300Tdi VGT project

[Turbocharger]

Right, a day's holiday booked and all the Chrimbo shopping done, so the new turbo's back on (I'll be the first person to wear out a set of manifold studs at this rate...), and Julian's diaphragm looks very sexy on top:

I bought a new oil drain hose to take the oil away again, replacing the cobbled-up one with a join halfway along it. Pirtek made me exactly what I wanted (and charged me £40 for the pleasure). It sits near horizontal at the lower end which I'm a bit worried about, but this seems to be the best balance to stop it kinking where the arrow shows. I tried trimming the length, this is best. Will this work?

Now for the water cooling:

Pirtek cobbled up this 10-16mm adaptor for the water hose by crimping some thick-wall tube to replace my copper B&Q reducer, but I'm not really happy with either.

John's link to info on the Garrett VGT suggests that their turbo has water cooling for applications where a heavy manifold will store heat and soak it back to the turbo on shutdown. Mine's just made of thin tube so this shouldn't be an issue - which solves the water cooling issue. Am I just seeking an easy answer to the problem?

Anyway, I'm quite ashamed that this is pushing on for a six-month project but I should be playing with the springing in the new diaphragm tomorrow.

[Turbocharger]

Right, my two questions above have been answered by the day's driving:

• the oil drain is fine since it hasn't seized
  
• the lack of water is fine since it hasn't caught fire
  

Now I'm wondering about springs to go in the diaphragm. In the diaphragm that Julian supplied there is a conical (progressive?) spring, and I've got another from a spare diaphragm (not shown) - ignore the scribbly diagram underneath, that's from something else:

In my mind I'm aiming for this logic:

But when I took it for a test drive, it doesn't move at all under boost. Some playing with an airline, a pencil and some scribbled maths bring me to this conclusion:

Which says it doesn't want to move until 3 bar or so with either of the springs I've got. The minus 23 mm is the distance that the spring is pre-loaded by screwing it into the diaphragm housing. I can increase this pre-load further by turning the adjustment screw that Julian's provided for me, but this diagram shows that'll send it further the wrong way for the springs I've got.

The tools I have are:

• adjust the end stop to preload the spring (make it shorter)
  
• chop the length of the springs down (but this will affect the rate)
  
• pick a weaker spring so it picks up off the end-stop at a lower pressure
  

I think the third option is preferable, and I think the my logic diagram means that I need it to sit still until nearly 1 bar and then extend fully - in practice it'll find a middle ground because extending it will reduce the boost that the turbo is producing. I'm in head-scratching mode now, not sure which way to go with it.

[Bush65]

Just some basics of the top of my head that may help.

Force from boost pressure acting upon the diaphragm, and apposed by the spring is:

F = boost pressure x area (where area is pi x diameter^2 / 4 and dia is inside the of the housing).

Spring rate is force to compress spring unit distance.

Now as an example say you want the vanes to go from closed at 1 bar to fully open at 2 bar.

Measure the stroke of the actuator that corresponds with moving the vanes from closed to fully open.

For sake of this example assume the inside diameter of the actuator (effective diameter of diaphragm is 50 mm.

And assume the required stroke is 10 mm.

Then the pressure difference is 2 bar - 1 bar = 1 bar = 100 kPa = 0.1 MPa

The effective area of the diaphragm is pi x 50mm x 50mm / 4 = 1963 mm2

Then the change in the force that will act upon the spring is 0.1 MPa x 1963 mm2 = 196 Newton

Then the spring rate required is Force / displacement

i.e 196 N / 10 mm = 19.6 N/mm

If you want the spring rate in kg/mm then

kg = 196 N / 9.81 N/mm2 = 20 kg (9.81 N/mm2 is gravitational acceleration)

Then spring rate is 20 kg / 10 mm = 2 kg/mm

If you want spring rate in lb/in

Change the boost pressure to psi, the effective diam to inches and the stroke to inches

1 bar = 14.504 psi

50 mm = 1.968"

10 mm = 0.397"

Then:

effective area is pi x 1.968" x 1.968" / 4 = 3.042 in2

change in force acting on spring is 14.504 psi x 3.042 in2 = 44.1 lb

spring rate is 44.1 lb / 0.397" = 111 lb/in

So now we have a spring rate, we can determine how much the spring should be preloaded so the vanes don't move until the boost pressure is 1 bar.

This is the distance that the spring will be compressed by the force produced with 1 bar.

This pre-load force is pressure x area

For the above example pre-load = 0.1 MPa x 1963 mm2 = 196 N (or 14.504 psi x 3.042 in2 = 44.1 lb)

And the distance the spring will compress with this force is Force / Spring Rate

i.e. spring should be compressed 196 N / 19.6 N/mm = 10mm

or 44.1 lb / 111 lb/in = 0.397"

Obviously then the chosen spring must be able to be compressed a distance greater than the pre-load plus stroke.

Also it is unlikely you will be able to buy a stock spring with the exact spring rate calculated.

So use the closest spring rate and work the calculations in reverse to determine what change in boost pressure will compress the spring the required distance to go from vanes closed to open. Then play around with the pre-load to adjust these other variables.

[David Sparkes]

I think you need to develop a greater understanding of the detail.

Here's my understanding of that detail.

The physical advantage of a conical spring is that when fully compressed the overall length of the spring is less (than if the coils were all the same diameter). Thus the valve can be made shorter.

You suggest the spring is "progressive?", but the question mark shows you aren't sure if that is correct.

In fact, all springs are 'progressive', because the tension increases (progresses) as the length alters. It is the manner of that progression you are interested in, and the values encompassed with the progression.

The wire diameter is consistent throughout the length of the spring.

As the spring is compressed, the wire in each coil twists, the overall effect is a reduction in length of the spring.

If you consider each coil separately, the first coil has the largest diameter (or the longest length), therefore the lowest rate (least rise in force for a given twist).

The second coil has a smaller diameter (thus length of the circumference) so has a slightly higher rate.

The third coil has a slightly smaller diameter, so the rate is slightly higher again.

The overall effect is that as pressure is applied to the spring, initially it is the first (longest) coil that twists the most.

As the compression loading increases the first coil becomes flat (and thus ceases to twist), but the smaller (shorter) coils start to twist more.

The overall effect of this is what you need to establish. I can't tell you what this is from my general knowledge.

I suspect the answer is one of two things.

A/ The rate of the spring (compression force over spring length) remains consistent over most of the compression length available.

B/ The rate of the spring becomes higher and higher as the spring becomes more compressed.

You need to research this, either by reading, or by empirical measurement.

Moving on, it is beneficial if you know the actual force required to compress the spring (and therefore the force exerted on the diaphragm), so this pushes you towards research by measurement.

Load the spring by a known amount of weight (This is the compressive force) and measure the resultant length.

OR

Shorten the spring by a known amount, measuring the resultant force.

You have attempted this using air pressure, but have led yourself astray, because you have forgotten to take into account the area of the diaphragm that interfaces between the spring and the air pressure.

An example.

If your diaphragm has an area of 2 square inches, and you apply an air pressure of 10 psi, the load on the spring is 20 pounds force.

If your diaphragm has an area of 1.5 square inches, and you apply an air pressure of 10 psi, the load on the spring is 15 pounds force.

Springs will be rated in terms of force (pounds, kilograms, ounces, whatever), and you need to multiply that by the diaphragm area to see the resultant line pressure in force per unit area (PSI, kg per sq cm, whatever).

If you follow the detailed method, and measure the progression line of the spring outside the valve assembly, you will be able to decide if you can use the existing spring, but add a spacer to the valve so the spring is less compressed when the valve is assembled. Because you can also establish how far the diaphragm can physically move, you can also pinpoint the maximum pressure the spring will exert.

Once you have established the mechanics of testing this spring, you can test any other spring you find, and 'know' what the results will be before you assemble the valve.

If you don't do the detailed method you are condemned to assembling and disassembling the valve for every spring you want to try, and for every spacer variation you want to try (just in case you have the right spring, but are using the wrong section of its' operating range).

I suspect (hope) that this post and that of Bush65 are complimentary, but I haven't read his in full to check this :-)

I suspect we have both picked up the point about considering Force AND Diaphragm Area, not just saying 10 Pounds Force = 10 PSI.

Cheers

HTH.

[Turbocharger]

Thanks both. I did the maths as John suggests before I posted, and I understand that a conical spring has a rising rate. I can see that I need to make the unit "stroke" around 1 bar (since that's the level of boost I want to run). My question is - do I want it to start moving as soon as any pressure's generated, or to select and preload a spring to move fully through its stroke around 1 bar.

[David Sparkes]

Thanks both. I did the maths as John suggests before I posted, and I understand that a conical spring has a rising rate. I can see that I need to make the unit "stroke" around 1 bar

I disagree, you do not want an on / off switch.

My question is - do I want it to start moving as soon as any pressure's generated, or to select and preload a spring to move fully through its stroke around 1 bar.

The experience I can relate is that the wastegate actuator starts to move on my BMW derived engine (Mitsubishi turbocharger) as soon as any pressure is developed by the turbocharger. If this is what a factory developed system does then I'm not sure you have an option about not wanting it to start moving as soon as any pressure's generated, if you are 'just' using a single actuator.

The alternative mode of operation you are suggesting is that the valve doesn't move with say 14 psi line pressure, but fully moves with 16 psi line pressure. With a single valve I don't think that's achievable, nor desirable. In your case it would mean the vanes go from fully closed to fully open, almost instantly.

There is also the point that, based on your current experiences, with the vanes fully closed the turbocharger may not achieve 14 psi boost anyway, as it simply won't be able to flow the exhaust gas.

I think you need to aim for a continuous movement of the valve (and therefore vanes) as the manifold pressure changes from 1 psi to 15 psi.

Surely this is the reason to have a fully variable vane system, to always have the optimum flow rate, both in terms of flow capacity and air speed, at any and every point of the turbochargers operating range?

You also need to bear in mind the point made in the other document about the benefits of keeping the vanes moving, to prevent them sticking.

HTH

[Turbocharger]

I can better show my question now. The spring I have is 60mm long and 22 N/mm (don't ask me in your furlongs and cubits). That's the thin black line on my graph, and it doesn't start moving until nearly 3 bar so it won't work.

Question is, what do I want?

• The green box shows my working 'area', 0-1 bar and 0-7mm stroke on the VGT control arm.
  
• The purple line suggests a spring with no preload, which would start to open the vanes as soon as it made boost. Spring 37mm free length, rate 24N/mm, zero preload.
  
• The blue line is the same spring as the purple line but with some preload. The vanes don't move until 0.5bar, but it needs 1.5bar to fully extend the vanes. Spring 37mm free length, rate 24N/mm, 5mm preload.
  
• The orange and red lines uses less stiff springs with much more preload to get the working range in the right place. The effect of these is to compress the operating pressures so that nothing moves until say 0.9 bar, and then hits full extension by 1.1bar. Spring length 60mm, rate 8 to 10N/mm, 23mm preload.
  

My question is - which do I want?

(The secondary question is "can I achieve it with the springs I've got?" although I suspect the answer is no since I need a shorter spring, and cutting mine will increase the rate excessively).

[David Sparkes]

I think I've already cast my vote -

... you need to aim for a continuous movement of the valve (and therefore vanes) as the manifold pressure changes from 0.1 bar to 1.0 bar.

Unfortunately I can't immediately suggest a source of springs you can mine to achieve this.

As a matter of interest, does Lara's valve use a diaphragm that is the same area as a standard 300TDi actuator?

Cheers

[Lara]

Have loads of springs,

Tell me what length you need,

what load at fitted length.

What stroke needed.

Lara

[TheRecklessEngineer]

As far as I can see this is a control theory problem. Control theory

In essence, you have what is known as a 'P' controller - this is just the proportional part of a PID controller. PID controller You want the turbo to produce as close to 1 bar as is possible - this is your reference signal.

The answer will lie somewhere between having a 'switch' that suddenly opens that vanes at 1 bar and having a progressive change.

A 'switch' system (high gain co-efficient) will produce a response something like the black line - note that the manifold pressure will peak beyond 1 bar before stabilising (if at all...)

A spring that progressively moves the vanes from 0.1 to 1 bar (low gain) will produce a response like the red line. No overshoot, but it won't get to 1 bar as fast as it should - acting against the point of having a VVT in the first place.

In this application I think it would be desirable to have a small amount of overshoot - your engine isn't going to blow up, and you will have an extremely fast response from the turbo.

It would be quite hard to analyse the system in control form as measurements of the turbo performance and the engine would need to be made - I suggest getting a manifold pressure sensor to check you don't get overshoot when you boot it and then play around with springs.

[Turbocharger]

James - I think you're right, but I don't think it matters here - at least not yet. I'm not into the realm of tuning the transient response yet, I just want steady state results. Question is: what response am I aiming for? Julian's VGT Td5 suggests that I need it to move gradually from 0.5bar to around my target pressure, and then use the diaphragm's adjustable preload to fine-tune the liftoff point, and use the adjustable linkage to set the zero point. I think (hope) that good transients will come out of this naturally.

[Bush65]

Have you done any trials to determine how much force is required to operate the vane control mechanism (including friction and forces from the exhaust gasses on the vanes).

I read in a vgt thread in a Cummins forum that the gas forces the vanes to open on their Holset vv turbos, which imposes considerable load on the control mechanism.

[TheRecklessEngineer]

*Re-reads post and ponders for a few mins*

I would suggest you want something like the orange line on your graph. You are, I think, actually asking what transient response you should be looking for.

If you have something that starts moving the vanes at very low boost (your purple line) then your steady state response is always going to be much less than 1 bar - you will have a large error value. This error value will be very large for less than ideal conditions - at low RPM for example.

Imagine you are making 0.5 bar - the vanes will be half open. Half the energy in the exhaust gas will be blowing past the vanes. If the vanes were closed however, you may be able to make 1 bar at this point (which is the whole point of having a VVT in the first place!) This is definitive of a P controller with a low gain value.

However, if you have something that starts moving the vanes at 0.99 bar and stops at 1.01 bar, you will have a very small error value whatever conditions you have. The vanes will remain closed (promoting max boost possible) until 0.99 bar, whereupon they will open almost immediately. This is quite extreme and I would suggest that you would never get a stable response - it will bounce from one limit to the other. This is definitive of P controller with a high gain value.

So - I would try something like your orange line. Bear in mind that you will always have some error - I would aim to have the vanes fully open at a little past 1 bar, maybe 1.05 or even 1.1 - then you can be certain that you will make 1 bar.

I doubt somewhat that you will be able to get this from the springs you have, but you should be able to work something out from your graph. My head hurts now, so I'm going to stop.

P.S. Merry Christmas and thanks for the excuse to escape the family for 20 mins while I ponder this!

[Bush65]

... the vanes will be half open. Half the energy in the exhaust gas will be blowing past the vanes...

With variable nozzle turbos that I know of, all of the exhaust gas flows past the vanes - unlike a waste gate system, there are no other gas paths.

At low engine speeds when the mass flow of gas is less, the vanes are positioned to reduce the area and increase the velocity. As mass flow increases the vanes are opened to increase the area and reduce the drive pressure.

[Turbocharger]

Thank you both.

John - it hadn't occurred to me that the vanes might not be balanced, so I'll have to think about that. You're right about the single gas path though - the vanes are capable of opening so far the turbo won't generate any boost at any speed or load, so it doesn't need a wastegate to throw any gasflow away.

James - you're pretty much mirroring my thoughts on what I can remember of control theory. I need a quasi-static response that follows the orange line, and then hope that this gives a sensible dynamic answer (or one that I can damp by clamping the tube to slow it down). Just damping it like this should sort out any problems because the feedback (moving vanes affect the boost that's generated, that moves the diaphragm that moves the vanes) should stabilise it in time.

Given the number of things that can affect it, I think a number of springs and a bit of trial and error is in order

[TheRecklessEngineer]

Bush65 - Yes, I didn't mean quite what I wrote. I meant with the vanes half open the turbo will not be gearing itself to produce maximum possible boost.

[Turbocharger]

More playing today, and possibly a conclusion that I'll look into tomorrow.

I tried to alter the length of the spring so it would pick up from the stop at lower boost, but reducing the free length by clamping the coils increases the spring rate, so this sends its operation in the wrong direction entirely.

I put a new linkage together with my pigeon welder (but no mask, so I used the look away & squeeze trigger approach - it'll do for now).

On a road test it shows 1.5bar boost at peak (it goes well ) but heavy surging again since the diaphragm doesn't move, even at that pressure.

In answer to Bush65's point above, I made a video showing the force required to move the vanes. The answer is, just about none (which is lucky, it makes the maths easier).

Next I went back to basics with the diaphragm, cut all the cable ties to return the spring to its original length and made a bench test setup to see what it does. The answer is: 1 bar = 7mm travel, which should be what I need to make the car work.

So:

• I have a linkage which is tight and doesn't have any wobbly slack any more
  
• the vanes don't need any appreciable force at low loads, so probably small amounts when under load
  
• The diaphragm does what it's supposed to at sensible pressure levels, on the bench
  

I think the last point means that there's probably an air leak somewhere, so I'll repipe it tomorrow and have another go.

[white90]

In the first video I'm thinking you mean force to open the wastegate rather than move Vanes?

where are you measuring the boost? when it is running? can you get the gauge plumbed into the inlet of the Diaphragm to see what pressure is being seen there under load.

[Turbocharger]

No, I mean vanes - there is no wastegate. The vanes are sized such that the turbo can flow all the gas from the engine without making mega-boost, so it just opens up instead of having a wastegate.

Good point about measuring pressure though - I have a pressure tapping on the back of the inlet manifold and this feeds the dash gauge, injector pump and diaphragm by T-pieces. I now suspect that the pipe onto the manifold is leaking so I'll replace it with a tighter fit tomorrow. Of course, everything wants a different bore of pipe so I need hundreds of adaptors or keep pushing pipes inside other pipes etc.

[white90]

Righto on the Vanes, next time I'll read the all the topic to save myself looking even more daft,

but loosing 15psi or above you would I reckon hear it

I had a split wastegate pipe and I could here it even at low boost,

very noticeable at higher boost.

[MECCANO]

Any progress?

[Turbocharger]

Blimey, everyone wants a piece of my time at the moment Yes, I have progress!

I repiped the diaphragm since there was a tiny air leak which was losing "not a lot" of volume but "oh plenty" of pressure. Once airtight, I pressurised it with my workshop airline:

and then measured the diaphragm's movement:

(See, top science in this game).

Then, since it drives, I went 100 miles to see a friend, recover another friend and do some shopping, so my passenger took a video of the dash while I drove. The rev counter is on the left, and the turbo pressure is on the right. 1.0bar is shown just before the end of the orange, and the turbo is quite laggy, doesn't hit 1.0bar until about 2300rpm, and stabilises at about 1.3bar by 3000rpm (by which point I'm munching along at quite a rate!).

This, along with the workshop 'pop off the push-fit hoses with the air compressor' game, gave me some results and allowed me the delight of ploughing through hours and hours of spring tables, selecting possible candidates that is the right diameter to go in the can, not so powerful I can't cram it into the diaphragm with my bare hands and, god forbid, suitable for the job of controlling the turbo.

Using these springs and the data from pressurising my own diaphragm, I now understand it more:

I can see now why it stabilises around 1.3bar at full chat, because the vanes don't move at all until 1.1bar and has travelled through the full 7ish mm available by 1.5 bar, so it stabilises somewhere in the middle. I've had to set the rest point where the turbo can make 1 bar without any vane movement so that it doesn't surge much earlier, so I'm not getting full benefit from the VGT which is why it's a bit 'laggy'.

I need a spring which will start moving at a lower boost level to allow me to set the 'rest point' much more aggressively, knowing it'll lift off as soon as it starts making any pressure. I think the spring shown in purple is my best option. If anyone has a spring that matches these specs, post up or I'll have to go and buy one (at £1.05 each, but minimum order of £10; the graph shows the other nearest matches and I don't see the value in getting more than one of each type )

I need:

Wire diameter: 3.25mm

Outside diameter: 25.4mm

Free length: 38.1mm

No of coils: 6

Rate: 24.5 N/mm

Solid length: 20.3mm

Maybe a phone call for an 'engineering sample' is in order...

[Timmy511]

More playing today, and possibly a conclusion that I'll look into tomorrow.

I tried to alter the length of the spring so it would pick up from the stop at lower boost, but reducing the free length by clamping the coils increases the spring rate, so this sends its operation in the wrong direction entirely.

I put a new linkage together with my pigeon welder (but no mask, so I used the look away & squeeze trigger approach - it'll do for now).

On a road test it shows 1.5bar boost at peak (it goes well ) but heavy surging again since the diaphragm doesn't move, even at that pressure.

In answer to Bush65's point above, I made a video showing the force required to move the vanes. The answer is, just about none (which is lucky, it makes the maths easier).

Next I went back to basics with the diaphragm, cut all the cable ties to return the spring to its original length and made a bench test setup to see what it does. The answer is: 1 bar = 7mm travel, which should be what I need to make the car work.

So:

• I have a linkage which is tight and doesn't have any wobbly slack any more
  
• the vanes don't need any appreciable force at low loads, so probably small amounts when under load
  
• The diaphragm does what it's supposed to at sensible pressure levels, on the bench
  

I think the last point means that there's probably an air leak somewhere, so I'll repipe it tomorrow and have another go.

no offence but that set up wont work correctly anyway. as soon as you pull that linkage it wont move back and forth in a streight line, it will try to pull the linkage in a way that creates the streightest pull possible causing a hell of alot of bind, considering just how little force the actoator is exerting, the binding effect of all the vg vanes covered in hot/dry soot and that linkage will not be helping.

id sugest taking the comp housing off, spining it to a position where you can fit your pipes and also where the actuator an pull in a direct stright line perpendictular to the arm on the turbo.

[rusty_wingnut]

two stage spring?

[Lara]

Blimey, everyone wants a piece of my time at the moment Yes, I have progress!

I repiped the diaphragm since there was a tiny air leak which was losing "not a lot" of volume but "oh plenty" of pressure. Once airtight, I pressurised it with my workshop airline:

and then measured the diaphragm's movement:

(See, top science in this game).

Then, since it drives, I went 100 miles to see a friend, recover another friend and do some shopping, so my passenger took a video of the dash while I drove. The rev counter is on the left, and the turbo pressure is on the right. 1.0bar is shown just before the end of the orange, and the turbo is quite laggy, doesn't hit 1.0bar until about 2300rpm, and stabilises at about 1.3bar by 3000rpm (by which point I'm munching along at quite a rate!).

This, along with the workshop 'pop off the push-fit hoses with the air compressor' game, gave me some results and allowed me the delight of ploughing through hours and hours of spring tables, selecting possible candidates that is the right diameter to go in the can, not so powerful I can't cram it into the diaphragm with my bare hands and, god forbid, suitable for the job of controlling the turbo.

Using these springs and the data from pressurising my own diaphragm, I now understand it more:

I can see now why it stabilises around 1.3bar at full chat, because the vanes don't move at all until 1.1bar and has travelled through the full 7ish mm available by 1.5 bar, so it stabilises somewhere in the middle. I've had to set the rest point where the turbo can make 1 bar without any vane movement so that it doesn't surge much earlier, so I'm not getting full benefit from the VGT which is why it's a bit 'laggy'.

I need a spring which will start moving at a lower boost level to allow me to set the 'rest point' much more aggressively, knowing it'll lift off as soon as it starts making any pressure. I think the spring shown in purple is my best option. If anyone has a spring that matches these specs, post up or I'll have to go and buy one (at £1.05 each, but minimum order of £10; the graph shows the other nearest matches and I don't see the value in getting more than one of each type )

I need:

Wire diameter: 3.25mm

Outside diameter: 25.4mm

Free length: 38.1mm

No of coils: 6

Rate: 24.5 N/mm

Solid length: 20.3mm

Maybe a phone call for an 'engineering sample' is in order...

If your spring is doing what you show in the graph, then that is fairly ideal,

The aim is to get the turbo to spin up as fast as possible ie. to simulate the smallest turbine possible, which it should do with the vanes directing gas inwards towards the turbine wheel.

Then moving to simulate a larger turbine as the boost gets to where you want it to be.

The control can staying on it's seat for as long as possible will do this best!

1 bar at 2300 may be all it will ever give you?

Try it with the can disconnected and see if you get 1bar earlier? if not, then that is all that this turbo will ever give you, reaction wise!

Maybe a larger compressor wheel is needed?

Lara.