300Tdi VGT project

[DaveSIIA]

Thanks Dave - I went through the calibration process you suggest with the original spring:

Yes, of course. That was a week ago (& quite a few posts back) so I have probably slept since then. SWMBO assures me that I frequently forget things if I've had a sleep

Keep up the good work. It's getting interesting.

[rusty_wingnut]

i still think a twin stage spring is the answer, the early stage closes up to produce boost at low revs, the latter then tails off the boost at higher revs. I don't believe a linear spring can achieve this.

[Turbocharger]

Thanks chaps. Here's the results of some reflection after I've lengthened the adjuster on the linkage (one LH, one RH thread, just visible below the jubilee clip above) so the vanes are about midway closed when the diaphragm's in the zero position, and I'm back to a 'laggy' VGT - to be honest, it's about as good as a wastegated turbo. When I shorten the linkage to set the vanes to 'closed' (small turbo) at the zero position, it doesn't build pressure or become fully open at a low enough pressure, and the turbine starts to surge.

I need to see what's happening by pressurising the system with the airline again to validate the spring extension figures I calculated, and then I can work out how to control it. I'm learning how the system responds, but I still don't know how I want it to respond - to move through its full travel as soon as it reaches 1bar, to lift off the stop and move progressively through the travel as soon as pressure builds, or somewhere in between. I've a nasty suspicion the answer is different depending on massflow (revs) which justifies the computer control on the original vehicle - a 2003 Mitsubishi L200, since someone asked.

[white90]

John

when the waste gate pipe split on mine it was producing 2bar happily

went well till I realised and fixed the pipe

[Lara]

much like Lara's dimensioned and toleranced CAD drawing above

If anyone is interested I can sell you a copy of "Autocad15/3 Paint" for ages 5 to 8, for a very reasonable price

Lara

[Turbocharger]

lol @ Lara

Today I broke it. I said before that the turbo sound is a 'dirty' sound, there's extra frequencies that mean it's not a pure sound and this has been troubling me for some time. Well, as I joined the motorway the boost suddenly shot off the scale and I backed off the throttle, hit the hazard lights and onto the hard shoulder. One of the nuts in the diaphragm linkage had gone AWOL, so the turbo stayed 'closed' as the boost rose - this makes sense. I rooted around, found a spare nut and reassembled it (in suit and shirt). The windy motorway hard shoulder, fingertip-contact only and red-hot turbo meant I didn't have the most thorough setup calibration, so I'm not really sure what point it's set too - I suspect I was a little conservative and I set the vanes further open at the diaphragm's zero setting than I previously had (ie bigger turbo setting).

Now it'll only make 0.5bar of boost at any throttle setting, which fits with the above, but the sound is much "purer", a pure whine/whistle/scream exactly as I've heard gas turbines sound in the past and how I expected this turbo to sound. The "dirty" rushing sound has gone. However, the EGT seems to be higher than I've seen previously (eg 400deg C when cruising at 40mph and rising to "above 500" when I try to hold 50mph) and it's certainly not as powerful at 0.5 bar boost. Maybe the EGT's the same and I'm just being paranoid but it'd make sense if the engine's running more back pressure in the manifold.

Anyway, I'm away on business for a couple of days so my dear girlfriend has to drive round it in the evenings for now; I'll have another play with it when I get chance.

[RedLineMike]

Hmm. During last night's forum outage I managed to post the above, but the picture of all my springs has gone AWOL and the topic's still on the second page. Hopefully this post will bump it back to the front.

The omitted spring pic:

As it says above, anyone know how much boost a 300Tdi will take before exploding?

my 200 runs quite happy at 1.5 bar

previous engine ran at 26psi but the turbo was pretty much flat out

[Turbocharger]

Had another brief play today - I adjusted the vanes to fully shut - nice 'smooth' noise still, 0.4 bar max at any point, big EGTs under any throttle still. I think I need to take it all apart again Maybe I'll have a go at changing the spring back to the one that Lara sent me. If I can't get it back to at least OEM power levels then the original turbo will have to go back on at the end of the month.

[MECCANO]

Doh!!!

Do you know what the wheel sizes are on the turbo?

[Turbocharger]

One of the happy benefits of being away from home all week is that I get plenty of time to think about the symptoms I've seen and consider possible diagnoses. Today's was pretty simple - no boost and high EGTs so I lifted the bonnet and found this tucked at the bottom of the intercooler.

Easy fix but back to square one now with calibrating the diaphragm settings after I ran round frantically trying to get boost back.

I've no idea what size the turbo is - it's off a 53-plate Mitsubishi L200.

[DaveSIIA]

Glad to see that pipe disconnect caused the loss of boost, and not a failed turbo. It would have been a pity to have brought this interesting thread to a premature end.

BTW: I had a similar problem with flexible hoses coming off pipes when I turbo'd my 2.25D IIA. My regular commute at the time (every week for 8 months in a IIA) was S Manchester, Macc, Leek, Ashbourne, Derby, M1 J26>J5, Stanmore so the system had plenty of time to get to working temperature. Hoses would regularly separate somewhere near the M1/M6 junction. Doubled up jubilee clips weren't much better. The final solution was to run a bead of weld round the pipe so the jubilee clip met a stop as the hose tried to slide off. It was a bit easier than trying to flare the end of a 2" thin walled tube.

Good luck with getting it back up to speed!

[muddy]

FFS john stop using thos ****e lr hose clips!

Get down the scrappy and all the german cars have decent quality 7mmdrive hose clips that you can really get nipped up

Would be a really shame to see this project shelved, do you get TC lock up when you select a gear (say 2nd) just wondered if that would allow you to measure things better?

Will.

[TheRecklessEngineer]

I've got a ford hose clip on one of my coolant pipes - I put that down as the root of all the problems.

[slug_burner]

turbocharger,

I think your answer is in the data you have on post 110 quoted here.

When you wound the thing up you pushed the response of the turbo down into lower rpm for higher boost, which is what you are after. So you appear to be heading in the right direction by winding in the adjuster. However if you keep pushing that way you will eventually choke the engine's exhaust by restricting the outlet too much. You would achieve the same by just blocking off the exhaust.

So the answer is in reducing the orifices that the gas flows through to increase gas speed. You can only do this until you start to get non laminar flow then the turbulence will stuff things up (Corky Bell's book on turbo tuning talks about air speeds of 0.4 mach resulting in turbulent flow when sizing intercooler hoses I maybe on the wrong tram here but it is a thought) . That maybe what is happening when you get to the flattening out of your curves at approx 2500 RPM. I don't believe you are going to get the turbine to stall or surge, I think you're over analysing that portion. The flow problems are likely to be occurring at the control surface before you get to the turbine.

You will need to open up the vanes as rpm rises to stop chocking the engine by building up excessive back pressure.

You need to experiment by changing one variable at a time not multiple variables.

Couple of things for you to consider.

If the engine this turbo came off makes approx the horsepower your after then it is sized correctly and there is no need to change turbines or compressors. (Unless it is part of a twin turbo set up)

Leave the control mechanism out of early experiments unless you think that the system will damage itself without the control mech present.

By removing the control mech you should be able to establish a map of boost vs rpm for various (fixed) settings of the lever that control the vanes.

The the lever that controls the vanes has some stops that you can screw in (not likely to be here if engine turbo came from is similar size to yours)

The diaphragm capsule made by "Lara" needs more space for the original spring (or the spring needs to be cut down), one of your curves shows this. The slope of the boost vs displacement of the diaphragm is the spring rate equivalent (all the people wanting you to consider the diaphragm area are just looking for you to apply a constant so you end up with a calibrated graph in a unit of force instead of air pressure). Someone already suggested that you needed some extensions for your diaphragm capsule I believe. If your spring rate is constant (assumption to start process) then judging from your air pressure vs displacemnent curve you need a 23.5mm extension on the diaphragm housing/capsule to allow the spring to be streched a bit more and apply zero displacement at zero boost/air pressure. If you examine your ideal response (I think it was orange line) the slope is similar to the one you already have in your original spring you just need to change the c, in the y=mx +c equation also results the changing the preload. If the spring rate is variable then it is likely to be different, but you should be able to iterate to a solution. (cut spring down or increase space for spring to iterate)

As the rpm raise you will need to open the vanes to stop the engine choking (the big turbo, little turbo analogy might be breaking down). With a constant geometry turbo the max gas flow before building up too much back pressure is determined by the turbine size, in your variable geometry turbo the max flow and back pressure is determined by the control vanes.

Levers can be used to change the displacement range. Select a pivot point and the length either side of the pivot of your lever will allow you to get a ratio of displacement.

The flattening out of the curve around where you have circled is an area that you want to avoid once you get your control system working. I suspect that this might be when you are choking off the engine maybe? or when gas speed is getting so fast that laminar flow no longer holds more likely therefore need to open vanes further.

Don't forget that the fuel pump on the 300 tdi also has some control system working (you already ran into this when you did not connect the boost compensator)

This will work, tidy up your set up and build things as you intend them to be when you finish ( that bit of all thread looks carp and the offset will be stuffing up your measurements providing you with artifacts of your set up)

Don't be so impatient to get out on the road and experiment until you know what is going on in theory at least.

For all that are applying control theory, don't forget that the engine is part of the system (choking exhaust stops engine from producing more bhp/rpm). That is why it would be beneficial to produce the map of vane position and rpm vs boost without the negative feedback system in place i.e., open loop response. Once the engine and turbo is characterised then you can overlay the feedback system to get the response you want. The dynamic (step) response can be sorted once you have control of the system.

If pneumatic control alone will not do it, then consider pulse width modulating a constant pressure or vacuum source. Have a look on the megasquirt web pages and you will find boots monitors that produce linear voltage response to pressure. A PWM circuit should not be that difficult. But if your going to go that far then get some more sophisticated control going by using something like the megasquirt controller or other controller circuit that you can program.

Mate you worked for cosworth, you should be eating this stuff up. I hope cosworth don't have experimental set ups with welded all thread.

Good luck, a good balance of theory of experimentation should get you a long way faster than experiment alone.

Well, I've done some more playing. This graph shows revs against boost:

The red line is how it was, my baseline setting.

After talking to my dear old Dad I wound the fuel up half a turn on the base fuel screw (didn't touch the injector pump diaphragm) - this gave the green line.

Then I wound the vanes closed by ~1.5mm, still with the fuel turned up. This gave the response shown as the blue line... but the area shown in the circle was definately showing surge, a fluttering sound as if it was making boost for half the time at about 2Hz, sounding a bit like a misfire might. Sorry Julian Any idea what the boost response from your Td5 might look like on these axes?

The blue line hits 1 bar at 1800rpm which is about what I got from the old turbo - progress, but not the quantum leap I was hoping for. It's worth mentioning, running the engine on the blue curve at high revs means it gets to sixty in about 13 secs, which is frenetic and exciting but it's surging hard and it's absolutely not what I was aiming for with this conversion. It does suggest what could be achieved by winding the boost and fuel up on a standard engine though, EGT allowing...

I'm pretty sure I need a more proportionate response from the diaphragm to move the vanes, and I should have a handful of springs turning up in time for the weekend, if Mr MOT doesn't condemn me to a weekend putting new brakes on the MG.

[SiWhite]

x2 on the threaded rod. There'll be so much stiction with that setup that the nozzle actuator won't operate smoothly.

Sort it out or I'll tell Jen your boss (we live in the same village!)

[slug_burner]

turbocharger,

I think your answer is in the data you have on post 110 quoted here.

When you wound the thing up you pushed the response of the turbo down into lower rpm for higher boost, which is what you are after. So you appear to be heading in the right direction by winding in the adjuster. However if you keep pushing that way you will eventually choke the engine's exhaust by restricting the outlet too much. You would achieve the same by just blocking off the exhaust.

So the answer is in reducing the orifices that the gas flows through to increase gas speed. You can only do this until you start to get non laminar flow then the turbulence will stuff things up (Corky Bell's book on turbo tuning talks about air speeds of 0.4 mach resulting in turbulent flow actually it was increase in drag when sizing intercooler hoses I maybe on the wrong tram here but it is a thought) . That maybe what is happening when you get to the flattening out of your curves at approx 2500 RPM. I don't believe you are going to get the turbine to stall or surge, I think you're over analysing that portion. The flow problems are likely to be occurring at the control surface before you get to the turbine.

You will need to open up the vanes as rpm rises to stop chocking the engine by building up excessive back pressure.

You need to experiment by changing one variable at a time not multiple variables.

Couple of things for you to consider.

If the engine this turbo came off makes approx the horsepower your after then it is sized correctly and there is no need to change turbines or compressors. (Unless it is part of a twin turbo set up)

Leave the control mechanism out of early experiments unless you think that the system will damage itself without the control mech present.

By removing the control mech you should be able to establish a map of boost vs rpm for various (fixed) settings of the lever that control the vanes.

The the lever that controls the vanes has some stops that you can screw in (not likely to be here if engine turbo came from is similar size to yours)

The diaphragm capsule made by "Lara" needs more space for the original spring (or the spring needs to be cut down), one of your curves shows this. The slope of the boost vs displacement of the diaphragm is the spring rate equivalent (all the people wanting you to consider the diaphragm area are just looking for you to apply a constant so you end up with a calibrated graph in a unit of force instead of air pressure). Someone already suggested that you needed some extensions for your diaphragm capsule I believe. If your spring rate is constant (assumption to start process) then judging from your air pressure vs displacemnent curve you need a 23.5mm 23 mm extension on the diaphragm housing/capsule to allow the spring to be streched a bit more and apply zero displacement at zero boost/air pressure. If you examine your ideal response (I think it was orange line it was the violet line) the slope is similar to the one you already have in your original spring you just need to change the c, in the y=mx +c equation also results the changing the preload. If the spring rate is variable then it is likely to be different, but you should be able to iterate to a solution. (cut spring down or increase space for spring to iterate)

As the rpm raise you will need to open the vanes to stop the engine choking (the big turbo, little turbo analogy might be breaking down). With a constant geometry turbo the max gas flow before building up too much back pressure is determined by the turbine size, in your variable geometry turbo the max flow and back pressure is determined by the control vanes.

Levers can be used to change the displacement range. Select a pivot point and the length either side of the pivot of your lever will allow you to get a ratio of displacement.

The flattening out of the curve around where you have circled is an area that you want to avoid once you get your control system working. I suspect that this might be when you are choking off the engine maybe? or when gas speed is getting so fast that laminar flow no longer holds more likely therefore need to open vanes further.

Don't forget that the fuel pump on the 300 tdi also has some control system working (you already ran into this when you did not connect the boost compensator)

This will work, tidy up your set up and build things as you intend them to be when you finish ( that bit of all thread looks carp and the offset will be stuffing up your measurements providing you with artifacts of your set up)

Don't be so impatient to get out on the road and experiment until you know what is going on in theory at least.

For all that are applying control theory, don't forget that the engine is part of the system (choking exhaust stops engine from producing more bhp/rpm). That is why it would be beneficial to produce the map of vane position and rpm vs boost without the negative feedback system in place i.e., open loop response. Once the engine and turbo is characterised then you can overlay the feedback system to get the response you want. The dynamic (step) response can be sorted once you have control of the system.

If pneumatic control alone will not do it, then consider pulse width modulating a constant pressure or vacuum source. Have a look on the megasquirt web pages and you will find boots monitors that produce linear voltage response to pressure. A PWM circuit should not be that difficult. But if your going to go that far then get some more sophisticated control going by using something like the megasquirt controller or other controller circuit that you can program.

Mate you worked for cosworth, you should be eating this stuff up. I hope cosworth don't have experimental set ups with welded all thread.

Good luck, a good balance of theory of experimentation should get you a long way faster than experiment alone.

[Turbocharger]

Right, I've no idea where the last four months went. I've had coolant "issues" with the engine for a while, and after running 1.5bar boost and chunky lumps of surge for a while I was leaving clouds of white smoke as I pulled away from junctions - around the same size and shape as the LR actually, it was quite impressive.

I replaced the head and chronicled it here. I've run it in and I'm happy with it, so I've got to try to break it again.

I've been running it just about static ie not using the vanes. It's pretty laggy that way, but it's safe. Today I've been playing with it again, tidying up some of the "temporary" measures I put in quickly at the beginning of the build. First off, it appeared the mounting for the diaphragm had been vibrating a little:

The actual rubber diaphragm looked a little worse for wear too, and it was leaking air:

I deluged the diaphragm in sealant and helped the integrity with an offcut from a Sainsbury's carrier bag:

I took on the wise words above and modified the adjuster linkage too. I'm confident that it was working (see static testing) but I had some time on my hands while the sealant was curing. I even broke out the hole cutters to add some lightness. Quite proud of this one.

Using the workshop air line I was able to pressurise the system and see what extension the diaphragm is giving at different boost levels.

The carrier bag cured the leaks entirely up to 2 bar, and then there was a loud 'pop' and a jumbo air leak - you can see the hole in the picture above, in fact. I pulled out all the sealant, replaced the plastic bag with the strongest flexible membrane I could find (the peel-off backing from a roll of lead flashing tape, no less) and it's running well. I've had to back it off from 'really responsive' because the diaphragm can't move until 1 bar but it's still pretty good, 1 bar of boost from ~1500rpm and up to 1.7 bar at 2000rpm. Goes well

Conclusion: I need different, lighter springs. Whatever I fit, I don't get any movement until >1 bar. Despite that, it's running well, performing better than a wastegated turbo and my understanding of turbos, back pressures, EGTs, revs, transient load, surge etc has all improved massively over the last twelve months so the project has certainly not been a waste of time.

[MECCANO]

Excellent, sounds like its gradually coming together!

[Lara]

Membrane is standard TD5 if you need a new one.

Lara

[Turbocharger]

I suspect I do - I'll throw a part number request on the other forum for Ralph (and probably wince at the price of dealer-only parts )

More driving this weekend experimenting with different positions ( ) for the linkage, I conclude that I need to get most of the movement on the lever between 0 and 0.75ish bar to give good response - it's quite acceptable the rest of the time with a quite conservative setting.

[Turbocharger]

This thread's approaching twelve months old, and I believe I now have the type of response I was looking for. I had an epiphany from a conversation with Rusty Wingnut. I've been playing with different springs to change the response of the turbo vanes. The differing springs have been changing the gradient and y-intercept of these lines:

This would bring the turbine into surge at low revs, because the vanes remain in the 'zero' position until the unit is boosting at around 1 bar - that limits me to set the zero position to that which will allow surge-free running at ~1800 rpm, which means the response is very similar to a wastegated turbo. So, I reasoned, I needed a response which would give some movement at lower pressures, maybe from zero. I'd need a very weak spring. I specced up what I wanted, and rang up a bespoke company in Bristol to get a spring made specially. £54. Since I wasn't certain that my numbers and reasoning were right, I bodged something up, using a tension spring outside the pressure capsule (this was the epiphany...)

This gave some very promising results, with a much more instantaneous response although the spring wasn't stiff enough to quickly return the vanes against the friction in the system. It gave me enough confidence to look at the other springs I had in my assorted bag from the £10 minimum order from said spring company. If the spring is 30mm long, it will sit in the capsule with no preload and give a response nearer to zero boost.There was one spring with a rate of 1.77N/mm and a free length of 100mm. I cut this down to 30mm, which means the rate is now nearer 6N/mm - not as stiff as some of the other springs I'd specced, but the bodge-spring effort from earlier

suggested that might not be a problem. Here's the chopped spring:

I tidied the ends and built it up again. Before I ran it out of the garage again, I measured the extension:

Just what I was looking for - time for a roadtest.

Result - :D Happy John :D There's a real 'shove' as it lifts off tickover, and it comes up to ~0.8 bar and holds steady, as it's working in a feedback loop. 0-60 is around twenty seconds with this boost level, which is really surprising as it 'feels' much quicker. I daresay this will come back as I work on it a little more to get the steady-state boost back to factory levels (or slightly greater), but the torque at lower revs is very pleasing, and it's exactly the effect I intended when I built it, rather than out-and-out power.

In a show of abuse of my vehicle, I revved the engine with my left-foot on the brake, to see what boost it'll make at low revs. For the viewing public's pleasure,

. Sorry about the unsteady camera work, the car kept moving despite discs all round...

[rusty_wingnut]

excellent news matey! The amount of rev's it takes for boost to begin is impressive, much quicker response than that of the TGV turbo.

Keep going, looks like you're getting there!

[will_warne]

Result!!!! At coming on for 12 months that's a time scale I'd be proud of

[Turbocharger]

Did a few more miles today, very pleased with most of it. The torque off the line is excellent, then it settles down to around 1.25bar. Transient response is good too, it overshoots briefly and then comes smartly back to the required level - the car picks up instantly and catapults strongly out of bends, roundabouts etc. There is a hole in the torque curve just where the torque converter locks which is irritating but there's plenty of toys left to play with. The only down side is that it appears to be quite "thirsty" when driven enthusiastically...

[rusty_wingnut]

these VGT's give you great overtaking ability too!