Differential rail pressure sensor for Common Rail systems

Ask for and share advice on using the PicoScope kit to fix vehicles here.

Differential rail pressure sensor for Common Rail systems

Postby steevegt » Mon Aug 15, 2016 8:54 pm

Hi there

I decided to share one of my adventures in finding an alternative way to diagnose the hydraulic function of a common rail injector in the vehicle. Or at least to get some clues of it...

As far as I know, so far, we can use 3 test for hydraulic diagnosing common rail injectors:
* Computer correction values to each injector, using serial data (and in some cases, if not available, current waveform).
* We can measure the return flow amount
* Return flow pressure pulses

The tests before are not always very conclusive and most of the time they don't identify very clearly if the problem is one or more injectors at fault, as we only use relative comparison.

Trying to find something that can give more information, I decided to test if we could get something useful from a piezo electric clamp on a common rail high pressure pipe.
This type of clamp, uses a piezo material, that produces a voltage when the pipe that it has attached to expands and contracts.
This type of sensor is usually used in tachometers, as an option for measuring RPM in old Diesel systems. Each expansion and contraction of the pipe produces a voltage signal in the clamp. This is than interpreted by the tachometer to show an RPM measure.

So, I first try to connect the clamp directly to a channel in the picoscope, in an old Diesel system (Bosch VE Diesel injection pump). We can see the pressure increase (as a negative voltage) and the pressure decrease (as a positive voltage) in the waveform below.

inj1 golf 3.PNG
Golf III 1.9 TD

We can see that this clamp produces a clean signal (with filtering), at least in this old Diesel systems.
This can be useful, at least, to synchronize captures with the scope. This is a real challenge in this systems... With this clamp, this becomes very easy...
I guess we can also detect with this, some problems with an injector that does not hold pressure, or an injection pump that does not produce pressure. Maybe more problems can be seen...

Ok, so now, can we get something on a common rail pipe?
In common rail the high pressure pipes don't get the pressure differences in an injection event as in this old systems, when we have pressure only near the injection event.

The next waveform was captured on an 2004 Smart For Two 0.8 cdi. I used the clamp on the number 1 injector high pressure pipe.
Channel A (Blue) is the injector current waveform using Pico 60 Amp clamp, and in channel B (Red) I connected the piezo clamp.

You can see the hardware setup in the next picture:

Piezo clamp setup

The waveform capture for each injector:

smart cdi inj 1.PNG
Smart 0.8 CDI Inj 1

smart cdi inj 2.PNG
Smart 0.8 CDI Inj 2

smart cdi inj 3.PNG
Smart 0.8 CDI Inj 3

The amplitude of the signal is very dependent of where you put the clamp in the pipe, closest to a bent or in a long straight part of the pipe.
Between injectors, the waveform looks very consistent. For example, in yellow, you can see two peaks and one valley at the main injection event. This is very clear between all 3 injectors.
The perfect example will be with a know good and know bad injector, but so far, I didn't get the change to capture that.
This may be a good example of a good pressure differential waveform for this injector in this system, but I believe we need to dig deeper, and get a lot more examples of this to conclude something.

I manage to get a capture in a 2005 renault scenic 1.5 dci (Delphi system). As before the waveform is consistent between captures and between injectors.

(Switched channels from before, A is piezo clamp and B is current clamp)

renault scenic 1.5 dci inj1.PNG
Scenic 1.5 DCI Inj 1

You may say that the waveforms I capture before may be produced from noise, may come from the magnetic field of the solenoid...
I think that it may be plausible that the waveforms before don't actually have nothing to do with pressure waves in the pipe, but something else is causing that voltage in the piezo material.

With that in mind, I did a test.
If the clamp is really showing the pressure increase and decrease amount in the pipe, so, the clamp is producing through voltage the measuring of the differential pressure. If this is correct, the integral of the signal must relate to the pressure waveform.

To do this we must have at the same time a pressure signal to compare.

The following capture was made in the same Smart 0.8 CDI engine as before, capturing a complete WOT event.

smart cdi inj 1 pressure compare.PNG
Smart CDI Pressure Compare

Channel A (Blue) is the current waveform of the number 1 injector (Pico 60 Amp clamp).
Channel B (Red) is the voltage output of the pressure sensor in the rail.
Channel C (Green) is the voltage produced by the piezo clamp in the number 1 injector HP pipe.

First, I must say that this engine has an issue, with some pressure fluctuations in the rail and a consequent little idle issue. This explained, lets get the point.
Black channel is a math channel that represents the integral of the C channel, as I had explained before.
As we have seen in the old Diesel system, the clamp produces a negative voltage when the pipe expands (pressure increase) and a positive voltage when the pipe contracts (pressure decrease). Because of this we must use the negative voltage of the signal (-C) in the math channel if we want to have a positive value representing a positive pressure.
To have a waveform that resembles to the pressure one, we need to have a "calibration" constant to the integral. In this case, using an error and try method, I get -0.001 as being one value that serves good in this particular capture.
I could see if the value was good or bad by comparing the value of the math channel in the start and in the end of the above capture. We know by the pressure sensor that the start and end values of pressure are almost the same.
This "constant" looks to be more dynamic than constant, but for a 20 second capture one value is enough to have a good match between the math channel and the actual pressure value from the rail pressure sensor.
For a different capture, in the same setup you may need to adjust this value, if not, you end with a positive or negative drifting signal.
The need to have this value, is something that I don't know for sure, if I know anything at all. It may be temperature related, it may be related to the vibration of the engine itself, or something completely different, like noise generated inside the scope...
Any ideas to possible solve this issue, or at least explain it, are more that welcome.

With this final capture it is possible to prove that the waveform that we see near the injection events, are in fact related to pressure pulses in the pipe.
We can also say that this type of piezo clamp can also be used to diagnose a pressure sensor signal problem, without connecting a second pressure transducer, or a pressure gauge.

This clamp is a little hard to find, as I only found it to be available as a spare or an option for some tachometers. It is also available as an option for some Diesel smoke meters and gas analyzers.
Posts: 17
Joined: Tue May 12, 2015 10:05 pm

Re: Differential rail pressure sensor for Common Rail system

Postby Steve Smith » Tue Aug 16, 2016 10:43 am

Hello and thank you so much for taking the time out to test, measure, document and post all the above.

I know it can take forever but it will help thousands of users.

I have no doubt what you have revealed with the Piezo Pick Up device is relevant to both injection events and of course the inherent vibration throughout the engine.

With that said, you cannot deny the repeatability in your captured signals during the main injection event.

I have carried out such tests using a similar device but more recently using the NVH accelerometer connected to PicoScope software (Not PicoDiagnostics NVH)

The results below speak volumes but also highlight the effects on the accelerometer from inherent engine vibration

Integral and Average maths with scaling

Using Integral A maths channel (x10 scale) reveals pulsations detected by the accelerometer relevant to cylinder contribution based on engine movement/vibration. (However, accelerometer activity relevant to diesel injection events can still be detected)

Using Average A maths (x4 scale) highlights pulsations relevant to diesel injection events only whilst pulsations attributed to engine movement/vibration are quenched and far less visible.

I believe a combination of vibration analysis within diesel injector pipework, combined with maths channel applications adds another string to your diagnostic bow in that all important Non-Intrusive fashion

Thank you again for the post, take care.......Steve
Steve Smith
Pico Staff Member
Pico Staff Member
Posts: 426
Joined: Sun Aug 25, 2013 7:22 am

Re: Differential rail pressure sensor for Common Rail system

Postby steevegt » Wed Sep 07, 2016 5:35 pm

Hi Steve

I have managed to find some captures I have made using this sensor, trying to validate if in fact we can also see engine vibrations using this type of piezo clamp, like the ones you have shown using the accelerometer...

To be honest, I didn't know if they were present or not, because I didn't look for that before you mention it. I was happy to see the correlation, in a long capture, between the rail pressure sensor signal and the integral of the piezo sensor output, not looking for detail in it.

So I gave another look:

First capture, short time:
scenic 1.5 dci integral.png
Scenic 1.5 dci Piezo Clamp Integral

Second capture, long time:
scenic 1.5 dci integral long capture.png
Scenic 1.5 dci Piezo Clamp tntegral WOT long capture

scenic 1.5 dci integral long capture zoom.png
Scenic 1.5 dci Piezo Clamp Integral WOT long capture (Zoom)

If engine vibrations/movements are embedded in the signal, they are not clear at all using this clamp. At least I can't see them...
I have more examples of this in other vehicles, but in nether of them we can see the sine wave like signal that the accelerometer produces.

The way you have used the average function, did the job in a very clever way I must confess. It has the disadvantage of only working around the trigger point... Multiple injection events in one buffer isn't an option...
In the other hand, it has the advantage of producing a more clear injection event signal around the trigger point, I guess...
Isn't the low pass filter an option for filtering this? I never found a practical use for that math channel, but in this case I believe it fits, right?
Posts: 17
Joined: Tue May 12, 2015 10:05 pm

Return to Diagnostic discussion

Who is online

Users browsing this forum: No registered users and 1 guest