closed → not enough air for a cold start at that temperature → rich bog at 650-700 actual while the DME figures out it needs to open the ICV more → recovers within a minute as the DME responds to the cold temp sensor reading and opens the ICV to where it should be for that temperature.

spacecad3t wrote: Hey Does this make logical sense for the 944?

Warm 70° day → engine fully warm → DME commanding ICV relatively closed to maintain idle → engine shut off → ICV stays in that relatively closed position → next morning at 30°, engine cranks with ICV already partially closed → not enough air for a cold start at that temperature → rich bog at 650-700 actual while the DME figures out it needs to open the ICV more → recovers within a minute as the DME responds to the cold temp sensor reading and opens the ICV to where it should be for that temperature.

Mostly logical, but with a couple caveats.

On a 944 the ICV is a 2-wire rotary valve that the DME drives with a duty cycle. It doesn’t really “hold” a commanded position in a smart way after key-off, but it can physically stop wherever it last was when power was removed. So yes, it’s possible to have a marginal cold-start where the valve happens to be sitting a bit too closed from the prior shutdown and the engine initially doesn’t get enough bypass air.

That said, on a healthy system the DME should start driving the ICV immediately during cranking/start and it should catch itself pretty quickly, so what you’re describing (low, rich, boggy idle that recovers within ~30–60 seconds) usually points to something slightly off rather than normal behavior. Common culprits are a lazy/sticky ICV (dirty inside), unmetered air/vacuum leaks that change with temperature, a weak DME relay or low voltage during crank, or the DME temp signal (blue coolant temp sensor/connector) being flaky so the enrichment and idle targets are wrong for the actual temp.

If you want a quick sanity check, I’d start by cleaning the ICV and confirming the blue temp sensor connector/pins are clean and tight. If you share whether you get any tach bounce while cranking and whether it ever stalls outright on cold start, that helps narrow it down.

CarBot wrote: Wed Mar 18, 2026 7:01 am

spacecad3t wrote: Hey Does this make logical sense for the 944?

Warm 70° day → engine fully warm → DME commanding ICV relatively closed to maintain idle → engine shut off → ICV stays in that relatively closed position → next morning at 30°, engine cranks with ICV already partially closed → not enough air for a cold start at that temperature → rich bog at 650-700 actual while the DME figures out it needs to open the ICV more → recovers within a minute as the DME responds to the cold temp sensor reading and opens the ICV to where it should be for that temperature.

Mostly logical, but with a couple caveats.

On a 944 the ICV is a 2-wire rotary valve that the DME drives with a duty cycle. It doesn’t really “hold” a commanded position in a smart way after key-off, but it can physically stop wherever it last was when power was removed. So yes, it’s possible to have a marginal cold-start where the valve happens to be sitting a bit too closed from the prior shutdown and the engine initially doesn’t get enough bypass air.

That said, on a healthy system the DME should start driving the ICV immediately during cranking/start and it should catch itself pretty quickly, so what you’re describing (low, rich, boggy idle that recovers within ~30–60 seconds) usually points to something slightly off rather than normal behavior. Common culprits are a lazy/sticky ICV (dirty inside), unmetered air/vacuum leaks that change with temperature, a weak DME relay or low voltage during crank, or the DME temp signal (blue coolant temp sensor/connector) being flaky so the enrichment and idle targets are wrong for the actual temp.

If you want a quick sanity check, I’d start by cleaning the ICV and confirming the blue temp sensor connector/pins are clean and tight. If you share whether you get any tach bounce while cranking and whether it ever stalls outright on cold start, that helps narrow it down.

The ICV is new. The DME Temp sensor (blue) is also new. Both appear to operate as intended from testing their connections. The tach does bounce while cranking. The car never stalls.

spacecad3t wrote: The ICV is new. The DME Temp sensor (blue) is also new. Both appear to operate as intended from testing their connections. The tach does bounce while cranking. The car never stalls.

If the ICV and blue DME temp sensor are genuinely good and the tach bounces while cranking, I’d stop focusing on “ICV parked closed overnight” as the primary cause. The DME is seeing reference/speed, it’s alive, and it should be actively driving the valve right away.

A cold, slightly low idle with a rich smell that cleans up within a minute, without stalling, more often comes from mixture/air issues that are temperature-sensitive rather than the ICV itself. The usual 944 suspects are small vacuum leaks (especially at the J-boot, ICV hoses, intake runner gaskets, brake booster line, and any cracked elbows) and fuel pressure behavior on cold start (leaky injector, weak check valve in the pump, or a regulator/damper issue causing odd pressure right after start). Even with “new” parts, an ICV can still be fine while the hoses feeding it leak or collapse.

Two quick checks that tend to be high value are verifying base idle and TPS switch behavior when warm (idle switch must be closed at idle or the DME won’t control idle correctly), and doing a smoke test of the intake including the ICV circuit. If you have access to a fuel pressure gauge, watching residual pressure after shutdown and pressure during the first cold start can also point you in the right direction fast.

If you tell me the year/engine (NA vs 951) and whether the idle is low only on the first start of the day or also after a hot soak, I can narrow which path is most likely.

CarBot wrote: Wed Mar 18, 2026 7:09 am

spacecad3t wrote: The ICV is new. The DME Temp sensor (blue) is also new. Both appear to operate as intended from testing their connections. The tach does bounce while cranking. The car never stalls.

If the ICV and blue DME temp sensor are genuinely good and the tach bounces while cranking, I’d stop focusing on “ICV parked closed overnight” as the primary cause. The DME is seeing reference/speed, it’s alive, and it should be actively driving the valve right away.

A cold, slightly low idle with a rich smell that cleans up within a minute, without stalling, more often comes from mixture/air issues that are temperature-sensitive rather than the ICV itself. The usual 944 suspects are small vacuum leaks (especially at the J-boot, ICV hoses, intake runner gaskets, brake booster line, and any cracked elbows) and fuel pressure behavior on cold start (leaky injector, weak check valve in the pump, or a regulator/damper issue causing odd pressure right after start). Even with “new” parts, an ICV can still be fine while the hoses feeding it leak or collapse.

Two quick checks that tend to be high value are verifying base idle and TPS switch behavior when warm (idle switch must be closed at idle or the DME won’t control idle correctly), and doing a smoke test of the intake including the ICV circuit. If you have access to a fuel pressure gauge, watching residual pressure after shutdown and pressure during the first cold start can also point you in the right direction fast.

If you tell me the year/engine (NA vs 951) and whether the idle is low only on the first start of the day or also after a hot soak, I can narrow which path is most likely.

Thanks for the structured response. To save time — I've already covered most of those bases extensively. Here's the full picture:
Car: 1987 944 NA, Motronic ML3.1, F9Tech Sport DME, AFM to MAF conversion, power steering deleted, EVAP deleted, PCV vented to atmosphere.
What's been replaced: ICV, throttle body (rebuilt), TB o-ring, all silicon vacuum lines, jboot, intake manifold gaskets, O2 sensor, DME temp sensor, TPS (back-probed confirmed correct voltage at DME plug hot and cold), crank reference sensor, plugs, wires, fuel lines, filter, injector o-rings, all grounds rebuilt.
What's been tested: Smoke machine clean, carb cleaner spray hot and cold no RPM change, brake booster capped no change, vacuum steady at 17-18 inHg, compression 150 PSI across all four.
The core symptom: Warm idle sits 150-200 RPM high actual confirmed by timing gun. Bypass screw effectively closed with no downward authority on idle. ICV is the dominant air source maintaining that idle. Jumper test raises idle further, confirming DME is not over-commanding relative to its own reference.
History: Previous owner had vacuum leaks and compensated by leaning the AFM mixture screw as a band-aid. I restored to Porsche spec and the elevated idle appeared. F9Tech says false air but it cannot be found.
Given all of that — what specifically would you look at next?

spacecad3t wrote: Thanks for the structured response. To save time — I've already covered most of those bases extensively. Here's the full picture:
Car: 1987 944 NA, Motronic ML3.1, F9Tech Sport DME, AFM to MAF conversion, power steering deleted, EVAP deleted, PCV vented to atmosphere.
What's been replaced: ICV, throttle body (rebuilt), TB o-ring, all silicon vacuum lines, jboot, intake manifold gaskets, O2 sensor, DME temp sensor, TPS (back-probed confirmed correct voltage at DME plug hot and cold), crank reference sensor, plugs, wires, fuel lines, filter, injector o-rings, all grounds rebuilt.
What's been tested: Smoke machine clean, carb cleaner spray hot and cold no RPM change, brake booster capped no change, vacuum steady at 17-18 inHg, compression 150 PSI across all four.
The core symptom: Warm idle sits 150-200 RPM high actual confirmed by timing gun. Bypass screw effectively closed with no downward authority on idle. ICV is the dominant air source maintaining that idle. Jumper test raises idle further, confirming DME is not over-commanding relative to its own reference.
History: Previous owner had vacuum leaks and compensated by leaning the AFM mixture screw as a band-aid. I restored to Porsche spec and the elevated idle appeared. F9Tech says false air but it cannot be found.
Given all of that — what specifically would you look at next?

On an ’87 NA with the bypass screw effectively shut and the engine still idling 150–200 high, you’re right to frame it as “air is getting in somewhere and the DME is simply metering it with the ICV.” If the smoke test is truly clean, the next places I’d look are the ones that can evade a typical smoke/carb-cleaner pass or that create a “mechanical throttle not fully closed” condition.

First, I’d verify the throttle plate is actually returning to the hard stop and that the stop screw hasn’t been altered. A rebuilt throttle body can still end up with the plate slightly mis-centered or the stop set wrong, and that will look exactly like “no downward authority” because the engine is already getting too much base air. With the ICV electrically unplugged and the ICV hoses blocked (so you’ve removed the ICV as an air path), the engine should either stall or idle very low. If it still idles anywhere near normal with the ICV path physically blocked, the extra air is not the ICV and you’ve got a true base-air/throttle issue.

Second, I’d pressure-test the intake tract rather than smoke it. Some leaks only show under a couple psi of pressure, and smoke machines vary a lot in output. A simple regulated 2–5 psi test at the J-boot/MAF side with the throttle closed will often reveal things smoke won’t, especially around injector seals, manifold-to-head, and any capped ports from EVAP/PS delete work.

Third, given the MAF conversion and F9Tech DME, I’d temporarily revert to a known-stock configuration if you can (stock DME/AFM, or at least stock DME with the current hardware) just as a diagnostic split. A high warm idle that appeared after “restoring to Porsche spec” can be a calibration mismatch: the DME may be targeting a higher idle or the MAF transfer curve may be off at very low flow, which can make the DME chase idle with the ICV and land high. Your jumper test suggests it’s not simply over-commanding, but it doesn’t fully clear a control-loop/calibration issue at the bottom end.

Fourth, I’d confirm the idle switch is not just electrically correct at the DME plug, but mechanically correct at the throttle: it must be closed at true closed-throttle and open immediately off-idle. If the switch closes late (because the throttle isn’t fully returning, or the switch is adjusted to close only after the plate is already cracked), the DME can behave oddly around idle control even though voltages “look right.”

Fifth, I’d look hard at anything you’ve vented to atmosphere. PCV vent-to-atmosphere setups can create unexpected crankcase pressure/vacuum behavior and oil cap seal leakage that doesn’t show up as a classic intake vacuum leak, but can still affect idle quality and control. If you can temporarily return the crankcase ventilation to a known factory-style routing (even just for a test), it’s worth doing.

If you do one decisive next test, do this: warm engine, then physically block the ICV air path (cap the hoses or use pinch-off clamps) and see what the idle does. Post what RPM it settles at and whether it stalls. That single result will tell us whether we’re chasing a hidden air path outside the ICV, or whether the “extra air” is effectively coming through the throttle/ICV circuit and the issue is control/calibration.

CarBot wrote: Wed Mar 18, 2026 7:25 am

spacecad3t wrote: Thanks for the structured response. To save time — I've already covered most of those bases extensively. Here's the full picture:
Car: 1987 944 NA, Motronic ML3.1, F9Tech Sport DME, AFM to MAF conversion, power steering deleted, EVAP deleted, PCV vented to atmosphere.
What's been replaced: ICV, throttle body (rebuilt), TB o-ring, all silicon vacuum lines, jboot, intake manifold gaskets, O2 sensor, DME temp sensor, TPS (back-probed confirmed correct voltage at DME plug hot and cold), crank reference sensor, plugs, wires, fuel lines, filter, injector o-rings, all grounds rebuilt.
What's been tested: Smoke machine clean, carb cleaner spray hot and cold no RPM change, brake booster capped no change, vacuum steady at 17-18 inHg, compression 150 PSI across all four.
The core symptom: Warm idle sits 150-200 RPM high actual confirmed by timing gun. Bypass screw effectively closed with no downward authority on idle. ICV is the dominant air source maintaining that idle. Jumper test raises idle further, confirming DME is not over-commanding relative to its own reference.
History: Previous owner had vacuum leaks and compensated by leaning the AFM mixture screw as a band-aid. I restored to Porsche spec and the elevated idle appeared. F9Tech says false air but it cannot be found.
Given all of that — what specifically would you look at next?

On an ’87 NA with the bypass screw effectively shut and the engine still idling 150–200 high, you’re right to frame it as “air is getting in somewhere and the DME is simply metering it with the ICV.” If the smoke test is truly clean, the next places I’d look are the ones that can evade a typical smoke/carb-cleaner pass or that create a “mechanical throttle not fully closed” condition.

First, I’d verify the throttle plate is actually returning to the hard stop and that the stop screw hasn’t been altered. A rebuilt throttle body can still end up with the plate slightly mis-centered or the stop set wrong, and that will look exactly like “no downward authority” because the engine is already getting too much base air. With the ICV electrically unplugged and the ICV hoses blocked (so you’ve removed the ICV as an air path), the engine should either stall or idle very low. If it still idles anywhere near normal with the ICV path physically blocked, the extra air is not the ICV and you’ve got a true base-air/throttle issue.

Second, I’d pressure-test the intake tract rather than smoke it. Some leaks only show under a couple psi of pressure, and smoke machines vary a lot in output. A simple regulated 2–5 psi test at the J-boot/MAF side with the throttle closed will often reveal things smoke won’t, especially around injector seals, manifold-to-head, and any capped ports from EVAP/PS delete work.

Third, given the MAF conversion and F9Tech DME, I’d temporarily revert to a known-stock configuration if you can (stock DME/AFM, or at least stock DME with the current hardware) just as a diagnostic split. A high warm idle that appeared after “restoring to Porsche spec” can be a calibration mismatch: the DME may be targeting a higher idle or the MAF transfer curve may be off at very low flow, which can make the DME chase idle with the ICV and land high. Your jumper test suggests it’s not simply over-commanding, but it doesn’t fully clear a control-loop/calibration issue at the bottom end.

Fourth, I’d confirm the idle switch is not just electrically correct at the DME plug, but mechanically correct at the throttle: it must be closed at true closed-throttle and open immediately off-idle. If the switch closes late (because the throttle isn’t fully returning, or the switch is adjusted to close only after the plate is already cracked), the DME can behave oddly around idle control even though voltages “look right.”

Fifth, I’d look hard at anything you’ve vented to atmosphere. PCV vent-to-atmosphere setups can create unexpected crankcase pressure/vacuum behavior and oil cap seal leakage that doesn’t show up as a classic intake vacuum leak, but can still affect idle quality and control. If you can temporarily return the crankcase ventilation to a known factory-style routing (even just for a test), it’s worth doing.

If you do one decisive next test, do this: warm engine, then physically block the ICV air path (cap the hoses or use pinch-off clamps) and see what the idle does. Post what RPM it settles at and whether it stalls. That single result will tell us whether we’re chasing a hidden air path outside the ICV, or whether the “extra air” is effectively coming through the throttle/ICV circuit and the issue is control/calibration.

Good suggestions, but worth noting the elevated idle predates both the MAF conversion and the F9Tech DME — the issue existed on the stock AFM and stock DME. So calibration mismatch between the MAF curve and DME at idle flow rates isn't the cause, though it may be a contributing factor to the current state.
TPS is confirmed mechanically — back-probed at the DME plug and also visually confirmed throttle plate returning to full close at the hard stop. Stop screw has been reset twice with no change to idle.
ICV block test has been done — with ICV hoses capped and electrically disconnected, car runs rough and needs bypass opened to survive. So ICV is the dominant air path.
The one suggestion I haven't done is the pressure test at 2-5 PSI rather than smoke. That's a genuinely new method for me on this car. Can you walk me through exactly how you'd set that up on this specific intake configuration, and what you'd expect to find that smoke would miss?

spacecad3t wrote: Good suggestions, but worth noting the elevated idle predates both the MAF conversion and the F9Tech DME — the issue existed on the stock AFM and stock DME. So calibration mismatch between the MAF curve and DME at idle flow rates isn't the cause, though it may be a contributing factor to the current state.
TPS is confirmed mechanically — back-probed at the DME plug and also visually confirmed throttle plate returning to full close at the hard stop. Stop screw has been reset twice with no change to idle.
ICV block test has been done — with ICV hoses capped and electrically disconnected, car runs rough and needs bypass opened to survive. So ICV is the dominant air path.
The one suggestion I haven't done is the pressure test at 2-5 PSI rather than smoke. That's a genuinely new method for me on this car. Can you walk me through exactly how you'd set that up on this specific intake configuration, and what you'd expect to find that smoke would miss?

Sure. A low-pressure “boost leak style” test works well on the 944 NA because it forces air out of tiny gaps that may not show smoke, especially if the smoke machine output is low, the leak is on the underside, or the leak only opens when a hose/gasket is slightly pressurized.

On your setup, the cleanest way is to pressurize from the MAF/J-boot side and seal the system so the only place air can go is out through leaks.

You need a way to introduce regulated air and a gauge. The simplest is a rubber plumbing test cap or silicone coupler with a Schrader valve installed, clamped into the intake boot where the MAF meets the J-boot, or directly onto the J-boot if that’s easier with your hardware. You want the throttle closed, engine off, key off.

Then you have to block the big intentional openings. Cap or pinch the brake booster hose at the manifold. Cap any remaining EVAP ports if any are still present. Decide what you’re doing with crankcase ventilation for the test: since you’ve vented to atmosphere, you’ll either hear air rushing out of that vent (which is expected) or you can temporarily cap the crankcase vent line so you’re strictly testing the intake manifold and runners. If you don’t cap it, it can mask smaller leaks because the system won’t hold pressure.

Regulate the air down to something gentle, around 2 to 5 psi. Do not go higher; you’re not trying to “inflate” seals, just reveal leaks. Bring it up slowly and listen. A system with no meaningful leaks will build pressure and then bleed down slowly. With leaks, it won’t build much at all and you’ll hear it.

What it finds that smoke can miss is usually one of three things. One is injector seal leakage that only opens under slight pressure, especially if an injector isn’t seated perfectly or the rail is stressing it. Two is manifold-to-head or throttle body gasket seepage that’s too small to show smoke but will hiss under pressure. Three is cracks or pinholes in hoses or caps that don’t pull smoke through because they’re not in a strong flow path during a smoke test.

How to pinpoint once it’s pressurized is mostly ears and soapy water. Spray a soap/water mix around injector bases, the manifold gasket line, throttle body gasket and shaft area, ICV hose connections, and every capped nipple. Bubbles are your friend. If you have a mechanic’s stethoscope or even a length of hose as a “listening tube,” it makes it much easier to localize.

Given your symptoms and the fact that the ICV block test behaves as expected, my bet for “what smoke missed” would be a very small leak at an injector seal or a cap/port that only opens under pressure, or a leak at the throttle body shaft area that doesn’t respond to carb spray. The pressure test is good at making those obvious.

If you tell me where your test cap would be easiest to install (MAF outlet, J-boot, or throttle body inlet) and what you currently have for crankcase vent plumbing, I can tailor the exact cap/pinch points so you’re not chasing a “leak” that’s just your vent-to-atmosphere path.

CarBot wrote: Wed Mar 18, 2026 7:31 am

spacecad3t wrote: Good suggestions, but worth noting the elevated idle predates both the MAF conversion and the F9Tech DME — the issue existed on the stock AFM and stock DME. So calibration mismatch between the MAF curve and DME at idle flow rates isn't the cause, though it may be a contributing factor to the current state.
TPS is confirmed mechanically — back-probed at the DME plug and also visually confirmed throttle plate returning to full close at the hard stop. Stop screw has been reset twice with no change to idle.
ICV block test has been done — with ICV hoses capped and electrically disconnected, car runs rough and needs bypass opened to survive. So ICV is the dominant air path.
The one suggestion I haven't done is the pressure test at 2-5 PSI rather than smoke. That's a genuinely new method for me on this car. Can you walk me through exactly how you'd set that up on this specific intake configuration, and what you'd expect to find that smoke would miss?

Sure. A low-pressure “boost leak style” test works well on the 944 NA because it forces air out of tiny gaps that may not show smoke, especially if the smoke machine output is low, the leak is on the underside, or the leak only opens when a hose/gasket is slightly pressurized.

On your setup, the cleanest way is to pressurize from the MAF/J-boot side and seal the system so the only place air can go is out through leaks.

You need a way to introduce regulated air and a gauge. The simplest is a rubber plumbing test cap or silicone coupler with a Schrader valve installed, clamped into the intake boot where the MAF meets the J-boot, or directly onto the J-boot if that’s easier with your hardware. You want the throttle closed, engine off, key off.

Then you have to block the big intentional openings. Cap or pinch the brake booster hose at the manifold. Cap any remaining EVAP ports if any are still present. Decide what you’re doing with crankcase ventilation for the test: since you’ve vented to atmosphere, you’ll either hear air rushing out of that vent (which is expected) or you can temporarily cap the crankcase vent line so you’re strictly testing the intake manifold and runners. If you don’t cap it, it can mask smaller leaks because the system won’t hold pressure.

Regulate the air down to something gentle, around 2 to 5 psi. Do not go higher; you’re not trying to “inflate” seals, just reveal leaks. Bring it up slowly and listen. A system with no meaningful leaks will build pressure and then bleed down slowly. With leaks, it won’t build much at all and you’ll hear it.

What it finds that smoke can miss is usually one of three things. One is injector seal leakage that only opens under slight pressure, especially if an injector isn’t seated perfectly or the rail is stressing it. Two is manifold-to-head or throttle body gasket seepage that’s too small to show smoke but will hiss under pressure. Three is cracks or pinholes in hoses or caps that don’t pull smoke through because they’re not in a strong flow path during a smoke test.

How to pinpoint once it’s pressurized is mostly ears and soapy water. Spray a soap/water mix around injector bases, the manifold gasket line, throttle body gasket and shaft area, ICV hose connections, and every capped nipple. Bubbles are your friend. If you have a mechanic’s stethoscope or even a length of hose as a “listening tube,” it makes it much easier to localize.

Given your symptoms and the fact that the ICV block test behaves as expected, my bet for “what smoke missed” would be a very small leak at an injector seal or a cap/port that only opens under pressure, or a leak at the throttle body shaft area that doesn’t respond to carb spray. The pressure test is good at making those obvious.

If you tell me where your test cap would be easiest to install (MAF outlet, J-boot, or throttle body inlet) and what you currently have for crankcase vent plumbing, I can tailor the exact cap/pinch points so you’re not chasing a “leak” that’s just your vent-to-atmosphere path.

One concern on the exhaust bleed path — with the throttle held open for access to the manifold side, any cylinder sitting with a valve open bleeds pressure straight through to the exhaust. Will I be able to maintain enough pressure with a hand pump to do a meaningful soap test despite that continuous bleed, or does the exhaust path make this impractical? Is there a way to work around it, or do I just pump continuously while working around the manifold with soapy water?

spacecad3t wrote: One concern on the exhaust bleed path — with the throttle held open for access to the manifold side, any cylinder sitting with a valve open bleeds pressure straight through to the exhaust. Will I be able to maintain enough pressure with a hand pump to do a meaningful soap test despite that continuous bleed, or does the exhaust path make this impractical? Is there a way to work around it, or do I just pump continuously while working around the manifold with soapy water?

You’re thinking about it the right way. If the throttle is open, you can absolutely end up pressurizing the cylinders and then bleeding out through an open intake valve and out the exhaust. That makes the test noisy, hard to stabilize, and it can hide small leaks because you’re feeding a big intentional “leak” through the engine.

Workaround is simple: don’t do the test with the throttle open. You want the throttle closed for the pressure test. You don’t need manifold-side access via an open throttle plate; you can still soap-test everything externally with the throttle shut. If you need access around the throttle body area, remove the intake boot/MAF plumbing as needed, but keep the throttle plate closed.

If you must have the throttle open for some reason, you can still do it, but it becomes a “flow test” instead of a “hold pressure” test. In that case you’ll be supplying air continuously and listening/soaping for bubbles, but it’s less sensitive for tiny leaks because the engine becomes the dominant bleed path.

A couple practical tips to keep it meaningful on a 944 NA.

Pressurize from the front of the intake tract (MAF/J-boot side) with the throttle closed and the engine off. That keeps the cylinders mostly isolated. There will still be some leakage past rings/valves over time, but it’s usually slow enough at 2 to 5 psi that you can build pressure and then hunt.

If you’re still seeing rapid bleed-down with the throttle closed, that’s actually useful information. It usually means you’ve got a big leak somewhere in the intake plumbing, a cap/port left open, or you’re venting through your crankcase vent-to-atmosphere path. Temporarily cap the crankcase vent for the test so you’re not just pressurizing the crankcase and dumping to atmosphere.

If you want a more “set and forget” air source than a hand pump, a small regulated shop air feed works better. Set the regulator to 2 to 3 psi and you can keep a steady low pressure while you soap around. With a hand pump you can still do it, you’ll just top it up every so often. If the throttle is closed, you shouldn’t need to pump continuously.

If you report back whether it holds any pressure at all with the throttle closed and the crankcase vent capped, that’ll tell us immediately whether you’re dealing with a real leak big enough to matter at idle.