What's the idle valve signal look like?
 

Rough idle on my 3.2 - surging and dying.

I've been through every test on the engine switches, sensors, fuel pressure, vacuum, AFM, etc., all to spec. Only possible exception is the crank speed sensor, which is showing a perfect trace during cranking but only 2V versus the 2.5V minimum. HOWEVER... the DME test plan says that 2.5V is with the engine running, and I'm only cranking at 200revs. I can see amplitude is increasing with revs so I'm confident it's within spec, at idle speed.

Swapping idle valve makes no change, and idle valve is vibrating at idle.

So I pulled the plug on the idle valve and scoped the signals with the ignition on. 12V centre is perfect. One end wire is giving a square wave with 50% nonzero and the other end wire is giving me nothing - 100% zero.

Does anybody know what these signals to the end-pins of the ICV should look like, and does this sound like one of the signals is missing? I expect there should be a square wave on both the open and close terminals, not just one of them.

Others will probably shime in and confirm or not, but I think you should indeed see pulses on both sides. So if one side is dead, this might be the sign of a dead transistor inside Motronic. One of the ON588 Darlingtons referred as T560 or T561 on the schematic.

I have a scope and an 88 Carrera. If I find time today I'll check it and report.
What's the frequency of the square signal?

I didn't catch the frequency and forgot to take a photo. Actually my daughters appeared and asked me to take them to play tennis and how could I refuse so I packed up in a rush.

But at a guess based on my memory of the scope settings and how many squares on the display I'd say it was maybe 50Hz.

But it was 50:50 on and off if that makes sense. This was with the car idling and the plug pulled off. If I get a chance I'll check again tonight.

I have to completely remove my seat to get to the DME otherwise I'd check it at the DME and look for any visible circuit problems.

Sent from my SM-G988B using Tapatalk

Hi, I scoped both together versus the engine as ground, it shows 0.5V pulses on one and not on the other. I was hoping to check it at over- and under- revs, to see if the other started showing a signal but before I could the engine speed started fluctuating wildly. I suspect the ICV piston was oscillating within the valve.

https://uploads.tapatalk-cdn.com/202...687bff1eeb.jpg

Sent from my SM-G988B using Tapatalk

the idle control should have a continuous 12V power supply and then two pulsed grounds controlled by the Motec. pin numbering at the valve for some reason is 3, 4 and 5.
Pin 4 should have constant 12V controlled by the DME Relay.
Pins 3 and 5 should be ground, supplied and controlled by the Motec.

When you are adjusting the base idle on the system it has you place a jumper between pins B-C in the round test plug located in the engine compartment electrical panel. My understanding is that when the jumper is in place it should drive the idle valve to the center position by supplying a 50% duty cycle to ground on the 3 and 5 pins at the idle control valve. for your testing purposes you may want to jumper pins B-C then use your scope to evaluate the signal. this will eliminate any engine performance issues from the idle control signal and let you evaluate the Motec independent to engine running condition.

Normal! One side needs to see the drive signal from the other side when connected, i.e. the signals are 180 degrees out of phase.
One transistor gets its drive signal from the other side. A better test is to back-probe the connector when connected with the key-on.

Read here for a idle valve test; Idle Valve Testing, post #19, http://forums.pelicanparts.com/porsche-911-technical-forum/360766-bench-testing-carrera-idle-control-valve.html

I had this on a 964 (same as a 3.2L) engine once. What finally saved me was when I had enough guessing and just bought a Air Fuel Meter, then welded a O2 sensor bung to the cat converter and plugged it in. The Air Fuel meter told me that I was running way too rich, and I adjusted the Air Flow Meter to bring it down and then it was perfect.

Stop the guessing.

What does "stop the guessing" refer to? To my eye, I'm working systematically through the DME test plan.

Sent from my SM-G988B using Tapatalk

Ok thanks for that suggestion. I was reluctant to pierce the insulation if unnecessary due to future wire corrosion.

Re the valve, I should have originally stated that I already bench tested the icv as working and also switched for another unit as per the DME test plan.

Sent from my SM-G988B using Tapatalk

Thanks. Worked a dream. Looks like I could be honing in on the problem. The icv bypass wire did indeed switch it to 50% duty cycle, but the dead terminal continues to be dead. I'll confirm continuity of the wire from the ecu next and assuming that's good, test the ecu output transistor.



Sent from my SM-G988B using Tapatalk

Here is the part of the diagram that can be of interest for you to troubleshoot these ICV output signals.

ADV11 is control signal from processor.
Both signals are output on pins 33 and 34.

So as stated before, one transistor (here T561) gets its control from the other transistor (here T560).

Now you need to find out why you only get one of these and not both. May be T561 is shot. http://forums.pelicanparts.com/uploa...1688315892.png

Again, you're failing to understand the signals. You only see a pin 34 signal (0.50V) of T560 because of the base to collector junction of Darlington transistor and signal thru R560 from the logic. The pin 33 Darlington transistor does not have the same base signal as the pin 34 Darlington. Both outputs (pins 33/34) NEED pull-up resistors (or the valve) as mentioned before to properly check the idle valve outputs!!!!!


http://forums.pelicanparts.com/uploa...1688336489.jpg

Study the schematic in post #11 to understand your scope images.

You basically answered your own question, initially! T560 needs a pullup resistor or the valve, which will provide the necessary base drive for T561 resulting in a waveform like T560 (.50V square wave).
He just needs to add two 1K resistors to simulate the valve at the connector to properly test both idle signals properly. It's not that difficult

Yes I did understand your point about the resistors the first time you stated it. I will deal with that. Thanks for your help.

Sent from my SM-G988B using Tapatalk

I did a little more on this today - ordered a couple of 1.5k ohm resistors as suggested, popped out the passenger seat, confirmed continuity of the pin-34 wire, and confirmed continuity to the transistor on the board. Inspected the board and pins for cracks and overheating, especially around T561. tbh I was disappointed the wire was continuous because I was hoping that would be the fault - hey ho, you can't win them all.

I attempted a test of the transistors in situ. T'internet said to diode test from base to collector and base to emitter, with positive on the base, and look for about half a volt drop. Well that didn't work with my multimeter, which was completely unresponsive - until I double-checked my assumption that the base was in the middle as per the circuit diagram and discovered it's at the end.

So with the positive on the base, it showed .545V and .619V drop across the two pins on T561 and comparable on T560 too. So all looks good there tbh, although to read too much into that basic test, and with the transistors in situ in the circuit, would be guesswork in my view.

All in all, having inspected thoroughly, tested a good many circuit board joins for continuity and checking resistors etc., I was unable to identify any defects so next step is to repeat scoping the ICV signal across reference resistors when they arrive.

Actually, the key test is from the collector to the base/emitter, collector positive, using the ohms scale (ECU not powered).
The transistor typically fails collector to the emitter.

Thanks for the advice. I can't really test ohms in situ due to the other components in circuit. The two transistors were giving different readings to each other across the other pin combinations. One reading was suspect but I'm not about to work our the impact of the other circuitry.

If I remember rightly, one was unchanged by reversing polarity which was the reading I found suspect.

Anyway, I fabricated the load resistor circuit this evening out of the resistors and a spare sensor plug so can retest the circuit at idle next.

Sent from my SM-G988B using Tapatalk

Can't find the datasheet for the ON588, but internal schematic shows a diode between C and E, so that may impact what you are trying to ohm.
I don't like checking resistance of components in situ, too risky for sensitive parts (probably not here) and a serious risk of wrong results because of other components interfering with what you are trying to check.
When I was a teen and diode mode was not yet available on multimeters, I used to roughly check NPN transistors by ohming between B and C and then between B and E in one direction, then in the other (swapping black and red probes). One direction would yield partial continuity while the other would show a very high resistance or even infinity. This was a rough indication of whether the transistor was more or less ok or completely shot. In many cases it seemed ok but was still shot for linear applications like audio.

Careful use of a VOM won't damage transistor checking! Here's the replacement idle driver;

http://forums.pelicanparts.com/uploa...1689091125.jpg

Because you are using 1.5K resistors, & not 22 ohms per side of the actual valve to simulate the idle valve, the scope signals will not both rise to 12V;

http://forums.pelicanparts.com/uploa...1689125584.jpg

Thanks for the tip!
Do you know if TIP12x and ON588 have comparable internal resistors? Have you had the opportunity to measure ON588 resistors from outside? If so, what are their values? Same as for TIP12x, i.e. 7 kohms and 70 ohms?

As for the test of Motronic output signals with pull-up resistors as a substitute for the ICV, why not use 150-ohm resistors? This would be closer to the actual 22 ohms and would only require 1-watt resistors, or even less, considering a 50% duty cycle. Only a suggestion. Maybe there's a good reason to use 1k or 1.5k resistors instead.

A TIP120 is a TIP120! You always go by the datasheet values. Unless you have a very very critical design, do you start to measure every little semiconductor parameter.

The lower the resistor value, the more you need to consider its power rating. A value near 1K is an easy first choice and available, and it provides the needed results.

The scope only draws a tiny current and it measures the voltage, so it can theoretically measure across an infinite resistance. It's generally whether the resistance is too small you would worry about, so as not to draw more current than the circuit can handle.

Although the icv may have lower resistance than 1.5kohms in a dc test, it's a coil and its supplies are switched and therefore AC. This brings the inductance of the coil into play and therefore its impedance at that frequency would determine the current flowing rather than its pure resistance, and its impedance will be higher than its resistance giving a lower current. I haven't calculated by how much but a using higher resistance in its place can help make up for the discrepancy.

Sent from my SM-G988B using Tapatalk

Actually, the ICV windings are essentially resistive, i.e. about 20 ohms per windings. So the Darlingtons just see a peak current of about .60 amps.
When using a scope with the valve connected, the signal is a pure resistive square wave, i.e. no inductive component.

Just to square up our mutual understanding, every coil has an inductance including the icv (in fact including any simple wire) and at the 10khz at which the circuit is switching, it requires an inductance of about half a henry to add 30 ohms to the impedance. So I assume your claim is that the inductance of the coil is much less than half a henry, correct?

Sent from my SM-G988B using Tapatalk

10kHz? Are you sure?
Your scope pic posted earlier shows 10ms/div. Signal period appears to cover 1.2 div, so that would be roughly 80 Hz.
As for the inductance that comes into play, don't really know about that, but the ICV behaves more or less like two small DC motors being pulsed.
Anyway it is not AC, I mean no sine wave here.

Right the plot thickens...

Is I mentioned I had a spare socket so I set the scope up across 1.5kohm reference resistors and it makes sense to use the 12V common as earth and read the switched earths as negative voltage.

https://uploads.tapatalk-cdn.com/202...c1015d57b9.jpg

Initially baffled myself having one of the probes set to 10X.

https://uploads.tapatalk-cdn.com/202...46e792f3de.jpg

Anyway, I did a variety of tests with the engine running and not running, with the icv valve bypass pins bridged and not bridged, and with the throttle adjusted to high, low and normal revs.

First thing that's apparent is that at all times, one terminal is getting the full -12V, I.e. switched to Earth while the other is getting much less, I.e. switched only to circa -7V. I switched the probe leads to confirm the difference was in the car.

Next, with the engine not running, we get perfect 50% duty cycle as seen here:


https://uploads.tapatalk-cdn.com/202...786e595a5e.jpg

Now bear in mind both the icv bypass connector and the throttle closed position switch are both definitely working, because the engine responds noticeably to both (in fact i can fix the surging by disconnecting the throttle closed valve). Also you can watch the motronic adjusting the timing with a timing light, when the throttle is closed.

However the duty cycle doesn't respond to engine speed. It stays at these percentages whether you manually bypass the throttle to get 500 revs, 880 revs, or 1100 revs:

https://uploads.tapatalk-cdn.com/202...d6c6966dda.jpg

My expectation was that the pulses would widen and narrow with engine speed.

This is perplexing because the engine speed sensor is within reference values and I scoped it at cranking speed.

So unless you have other suggestions I think the next steps are to recheck the flywheel speed sensor, then make sure its signal is getting to the ECU, and finally it might be a circuit defect in the ECU itself not reacting to engine speed.

Sent from my SM-G988B using Tapatalk

You're right I misread the scale so I'm out by a factor of 100. So we would need the inductance to be much less than 50 henry for the impedance to be very close to the resistance.

Square wave is still AC, in fact it has higher root mean square power than the same sine wave by a factor of sqrt 2.

Sent from my SM-G988B using Tapatalk

Responding to your post above with new scope pics.

First, the difference in voltage (7 volts vs 12 volts) was expected, as explained and illustrated by Dave on previous page of this post. That's because you are using high load resistance values and not real low values, which would require 5-watt resistors.

Also, I don't quite understand what you are trying to test or do. If you disconnect the idle switch on the throttle or bypass the ICV by jumping terminals B and C on test socket, in both cases the Motronic stops adjusting duty cycle. It will stop controlling the ICV. My guess is that it will keep sending two 50% duty cycle opposite signals. And that's what you see on your scope.
If you want to see the Motronic in action, i.e. adjusting the idle by varying both signals duty cycle, you need to plug everything back with you scope probes connected, run the engine, keep it idle (idle switch connected and closed), and then make the engine load vary, like maybe switching on and off the AC or high beams or both and then you should see the ICV signals duty cycle vary, as the Motronic is trying to compensate the extra load from the alternator by adding more or less air to maintain the idle speed at nominal.

I'm seeing 50:50 with ignition on and engine not running, and about 70:30 at all times the engine is running, whether bypass is in place or not. When the bypass is removed and the motronic is adjusting idle with ignition timing only, it's still 70:30 whether revs are adjusted to 500, 880 or 1100 plus by manually allowing air to bypass the throttle.

Are you able to confirm you can see the 70:30 duty cycle it operates at, at all states, while running? It's the last image.

Sent from my SM-G988B using Tapatalk

Righto, as icv duty cycle isn't responding to engine speed I decided to double check the speed sensor at idle speed. You may remember I had 1.5-25V at cranking speed, and assumed this would amount to the required 2.5V minimum at idle.

Also checking at the ECU now, to rule out loom issues - pins 8 and 23.

https://uploads.tapatalk-cdn.com/202...35dce60b64.jpg

And got a healthy signal over 5V with no significant interference.

https://uploads.tapatalk-cdn.com/202...e21865c18c.jpg

Then I spotted a schoolboy error, any guesses?

https://uploads.tapatalk-cdn.com/202...0d7a1608bf.jpg

So it turns out the duty cycle results are all meaningless and I need to repeat the original tests!

Sent from my SM-G988B using Tapatalk

So if you said the throttle closed valve is unplugged, you were right. Now, repeating tests with it connected. [emoji2357]

Now here we are the same with icv bypass circuit in place:

https://uploads.tapatalk-cdn.com/202...0563ecf51a.jpg

And now it's clearly responding to idle speed, seen at low, normal and high idle speeds:

https://uploads.tapatalk-cdn.com/202...02fd62da71.jpg

https://uploads.tapatalk-cdn.com/202...547fd309d1.jpg

https://uploads.tapatalk-cdn.com/202...15313d50fe.jpg

https://uploads.tapatalk-cdn.com/202...d38dec8821.jpg

So I think I can finally rule out faulty icv signals as a possible cause of the rough idle.

In my view this now points squarely at the missing O2 sensor as the only parameter which is not how it should be - unless anyone has a different point of view?

Sent from my SM-G988B using Tapatalk

Here are some reference measurements of the icv duty cycle from my car - any pointers as to whether they're correct or not would be appreciated. Bear in mind mine is a 1988 so target idle is 880 rpm.

1200rpm and over 72%
900rpm 60%
840 48%
820, 700 and 650 rpm 43%

That operating range of 72%-43% seems narrow to me, but that's what it measures as.

Sent from my SM-G988B using Tapatalk

Not sure to understand you setup, but since it appears that one signal is 12 volts in amplitude and the other one is 7 or 8 volts, that means you are testing with the two 1.5k resistors in place of real ICV. Therefore this is open loop. You are varying the rpm and the ECU is trying to maintain it to 880 by changing the duty cycle, and it has no effect since the ICV is not plugged in.
But at least you see correct signals of the proper shape and "phase", with a varying duty cycle showing that the ECU is responding correctly.
It seems to me that you can leave the ICV and its control alone, they are doing well.

Yes, that's right. I was testing the icv signal was responding correctly to engine speed.

The latest news is... I fitted an O2 sensor and the car is now running like a dream. I knew no O2 sensor made it run rich, and some said Euro cars were shipped with no O2. Mine certainly hand no O2 sensor, but fitting one has fixed the idle problem.

It feels so great to finally have fixed this.

Sent from my SM-G988B using Tapatalk