2008 BMW 335i Piezo Injector Waveform Explained
This vehicle is fixed but while we were trying to pin down the actual cause for the excessive fuel trim (multiplicative) DTC's, we gathered the piezo injector waveform as part of our analysis and since this is the first time scoping one of these, I was somewhat puzzled. And since I make it a habit to store data for comparative analysis, I’m here to share what I’ve found and learned thus far about this injector and how its controlled. I'm especially interested in hearing from anyone else who has more familiarity on this. Later models are using solenoid injectors and for those who haven't seen the difference, I was able to take a comparative picture recently while I was in Chicago at Scot Manna's shop with a few others such as Bob Heipp, Matt Fanslow, and Eric Ziegler.
The above vehicle is the N54 inline six with the front and rear cylinder sets paired as B1 and B2. The vehicle arrived with the MIL on and the following DTC's stored:
- 29E1 (P0174) - SYSTEM TOO LEAN BANK 2
- 29E0 (P0171) - SYSTEM TOO LEAN BANK 1
LTFT was ~29% on both banks.
After much testing, we ended up pulling the trigger on condemning the injectors which indeed solved the lean condition setting failure codes.
My reasoning for the post is to start some discussion about how these injectors are controlled (inquiring minds would like to know more) and after talking with a friend/colleague, (Jorge Menchu, whom many of you likely know well) he pointed me in the direction of seeking out the information disclosed in patents and other papers that might have a better description as to how Piezo injectors work.
Part of this research involved looking at Delphi Patent #US6760212B2 as an example. Initially, it appeared to have what we were looking for but we weren’t sure.
Within that filing you'll see schematics I tried to use to support the waveforms I recorded from the subject vehicle which I’ve included below. After spending a few hours reading though all of the links Jorge and I found, I have to say that I know a little more about the device and found this system to be very interesting. And for this particular injector and operation, the patent listed above does not appear to support how this particular injector works. Within the Delphi patent you’ll see reference to an H-Block driver which appears to be used to reverse the current on the circuit. However, in this case, how can the current be reversed without it affecting the voltage drop across the injector?
If you’re still with me at this point, I’m happy to report that we found a document for a device that is used to test Piezo injectors. The waveforms and descriptions provide a goldmine of information and intelligence on this Piezo injector which closely matches the injector out of our vehicle.
Fyi: Piezo is Greek for "Push" and was discovered back in 1880.
What we’ve gathered thus far is that a crystal stack is used to push the injector open and subsequently pull it closed. Looking in the factory service information under “description and operation” I was left to think that the injector was only driven in one direction and spring loaded to return to the closed position? Looking at the MFGR wiring diagram, you can see that there are drivers on both sides of the circuit with one leading to terminal 15 and the other to terminal 31 (boosted voltage) (I drew another image based on what I saw on my scope as well). Comparing that to what I recorded with the scope, l was confused by the apparent change in current flow when compared to the voltage drop across in the injector. The current is definitely reversing direction, but, the voltage signal doesn’t support it. How could that be? (Thanks goes out to Robby for sharing his knowledge on the subject as well)
My connections for measuring the injector operation (voltage vs amperage) were as follows:
- Ch A - #1 Cyl + lead = White Wire | - lead = Org Wire
- Ch B - Current probe = White Wire
Refer to Fig 1 for scope trace identification.
From what I’ve read on the subject and found within the operation manual mentioned above provided by National Instruments on the Drivven CompactRIO DI Driver Module, Jorge and I were able to form a hypothesis. But first, a few baseline items to note:
- The Piezo injector is opened by pushing the pintle it into the combustion chamber.
- Current only flows when the voltage level across the stack is changing.
- Once the opening has been achieved, the voltage is rolled back slightly to hold the injector valve open.
- To close the injector, the power supply is removed and the voltage charge in the stack is drained through the ground side which pulls the injector closed.
I’ve illustrated the three periods within the waveform sampled from the new injector as follows:
- Charge period: (Fig 2) 350ma (avg current) 137.5v for 197µs
- Hold period: (Fig 3) 0 ma 139v for 281µs
- Discharge period: (Fig 4) -249ma (avg current) for 262µs
The injector is pushed open during the charge period (voltage change = current flow) where the Piezo stack (plates) are loaded with electrons. There is a critical voltage level needed to create an attraction between the charged plates and the uncharged plates in order to “expand” the stack of plates surrounding the injector pintle to the open position.
Once the critical voltage level is achieved, the voltage is reduced slightly to place the stack into the “Hold” period where we have little to no change in voltage.
When it’s time to close the injector, the power side is removed and the Piezo stack is discharged through the driver connected to the ground source 31.
Looking at the new injector waveform, (Fig 5) you can see that the injector on-time is 740µs.
Looking at the old set during duplicate conditions, you can see the injector was open approximately 28% more time at 949µs.
As I began to wrap my head around what the injector current was telling me (and I didn't have this as deeply understood at the time) what the on-time was and I could see the current reversing but the voltage trace I couldn't quite understand.
Does this make sense to you? I’d love to hear your perspective.
This case study is very interesting and I would of expected a change in the voltage as well. I would be curious into how they would compensate for a lower volume of fuel or even lower than desired pressure. Wonder if they can extend the on time even more or if the 949 micro seconds is its max limit. Maybe start injecting earlier and longer to keep it as smooth as possible. Guess will have to
Scott, It is interesting that the charge and discharge period (Voltage ramp) is controlled to protect the plates. While reading your article I was trying to come up with a way to describe the current. I am thinking that an analogy to a diaphragm chamber. A pull on one side brings in the electrons. Remove the pull and the electrons reverse direction. Your thoughts?
After seeing the graphic Steven drew up below, I now have a clearer picture of what's happening. I'll add another analogy, think of this piezo device as a sponge and your hand is the electrical part but you'll have to reverse some of the stages in order for it to make sense. The sponge starts off squeezed in your hand in water. As you release hand pressure, in flows water (current) (charge
I made some marked-up images to describe what I believe is happening with the current and voltage during the three phases (charging, holding, discharging). I don't really know how to add these images properly. I'll just try adding them
Hi Steven, Well done, your illustration talent is remarkable. Image #3 says a lot and isn’t what I actually described. You can see where the factory diagram would leave one hanging... Thanks or the additional perspective, next time I see one of these I’ll have two current probes installed, one on either side of the injector and will share what I find.
Actually, my third image that shows the discharging could be improved. I could have included orange arrows from the ground side transistor in the DME to the negative side of the injector. Just like discharging a capacitor or battery, I'm sure there has to be a complete path for the current. Thanks for making a post on this topic. That document you included had some great information.
Well thanks a lot Scott! That was a three hour trip down a rabbit hole of white papers and engineering documents;-) You had me scratching my head for a bit but I think I get it now. If you consider the piezo stack as a capacitor I think it makes sense. The rise in voltage essentially "charges" the capacitor (the piezo stack). Once it is "charged" it's charged, it's full of electrons so even
Hey Jon, Great to hear from you. Yeah, this one had me puzzled and while we were under the gun, I didn’t have the time to investigate how this thing was working. Yeah it was a tough call to make. We made the call on the injectors after double/triple checking everything and we were quite relieved when we made our post repair observations. I suspect that we will see this vehicle back again in
John, Great to see your picture and be chatting with you here. Yes, it would also be interesting to see a driver control timing graph.
Yeah ineed! It’s been a ling time. Hope this finds you and yours doing well. It appeas we may have some questions in common regarding the “hold” portion of Scott’s captures. It appears that the drive circuits contain the load, inductors and diodes. It looks like there are different ways to control the current and that in some cases they are trying to recoup some of the energy applied and use it
Jonathan, I had to readjust my thinking regarding the drivers to "switches to charge and discharge the injector" (see fig. 1 in PDF or Robby's post - bipolar configuration). One thing that makes this circuit so fun is that the electrons must not go through the injector (as indicated by the somewhat steady 'charged' line between the rising and falling ramp). And, as in the Fig. 1 circuit design…
Awesome story Scott! Really cool how your scope captures line up with the fuel system data. Piezo actuators are difficult to wrap your head around. When you first look at the waveform, it doesn't make sense how you can have power & ground to the injector, but no current flow. You and Jorge agreed on this statement: - Current only flows when the voltage level across the stack is
Thanks Robby, It is ironic how the % of time correlates to the trim adjustments. In regards to the current only flowing when the voltage is changing, I credit you with that statement as you raised it when we had a private discussion a month or so ago on this very subject. As for the deeper dive material you provided, wow, I appreciate sharing this. I suspect that for anyone coming across this
Yes, Scott fun info. I now have may question answer in regards to the power being constant or just for opening (charge) and closing (discharge). The waviness in the waveform between the rise and fall indicated no external voltage applied as well as the overshoot at the top of the rising edge.
Hi Scott, (and everyone)! This has been an ongoing area of study for me since piezo injectors were introduced. I will add some thoughts in here which pertain to both gas and diesel engine piezo injectors. Like most of you, we've all heard or experienced different versions of what is actually happening. In one of its 6.4L publications, Navistar (international) comes right out in print and
Some additional info from a Smog update licensing renewal class I took (I will paraphrase): When the piezo stack is charged, it "distorts" into an hourglass shape, changing and lengthening its dimension. Since reverse current is needed to close the injector, unplugging one on a running engine to use for misfire diagnosis (read: Power Balance) is to be avoided due to the possibility of keeping
Mike, Thanks for the info. "doesn't really mention this polarity reversal ". According to the general documentation, the driver can be set up uni-polar or bi-polar. Either way, it is just a charge and discharge just like Steve mentioned. The difficulty, the DME wiring diagram examples above, there is not enough information to understand operation. And it appears to be more of a traditional
I'm lost, wouldn't the increased hold time be normal pcm making up for lean condition?
It was trying to do just that and reached its limit of roughly 30%. The new injectors reduced the open time by about 28%.