Driving Aquamist 2c high speed valve with stand alone EMS

[max legroom]

I have a highly modified 2002 Subaru WRX, with forged internals, 9:1 compression ratio (stock is 8:1), and lots of other supporting mods. I'll be installing a Tec3 EMS (using full-sequential fuel injection) very soon and have an Aquamist 2C ready to go in at the same time.

There are 2 possible ways I can control the high speed valve. The Tec3 has a staged injection feature which links injector drivers 5,6,7 and 8 with drivers 1,2,3, and 4, respectively, using a load/rpm map to determine when and for how long to open staged injectors. Using this feature, the Aquamist high speed valve would be opened in sequence with only one of the cylinders, since only one driver can be connected to it.

The second possibility is to use one of the General Purpose Outputs to drive the valve, which could be configured to fire the valve with every cylinder.

Are there pros and cons to these two approaches? Which one would be best, and what size nozzle would be appropriate for each (assuming 15% water/alcohol mixture)?

[Charged Performance]

Yeah I can imagine if you are running any boost at all on the flat4 with that CR you are dying to get some detonation suppression or have a good account with a race gas supplier.

Theorectically, I suppose you could use the signals from drivers 5-8 even. Isolate the signals with diodes. Though at higher rpms the frequency may be too high for even the hsv. Maybe do a 5 & 8 or 6 & 7 signal this way.

Anyway I digress to your question. If you use just one driver, you will of course only be pulsing once every two rpms. On moderately tuned Subarus this should not been problematic given the long intake runners. However you are likely pushing much larger volumes of air and the related velocity may not allow for a uniform mixture. You could possibly get water leaner and richer cylinders, probably not the same ones every stroke while rpms are changing but still each cylinder would be water lean or water rich at one point or another especially at high rpm, high boost. This isn't good for detonation suppression or finding a stable timing setting. Again not a problem on moderate setups but high cfm setups could face this problem. The method I described above as well as your second method would mitigate this. My response here also assumes you are injecting at the throttle body. Clearly if you are injecting at an earlier point on a FMIC system for instance the water will have more time to become uniform before the plenum. Also the above is speculation since the plenum design and runner length has not been shown to cause a problem with getting uniform flow.

Provided the duty cycle could also be controlled the firing twice per rpm with each cylinder would be ideal since the mixture to each cylinder would be much more uniform. The issue again becomes whether or not the hsv could keep up with the frequency. It is rated at 250mhz or about 7500 rpms. However this also only leaves 4ms duty cycles. Take away 1ms to open and 1ms to close and you are down to 2ms of resolution for flow control. Not a great situation.

I think the best solution may be to go to 2 hsv's. Drive one with the 5&8 wires and the other with the 6&7 (using diodes to isolate betweent the paired driver wires). At a minimum that would result in 8ms cycles for each hsv allowing for 6ms of resolution for the flow control. In theory you could try drive these through one jet but are probably better of driving them through at least one jet for each hsv.

Just some thoughts.

[max legroom]

Thanks, Ed, for the interesting discussion. I will be raising the redline to take advantage of the new engine components and it is good to know that the hsv will not handle that frequency. I like your idea about using 2 hsv's and alternately firing them. Do you think the single pump that comes with the 2c will handle 2 valves and 2 jets?

[Richard L]

Hello John,

I like to point out the conclusion of the HSV not handling the frequency is quite misleading, here is my approach to the situation. The HSV's frequency response is almost identical to the WRX fuel injectors, if not slightly faster and higher flow.

Your WRX, an unusual approach to the norm - EMS-T3 is a complete takeover as I can see . Most people will be happy to buy a read-made (utec) or semi-made (reflash) package using existing downloadable map or asking a tuner to map it for them. Unfortunately 99.9% of the tuners out there are not adventurously enough to approach WI head-on, some use it as a safeguard for bad fuel and some just dis-approves it outright.

Injector speed: all 16 ohm fuel injector (wrx/sti) has an average opening and shutting time of around 2ms at 12V, the aquamist HSV is no different -this is not a design fault - it is to do with limited energy feeding into the coil to perform a mechanical action. The amount of energy applied to the valve is limited by voltage and coil impedance (resistance). As energy is power over time (V^2/R).

Assuming your are tapping into one cylinder of the T3, running sequential injection. At 6000RPM (100Hz), one of your fuel injector is firing at every two engine revolutions or an interval of 20ms (50Hz). The control-able range is between 2-28ms (taken into account of opening and shutting time). I cannot see any problems for the HSV or your injectors to run out of injection time. So far so good.

The main perceived problem is not the HSV unable to run fast enough but out-of sequence of the induction phasing intervals. In other word, the HSV is firing in phase with one injector and no water for the other three cylinders. This conclusion may hold true if the water jet is close to the inlet valve of the plenum chamber.

Let say if you place the water injector before the intercooler and immediately after it, the water droplets will take sometime to arrive at an individual cylinder, after passing through the satellite distribution chamber. The "no-water" gap is no longer clearly definable. Further factors such as rate of evaporation of drop size travelling at different speeds, rate of evaporation in the inlet tract and some speed reduction of the drop hitting the wall of the inlet tract. My conclusion is that it will not be a problem triggering the HSV with pulse from one injector. If you are still concerned about cylinder distribution, place two small water jets at few inches apart so that it will have the same effect of "filling-up" the gaps. One good method is placing another small jet in front of intercoolern and one after.

Your second choice of using the GPO - no problem at all - run at a fixed frequency of 32Hz or 64Hz, the HSV will deliver water effectively and evenly across the entire RPM range.

Why not running both (your suggestion) if you still have concerns, two HSV firing at opposite cylinder phase- intervals - cheaper to phase the mechanically (not bad) than electronically.

We need to talk about the faulty diagnostic functions you can perform with the GPIs later. If you are happy with my reply and would like to add some comments.

Even at 8000RPM, the inteval of the fuel injection pulse is approximately 14ms. The HSV will handle that with ease.

[Richard L]

my post merely reflect and comfirmed "charged-P's"

[max legroom]

Hello Richard,
Thanks for that clarification. Using two jets located a little distance apart in the intake path sounds like an ideal solution. I guess using 2 x .4 mm jets would be best, as that would give .32 sq-mm area, a little less than a single .6 mm jet. Do you think that's reasonable?

I thought about the utec and reflash options, but in both cases I'm stuck with the limitations of the stock ecu. I want to add wideband O2 and use MAP for load calculations, neither of which are possible with the stock ECU. I will also probably use some of the tec3's more esoteric features, like "Throttle position rate of change"... Sure, I could add a few thousand dollars of controllers (and design some custom electronics myself) on top of a reflash or utec, but putting all that together in a single unit like the Tec3 makes life simpler.

But the main reason I'm going in this direction is that I want to understand eveything about engine design and tuning. I did a lot of research last year, decided what I wanted the torque curve to look like when I'm done, then pulled my car into my garage and rebuilt the engine last winter. Now I'm ready to start tuning it to take real advantage of "my" engine. It runs well enough with the stock ECU (although open loop is less than optimal because the map is way too rich for the smaller combustion chamber) that I can drive it while installing the Tec3.

As for tuner's attitudes toward WI, i think in general you're right, many of them just don't get it.

[Charged Performance]

The two 0.4 might be a good place to start - step wise tuning is best unless you have a lot of WI tuning experience on a set up. But I do suspect that eventually you will eventually increase the size to two 0.5 or more. Remember you will have both pressure and jet size to adjust your flow rates.

Just a clarification by no means was I suggesting that the hsv is at all lacking as Richard said its performance is on par with most quality fuel injectors - just using one hsv to keep up with 2 injector pulses per revolution is asking too much of it. It should keep up with 1 per revolution fine. Hence the recommendation on the cross cylinder cycle injection.

I will be interested in seeing how the Tec3 tuning goes. If you haven't gotten it yet have you looked at the new hydra system from element tuning at all? http://www.elementtuning.com/main.htm

[max legroom]

Hi Ed,
Thanks for the advice on jet size. You're right, it would make sense to go with .5mm jets. I'm leaning toward using a GPO for hsv control, since that would allow me to stop firing the valve completely under light load/low rpm conditions and customize a map over the rest of the matrix. The .5mm jets should flow plenty of water, especially at high rpms when there's less time available, and if they're off at low rpms then I don't have to worry about exceedingly short on-times to keep water flow at a low rate.
It'll probably be mid May before I get everything installed and ready to tune (got other things I also need to do. Bummer...). I'll post my experiences as interesting things happen. In the meantime, other people's comments and experiences would be welcome.

[slowMX5]

I am running a TEC3 on a turbocharged MX5 (1.6L) with a stock 9.4:1 CR and 15psi of boost. I've been experimenting with a 2C over the last couple of months - with Richards help. I have ended up using 2 jets, one just post IC and the other just before the throttle body. I also started out with 0.4mm jets - but have since gradually moved up in jet size to deliver more water. Main reasons were the improved spray pattern once using 0.6mm jets and larger and the fact that when using the TEC3 to drive the HSV one can adjust duty cycle and hence flow. I am also experimenting with higher water flow rates (upto 400cc per min so far). At present I am using a 0.9mm jet and a 0.7mm jet, but will most likely replace the 0.7mm with a 0.9 or 1.0mm jet.

I have plenty of flow rates versus duty cycle for differing combinatins of jet size. My 0.9mm and 0.7mm jets will provide upto 470cc per min of water flow at 90% duty cycle.

Also when testing the flow I discovered that the jets continue to provide a mist down to about 15% duty cycle, any lower and the pattern suffers severly.

I have set things up using the GPIs so that the TEC3 only adds timing and leans the fuel when the TEC3 sees water flow. Currently it's flow or no flow - but I intend to try a linear relationship relating timing added/fuel pulled to flow rate soon.

If I can help any further then please ask.

[max legroom]

Hi Steve,
This is really interesting. It sounds like you've got a great car! I don't want to ask too many dumb questions, but it sounds like you've got a very well designed system using a GPO to control the hsv (I'm new to the Tec3, but I can't see any other way of tuning the water injection as you've described it except by using the GPO duty cycle table). My WRX has only a couple of inches between the intercooler and the throttle body. I apologize for my ignorance of the MX5 configuration, but how much distance is there between the intercooler and the throttle body, and what is the pipe diameter? I would also love some duty cycle information to understand how much water you're injecting and where in the load curve you're injecting it. Most people would think those jet sizes are very large for a 1.6 liter engine (and they would be right if you just pumped water through at a constant high pressure), but it sounds like you've done a great job configuring the injection duty cycle to control injection rate.
I'll post the findings with my configuration as I work things out, but I'd really appreciate more details on your setup.
Thanks for posting!

John G.

[b_boy]

Please excuse my utter ignorance of most things engine tuning.

I'm going with the the ECU that Ed mentioned, the Element-Hydra, with maps from Phil Grabow--former TurboXS employee. I just bought a Aquamist 2d from Ed. I have the USDM WRX STi 2.5L EJ257, and I too would like to use the ECU to drive the water injection system.

The Element has fuel injection controls as well as a few "extra" controllers for things like water injection.

I'd like to set up a 3D map for water injection but I am at a loss as to what parameters will be best for the ECU to track and alter water injection to suit the engines needs. RPM and boost are givens for water injection, but I'd like to add in something like EGT temp or air temp at the throttle body, scaling water injection when temps rise.

Let's say we use: Load x Boost x RPM x Temp at Throttle Body (TTB).

The water injection map will mirror (scaled down) the fuel injection duty precisely according to the tune without WI plus additional water as boost increase. So load and boost go up so too will the percentage of water (I think). Let's say at a load value of 2, boost value of 3 psi we will inject 5% water (as a percentage of fuel) across the RPM range (that is the 5% value holds, but more water is injected as the RPM increases with the fuel injector duty). At full boost, say 17 psi, and a load value of 5, we will inject 15% water (the load and boost have consorted to raise the percentage from 5% to 15%). Now I'd like to scale water injection such that an addition 1.5% of water is injected for each 10 degrees C above 40 C TTB. So, if the outside air is 40 C (100 F) instead of our usual 20 C (here in Seattle), the controller would be injecting an additional 3% water changing the injection range from 5-15%, to 8-18% of total fuel.

I have three basic questions:
1) Am I crazy, does it matter to scale with outside air temp, boost, and load?
2) Where do even begin with the 3D map values (I can't find any 3D water maps)? How does one develop such a map?
3) Let's say I manage to get a 3D map from someone (maybe from Richard L) do I need to worry that this 3D map is good enough for my engine, with my mods, etc?

I basically think that the "customization" to my mods will come from paralleling the fuel map (and anti paralleling the fuel dumping). So I think question 3 is answered to some degree.

EGT seems to be a means of developing a water injection map, but the number of test points to create such a map is just mind boggling. I can't imagine injecting water at each load, boost, RPM value to the point where EGT is lowered, and then backing off on fuel to reach stoichiometric, then adding water back to lower detonation if needed.

From what I have read a fine tuned map like I describe using EGT is not necessary; there is a fairly wide range of water percentage for each point that will "work". None the less I think a finely tuned map would not be unwelcome to the community. (Cylinder pressure based map has been discussed, but it's a pipe dream right now).

The first two questions are totally up for grabs. I'd be interested what percentages of water injection the TEC3 users are using and why/how they decided to go with those values.

[Charged Performance]

The closest public one I am aware of right now is the system 2s document from ERL/Aquamist. They have an example of how a potential 3d water map would look graphically with explanations of the increases and decreases along the axis.

Using that as an example of the shape a tuner would start with the set up non-WI tune and add water in on the points most stressed and requiring the richest fuel and least ignition advance. The water added should ease those requirements permitting fueling to be reduced and requiring ignition to be advanced. It becomes an iterative process at this point until the tuner is satisfied that the tune is achieving the sought after objectives.

In this case with Phil I would expect it to be some significant power improvement with good realiability and stability - not the ragged edge. But I am pretty sure that is what you are looking for.

[Richard L]

A simple map created by the MF2 based system (system2s) fro general use on a turboed engine.

http://www.aquamist.co.uk/forum/mf2-map.gif



Here is a MAP created by our discontinued, 12-year-old flash-based, 16-bit mirco-controller WI system for the turbo Bently, it has 128 rpm locations, 64 load location and 256 level of water injection %. Flash memory was the "State of the Art" technology at the time - we were one of the first company that employed it commercially.

The system was completely self-learning due to the huge memory loactions: 128 x 64 x 8-bit word, it was naturally impossible to map manually (8192 cells). The follwing map was created in just 15 minutes of driving on a test track when boost was increased from 6 psi to 15 psi. The only input was a dedicated knock signal line provided by the factory ECU.

If anyone want us to re-manufacture this system, let me know. Please don't ask the price.

http://www.aquamist.co.uk/forum/sys3map.gif

[Charged Performance]

Another starting point would also be the fuel maps being employed.

Using the IDC of the injectors vs RPM & MAP for the initial duty cycle of the high speed valve with jets sized to a % of the injectors' potential would start you at approximately the same point as the 2d. From here the WI map could be modified based on torque and knock feed back as fuel and timing are adjusted to the targets. Of course from this standard starting point fuel and timing need to be adjusted to make use of the starting injection map.

If the fuel and timing in an area of the map are at desired levels but torque has fallen off from prior results, reduce the water injection rate at that point assuming knock isn't being encountered, which it shouldn't be.

If knock is being encountered prior to the fuel and timing desired for an area of the map, increase the water injection rate further at that point to permit tuning to the desired lambda and timing advance.

[b_boy]

Thanks for the replies guys.

Well I can see why you went to a more simplfied map with MF2, that matrix is dizzying.

I have studied the graph in the MF2 manual many times. It's easier to read in color ;-) Thank's Richard. Here is what I don't understand: the "injection rate %" description of the Z axis.

A: Is that the percentage of water injected compared to fuel?

B: Or is it the percentage of time that the water injection valve is open, thus the "rate" would depend on the jet size/flow and pressure?

Based on what I have read, 65% of fuel (the peak at 1 Bar boost) as the amount of water injected is very high, and I would have to assume no intercooler in the works. Option B seems the more likely interpretation, but I really don't know.

If it's option 'B', then the amount of water injection is not readily determined from the graphs. Is the water injection rate the maximum 150 ml/min from the standard 0.5 mm jet at 4 bar pressure? Thus we are looking a percentages of this rate?


So Ed, I would like to do just as you say "with jets sized to a % of the injectors' potential".

In either case I don't know what the "% of the injectors" number is-- 15%, 25%, 50%.

If the max water rate is 25% of the max fuel injection rate, and Richard's graph is showing 65% of 25%, that's a 16.5% water to ratio to the fuel. That seems about right.

Let me know the many ways I am confused.

[Charged Performance]

I believe that is B. or the water injection duty cycle you see there. Again just gives you an idea of the shape not the exact flow rate.

% of potential is based on your fuel injector's flow rating at the pump pressure being regulated.

For your goals depending on the load point you would find that from 12% to 20% water to fuel will likely be the result.

If for instance you have 400cc injectors your potential on the fuel injectors is 1,600cc. To make sure you can reach 20% if you needed to you would select a jet combination that have a potential of 320cc or 20% of the fuel potential.

This can be achieved with an Aquamist pump using a 100cc accumulator (even more easily using the STi water tank pump as a primer) and a 1.0mm jet. You could bench test the system to play with the water pressure switch to get that flow by engaging the hsv fully for one minute with the jet in a measuring container.

Now if a particular point is at 70% IDC for the fuel injectors and you put the water injection cycle at 70% also you will be at 20% water to fuel. 52.5% water injection cycle will put you around 15% water to fuel and 35% WIC would be at around 10% water to fuel. The exact percentage on a point by point basis is going to be trial and error. This is because there are very few identical setups to copy across several of the same car and almost none at this point for the STi as to my knowledge in the US none are on stand alone yet.

Starting from the optimal nonWI tune I would advise Phil to use water injection rates at 50% of the initial fuel injection cycle map - this would give around 10% water to fuel across the water injection map. Without fuel and timing adjustments the car will have lost torque at this point. Reduce fuel and increase timing to the new knock threshhold and see where that ends up. As fuel is being reduced the water to fuel ratio will be increasing as well since they are still at 50% of the initial fuel values.

In either case try not to get below 25% water injection cycles - if any points need this low an amount you probably need to use a smaller jet instead and increase the cycles.

Phil has the experience to feel out where adjustments are needed between the three variables from this initial point.

Treat it kind of like race fuel except that you can actually adjust the octane of the fuel itself - the water injection rate.

Invite him over to read this thread as well.

[b_boy]

Thanks Ed,
That makes more sense that the graph is reading % duty of the HSV. I understand it's an iterative process, I just want a base to start from and I seem to have it now.

Using the 3D graph from Richard if I set his maximum water injection rate of 65% duty at somewhere around 20-25% water:fuel and scale the rest of the map I'll come out fairly close. So at very low boost and medium load you are injecting (~5% duty on graph) or 1.5-2% water:fuel.

Why do you encourage water duty >25%? If you drop below that duty are you getting pulses of water such that each cylinder is not seeing a homogeneous mix of water:fuel. I'm cool with doing it, just wondering want the reasoning is.

[Charged Performance]

There is another post around regarding this. But when the duty cycle is low the high speed valve is not open long enough to fill the hose before jet and give good pressure during the cycle. If you have a need to use a duty cycle less than 25% you probably should go to a smaller jet or lower pressure and raise the whole map.

[BMWTurbo]

Hi Guys,

I'm sorry to mine an old thread, but it makes more sense then starting a new onw for a similar question.

I have been using a home brew W/I system for a little while now and think I'm approaching it the wrong way.

I have the nozzle fitted cirectly into the throttlebody of my 2jzgte engine and I'm injecting through a '400cc/min' nozzle via a solenoid valve that is either On or Off. I don't have any fancy control over this only an On/Off switch.

The engine makes somewhere over 400rwhp and I'm at around 70% injector duty on 650cc/min injectors (6-off).

My trigger point is 3800RPM where the second turbo comes on and both turbo's run at about 23/24psi boost.

Working through the above I'm starting to think that I'm injecting way too much at this point (prompted by the fact I changed the plugs last night to NGK BKR7E's gapped to 0.7mm and I'm having the engine miss from the opening of the W/I valve until about 5500RPM where is cleans up).

The 10%-20% stated is a percentage of actual fuel injected and cannot be derived back from injector duty and injector size due to the fact the injector duration isn't enough to use this as a percentage.

IE I can't say 70% duty of 650cc/min injectors = 455cc/min roughly x 6 = 2730cc/min

So 10% of 2730cc/min = 273cc/min.

However that 273cc/min would be if the injector duration was the whole revolution but this changes with RPM.

With a single trigger system, is the calculation different then the usual percentage I see generally referred to?

I don't really want to gap down the plugs if I don't have to the engine feel much nicer at light loads with 0.7mm gap then it did with the old plugs gapped to 0.6mm.

I have what I beleive are decent coils and I'm also running a HKS Dli.

[Howerton Engineering]

One calculation we have used for single stage systems is this: You want to start injecting before the torque peak. So on the dyno plot, find the RPM where the TQ is climbing to peak, and look at the HP at that point. Multiply the HP number by 1.5 to get CC/min of an injection amount.

We usually use crank hp numbers for the calculation.

Now, if the car were at 250hp during the ramp, then the jet would be a 375cc as a starting point. Not far from where you are. You can run your numbers to see what it actually is.

One more item to consider, injecting at the TB puts a lot of large droplets in the chamber which can blow out the spark. One thing I would try is moving the jet further away from the manifold, as close to the exit of the IC as possible.

[BMWTurbo]

I have taken this post and started a new thread (http://www.aquamist.co.uk/vbulletin/showthread.php?p=15861#post15861) as it was drifting off topic.


One calculation we have used for single stage systems is this: You want to start injecting before the torque peak. So on the dyno plot, find the RPM where the TQ is climbing to peak, and look at the HP at that point. Multiply the HP number by 1.5 to get CC/min of an injection amount.

We usually use crank hp numbers for the calculation.

Now, if the car were at 250hp during the ramp, then the jet would be a 375cc as a starting point. Not far from where you are. You can run your numbers to see what it actually is.

One more item to consider, injecting at the TB puts a lot of large droplets in the chamber which can blow out the spark. One thing I would try is moving the jet further away from the manifold, as close to the exit of the IC as possible.

Great info thanks for that.

It's a sequential twin turbo system, ie, one turbo until 3800RPM and then both turbo's after this. By about 4200RPM I have full boost on both turbo's and roughly peak torque @4500RPM odd(from memory). Power would be a modest 400 engine around this area.

I gapped the plugs down to 0.65mm last night and found I still had a bit of blow out again this morning. I can easily enough move the nozzle into the I/C piping for a trial, but it will be in a vertical section of pipe about 2 pipe dia from a 90 degree bend. I was wondering if possibly the nozzle location wasn't the best as the spark blows around 4500 and it clears up 5500rpm odd on the 400cc/min nozzle.

http://img38.imageshack.us/img38/7966/cooleroutletpiping.jpg
Here is a 1/8 BSP fitting that I have used for testing I can easily relocate the nozzle to.
http://img252.imageshack.us/img252/5989/img3909medium.jpg
The engine bay, pipe in the photo above is vertical into the large rubber bend heading into the throttle body.
http://img198.imageshack.us/img198/7484/dscf9432medium.jpg
Currently mounted like this in the lower part of the TB.

I'd like to run the 0.7mm gap on the plugs again if not a fraction more as the engine is more responsive with this gap, but I want the spark to stop blowing out.

My trigger ponit for the solenoid is when the 2nd turbo is told to come on-line, so it's injecting as #2 comes on-line and stops at 2psi manifold pressure, ie when I back off.