219 replies to this topic

[acres]

After a bit of stuffing around I think I may have worked out the flaw in the system that is causing the major lack of boost. The 2 valves I have utilised as the ECV and ICV do not have a 100% seal when closed, I assumed this would be no issue for the ECV as the flow loss is minimal and may only be a minor issue for the ICV. Since running some tests I now believe this lack of seal is a bigger issue then I originally anticipated.

After closing up the cracks in the headers I decided to start taking measurements of the compressor outlet pressures of both turbo's under various conditions. I started by attaching a map sensor to the primaries compressor outlet to track it against MAP.

The measurements in these graphs are in KPA and are absolute, so 100kpa is atmo, 200kpa is 14.7psi boost, etc. The red line up top is MAP, the green line up top marked sensor01 is the compressor outlet. As expected both lines are basically tracking each other through the rev range and have near equal readings.

 

 

 

A reading was then taken from the secondaries compressor outlet. At this stage the ecv has been set to remain full closed until 4250rpm where it will switch to full open. There should be no pre-spool at all however in the graph you can see that secondary pressure is building with rpm, although at a reduced rate compared to MAP. The vertical blue line in the center of the image has been placed at 4250rpm, you can see to the right of this line that the pressure from the secondary has begun to ramp up, indicating the ecv has opened fully as it was supposed to. To the left of the line it should be closer to 100kpa as the ECV was closed so the higher pressure reading indicates that either the ICV is leaking too much boost down into the secondaries charge pipe or the ECV is leaking too much exhaust and is robbing the primary of the energy it requires.

 

 

Next test was to apply vacuum to the normally positive port of the ECV actuator in an attempt to 'lock' it shut. In the graph pressure from the secondary builds very similarly to the image in the previous graph. Considering the ECV was not meant to open in this test, it adds further weight that the valve's are leaking too much.

 

 

The ICV was then removed and replaced with some wooden plates to 100% block the flow of the secondaries intake. Again the secondary measured positive pressure and this time surging was recorded in the graph as the flow had nowhere to go but back through the turbo. Primary boost was still unacceptable at this point.

 

 

The same thing was then done with the ECV as well. It was removed and the exhaust was blocked with a plate. The primary now finally came to life, building boost very rapidly in comparison to previous results. Boost, and power, was picking up strongly from 2500rpm and was easily hitting max boost at 250kpa. The wastegate finally got to do something at this point too.

 

 

So it seems that these results are indicating that either the ECV was being pushed open prematurely by the exhaust pressure (unlikely being a butterfly valve) or that more likely the fact they don't seal 100% is stealing too much energy from the primary. Going with the latter I now need to either work out how to somehow modify the valves so they seal better, or source some better units that will seal better in the first place.

This is an image of one of the valves, you can see the small ring of light around the plate that shows how they don't seal correctly.

 

 

I also took an image of a run I did with the ecv set to remain open so the system was running as a parallel setup. The green line is also measuring the secondary to gauge it's response in comparison to the primary. It doesn't begin to ramp up until approximately 4700rpm.

 

 

I also pulled close to 10% out of the fuel map in most boost regions. This changed the AFR to read somewhere between 13-13.5 in most areas but the engine is definitely happier at this AFR. The plugs are now healthy looking too instead of being caked in soot.

 

I am all ears if anybody has any recommendations on where I can source some better valves or how I can modify these ones. 

[glenisanutter]

Fabricate a new butterfly to suit. You may been to machine the internal wall to improve tolerances.
Base the dimensions off a similar sized throttle body where tolerances are much closer

[glenisanutter]

Just had another idea... Shave the top of the existing butterfly until the bottom edge hits

[acres]

Fabricate a new butterfly to suit. You may been to machine the internal wall to improve tolerances.
Base the dimensions off a similar sized throttle body where tolerances are much closer

 

Thought about this but the shaft is pressed onto the butterfly and the shaft is held in by a bush that is pressed into the housing. I tried to dismantle the original, failed valve as a test but the bush was too seized into the housing. With some persistence  I think I could probably dismantle the new, replacement valve though.

 

I thought I'd try this first though:

 

 

Using some left over 63mm pipe I cut out 2 sections that could sit inside the housing and but up flush against the butterfly when it was closed - 1 piece on each side of the butterfly. Clamped and welded them in (carefully as the hosuing is cast iron) and bench tested. The valve leaked alot less but it was still leaking from the shaft as there is the same small gap between the shaft and the housing. The shaft reduces in diameter where it enters the housing.

I still put the ecv back into the exhaust though to try it out and it was definitely an improvement but still not good enough.

Here's an image from that run:

 

 

Just had another idea... Shave the top of the existing butterfly until the bottom edge hits

 

I'm not quite sure by what you mean by this. Do you mean shave the edge so that it reduces in diameter?

[allpaw4]

I think I'm understanding, shave the leading edge a smidge so the flap shuts to a ninety degree edge rather than the current one that will close up on the leading edge, effectively making it close on a knife edge, rather than a solid one if you shave it back on the respective angle.

[allpaw4]

The other way to do it would be weld a line around the close point of the housing, and then use something like a dremel to make a flat edge for the flap to sit against. Would require a lot of care to be taken if it's cast, but you might be able to do it.

[Reevesy]

What kind of gas temps are running through the valves? I maybe able to find something to suit at work but different valves have different temp ratings.

[acres]

I think I'm understanding, shave the leading edge a smidge so the flap shuts to a ninety degree edge rather than the current one that will close up on the leading edge, effectively making it close on a knife edge, rather than a solid one if you shave it back on the respective angle.

 

Ok I get it now. That wouldn't have worked though as the valve doesn't seal correctly at the shaft either so would still leak at that point.

 

The other way to do it would be weld a line around the close point of the housing, and then use something like a dremel to make a flat edge for the flap to sit against. Would require a lot of care to be taken if it's cast, but you might be able to do it.

 

This crossed my mind briefly but I soon realised that I don't have the talent with the welder to pull that one off. I guess though it would have offered very similar results to welding in the 2 bits of pipe to but up against the flap.

 

What kind of gas temps are running through the valves? I maybe able to find something to suit at work but different valves have different temp ratings.

 

That would be awesome if you could find something! I am not sure of the exact temps at the valve but I have the EGT probe placed in the runner of cylinder 1, about 25cm from the port and with current tune it will read peak temps between 660-730C under boost. So it would surely lose a couple of hundred degrees between that point and the ecv? Between the sensor and the ecv the exhaust has to travel the rest of the runner, the up pipe, the turbo and then the full drop of the dump so it's a fair distance. I could always place another bung in the dump and move the temperature probe to get accurate results but if I had to guess, maybe 450-500C max?

 

I pulled the valve back out last night and had another look at it to see if I couldn't reduce the gap between the shaft and housing. I figured if I could dismantle the valve maybe I could add some washers between the shaft and housing to close up the gap. I couldn't get the bush out though so couldn't remove the shaft however by using a bit of steel tube and a hammer I managed to smash the bush in a bit further so that it protruded into the housing slightly. This wedged the shaft in a bit tighter and closed the gap almost completely. Bench tested and was a further improvement but still had a leak, chucked it back on the car anyway and again was a good improvement as primary was able to hit peak boost but it's response is still to sluggish to be acceptable. This valve is just too much trouble I think so I got on the google last night and came up with this.

http://www.aliexpres...2288307447.html

 

Sent the seller a message to ask if the valve seals 100% and to find out it's temp rating. Price is good but seller is in China so I would rather source something locally if I could.

[duncanm]

wow - this is some fantastic data you're gathering.

 

Are you sure the ECV stayed closed during the run with it 'locked shut' ?  I thought the wastegate function was performed by the ECV in primary-mode -- which would explain why your primary boost went so high with the ECV blocked off by a plate.

[acres]

Pretty positive it stayed shut. The wastegate function was disabled as the positive port was replumbed to vacuum so couldn't be pushed open from the actuator. I actually went back and re did this test the other day though because it was playing on my mind too that perhaps the flap was pushing open just enough to let a small flow of gas through. This time I first checked that applying vacuum would actually 'lock' the valve by plumbing it up to the manifold while the valve was not installed and starting up the engine. It definitely was held shut as it was more difficult to open the flap manually by turning the shaft with a ratchet. I then bench tested the flap again to double check how the flap behaves when closed and as expected when closed and pressure is applied to one side of the flap it will resist opening. It actually more so tries to close itself as tight as it can.

To confirm further the valve was refitted and I re-ran the test by again applying vacuum to the positive port and ziptied the flap shut. Then did a run and the results were exactly the same as before.

 

When the ecv was blocked off I reset the settings in the ecu so that the wastegate would control all boost. I hadn't re-adjusted the boost targets though which is why it boosted so high. I've since turned it down but I think better turn it down again as it is still pushing past 1 BAR.

 

To bench test I simply remove the section of intake piping that has the flange for the ICV as it is a short section of 50mm pipe. This is bolted to one side of the valve and I use an animal dryer (very high velocity, high flow & high pressure dryer) to add flow/pressure to the pipe. Any leaks are then very obvious.

[glenisanutter]

that valve looks alright! put an adjustable wastegate diaphram on it to make sure it closes tight!

[acres]

It needs to be vacuum actuated though, so I would either mod the standard actuator again to fit or just use the vacuum actuator that comes with it.

 

Received a reply from the seller, they need to ask the factory to confirm whether it seals tight and it's max temp. The factory is on holiday until the 24th though so I wont receive an answer for a few days. I imagine temp wont be an issue though as they confirmed it is all stainless.

[glenisanutter]

The top of my turbo smart external gate has hose connectors on both sides of the diaphram!

I dont see why it couldnt run on vacuum and not pressure

[acres]

The external gate was an interesting idea as they do make a 60mm gate so would be suitable as an ecv. I was going to give Turbosmart a call to see what they thought about the idea but I thought I better check the dimensions of the unit first to see if it would fit and unfortunately I couldn't see an easy way of making it fit. If I was going to place it low in the dump where the ecv currently fits then the actuator section would be pointing down at the road, which would end tragically. I could have possibly hung it straight off the flange on the back of the turbo but it still would have been a tight fit. In the end though I managed to hear back from the chinese seller for that valve on aliexpress and they confirmed the valve will seal tight so I went ahead and placed the order, they were cheap enough to throw out if the units end up being shit.

Took delivery of them today and they actually look really good, they seem to have a decent seal on them and the machining is really neat and tidy for cheap china stuff. I'll try and get them fitted over the next couple of days and see if they are any good.

 

[acres]

Bench tested the new valves over the weekend to see how well they would seal and unfortunately they don't seal 100%. Their leak is only very minor though in comparison to the first valves that was tried however so I thought I would still whack in the unit for the ecv and see how it performed. These don't have a flange and can't have a flange welded on without adding some short pipe extensions to their body, so I decided to relocate the ecv further up the dump (closer to the turbo) and just weld it in solid. So yesterday arvo I removed the secondaries dump, cut it in half and welded the new valve into the top half of the dump. I didn't bother welding the bottom half back in as I just wanted to test it out and see if it would still be any use.

 

Results were promising. Primary was more responsive and had no issue with hitting full boost so I will weld in the rest of the exhaust and also weld in the replacement ICV and see how they go. I would still like it to spool alot faster but I think that will come down to getting a good tune sorted as looking at the graphs from this run and from when I had the ecv blocked with a plate, primary response doesn't seem to be all that indifferent.

 

[acres]

So new valves work a treat. Finished modifying the pipes to suit the new valves and the difference is night and day. The engine sounds awesome, hits boost when it should and is operating the sequential system nicely.

The ecv is no longer using the standard actuator with the positive and negative ports. I just left the vacuum actuator that came with the new valve so there is now no pre-spool and no boost control via the ecv. The original theory seemed good but just as shadoxity mentioned in some earlier posts, it seems that the system responds better if the ecv is either open or closed. Interestingly though the data records an increase in pressure from the secondary even when the ecv is closed, I am unsure if this is from a minute amount of exhaust leaking across the valve or simply the heat energy itself driving the turbine.

The ICV is still opened by a positive actuator plumbed directly to the secondaries compressor. 

 

Pic of the ICV plumbed in. Still got to pull it back out and paint.

 

 

 

Good ol' VOD

 

 

more VOD

 

 

and after a shift so still in parallel

 

 

I am pretty confident of being able to tune out the vod. The ecv is opening at 4250rpm in these shots and you can see that boost flat lines but doesn't drop, it only drops once the ICV opens at about 4750rpm. The actuator on the icv opens when the secondary hits ~12kpa if this is delayed to a higher pressure by use of a different actuator or a bleed valve/solenoid then the vod could be reduced or even eliminated. I will have to experiment with a few different settings to find the best results. It may even be able to be reduced by moving the ecv open point higher up the rev range.

[Jimbo]

Holy smokes Batman!


What does it sound like with the turbos sucking in air like that? 747?

[acres]

Yep pretty much. The spool noise is very loud, I like it. No blow of valve either so there is a whole lot of funky noises being made under there.

 

Did another couple of runs with the ecv set to open at different points. First it was moved up to 4650rpm, vod was the same just a few hundred rpm further up the rev range. What was interesting though was the response time of the secondary once the ecv opens. It is approximately 500rpm before the secondary ramps up which was the same response time when the ecv opens at 4250rpm. Like the other graphs the green line marked sensor 1 is the secondary.

 

 

 

Next, the changeover was reduced to 3650rpm. This moved the vod lower into the rev range but delayed the response of boost building back up. Boost didn't begin to ramp up until between 4500-4700rpm, the same results as when the engine is operated in parallel mode. 

 

 

From all the runs that have been made under different conditions it would seem that the sweet spot for the secondary coming online is approx. 4600rpm. Assuming a response time of 500rpm after the ecv is opened, than that would indicate the ideal changeover point will be 4100rpm approximately.

So from this I am going to try moving the changeover point to 4100rpm and using an old bleed type boost T I have in the shed I am going to experiment delaying the opening of the ICV until the secondary is building more boost, hopefully deleting the vod. 

[Robbks]

Just to refresh (it's a huge thread)

What primary and ECU are you now using?
MAP or MAF sensed?

[acres]

Turbo's are still the original hybrid's from the start.

 

Primary -> TD04 center, 13T Compressor, TD04 turbine wheel (smaller tip height than TD04L), VF25 P12 Turbine housing

Secondary -> TD04 center, 13T Compressor, Td04 turbine wheel, VF14 P16 Turbine housing

 

I do plan on replacing the compressor's with some larger wheels. Was thinking 18T originally for both but may end up going bigger on the primary.

 

ECU is a megasquirt 3 with expander board. DIY assembled and modified so that it will actually run using the subaru cam/crank signals. MAP based tune.

 

 

and this is the gizmo I use as det cans. Just a simple amplifier circuit that google supplied where I can plug a set of headphones in and listen to a mic clamped to the block.