911SC CIS frequency valve control signal

[mopydick]

Hi Everybody,

I'm a digital controls engineer and this winter's project is building a CIS test bed.

I'm trying to determine what sort of control signal I ought to drive the frequency valve to simulate normal operation. From the various prior posts, I've gathered that its a PWM signal with a 70 Hz (14.3 ms) envelope with a 12v amplitude signal.

What I find odd/interesting, is that the modulation is taking place in the second half of the period. And the actual modulation period is on the order of 500 usec. Its a weird modulation pattern to be sure. But it does explain why we (as humans) hear the frequency valve 'buzzing' since 500 usec => 2 kHz, which is right in the sweet spot of our ear's sensitivity.

If my observations are correct, it raises some questions:

1) Does the true 500 usec modulation always stay in the 2nd half of the 14.3 msec period?

2) Is the initial 'on' time at the beginning of the second half of the 14.3 msec period always 1 ms? Looks like its a kick in the pants to get the valve to open and then the modulation starts thereafter.

Thanks for any insight,

Thurmond

PS. Thanks to whomever took the scope shot below. I wish I had done this sort of surveillance prior to pulling the engine out.

[Walt Fricke]

That was me. Attached are a couple of other scope pictures. The blue and blue/yellow two channel ones are mine. The diagram and the CRT scope annotated one are someone else's. My understanding of what I see is pretty limited, though I understand why there is a spike - the FV, like injectors, is a solenoid, and turning it off creates an inductive spike, like a spark coil does. Porsche's diagnostic plug gets a cleaned up signal from the mini-ECU. I routed both signals to a little box with BNCs so I could look at them in the car.

One of the many things I don't understand is why the on signal, on inexpensive digital scopes, shows up as a bunch of closely spaced up and down traces, while on a CRT they look like a fairly stable voltage level. I understand a little fuzz on a horizontal trace - inexpensive, hardly lab grade, instruments. But the blue lines doing so much of this I don't understand, though maybe it is an artifact of the Sainsmart and some operating frequency. The picture you posted is of an even cheaper scope, with its own small screen, not run through a laptop as with the Sainsmart. It also has the up and down traces, though not so closely spaced.









I tried to modify the system using a PWM generator and two MOSFETs (based on advice from an EE). I ran the FV ground (the ECU grounds the FV) through the MOSFETs. In normal operation one MOSFET passes connects the ECU ground. At WOT, the 3 pin switch on the throttle body tells that MOSFET not to work, and tells the other one to pass the WPM signal instead. I used a CMOS NAND suitably configured to tell the MOSFETS which should be on, and which off.

Alas, it didn't work. I may have wired it up wrong, I may have miswired the DPDT switch which allowed me to change from normal operation to this special WOT operation, and I may just have failed to tell the PWM to output a high enough voltage. And I didn't test with a scope. But the Air/Fuel meter did not change its behavior when I switched the circuit in. Nor did another, simpler switching involving a regular input, as suggested by Andrew, appear to do anything. So probably some goof on my part.

[mopydick]

Hi Walt,

I was hoping that you'd jump on this thread. From all the prior threads I've read, you've been in the thick of this sort investigation for a while.

I'd like to sort out the differences in scope traces first (and hopefully answer your question as why a signal appears to be different on different scopes). Please correct me if I'm wrong when I paraphrase your statements.

It scope traces appear to me be:

Ch1 (blue) : frequency valve control signal (pin 15 from ECU)
Ch2 (yellow) : diagnostic signal (pin 17 from ECU)




Its my understanding (and I'm not 100% sure here) that the diagnostic signal (pin 17) is an indicator of how well the system is achieving the correct Fuel/Air ratio with a 50% duty indication a correct ratio, and any greater or less indicates a rich or lean condition.

If I'm correct, your first scope screen shot makes perfect sense to me. Ch1 is showing the weird PWM control signal and Ch2 is showing about about 50% indicating that the system is working correctly.

Another comment about the signals - I think the Ch1 signal is real because its the only one that is in the 2 kHz regime and that's why we hear buzzing. If the real PWM period was 70 Hz, that's a relatively low frequency sound and wouldn't be noticed as a buzzing sound.

The second plot below concerns me because it clearly defining the PWM frequency as 70 Hz (it has none of the 2 half period 500usec modulation). Yet, it would be a more traditional control signal and what I would have expected.

I'm left wondering why does the weird 500usec signal appear on the first scope shot and but not on the second? Do you know where the 2nd plot came from? It is conceivable that a heavy duty analog filter was applied to filter the 500usec signal out with the thought that is was just noise.

Thurmond

[mopydick]

The notes in the 2nd plot say that its signal is from pin 8.

I'm not sure what pin 8 does other than its powered only when the fuel pump is running so either its a power source, or a enable signal to the controller's logic.

Even more mystery ...

[AndrewCologne]

Its my understanding (and I'm not 100% sure here) that the diagnostic signal (pin 17) is an indicator of how well the system is achieving the correct Fuel/Air ratio with a 50% duty indication a correct ratio, and any greater or less indicates a rich or lean condition.

Yes, at pin 17 the same dutycycle is present as at pin 15. So at the testport you know which duty cycle is present at the FV.

The notes in the 2nd plot say that its signal is from pin 8.

pin 8 in regular is the power surce of the ECU.

[LIRS6]

WTF are you guys talking about????

Jason
(Econ major)

[3rd_gear_Ted]

Possibility?
Old school PWM signals for hydraulic solenoids would have a hysteresis component to keep the solenoid responsive because of the high system pressure.

[Walt Fricke]

Jason - this discussion concerns applying Poisson's Equation to the inverse of Maxwell's 3d theorem. Isn't that clear?

To get my digital scope traces I opened the mini brain and soldered pieces of small coax to pin 17 (which is the test signal which emerges back in the engine compartment in a three pin connector aft of the relay box) and to pin 15 (which is where the positive current from the FV enters the ECU, where it is grounded. They terminate in two BNCs I mounted in a small box and attached to the acceleration enrichment unit, as that was handy.

You can see why, following Jim Williams' supposition, it looks like pin 17 either is the clean signal sent to some other components in the ECU to cause the pin 15 circuit to go to ground, or it is a filtered/cleaned up signal from pin 17 itself.

I understand the spike on the one but not the other. What I don't understand is why does the CRT trace show two parallel lines, while the digital scopes show a very rapid up and down. I'm guessing it is an artifact of inexpensive digital scopes - or alternatively they are actually faster than an ordinary CRT? For that matter, is my single port battery powered standalone Sainsmart's display slower (or has a lot fewer pixels in its display) in this picture:



Than my two port Sainsmart box which uses the computer screen for its display? The screen graphics card surely is a lot faster and has better definition than what is on the little portable? Is that what causes all those blue up and down lines, instead of the steady voltage levels shown in this CRT image:



This is using two channels and displaying both traces at once. However, I don't understand why there are five traces, so to speak, and not just four. Perhaps if I had any formal training in electronics I would know.

70 Hz is what I measured for the frequency. The pulse widths varied as expected given the static inputs I gave them - I didn't have the oxygen sensor hooked up, so there was none of the variation associated with response to changed AF ratios.

I don't know why there is reference to pin 8 in that nice diagram, nor do I recall where I found and copied it. But you aren't going to get anything much more than a constant voltage level on your scope from pin 8.

[mysocal911]

Originally Posted by Walt Fricke

You can see why, following Jim Williams' supposition, it looks like pin 17 either is the clean signal sent to some other components in the ECU to cause the pin 15 circuit to go to ground, or it is a filtered/cleaned up signal from pin 17 itself.

Pin 17 just replicates the FV signal without the inductive voltage ringing of pin 15.

Originally Posted by Walt Fricke

Than my two port Sainsmart box which uses the computer screen for its display? The screen graphics card surely is a lot faster and has better definition than what is on the little portable?

The sampling rate of the Sainsmart box is too low to provide the same resolution as an oscilloscope.

Originally Posted by Walt Fricke

However, I don't understand why there are five traces, so to speak, and not just four. Perhaps if I had any formal training in electronics I would know.

There're really only two traces on the oscilloscope. If you change the time base on the oscilloscope you'll see the high frequency ringing of the FV (pin 15).

[LIRS6]

Originally Posted by Walt Fricke

Jason - this discussion concerns applying Poisson's Equation to the inverse of Maxwell's 3d theorem. Isn't that clear?

My bad

Jason

[Walt Fricke]

Dave - It would be nice if Porsche would release the schematic for the Lambda CIS ECU.

I understand the inductive spike. But why is there high frequency ringing? There are 70 pulses to the FV per second, which vary only in their on and off time (PWM). 70 Hz must be within the ability of the solenoid to open and close for varying parts of its constant period.

My (limited) understanding of ringing is seeing a spike or spikes at the start and stop of a square wave, but not through the entire on time.

The "little" scope, Sainsmart DSO138, claims 0-200 KHz bandwidth, 12 bit, and 10us.
The DDS120 claims a sampling rate of up to 50 MS/s, 50ns, and 6/60MHz bandwidth.
The specs I found didn't quite line up as being the same parameters.

[mysocal911]

Originally Posted by Walt Fricke

Dave - It would be nice if Porsche would release the schematic for the Lambda CIS ECU.

Except for the hybrid circuit (white ceramic I.C.), it's fairly easy to reverse-engineer.

Originally Posted by Walt Fricke

But why is there high frequency ringing? There are 70 pulses to the FV per second, which vary only in their on and off time (PWM). 70 Hz must be within the ability of the solenoid to open and close for varying parts of its constant period.

It's "peak-and-hold" type of a driver circuitry;
https://autoditex.com/page/peak-hold-injector-53-1.html

It's also used in the 911 3.2 DME ECU. Starting with the 964, Porsche-Bosch no longer used this design approach in the injector circuitry.