Electronic VNT controller design - Updated!! Rev 0.1 :)

[Sprockets]

Hi all.

Just been reading this forum for the past few hours and your ingenuity and skills are a pleasing difference to most car sites out there.  I mainly deal with performance cars and bikes in my day job, but never did to much on the mechanical diesel side of things.  That changed a bit when i bought a 300TDi landrover discovery, and did all the usual pump and turbo mods, but i felt the limiting factor was the exhaust manifold design and turbocharger.  That got me designing a tubular manifold and looking at the vast array of turbo choices.  I had quite early on gone for the VNT choice but was struggling with the control mechanism.  There were lots of variations, but i always felt that this turbo was designed with total control in mind, so was never 100% happy.

I get to see lots of the modern VAG diesel engines, and play with em diagnosing all sorts of faults which involves a fair bit of reverse engineering some times  During this time i came up with a lot of conrol ideas and what not, but unfortunately I had to sell the landrover and gave up with the entire idea. (BTW, the new owner couldn't believe how well it went, said it was the quickest one he'd ever driven!)  But now, looking at this site and various posts it seems there may be a market for a simple, low cost controller.  I have recently just built an ignition controller for a single cylinder 4 stroker engine, and feel i could reuse a fair chunk of the design!  I will post up a general overview/flow diagram along with basic specs soon, as it's currently sat on several pages of my workbook

There comes one problem tho....i currently do not have a real car to protoype on.  I will be building a test bench, but nothing can replicate the environment of a running engine!  I do not mind sending demos out to people who don't mind testing and are not worried about using a PC!  This will be a while yet, so no excitement, as I like to design everything right first before i get my protyping kit out.

Sorry for the long first post guys, as I say, give me a while, and i'll post some ideas up and I'll gladly recieve any input.  As I say the main aims are to be low cost, ease of use and good functionality.

Cheers all

-Gavin

*UPDATE*

Well guys, I know I'm a tad late with replying, but here is Version 0.1 (pre alpha alpha lol) of the schematics.  Please note, currently a lot of component values are missing, will get round to it later on in the day! Also, I'm not happy with the power and grounding situation, again, will be rectified.  And the last thing, there are a few little features sneeked in.  Firstly, the 'state conf' pin, will be for selecting different maps (high and low boost etc), and also can be used for the second little feature, which is something I have implemented as I want to play with it.  Notice there is a second high power output, this will be for experimenting with LPG injection.  Seeing as I have access to RPM, Boost and TPS, thought it'd be rude not to, also have spare ADC for CLT.  But, this is not it's only use.  It could be used for any output you like.  Water injection, NOS, you could even get it to trigger your wipers every time you put your foot down if you liked lol.

Final thing is the use of the LM1815 for RPM.  This gives people the choice of W terminal or a modern VR sensor.  Will just require a few component changes, I want to get it jumper configurable, but thats later on.

Code and GUI are still in my head and on paper.  Will give some sneaky screenshots etc soon.

Comments and criticisms welcome!

[gigaz2]

I wondered why nobody had already come up with something like that

on the hardware side we would need:

INPUTS:
-pedal travel (analog)
-manifold pressure (analog)
-rpm (digital pulse)

optional:
-turbo pressure cold side (so it could detect spikes more easily and leaks on the intercooler)
-air pressure (to do altitude correction on the fly)
-air temperature exit(turbo)
-air temperature (manifold)
-air temperature input(turbo)
-turbo pressure hot side (so we can directly use a turbo map)
-EGT pre-turbo (hell why not? plus we can't calculate flow without it )


OUTPUTS:
-vacuum control valve (N75 alike, PWM drive)

optional:
-General purpose outputs, for driving other stuff like nitrous/propane/water...
-a serial interface, so we can easily program maps and variables, and look at the nice readouts on the video interface I'm designing...(hush hush)
 

a PIC would do it, especially the last generation, but there are millions of alternatives that comply to the requirements

what do you already have?

[Sprockets]

Thanks for your reply and input gigaz2!

You're thinking is more or less what I have in mind.  As I said in the previous post, this needs to be a simple controller.  I like your extra ideas, especially something along the lines of a K-type for EGT sensing as a safety feature to drop boost in case it climbs out of spec.  All this and more is posssible, but it drives up complexity (not so much of an issue) but also unit cost and size.  The big thing i seem to see is 2 schools of thought.  The first is "I don't trust electronics, i want a mechanical system" and the second is "Wow! How many features does it have?!?!? Unless it can program my toaster and tell me the exchange rate in Japan, I'm not bothered!"  I want to proove to the first group that it is nice and simple, and to the second group optional functionality can be achieved further along the development path.

On to the geeky techy bit.  The MCU I have been working with is the PIC 16F88.  It has all the needed features, can write it's own memory, PWM out, 2 timers, ADC's etc.  Has a nice small pincount (18 in the DIP package for prototyping...be nice to go SMD if theres a market).  

It will be configurable via a simple PC based app using the serial port on early models. Again, if put into a small run, will look into USB as very few PC's/laptops come equipped with that wonderful socket!  But a MAX232 chip is simpler to start with to get a working model.  The app will be a simple 3D table RPM Vs TPS Vs Desired boost and diagnostic/configuration/live reading display.

The output will obviously be a PWM signal to drive a standard run of the mill N75 valve.  The complex part comes from obviously us stating we want X boost pressure, when really, what we actually need is I want this duty cycle!!  This is going to be the complex part of the coding, the required PID loop.  Lots of good examples about, but will be the key to it all.

The RPM input is a tricky one, as being old school diesel, the only way really will be from the W terminal of the alternator.  I have done some research, but found no scope outputs or verifiable documention.  CLosest i got was a web page stating that on his scope, it was almost square wave 0-12v at approximately 25 pulses/rev.  This will have to be investigated and also be software configurable.

An interesting note from this is that this could quite happily work on a "D system utilising just TPS, but it seems in the OEM world, people are now mapping not for a flat boost pressure, but for a flat torque curve, which gives a nice linear HP graph, rather than the typical glob of power in the middle graph

The pressure sensor I am thinking to position either on the controller itself so as there is one less thing to wire up, making it simpler, and also no confusion on any replacement parts/calibration.  was going 400kpa which should be enough for all you power merchants out there! (400kpa being 300Kpa above atmospheric = 3 bar = 44psi )

The hardware is the simple part, the code is the real hurdle.  Hopefully tomorrow I shall have the basic schematics done, and have a flow system for the code.

Cheers for reading this far!!!  I ramble to much lol, but my brain has lots of ideas all the time, need to bleed some off somewhere, forums are an ideal place lol.

-Gavin

[gigaz2]

I just raided my microchip drawer, and I only have 16f627A's and a 16f872, and a lot of the smaller ones, I use pics for the cheap, low power designs.

the W terminal is simple to understand:
the alternator is a three phase machine, so the stator has three coils that are fed to six diodes in order to make the direct current we need.
the rotor is what creates the magnetic field, it is fed DC via the regulator in order to sustain a correct voltage output.

the W terminal is just taken from before one of those diodes, its a large, sinus-shaped signal, amplitude depends on rpm and alternator load, huge spikes are normal there :S

[gigaz2]

problem with W signal is that because of belt slip, its not that reliable.
I would design something that could use both the W signal or a crank sensor like the TDI's use

[Sprockets]

The problem with the 2 MCU's you have is they can't self write

With regards to the W terminal, thanks for the clarification, not to much of a problem with the noisy output, should be able to sort something.  And yes, i agree getting a reliable signal from it will be tricky.  Only decent method would be as you say TDI setup, or just a simple hall sensor on the cam/crank/IP/flywheel/anything that turns!  Would make the electronics design nice and simple, but adds an installation hurdle.  This isn't a great concern tho, as I said, can happily run in 2D mode if require for the user who doesn't want the added complexity.

-Gavin

[arb]

The RPM input is a tricky one, as being old school diesel, the only way really will be from the W terminal of the alternator.  I have done some research, but found no scope outputs or verifiable documention.  CLosest i got was a web page stating that on his scope, it was almost square wave 0-12v at approximately 25 pulses/rev.  This will have to be investigated and also be software configurable.

Gavin, welcome ! Are you in the UK ?

You could use a magnetic or optical coupler for your tach if you don't have a W terminal on your alternator (I'm using a Chevy alternator)

http://www.qsl.net/m0pmb/tacho.html

There have been several posts by guys doing it this way...

[gigaz2]

no problem on the self writing, we put all variables on the eeprom and thats it, firmware on the flash and tables/variables on eeprom.

on the W signal, thats not so hard, resistor/zener to clip the signal to a lower amplitude, then a ttl converter out of a transistor.(just one way to do it)

all the filtering/interpolation would be done by a routine that discards very different values (like going from 1000rpm to 4000rpm in 1/10th of a second.. our engines can't do that)
allowing enough filtering even belt slip issues could be taken care of.

[Sprockets]

Cheers for the responses guys!

arb: Yes mate, I'm from little old England  not to bad a place, just wish we had some good weather!!!  It's OK for my off roading fun, but not for my bike racing

To get a decent RPM reading, the only simplistic and reliable method will be hall sensor.  Great thing with these being they generate a clean signal, and only really need something as crude as a bolt to read from!

libbybapa:  Thanks for the interest  Yes, I have read that thread, and interesting with regards to the limit switch to detect stuck vanes.  I know this is a major problem as i have rebuilt and replaced enough of the bloody things!  Again, it is a reliabilty issue.  Would need some investigation, but is an easy addition hardware and code wise.

As for the stepper motor, would be a tricky thing to figure out how much power is required to operate the vanes under normal load, and when to say "Ooops! we can't move anymore, do we keep trying and possibly burn ourselves out or are we starting to stick?".  Another issue as you pointed out is cost, the VNT turbos already come with the vacumn setup, and if it's good enough for VAG, I'm happy enough to use it  If it is implemented with the limit switches you mention, i think it will be fine.

Thanks for your input!

-Gavin

[gigaz2]

Because the resistance to motion varies, trying to control the vane position based on other inputs without a vane position input would be unreliable at best.

haven't thought about that well, even VW haven't thought that would be an issue as they don't use any kind of feedback (except on the newest turbos that have a specific electronic actuator)

any of those solutions would put the average enthusiast down, at most, the microswitch fully open/closed position sensor would be the simplest.
the unit could calibrate itself everytime the engine started and flash a light to let you know that the vanes were stuck, or at least that it wasn't capable of moving them as it should (vacuum leak for instance)

[Sprockets]

gigaz2:  I like the self write mainly due to have bootloader and firmware upgrades in field.  If a protoype goes out, more than likely not going to be able to have access to a programmer even if i put ISP headers on!  And yes, W terminal output condition is no problem, and i hadn't thought of the software filtering method you described.  Will put it in my notes and have a think!

Cheers

-Gavin

[Sprockets]

Interesting device there Andrew.  Have you looked into it's function at all?  Ie is it a simple 2 pin switch that says it's at one of it's extremities, or is it a 2 position 3 pin switch detecting open and closed?  Or (which i doubt) a linear potentiometer to feedback exact position?  Do you know the turbo manufactuer it came off?

Sorry for all the questions, i just like figuring out new stuff i haven't seen before

-Gavin

[Sprockets]

Top information!!  Nice one, cheers Andrew.  Thats a very nice part then, as it is already automotive grade  I have enough ADC's to use the info it can give us.  Good thing being the controller doesn't have to be dependant on this, but is an easy upgrade for people wanting to know the turbo is still sound.  IE, an optional part for people that don't want the hassle of installing.

Right then guys, thank you all for your feedback, now off to the pub for some more brainstorming with a few other automotive geeks.  Shall return tomorrow with more ideas/news/drawings/schematics/hangover :p

Oh, Andrew, is there a part number on there?  I would like to find out if it is obtainable, as I had a VNT actuator die on a 3.0tdi Audi A6 once, and no one would supply one evn tho the turbo itself was sound, as everyone said it was calibrated with the turbo.

-Gavin

*EDIT*  You already gave the numbers, my bad!!! */EDIT*

[aidan]

http://www.vwdiesel.net/phpBB/viewtopic.php?t=11793

If you fancy making the controller as on that thread, i'd buy it, Its not for a VW and has an electronic pump, with TPS and MAP already on there, turbo is standalone wastegate, no boost solenoid or anything.

[ahto42]

google translate polish to english