2001 Citroen C5 HPi 2.0 HPI (Direct Injection)

Your chance to introduce yourself and your vehicle
Post Reply
mzk
Posts: 21
Joined: Fri Jun 02, 2017 3:44 pm

2001 Citroen C5 HPi 2.0 HPI (Direct Injection)

Post by mzk »

Since my welcome topic became a bit technical I decided to start a new one dedicated to my project.
The first topic is here: http://rusefi.com/forum/viewtopic.php?f=2&t=1245

In few words: I have Citroen C5 2.0 HPi (direct injection). I would like to let rusEfi control the engine. Main reasons are:
- too expensive engine for support - not well designed high pressure fuel pump which fails and costs a lot - possible elimination of the high pressure fuel pump after failure or in future experiments;
- too delicate for normal LPG injection kit. 1st these LPG kits are expensive. 2nd they rely on tight communication with the factory ECU in order to keep the fuel mixture precise. Almost no DI LPG ECU suitable for this engine;
- Faulty wideband O2 sensor - while it sounds like a great toy to play with, it is also expensive. I plan to cut the engine from operating in lean mode and to use ordinary O2 sensor. Also the engine burns some oil and this means that any O2 sensor installed will break.
- LPG would lead to gasiline injectors burnout. Therefore a short amounts of gasoline should be injected together or in separate cycles in order to keep them in good condition.
- in the future, if the high pressure pump or any of the DI injectors fails I can mount single or four ordinary fuel injectors at the manifold and convert the engine to normal port injected. This may be done also for solving cold start problems typical to this type of engine.
- and beyond that - if it happens to live - I would like to experiment with LPG direct injection, probably I'll mount port injector(s) for petrol and feed the installed fuel rail with LPG.

So you can see there are several configurations of the fuel management that I would like to play with.

Table of contents (to be filled on each successful step):
1. First steps - this posts - shows some oscillograms of the fuel injectors.
(next)
- driving circuity (not yet completed)
- ETB - throttle body motor control (not yet completed)

But first, let me write all the periphery connected to the factory ECU with my view of what should be driven by rusEfi, what can or should be left to the factory ECU and what should be "emulated" to prevent the factory ECU from inappropriate work.

--- to be disconnected/proxied from the factory ECU ---
Fuel pump
Injectors (x4/x8)
Input Air Temperature (IAT)
Engine coolant temperature (ECT) sensor
Fuel rail pressure (FRP) sensor
Knock sensor (KS)
Manifold absolute pressure (MAP)
Engine coolant blower motor
Accelerator pedal position sensor (APP)
Throttle body motor
Throttle motor position sensor
Crankshaft position sensor (CKP)
Camshaft position sensor (CMP)
Ignition coil (x4, each cylinder)
Camshaft position actuator
Fuel pressure regulator control solenoid

--- under consideration ---
Pre-cat (wideband) O2 sensor (5-wire?)
Post-cat (narrowband) O2 sensor
Exhaust gas temperature sensor 1
Exhaust gas temperature sensor 2
Exhaust gas recirculation (EGR) valve position sensor
EGR solenoid
Evaporative emission (EVAP) canister purge valve - ???

--- not to be touched ---
Brake system vacuum sensor - TO BE TESTED AFTER rusEfi installation! Incl oil pressure sensor, etc...
ABS conntrol module
AC refrigerant pressure sensor
Power steering pressure switch (PSP)
Clutch Pedal Position (CPP) switch
Cruise Control Brake pedal switch

=========================================================================

Today I did a little research on how the ECU behaves. I took the signals from the CMP (Camshaft Position Sensor), CKP (Crankshaft Position Sensor) and the voltage over 1st cylinder's injector.

First of all, yes, this is true DI engine. It has a high pressure pump and drives the injectors with high positive and negative voltage in order to open and close them.

The injector timing is divided in several steps as follows (engine idling):
1. 400 us (micro-second) "start-up" time with 12 V DC voltage.
2. 600 us period follows after that which is PWM modulated +12V. It looks like the duty cycle is higher in the beginning and lower in the end of the period, but I can't guarantee it.
3. 200 us high voltage (80 V DC) (opening) period.
4. 200 us +12 V DC period (probably keeps the injector open).
5. 30 us period follows with -80 V DC negative (closing) voltage.

------- till now the period was fixed. Remember that these readings were made on idling engine w/o any load (headlights, AC, fan, etc.)
The 6th period changes with each revolution and the engine load (even when idling). Keep in mind that the condition of the injectors is unknown, also the pre-cat O2 sensor is broken (wideband).

6. 100 us period with 0V (nothing happens). It may continue with 0V followed by second (-80) negative closing peak which ends the injector's cycle OR it may put some 12 V PWM modulated (it was always changing when I took the scope shots) after which follows the second -80 closing peak.

You can see a video https://youtu.be/xZYNPauRSJ0

Please make note of the ugly noise which disappears when the injector time decreases. I guess this is the moment when the ECU switches the operating modes (normal mixture / lean mixture). In normal mode the gasoline is injected in the 1st stroke and in lean burning mode the fuel is injected about the end of the 2nd stroke.

---------------------

And a scope shot of the camshaft and the crankshaft position sensors:
Blue channel - CMP, red channel - CKP
Blue channel - CMP, red channel - CKP
cmp-ckp.jpg (178.79 KiB) Viewed 16984 times
I'm not sure about the timing of the camshaft sensor - should it coincide with the crankshaft? I guess there is difference since I think the engine has something like a camshaft actuator... :roll:

---------------------

I think there is a chance for rusEfi to be used, but some hardware should be made to drive the injectors...
Last edited by mzk on Sun Jun 04, 2017 8:24 am, edited 1 time in total.
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: 2001 Citroen C5 HPi 2.0 HPI (Direct Injection)

Post by kb1gtt »

Do you intend to control the DI or just the LPG? Is the LPG going to be manifold or direct injection?

There was some talk about DI here http://rusefi.com/forum/viewtopic.php?f=3&t=937

Also I've made this simulation, which is yet to be tested.
Image
See files found here https://sourceforge.net/p/daecu/code/HEAD/tree/wiki/simulations/QUCS_souce_files/picture_docs/DI_3.PNG?format=raw
Welcome to the friendlier side of internet crazy :)
User avatar
AndreyB
Site Admin
Posts: 14292
Joined: Wed Aug 28, 2013 1:28 am
Location: Jersey City
Github Username: rusefillc
Slack: Andrey B

Re: 2001 Citroen C5 HPi 2.0 HPI (Direct Injection)

Post by AndreyB »

MC33816 comes up as the chip to try for direct injection - hopefully that would cover the hardware side?

But there is also firmware side - at the moment we do not have the neccesary precision. Firmware would require some changes.
Very limited telepathic abilities - please post logs & tunes where appropriate - http://rusefi.com/s/questions

Always looking for C/C++/Java/PHP developers! Please help us see https://rusefi.com/s/howtocontribute
mzk
Posts: 21
Joined: Fri Jun 02, 2017 3:44 pm

Re: 2001 Citroen C5 HPi 2.0 HPI (Direct Injection)

Post by mzk »

1st I've updated the first post a little. A list of the periphery connected to the ECU is included. May I ask you to have a look and comment? Thanks!
MC33816 comes up as the chip to try for direct injection - hopefully that would cover the hardware side?
Unfortunately it seems hard to find this chip here, Farnell or AliExpress...
But there is also firmware side - at the moment we do not have the neccesary precision. Firmware would require some changes.
If it is up to me I'll probably make the driver side together with STM8S103 which can be used as programmable driver? The ECU sends the calculated injector time and the driver is responsible for driving the injector properly. Also please note that the current should be controller in this type of injectors, more information follows.
Do you intend to control the DI or just the LPG? Is the LPG going to be manifold or direct injection?
Both. LPG will be manifold.
Also I've made this simulation, which is yet to be tested.
Thanks! I have doubts on the energy stored in C1. I like that you kept the schematic simple!

Please refer to the following oscillograms I've found on the Internet:

1. Voltage and current of DI injector.
Source: http://www.tiepie-automotive.com/en/Measurement_examples/Actuators/Direct_Petrol_Injection_Voltage_Current
Source: http://www.tiepie-automotive.com/en/Measurement_examples/Actuators/Direct_Petrol_Injection_Voltage_Current
DirectInjectionPetrolVoltageCurrent.png (42.8 KiB) Viewed 16968 times
Here it can be seen that the PWM part of the voltage is actually current regulation.

Another example:
Source: http://www.underhoodservice.com/tech-feature-common-rail-systems-may-soon-be-commonplace/
Source: http://www.underhoodservice.com/tech-feature-common-rail-systems-may-soon-be-commonplace/
di-voltage-current.jpg (38.58 KiB) Viewed 16968 times
Now the video that I've posted in the first post makes a lot more sense.

2. "Waste spark"
From http://www.underhoodservice.com/tech-feature-common-rail-systems-may-soon-be-commonplace/
Unlike non-GDI systems, each injector is controlled on both the positive and negative side of the circuit. In addition, most systems are designed with two injectors sharing two positive drivers.
One of the advantages to pairing two injectors together is the ­utilization of the sister cylinders inductive kick to recharge the capacitor.
Source: http://www.underhoodservice.com/tech-feature-common-rail-systems-may-soon-be-commonplace/
Source: http://www.underhoodservice.com/tech-feature-common-rail-systems-may-soon-be-commonplace/
inj-double.jpg (52.69 KiB) Viewed 16968 times
http://www.underhoodservice.com/tech-feature-common-rail-systems-may-soon-be-commonplace/
http://www.underhoodservice.com/tech-feature-common-rail-systems-may-soon-be-commonplace/
waste-injector.jpg (37.14 KiB) Viewed 16968 times
Actually I took a vide of this yesterday (after 00:59): https://youtu.be/5dQf4Qvx-oA
Please note that I didn't connect the oscilloscope GND to the injector cables this time in order to prevent ground loops.


Homework:
- take current measurements of 1st injector together with CKP reading; measure the injector timing against the current stroke - does it inject the fuel in the suction or the compression stroke?
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: 2001 Citroen C5 HPi 2.0 HPI (Direct Injection)

Post by kb1gtt »

I both like and dislike a separate STM approach. I like the flexibility of being able to change things with software updates instead of soldering iron. However I dislike having 5 firmware's to update if you want to make changes. I guess one STM chip could potentially handle 4 or 8 injectors. So perhaps it could be made into just 2 firmware's. For starters I would suggest one STM per injector to simplify things, and avoid complications of cross talk, and cross timing critical events, etc. A one STM for one injector approach makes for easier software development. Perhaps one STM with 4 channels, but only plan to use one channel to start with. Also longer term I would like to see a discrete approach, perhaps using 555 timer chips to do the PWM, and other such steps. This would avoid the need to program multiple chips.

Would you consider drafting a 555 or discrete base schematic, then build the first one with an STM chip? Such that we can learn the PWM and high voltage details with an easy to change platform, then once we know the general specifications, we can change to a discrete circuit?

I see other chips at aliexpress for DI. Perhaps we can find a datasheet for one of these chips. I found this one, and others probably exist.
https://www.aliexpress.com/item/30554-diesel-efi-and-direct-injection-gdi-engine-power-supply-chip-computer-2PCS/32745902262.html?spm=2114.40010508.4.92.8eaJR6

I see stock of the MC33816AE in multiple international supply companies.
https://octopart.com/search?q=MC33816AE&avg_avail=(1__*)&start=0

How hard would it be to make a small engine like a lawn mower into DI? If we can get a small low cost engine, then either @ or myself would be more likely to offer some more hands on help. A key problem for @ is having hardware which he can work with. It becomes really hard to develop stuff if you do not have hardware to see the real world details.

On a side note, I have a DI petrol car. If I could find an OEM connector, there is a chance I can experiment with my daily.

You may find this project interesting. It was specifically designed to be mostly MFG'ed by sky-macau.com's inventory list. I have to populate some chips like the hall based current sensor, and other less generic ships. The H-Bridge chip has an internal current sensing which may or may not suffice for my needs. So I have the hall sensor as a backup plan.
https://sourceforge.net/p/daecu/code/HEAD/tree/Hardware/trunk/Stervo/

On a side note, I dream of a Cummins 4BT with common rail injection for my Jeep.

Also beware the carbon build up of DI engines. On port injection your fuel and wall wetting tends to dissolve carbon build up off the valves and intake runners. On DI you do not have this. If your engine is consuming oil, and if you can not account for this lost oil on the ground, you probably have blow by which is being dumped into the intake. This oil will then cause carbon build up problems as it sits there and burns slowly in the intake. You may want to consider adding a "catch can" or periodically cleaning it with intake cleaners. Also keep in mind that intake cleaners can backfire and cause damage, so it's best to do those are idle which prevents the chances of back fire.
Welcome to the friendlier side of internet crazy :)
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: 2001 Citroen C5 HPi 2.0 HPI (Direct Injection)

Post by kb1gtt »

Welcome to the friendlier side of internet crazy :)
User avatar
AndreyB
Site Admin
Posts: 14292
Joined: Wed Aug 28, 2013 1:28 am
Location: Jersey City
Github Username: rusefillc
Slack: Andrey B

Re: 2001 Citroen C5 HPi 2.0 HPI (Direct Injection)

Post by AndreyB »

mzk wrote:with STM8S103 which can be used as programmable driver?
that would work
Very limited telepathic abilities - please post logs & tunes where appropriate - http://rusefi.com/s/questions

Always looking for C/C++/Java/PHP developers! Please help us see https://rusefi.com/s/howtocontribute
mzk
Posts: 21
Joined: Fri Jun 02, 2017 3:44 pm

Re: 2001 Citroen C5 HPi 2.0 HPI (Direct Injection)

Post by mzk »

Thank you very much for your activity! It seems that the hard work is going to begin soon.

I'll be away for some time, during the week, but I'll do my best to sketch something for the driver.
How hard would it be to make a small engine like a lawn mower into DI? If we can get a small low cost engine, then either @russian or myself would be more likely to offer some more hands on help. A key problem for @russian is having hardware which he can work with. It becomes really hard to develop stuff if you do not have hardware to see the real world details.
As you have written in the other posts, the high pressure pump is quite expensive. Also the DI engines you can find at the scrapyard aren't as much as the normal engines. Actually I'm interested to keep the DI components only to ensure minimal hardware rework of the injection system. However I'm not a big fan because I believe it is too complex and expensive, unreliable solution. Years ago I was interested in hydrogen/oxyhydrogen/water as a fuel additive and I believe it may contribute to decrease exhaust emissions. If I have some experience/understanding of electronics and software, I have absolute no idea of mechanics. Probably it will be hard to couple the high pressure pump with the camshaft (as it is on my engine). I'm not sure if there are electrical high pressure pumps, but it would be definitely easier.

I agree that it will be interesting to combine the same injection system for GDI/diesel, however I'm not sure how the newer piezo injectors work and if the driving circuity is the same / similar.
Also beware the carbon build up of DI engines. On port injection your fuel and wall wetting tends to dissolve carbon build up off the valves and intake runners. On DI you do not have this. If your engine is consuming oil, and if you can not account for this lost oil on the ground, you probably have blow by which is being dumped into the intake. This oil will then cause carbon build up problems as it sits there and burns slowly in the intake. You may want to consider adding a "catch can" or periodically cleaning it with intake cleaners. Also keep in mind that intake cleaners can backfire and cause damage, so it's best to do those are idle which prevents the chances of back fire.
Thanks for your suggestion! I didn't think about that...

I'm watching the Stervo's schematic now. Actually I'm inexperienced in noisy systems such as the combustion engine. I'll keep it on mind. You're very experienced, but I'll do my best to keep on track :)
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: 2001 Citroen C5 HPi 2.0 HPI (Direct Injection)

Post by kb1gtt »

I have interest in di I can offer help in a layout. Are we thinking STM chip with high side and low side drive? Also current sense. Should we plan on ball sensor chip?
Welcome to the friendlier side of internet crazy :)
mzk
Posts: 21
Joined: Fri Jun 02, 2017 3:44 pm

Re: 2001 Citroen C5 HPi 2.0 HPI (Direct Injection)

Post by mzk »

I prefer the STM solution for experimenting. Later we can discuss another, more robust (analog) solution.
Maybe it is fine to start with current sense resistor and an op amp?

A source of carbon buildup is also the EGR valve. I've seen pretty ugly things with diesel engines, but this one (EW10D) also has the EGR... I guess that your DI car also has it.

PS: also if you agree optocouplers can be skipped. I have several of these boards: https://www.aliexpress.com/item/1pcs-lot-STM8S103F3P6-system-board-STM8S-STM8-development-board-minimum-core-board/32653059550.html?spm=2114.13010608.0.0.QxyEXr and together with the mosfet drivers they can be soldered on experimental pcb.

I'm curious if I can disconnect completely one of the injectors. In its place I can place an equivalent inductor. The only thing I'm unaware of is how "read" the signal of the ECU. I need to measure the current...
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: 2001 Citroen C5 HPi 2.0 HPI (Direct Injection)

Post by kb1gtt »

Keep in mind that the injectors are liquid cooled. An equiv inductor may have overheating problems.

Yeah, that board is potentially handy. We once build the below board, which is very similar to the board you posted.
http://rusefi.com/forum/viewtopic.php?f=4&t=769&p=15434

How do you see a board being made? Are you thinking bread board and thru hole, or are you thinking SMT board? I see Sky-macau has that STM chip. As well they have op-amps and most of the other SMT stuff. They are fairly low cost assembly. I have them build Stervo right now, that run they had a minimium of 5 boards.
http://www.sky-macau.com/Products/All-C168/STM8S103F3P6-STMICROELECTRONICS-STM8S-Series-16-MHz-Microcontroller-TSSOP-20-P7398337169.html

Also should this be one board with like 4 or 8 injectors? Or should it be just one channel per STM?
Welcome to the friendlier side of internet crazy :)
mzk
Posts: 21
Joined: Fri Jun 02, 2017 3:44 pm

Re: 2001 Citroen C5 HPi 2.0 HPI (Direct Injection)

Post by mzk »

The inductor has an inductance of 3.3 mH and resistance of 2 Ohm.
The measured current across 0.1 Ohm resistor was ~1A for the steady state (+14V) opening cycle.
The high voltage (opening) pulse has current rising to 12A (for the 200us ON(HV) period).
The continuous opening current (after the applied high voltage) is about 2.25A
CH2 shows 2V but it actually 200mV/div because I changed the probe to x1 but forgot to change the scope setting from x10 to x1
CH2 shows 2V but it actually 200mV/div because I changed the probe to x1 but forgot to change the scope setting from x10 to x1
SHS00005.BMP (219.43 KiB) Viewed 16912 times
CH2 shows 500mV but it actually 50mV/div because I changed the probe to x1 but forgot to change the scope setting from x10 to x1
CH2 shows 500mV but it actually 50mV/div because I changed the probe to x1 but forgot to change the scope setting from x10 to x1
SHS00008.BMP (219.43 KiB) Viewed 16912 times
The emulated injector can have similar inductance and resistance in parallel. The signal feeding the STM8 can be taken using an op amp or comparator with threshold from the resistor side (rising and falling edges).

According to the board - everything is fine. I prefer thru-hole bread board, but SMT is also fine. It will take some time to order the PCB here, or delivery time from sky-macau. If you're okay we can only discuss the schematic, prepare very early variant on thru-hole bread board and later the PCB can be made with SMD components.

I'm thinking of one board per injector, so as to take measurements and proof of concept (first only on one cylinder). Later 4 boards can be used for proof of concept with emulated injectors and the factory ECU, then if it works well rusEfi can replace the ECU.

I think there are some current regulator IC's with output enable. The current sense resistor may be switched "on the fly" in order to change the current required by the injector. What do you think? Or maybe one of the before mentioned specialized ICs can be used. But this is some of the next steps.
Last edited by mzk on Wed Jun 14, 2017 12:34 pm, edited 2 times in total.
mzk
Posts: 21
Joined: Fri Jun 02, 2017 3:44 pm

Re: 2001 Citroen C5 HPi 2.0 HPI (Direct Injection)

Post by mzk »

PS: the only changing parameter in time is the injector keep-open time (after the high voltage pulse). If the engine is under load or forced, the energizing and HV cycles stay the same. Or at least that I measured with the car at the parking, no driving conditions.

PS2: now I'm curious where to buy Frankenso from? Do you have some seller in Europe? I would like to skip the taxes. I can't find company here that can produce the board at reasonable price. Other way is pcb-way.com, it is more expensive, but they'll prepare 5 boards and I plan to experiment on some other engines, so I will need more boards.
Rhinoman
contributor
contributor
Posts: 256
Joined: Thu Sep 24, 2015 2:14 pm
Location: Wiltshire, UK

Re: 2001 Citroen C5 HPi 2.0 HPI (Direct Injection)

Post by Rhinoman »

You may find that at low battery voltages additional high voltage pulses are inserted to maintain a constant current; there can be quite complex strategies with HV injectors.
Switching the injectors on the high and low side is standard practise for a low-z injector. The upper transistor is switched, the lower transistor is turned on for the injection period, the current sense resistor is between the low-side switch and the ground rail.
mzk
Posts: 21
Joined: Fri Jun 02, 2017 3:44 pm

Re: 2001 Citroen C5 HPi 2.0 HPI (Direct Injection)

Post by mzk »

Thank you for your clarification! I'm doing the buffer now and once it works in normal conditions I'll think for the critical conditions. Till now the low current opening cycle and the HV current cycle were always the same.
User avatar
kb1gtt
contributor
contributor
Posts: 3758
Joined: Tue Sep 10, 2013 1:42 am
Location: ME of USA

Re: 2001 Citroen C5 HPi 2.0 HPI (Direct Injection)

Post by kb1gtt »

For current sense, a hall sensor chip like the ACS712 or ACS709 would likely work well. They have low internal resistance, which avoids a bunch of problems, as well they can be obtained on eBay and aliexpress as prototype board for low $

http://www.ebay.com/itm/New-20A-range-Current-Sensor-Module-ACS712-Module-Arduino-module-/401166650666?epid=936577141&hash=item5d67654d2a:g:7QYAAOSwnDZUA1XJ

I'm using a ACS725 on servo.
Welcome to the friendlier side of internet crazy :)
User avatar
AndreyB
Site Admin
Posts: 14292
Joined: Wed Aug 28, 2013 1:28 am
Location: Jersey City
Github Username: rusefillc
Slack: Andrey B

Re: 2001 Citroen C5 HPi 2.0 HPI (Direct Injection)

Post by AndreyB »

Any news?
Very limited telepathic abilities - please post logs & tunes where appropriate - http://rusefi.com/s/questions

Always looking for C/C++/Java/PHP developers! Please help us see https://rusefi.com/s/howtocontribute
mzk
Posts: 21
Joined: Fri Jun 02, 2017 3:44 pm

Re: 2001 Citroen C5 HPi 2.0 HPI (Direct Injection)

Post by mzk »

Unfortunately not enough time to work on my hobby projects now. The last unpublished stuff is stm8s103 based digital buffer that samples the input (active high from rusEfi) and does 80 kHz current-mode PWM to drive the injector. It is in early development phase and only some of the software works and no hardware was developed.
I'll try to continue my work this summer.
Post Reply