Project Potential... variable vane turbo on 1.6l

[veeman]

Is there anything specific about the form factor of the "can" other than the fact that it bolts to the turbo assembly?  

I was thinking that maybe you could substitute some other type of vacuum chamber / actuator in its place.  What about using some sort of external wastegate or BOV body and customizing it to do what you want?    

I have several Audi 5kt wastegate bodies at home and I know they have a removable cap with a replaceable diaphragm.  You can switch springs no problem and it all still seals up.  Thinking again, that might not give you the linear "pull" you need to make the mechanism actuate.

Just an idea.

You know, that can looks like those actuators on the vac controlled vents/flaps in the VW's.  Maybe you could practice opening / resealing with those?  I've got a couple you could use.

[Mark(The Miser)UK]

Interesting...
So how far up the power curve can you go with an N/A  before meltdown; if there is no oil cooling to pistons, and no cooling through the oilcooler; and a weaker crank? What about the lack of boost from the aneroid setup?
When I disconnected the aneroid in search of the ultimate economy on my 'Q' ; the response was akin to a slug  :lol: which was OK when passing a snail but not when trying to get past a smoke belching truck :?

(That reminds me my 3rd and final TD pump had the same '34' on the cone)

[MrDave]

I'm a little late to the party, but I'll share a few things regarding passive control of the VNT.

Alot of folks with high power TDIs will run a boost valve on their setup.
It's basically a spring loaded pop valve, and is used to control boost spikes.

It's connected between the intake manifold and the vacuum line controlling the VNT actuator.   Under normal conditions, the valve is closed.  If boost gets too high, it will bleed boost into the VNT line, and close the vanes, reducing the turbine speed, and reducing the boost.

One could, technically, run the VNT on full vacuum, then use a boost valve to limit the top end.  The problem with that is the vanes would end up fluttering under full power, which wouldn't be good for the VNT.


Another passive boost control is 'the boost elbow'.
Connect the VNT actuator to full vacuum, with the boost elbow teed into the line.  The other end goes to the intake manifold.  The design of the elbow is such that you have a small set screw which controls the amount of boost which is bled into the VNT control line.   At lower RPM, with little boost, the vanes are wide open, then as pressure builds, the vanes slowly back off as more pressure is bled into the VNT conrol line.

On an electronic TDI, (IE: ECU controlled), you can bypass the N75 valve and run the turbo under mechanical control without setting off DTCs.

The boost elbow ends up giving the diesel engine a dyno plot that looks very much like a gas engine.

TDIclub hosted dyno:  (scroll down the page for the dyno)
http://forums.tdiclub.com/showflat.php?Cat=0&Number=802711&page=&fpart=2&vc=1


I say all this, but I have no idea what would work for a TD.


-Dave

[fspGTD]

MrDave - thanks for posting your thoughts... I am still mulling and processing the information that you have contributed.  (Edit: see below post for my reactions to these ideas.)

Mark the Miser - Re: thermal control for an engine lacking piston-oil squirters... I guess you missed the part of the plan which is to not set aggressive fueling levels.  In any event, yesterday I ordered a VDO pyrometer to make sure the thermal limits are safe.  I am planning on installing it pre-turbine into the exhaust manifold / turbine housing and will consider a red flag being raised if it approaches 1300 deg F (which is 200 deg. F cooler than the factory 1.6lTD.)  I also never said I wasn't planning on running an oil cooler, or a TD fuel pump.  (I was only in question about which type of oil cooler to run.  I'm leaning towards the factory version at the moment by the way, mostly due to the fact that I already have it, and that it might be hard and expensive to track down a thermostatically control sandwich plate which I'd want if going external oil cooler route.)  Also, re: your contention that the crankshaft differs between an '84 1.6lNA vs TD, show me some evidence that supports this notion, because according to ETKA as discussed on this board already, they have the same part number.  I am planning on installing the TD harmonic balancer pulley.

I also wanted to update you all on where I am with the VNT can...

I first took some more measurements to determine the viability of my idea that wouldn't require modification of the can.  I found a vacuum gauge in the shop that I could use so I can measure actual vacuum values, rather than estimate them from diaphgram area and forces.

The vacuum pump (this is an '84 Rabbit, with the compact vane-type vacuum pump not the diaphgram style by the way) pulls a very healthy -29 in Hg (that's -14.2psi.)  Although that is for zero air flow, and I found that the vacuum strength gets weaker as flow increases.

The actual, as-installed amount of vacuum necessary to get the VNT mechanism to begin closing is -7.5 in Hg (-3.7 psi).  -17.5 in Hg (-8.6 psi) is required to reach full closed.  Then due to friction in the mechanism, the valued required to get the vanes to open are different: -16 in Hg (-7.9 psi) required for it to begin opening, and -2.0 in Hg (-1.0 psi) required to reach full open.

Putting all these numbers into my spreadsheet I built to predicts the vane movement for various boost levels and R2/R1 relationships, I found I'd need an R2/(R2+R1) to be 45% or less to make the vanes remain fully closed after the engine is started up when there is zero boost, and 40% or less to make the vanes return to fully closed after the engine has gone through a boost cycle.

At 45%, the vanes would reach full open at 15 psi, and wouldn't start to begin closing until boost dropped below 10psi.

At 40%, the vanes would reach full open at 19 psi, and wouldn't start to begin closing until boost dropped below 12psi.

I was hoping that there may be a relationship that would allow the vanes to be completely closed, and then open the vanes over a tighter boost range.

What does this mean?  I decided I am not confident this scheme would reliably control the vanes in a narrow boost range.  There is risk that this method may give boost creep, due to the wide range of boost pressure change necessary to get the vanes to move from full close to full open, and vice versa.  And I'd rather not install the turbo with everything hooked up to find that it has to come off because I need to open up the can.

Result of this... gave me another reason to make the can open/closeable, and assess the viability of QuickTD's idea to convert the can from vacuum to pressure.  Using a grinder, I thinned the metal around the outside rim or the crimp, and then I finished off removing the crimped on ring by carefully chiseling it off.  After filing off any remnants of the crimp ring missed by the chisel, so there was nothing left to hold the can closed, I gently pryed the can open.  For modification of the diaphgram to take boost rather than vacuum, I think the diaphgram will need to be flipped over along with it's rigid metal support behind it, to keep the metal support on the low pressure side and prevent the diaphgram from "balooning".  To flip the diaphgram required that the linkage rod be removed (which was permanently riveted on.)  The flipped over can with spring on the other side will have less room in the can to move around, due to the can and the diaphgram support being different on one side versus the other.  However, I think there would end up being enough travel to completely move the vane mechanism, if the linkage were tuned correctly.  I bought some various 6mm studs, threaded on both ends with a smooth center area, to experiment with replacing the riveted on linkage rod.  I was able to install them to the diaphgram using nuts on either side of the diaphgram.  I first tried a stud about 3" long but found it was too long, so went one step shorter.  If I need to change anything, IE the rod length, change the diaphgram or can somehow to change it's travel of the spring pre-load, or change the spring, I wanted the can to be serviceable by just undoing some bolts, but I wanted it to also seal reliably.

I set about solving the problem of finding a clamping ring to seal the can back up.  I bought a 4" wide plate of thick aluminum.  And using my new dremel tool with router attachment that can do circles, I already precisely dremeled through a hole through the thick aluminum that has the needed 2 5/16"ID.  My next step may be tricky, but I want to try and machining a relief in the edge of the aluminum, so that a shoulder sticks out past the outside of the can, so that the bolts I'm going to install holding the can together have a shoulder on the outside so they are evenly tensioned and not torqued.

I'll post pics of it later if I am successful, with me luck!

(If not, DVST8R, maybe I'll shoot you the details specs of the piece I am trying to fabricate and take advantage of your offer.)  I am planning to clamp the can together with either 6 or 8 little bolts (6mm with wide bolt head).

PS - although I haven't taken any measurements of the pressure required to get the modified can to move the VNT mechanism, I did (with the can clamped together with a couple c-clamps) apply some pressure to the uninstalled can and determine the pressure required to move it's linkage.  The results so far are encouraging.  The resulting range to make the linkage start moving and reach full travel are: 6-7 psi to start moving out, and 10-12 psi to reach full extension.

I hope the range to actually pull the VNT mechanism open and closed is somewhere near this tight!    I can always plumb in a bleeder valve if I want more boost pressure, so I was glad to see some pressure numbers that were in a reasonable or maybe slightly low range (I'd like to eventually target running around 10-14 psi boost level on this thing.)

[fspGTD]

Another passive boost control is 'the boost elbow'.
Connect the VNT actuator to full vacuum, with the boost elbow teed into the line.  The other end goes to the intake manifold.  The design of the elbow is such that you have a small set screw which controls the amount of boost which is bled into the VNT control line.   At lower RPM, with little boost, the vanes are wide open, then as pressure builds, the vanes slowly back off as more pressure is bled into the VNT conrol line.

OK, I now understand what you're saying about the "boost elbow".  This is essentially the same idea I had, although omitting R1.  This might have some adverse effect on vacuum pump life and there may also be a loss of vacuum if the vacuum pump flow required to set desired boost level is too high (In effect, it is equivalent to my scheme, except R1 would be hard-set to a very low resistance and not be adjustable.)

The effect you describe, "similar to a gas engine", is what I was worried about.  And there basically just being too wide of a manifold pressure range required to move the vanes from full closed to full open.  Essentially this would manifest itself as boost creep.  In some settings, the vanes may never even get completely closed at zero boost.

The other idea you mentioned in your post, which is basically adapting an on/off flow valve into the system, would indeed solve the problem about there being too wide a manifold pressure range to get the VNT to move between full open and full closed, but indeed is also problematic because it makes the controlled pressure range too narrow.  I would expect to see the opposite extreme of boost creep which is also undesirable... boost spike.

[DVST8R]

Sounds like some great progress. If you could snap a few pics to aid us slow people challanged by visulizing just words that would be great too.

As for the pieces just let me know. If your good with auto cad you can just send me a 2 dimentional .dxf file with what you need and I can import that directly. If not just a dimentioned drawing and I will autocad it.

[fspGTD]

DVST8R - thanks.  I have already snapped a few pics of the disassembled VNT can... I have just left the digi cam out at the shop right now and have not downloaded the pics yet, and it is not currently accessible from home.  I will make a note to grab the camera next time I get to the shop so I can show you all the disassembled VNT can (and maybe snap some shots of my aluminum plate washer fabrication in progress )

[MrDave]

The effect you describe, "similar to a gas engine", is what I was worried about.  And there basically just being too wide of a manifold pressure range required to move the vanes from full closed to full open.  And in some settings, the vanes never get completely closed at zero boost, which may adversely effect low-RPM torque and turbo lag.

The other idea, which is basically adapting an on/off flow valve into the system, indeed raises concerns in my view about flutter due to the system controlling VNT movement over too narrow of a boost range.

Actually, at low rpm or zero boost, the can would see full vacuum, so there is minimal turbo lag once you get on the go pedal.   As vs. the ECU control where the can would be under atmospheric pressure, till you tell the ECU you want to go, THEN it puts it under vacuum.


One idea I had was to connect multiple boost elbows in parallel as a passive system, and a boost valve to limit max boost.    Multiple elbows should essentially allow you to tune your boost curve.  

Another idea would be to have a boost valve bleed into a boost elbow.
ie:  at 5lbs start bleeding in.  Under 5lbs, no bleed at all.

Other ideas would be for an electric vacuum source.  Think electric mitivac...

[fspGTD]

Actually, at low rpm or zero boost, the can would see full vacuum, so there is minimal turbo lag once you get on the go pedal.   As vs. the ECU control where the can would be under atmospheric pressure, till you tell the ECU you want to go, THEN it puts it under vacuum.

Gotcha - the TDI computer keeps vanes open at low throttle settings, probably to increase fuel economy.  Although if the restrictor between manifold and VNT can is open enough, vacuum inside the VNT can may not be high enough to get the mechanism completely closed.  For example, if you had equal restrictions between vnt can and vacuum pump as there is between vnt can and manifold, according to my calculation there would not be enough vacuum in the VNT can to fully close the vanes.

One idea I had was to connect multiple boost elbows in parallel as a passive system, and a boost valve to limit max boost.    Multiple elbows should essentially allow you to tune your boost curve.  

Another idea would be to have a boost valve bleed into a boost elbow.
ie:  at 5lbs start bleeding in.  Under 5lbs, no bleed at all.

Other ideas would be for an electric vacuum source.  Think electric mitivac...

Re: Multiple "boost elbows" in parallel... I might be picturing something different than you're describing, but connecting two lines in parallel I don't see offering any more flexibility than a single line with a variable restriction, where if you want more flow you just set the restrictor to be more "open" (less restrictive.)

By "boost valve" are you referring to a boost pressure controlled airflow valve that is either on (low restriction) or off (high restriction)?

Is a spring loaded pop valve a restriction that is closed until pressure exceeds a certain differential, and then it varies flow to try and maintain that amount of differential?  Or do I have these two mixed up?

In any event, I don't see the laws of mechanical VNT boost control being much at all different 1.9l TDI vs 1.6l IDI...  So if you guys have already tried stuff that didn't work, I'll be sure not to duplicate those experiments!  

Also, do you if anyone has yet tried QuickTD's idea of adapting a VNT can to be controlled by boost pressure?  Thanks for your help!

[lord_verminaard]

Didnt some versions of the Dodge Daytona Shelby (1990-ish) cars use a VNT turbo?  VNT 20 or VNT 25 sound like they ring a bell.  I highly doubt that during the few years they were made, they had any sort of computer controlled boost device.  I dont have a clue how they worked, but maybe you could look up how they worked and copy the idea.  

Just a thought, as I saw one on the way home from work today.

Brendan
84 Scirocco 8v
00 Camaro L36 M49

[fspGTD]

I've made some GREAT progress today!  No time for a very detailed update right now, but here's a pic for you guys...

(more pics plus testing results will be coming later...)

[fspGTD]

OK, on to some more pics!

Disassembling the VNT can:

Ground / filed off the crimped-on lip off the VNT can:


Here is what was inside...



Modified the diaphgram replacing riveted-on stud with a different one that comes out the opposite side (so diaphgragm can be flipped over...)

[fspGTD]

Putting the VNT can back together

1/4" 6061T6 aluminum plate.  Dremel-routed out a hole and then a step around the inside of the hole:


result:


test fitting step for proper size:


drilled 8 holes where I wanted them:


The half of the can with the narrower lip fits perfectly; did not need to be notched:


notched the half of the can with the wider lip to align with the holes and allow bolt heads to have more clamping surface:


Almost done cutting out the fabricated ring piece from the plate:


Just the fabricated retaining VNT can retaining ring, cut out and cleaned up:


Reassembled VNT can modified to actuate with boost pressure:


Widened out the heat shield:


... and all assembled!

[fspGTD]

Testing...

For testing I used a brake power bleeder to build up and measure simulated boost pressure applied to the VNT controller can.  Pic of the test rig:



Results:
First thing I did was pump 'er up to 25psi and see if she holds.  She did!  No leakage at all.  The VNT can retaining ring fabrication project was a success!

Next, I went on to make several simulated boost cycles while observing changes to VNT geometry with respect to changes in the simulated boost pressure.

I found that the VNT mechanism is sticky when it initially moves toward open from fully closed position.  Tapping on the turbo might be a good simulation of engine vibrations, and helps the mechanism open earlier.  Without tapping, it took in one case up to 14 psi before the vanes started opening (in which case they snapped open almost all the way.)  In two other trial boost build-up cycles without tapping, it took 10 and 11 psi before I observed the vanes started to open.

The results that I feel are most accurate however were from my last trial which I made while tapping the turbo (sort of jiggling the turbo on the ground).  While tapping, the vanes started to open at 9 psi.  After that point they opened smoothly thereafter (with no further tapping/jiggling necessary) until about 12 psi, when the rate of opening seemed to markedly slow (was almost completely open by that point though.)  By 15 psi the vanes were really all the way open.

When bleeding the pressure down, the vanes started reliably closing when pressure dropped to 12.5 psi.  No "stickiness" was observed on any of the pressure bleed-down cycles, and so tapping was not attempted during the bleed-down cycles.  The bleed down cycles moved smoothly and evenly until they were completely closed at 6 psi.

I am pretty encouraged with these results!    (Besides the stickiness part... because if it's too sticky, it could cause the boost pressure to spike.  But if it's not too sticky, I suppose it might actually work out favorably, helping build boost pressure to full more quickly. :twisted: )

I am excited about trying this contraption out!

[fspGTD]

By the way, DVST8R, fabricating that ring ended up taking a lot of work with my crude shop tools.  I used a dremel to cut holes and machine steps for the first time with that fabrication project!  I am just happy it turned out as good as it did.  But next time though I think I'll send you a drawing and have your CNC mill go at it!  :wink: