Micro controlled VNT project

[regcheeseman]

Blue Mule,

Sorry but you've lost me completely now.  

There is no spring and no can in my design.

I am monitoring servo position and therefore vane position - Is that not AR? I really don't understand it's relevance as a measurement when I have the other conditions being monitored?

At off idle the three cylinders would not produce enough volume/velocity of exhaust gas to turn a larger turbo, thus you get turbo lag. So with the VNT you can adjust the A.R., thus allowing for a smooth transition between off idle up through the full RPM/Power band.

So the bottom line is that with a variable A.R. you can have a large turbo, which at lower RPM’s, can act as a smaller turbo, virtually eliminating turbo lag. This is the genius of the design. So you might want to focus on programming the aspect ratio rather than pressures.

I had thought this is what I was doing?? In that the thing being controlled is the AR. The criteria for it's position being the throttle position/revs condition.

I.e If I'm nailing the accelerator to the floor the AR will be forced to push the boost as high as possible
I.e If I'm light on the accelerator then the AR will be relaxed

My only use for the boost measurement is to implement a limit (in fact 5 limits) - low rev boost will be severely limited as I understand too much at low revs will kill the engine?

The map is easily modified with a usb lead and laptop and all the boost limits can be reconfigured in situ with no additional equipment by pressing a button during the units start-up diagnostics check.

[regcheeseman]

I think you're totally on the right track.

Thanks for that, it's a bit to late in the day for any u-turns! Time will tell I think.....

I'll probably run the car with the servo un connected and the VNT fully open for a bit and rig up a servo position indicator just to monitor what the turbo vanes would potentially be doing. It's fine testing on the bench but until the unit is properly hooked up to an engine I'll not really know how it will work.

[BlueMule]

  Reg, I agree with Andrew that what you are designing will work to control boost, and the vehicle will no doubt be very drivable. Just to add I think you are doing an excellent job in your set up including the Tach. Please understand, my purpose is not to be critical either, it was just to point out that the real genius in this type of technology is in the drivability aspect. Also please keep in mind I am a Southerner with German born grandparents on both sides, so I tend to be a bit direct in my speech, and I don't mean to offend

So if I may, allow me to illustrate. A turbocharged vehicle with a direct link to the nozzles is driving at 60 mph at 3k RPM, it encounters a large incline, like Cheddar Gorge, the engine "feels" the load and the RPM's start to drop a bit, the driver steps on the gas to add more fuel to keep the RPM's and speed of the vehicle up, also since the AR is linked directly to the throttle, the "system" makes the AR larger, this may sound great, but in reality under these conditions you need the aspect ratio on the turbine to get smaller so that the turbo will stay spooled with the lower RPM due to engine load. Now someone will jump in here and say "Blue all you have to do is downshift", yes this is true, but then you are having the same situation as a fixed AR turbo. One of the drivability aspects of a VNT is to cut down on the shifting in the manual trans and auto trans. There are other conditions of course, but I think this one is the most clear in real world application.

Now as far as the sensor situation, the RPM, Throttle, VSS, are a very good start, but think how much more you could get out of a system that uses Air Mass to help filter the other three inputs, again this is just a thought, and I am not saying it is absolutely needed to function.

As far as control, I see you have replaced the “can”.  I do have a thought for future reference; you only need a vacuum solenoid and a circuit that uses a TI 555 timer to alter the duty cycle to the solenoid this would allow a very precise sizing of the AR while using the original can.

Anyway, much respect to you and your project, and I look forward to you further installments.


“Spring little cobra with all your might”

[anto]

Reg are you planning to develop a few more of these sale?
Id be very interested in it and on my 306 i already have the boost input, revs input and throttle poistion input.
The programming etc i just wouldnt have the hardware or the brain power to even begin to start making one.

[gldgti]

So if I may, allow me to illustrate. A turbocharged vehicle with a direct link to the nozzles is driving at 60 mph at 3k RPM, it encounters a large incline, like Cheddar Gorge, the engine "feels" the load and the RPM's start to drop a bit, the driver steps on the gas to add more fuel to keep the RPM's and speed of the vehicle up, also since the AR is linked directly to the throttle, the "system" makes the AR larger, this may sound great, but in reality under these conditions you need the aspect ratio on the turbine to get smaller so that the turbo will stay spooled with the lower RPM due to engine load.

True - but: When you add fuel (as the incline increases and you put your foot down) you are increasing exhaust gas flow aswell. Infact, by adding fuel you are able to increase exhaust gas flow by an incredibly large amount compared with the change in rpm.

In an ideal scenario, (lets take your getting to a hill example) the vanes will go wherever they need to in order to keep an optimum (caution, gasser terms here) Air Fuel Ratio. In a turbo-diesel, we need to look at it this way - for high load application, we want to keep EGT pegged to a critical value of our own choosing. This ofcourse is largely governed by the AFR at any given time. What we want to achieve ideally is to increase boost with fuelling to keep the EGT in check, up to the point of maximum torque output for whatever rpm you are at.

What consequences does this have for our controller design? We need to consider what it is the vane controller is actually doing. The controller shown in this thread uses a "map" (the matrix of numbers) that depends only upon 2 inputs. With 2 inputs, we need to choose exaclty which inputs are going to be the most useful to achieve a given outcome. We can say that we will manually tune the EGT at full load with the LDA, so clearly the best inputs should be boost pressure and rpm, as these allow: 1) a controller feedback; and 2) idle/midrange/full speed tuning capability.

 

[regcheeseman]

Also please keep in mind I am a Southerner with German born grandparents on both sides, so I tend to be a bit direct in my speech, and I don't mean to offend

Man, you'll have to try a whole lot harder to manage that!  

also since the AR is linked directly to the throttle

no, not exactly it's a function of revs and boost too - so in your scenario the vanes won't just open up and create lag  

(was the 'cheddar gorge' reference purely for my benefit? - like it!)

As far as control, I see you have replaced the “can”.  I do have a thought for future reference; you only need a vacuum solenoid and a circuit that uses a TI 555 timer to alter the duty cycle to the solenoid this would allow a very precise sizing of the AR while using the original can.

Why the fascination with the diaphragm can? It's a poor solution with considerable hysteresis and limited controlabilty. It would also shift the system from being electro/mechanical  to pneumo/electro/mechanical - (ignoring the boost input) , this would add extra complication and potential failure modes, lots of leaky pipes, clunky solenoid valves and other stuff best avoided.

My servo position is adjustable in 256 steps - how much precision do you require in a 100(?) degree arc.
The servo also has feedback built in.
My only reservations with the servo is it's response rate, however I have the option to overclock the processor by 4x if required - currently running at 8MHz.
 

Reg are you planning to develop a few more of these sale?
Id be very interested in it and on my 306 i already have the boost input, revs input and throttle position input.

In keeping with this forum ethos (and my ranting about the covert nature of some peoples knowledge regarding mtdi work ) I will publish everything I can, when I am happy with it.
I may consider making some more - but a system for a peugeot may have to be tailored. The input amp gains may not suit peugeot sensors....but it's merely a matter of three resistors to change.

5v  pulse from pump to get revs signal.
1K - 1.5K variable on throttle position
The servo is currently from a Honda NSR250 but I may switch to a more available yamaha unit.

[BlueMule]

Reg, OK,

As far as the can is concerned, since its already on the turbo I'm just thinking out loud that it would be one less thing to have to modify, but I know that dodge sprinters use a system that is completely electronic, so as long as your solenoid can function with the heat, then it can be very precise in its metering.

Mule Out

[Vincent Waldon]

I think what we are seeing here is the good old "80/20" rule... 80% of the problem can be solved with a simple 20% system... the remaining 20% finesse takes 80% more engineering.  ;-)

Andrew's mechanical control has demonstrated that good drivability can be obtained using a vane controller with only two mechanical inputs:  throttle position and boost.   Reg is taking it to the next level by adding a third input: RPM... as well as putting electronics into the mix. Mule is talking about even more sophistication, more inputs, arguably more finesse from a control perspective

All of these designs are valid...it's up to the reader to decide when they are on the other side of the 80/20 curve... ie, when is further finesse not worth the additional time, effort, expense??  Guys happy with wastegate turbos would probably argue "too late... who needs vanes in the first place!!!"   

[regcheeseman]

The primary reason for chosing an electronic solution over a mechanical one was cosmetic. I'm aiming to get a 'scene' accepted engine bay with a diesel engine.

Every thing that could be relocated has been, wiring hidden, holes plated etc - to fit a bunch of cables, pulleys, levers and spring would ruin the look.

The servo will be remote mounted in the car (well away from heat) and two discreet bowden cables through the bulkhead/firewall directly to a pulley on the turbo.

Incorporating the boost gauge into the factory rev counter was also important in maintaining the 'look'

It's always my ethos that 'Form follows function' but having both is a bonus...

[regcheeseman]

Libbybapa,

I'm really impressed that you seem to know better than me how my system operates! Having mentally converted the systems operation into a set of variables and then translating those variables into a variety of voltages and then translated those voltages into PWM values of frequency and duty cycle - I've actually forgotten what the design was supposed to do in the first place!
Now I've got to final testing, I've had to go back over the original scribbles and posts on here to jog my memory!

[regcheeseman]

Neat! Dimmable?

I've finally managed to get some code loaded (test routine) into the processor and now bacause it is running from a PWM signal from the processor the display actually increases in intensity as the boost increases!

Way cool 

Oh yeah, it wasn't just a fluke - I meant that to happen honest!

[arb]

Very awsome project !!!  Do you have an EGT to at least see how hot your turbine wheel it getting, if not an input to V2 of your project ?

[745 turbogreasel]

It seems like any further inputs would actually be redundant.  Reg is already measuring air mass in that it is a function of rpms and boost.  He's already got servo position which is in turn A/R.  I can't see how VSS would be worth measuring as RPMs are already measured (do the vanes care what gear you're in?). 

Might be worth something if you are programming for reduced output in lower gears to save you made of glass transmission?
Turbine speed would be a handy input from a diagnostic standpoint, but not easy to implement on our R&D buget.

I like where this is going:)

[regcheeseman]

Are the gearboxes that weak then? I knew the clutch was weak but the 020 seems to take some abuse, I'll be running the cable change 02A TDI box anyway.

Tonight I've been thinking of whacking a shift light in, it would require a small circuit board to be added and a little bit of code, but would be user configurable at startup - as are the boost limit settings - you just have to hit a button within 5 seconds of turning the ignition on to enter the config mode or the unit continues into normal mode.

I was thinking of putting the shift light into the revcounter but it may be too crowded so add it as a remote output switched 12v line.

[745 turbogreasel]

On an IDI, I don't really think so, but the ALH TDI with an auto sure seems to do some funny stuff in an attempt to preserve itself.  A lot of Honda products, you can't manually select 1st gear...
More of an issue with engine converted vehicles, running slicks, etc