An In Depth Analysis of The MZR DISI Head

[Mazdazilla6]

Okay, so this has taken me a while but I've finally gotten enough time to sit down and compile my findings to share with everyone. I've spent the past couple of months conducting many, many tests on our cylinder head. Overall, I'm very pleased with the results and I'm going to explain why things are so awesome.

This first post is only going to cover the intake side, I've got lots more testing to do on the exhaust side but the time frame for that is still undetermined. I'm going to start with talking about the port itself before I go into actual flow data because it's easier to see why things are the way they are with a proper image.

So here's what it looks like looking down our ports:

port.jpg

The Good: As you can see most of the head of the valve is visible, this is a good indicator that the air charge does not have to bend very far to get around the short radius turn on the port. That is good because a sharp short radius (like most old V8's have) will cause the air to detach from the port wall and cause turbulence going into the cylinder. That has a very large negative impact on airflow.

The Bad: Well this isn't really bad, but looking down a port like this doesn't really give a great idea on what the thing actually looks like. It doesn't really give us any apparent weak points in the design of the port shape that could be improved. So, how do we get a better picture? Simple, by making a silicone mould of the port!

mold 1.jpg

So that is exactly what I did. It took only about 20 minutes to mix it and pour but I gave it 4 days to fully cure. It took me a solid hour to remove it from the port without destroying it but here is what I was rewarded with!

mold 2.JPG

mold 3.JPG

mold 4.JPG

So, now that we have a 3 dimensional view of what our whole port looks like we can do some analysis of the port shape as well as measure it in certain areas. This should hopefully tell us everything we need to know in order to make a good flowing port. But before I go into any sort of analysis I would like to take a step back and talk about some theory stuff first so you can gain a better understanding of what is desirable for a port as well as why I come to the conclusions that I have.


Here is what an ideal intake port design looks like:

ideal design.jpg

Theory says that an ideal engine has the port be parallel with the valve, essentially flowing straight down upon it. This is not possible because the valve train components would get in the way. Realistically, it’s not possible to get the port any closer to the valve than ~45*. 45* is still very good because the short turn radius (the angle at the bottom of the port where it meets the valve on the right side in the above pic) is far less sharp than your average V8 port which is the picture below (notice how the port approaches the valve at almost 90*).

v8 port.jpg

Below is a look at another head, this one is off a 4G63, an engine that is well known for having an extremely good flowing head. Notice how the head port starts in the middle of the head. The reason why it can’t be moved any higher like ours is because the injector has to sit above the port so it can’t be raised any higher. Since our engine is DI we don’t have that issue and the port can start at the very top of the head, giving the air stream a very straight shot down at the top of the valve.

4g63.jpg

So now that a bit of shape theory is understood we can examine the quality of our port. To start, on aspect that makes our port great is that it begins at the very top of the head, meaning it can get closer to being parallel to the valve, so it’s right around that 45* angle which is about as good as it gets. Our short turn radius almost doesn’t even exist, and that is because the port angle is 45*. This is great because it means that the air is much less likely to get detatched from the port wall and get very turbulent (a very common effect on most stock V8 engines). Since there aren't turbulence issues going into the port, then more air can make it into the cylinder.

Another thing that is noteworthy is that the port is tangential to the valves. What this means is that the ports do not line up to the valve, but rather they are canted off to the side a little. Here is a photo of what I mean, I've traced the centerline of the cylinder with a red line to show where the center of the cylinder is relative to the center of the port:

red line.JPG

This is important (and beneficial) because of this tangential design, the air charge into the cylinder will actually swirl after passing the valve. This swirl is a form of turbulence, which is good because increased turbulence inside the cylinder leads to better atomization and more importantly, better combustion quality. +1 for Mazda

Here is a side profile of the port:

side profile.JPG

In the above photo, the nice steep port angle can be seen. Another thing that can be seen is the shape of the ceiling and the floor of the port. Two important things can be learned from this. One is that in stock form, the floor already has a fairly desirable shape which gradually connnects the port entrance to the valve and really does not need it's shape changed at all. Two is that the ceiling actually lowers further down in the port and this is important to note because the CSA of the port can greatly be increased by raising the ceiling of the port. This is also important because the majority of the intake air charge rides along the ceiling of the port and by raising the ceiling you're actually helping the transition from the port inlet to the valve, which would in theory increase airflow.

Here is something else for you to think about:

large inlet.JPG

small inlet.JPG

As you can see the two ports are different sizes at the inlet. This is true for just about every dimension between them (large first, small second):

large height.JPG

large width.JPG

small height.JPG

Small width.JPG

So now you might be wondering why they are different sizes. The answer lies in something that many people get rid of. The Variable Tumble Control System (VTCS) is often removed from the intake manifold in order to pick up a bit of top end for performance. It's mostly used during light load and cruise events so getting rid of it doesn't really affect too much but what's important is that the flapper is designed to cover the larger port. The reason for this is that all of the airflow during these light load events will have very high velocity because it is only going through the smaller port. This is good because it means that the volumetric efficiency of the engine at lower RPM/low load will be very good, and get better gas mileage because less throttle input is needed.

But this design is somewhat detrimental when it comes to making power. As I said it's a design that really isn't in use during higher load events. And since it's not in use, it's in the way. So removing it is good and all, but you're still left with one port that is smaller than the other. This isn't a horrible thing (something that I'll brush on later) but it also isn't helping us at all because if the ports were the same size, each cylinder would flow a good amount more airflow. Let's take a bit of a closer look at why I say this.

shitty valve entry 1.JPG

shitty valve entry 2.JPG

shitty valve entry 3.JPG

shitty valve 4.JPG

Above are a couple of photos of the port bowl area (where the port meets the valve). As you can see, the connection between them has a VERY sharp bottleneck. This is horrible for a couple of reasons. One, it's a bottleneck and that means it will restrict airflow. Two, it's a sharp transition. Airflow does not like sharp transitions as it will cause the airflow to detach from the port and become turbulent which is not desirable.

So the remedy for this would be to smooth it out with some bowl work. That alone will have some very positive effects on airflow without much of any downsides. For those wondering why it is so shitty it's because the head is a mass produced part and making this area smooth is not something that is done by any manufacturer except for high end performance vehicles.

While I'm talking about the bowl area, there's something else I'd like to note. And that is that our head actually has a 3 angle valve job from the factory. It's not anything amazing but it certainly is nice to see. A 5 angle would certainly be an improvement but what we have is a pretty nice start.

3 angle.JPG

So going back to the mould there are some other things I would like to touch on. One is that the valve guide extends into the port not much really and this is good because valve guides can spilt airflow and cause all sorts of trouble is they're not shaped right but ours are not very intrusive so there isn't much to worry about.

valve guide 1.JPG

valve guide 2.JPG

Also, the overall length of the port is only 5.85". This is important because port length changes things.

port length.JPG

A 6" port isn't very long and short runners are good for high end performance but not so much for low end torque or response. So how does one extend the runner? With a manifold of course!

As some of you know, I've been doing some manifold testing and part of those tests involved connecting the manifold to a head to see how it flowed because the more of the system that is tested together, the better of an understanding that can be gained on it. Yes, it is useful to flow individual components but if you can connect them and test them together, it is even better.

For my testing, I didn't test as many manifolds as before because this testing takes a very long time and I wanted to examine some certain aspects which led me to these manifolds: JMF V2, JMF V1, Stock and the head by itself. No I did not test any ported stock manifolds, it is not worth my time to test these because the results are predictable and I would rather spend my time testing other stuff.

Before I dive into the testing and results I'll give a bit of info on the setup so you can sort of see how it's done. As before, the test pressure was set at 25" of H2O in order to keep things consistent. Please don't compare my tests to other tests that people have done, different test pressures give different results and many people conduct tests at 28" of H2O and the pressures do not exactly convert. I chose 25" because that is what the bench was calibrated at. To minimize leaks, the exhaust valves were installed with their springs to hold them closed and the spark plug was installed as well as the injector bore was taped off. All connections on the manifold were taped off such as the MAP sensor hole and all other holes for vacuum and boost sources. The intake valves were installed with light springs in order for them to be opened easily. A dial indicator is installed above each valve to keep track of the lift that they are at. Lift points were done in increments of .05 L/D and ended at .035 L/D. This was done to give a good idea of how the head performs with camshafts. A bore adapter was used to simulated proper cylinder conditions so the valves would be shrouded properly to keep flow numbers from being inflated. Unfortunately the closest adapter I could find was 3.5" and our bore is 3.44" so bear in mind that my results might be a tad inflated.

setup 1.JPG

setup 2.JPG

setup 3.JPG

setup 4.JPG

On to the tests! So to set a baseline I flowed just the cylinder head by itself. The results are below and they're really not too bad for a totally stock head. The port itself is restricted to right around 190 CFM but what's important is that they are all exactly the same shape and so they all flow the same amount at pretty much all lift points. +1 for Mazda again.

Bare Head.png

Then I moved onto the stock intake manifold connected to the head. I decided to test with the VTCS open as well as closed. What I found is that with it closed, they flow fairly even, which is nice for cruise and low load events but really not useful for making power. When the flappers are open it can be seen that the best the imbalance gets between cylinders is 12%. That is fairly close to the 14% that I found when flowing the manifold by itself. Why is that important to note? It's important because it tells us that despite the head flowing perfectly even, the manifold imbalance upstream from the head will still cause flow imbalances and that in order to have an engine with balanced flow across the cylinders we need little variation between the ports in the head as well as the runners on the intake manifold.

STOCK.png

After that my next curiosity was between the two JMF manifolds. From my bare head test and the flow tests of the JMFs I knew that they would easily outflow the head but what I wanted to find out was how much the runner divider in the V1 made a difference compared to the V2 which didn't have one. My results were a bit surprising in the fact that there really wasn't much different. My guess was that the longer divider would aid in low lift flow but it seems I wasn't quite right on that. More importantly, what was learned from this is that the flat face on the head of the runner divider really doesn't impact flow much at all. So contrary to belief, it's okay to run a manifold that doesn't have a divider on a stock head. Another thing that was learned is that the JMFs don't flow evenly at low lift and are decent at high lift. I believe it tends to smoothen out at higher lift is because the runners are so short and the air has a higher velocity. One other thing that I just feel inclined to say about the JMF manifold(s). While yes it is far better than stock, it leaves a lot to be desired. The JMF flows enough air for 1200 horsepower. So yes, it's pretty oversized and not only that but the runners are extremely short. This is fine for high RPM power but it doesn't do anything for low down torque and doesn't really help spool a turbo any faster. Flow results are below:

JMF.png

So after looking at the results I'd like to go back quickly to the topic of cams. Our stock cam has a max lift of 0.372". That's right around the area where flow levels off so what this tells me is that the stock cam is matched nicely for a stock head. It also tells me that there's no point in dropping in a hotter cam without putting in larger valves. The issue with larger valves is that airflow gained is minimal because they will be shrouded even more by the cylinder wall and the other valve. So IMO there's not point in upgrading cams unless you are opening the ports up UNLESS you are moving to a fatter profile grind that offers similar lift but has a greater duration at those higher lift points.

Some other things that I would like to brush on are once again the difference in port sizes for each cylinder. The reason I'm going back to this is because it is not desirable in terms of power to have these be different sizes. The reason for this is because the larger port can actually start to pull air from the smaller port once it has reached its max flow potential. I know this to be true because at high lift on the bare head tests the noise was unbearable. This noise is from the larger port pulling air across the divider and out of the smaller port. The two remedies to this would either be to eliminate the divider or make both ports equal in size.

One last note before I wrap this up is that the single greatest quality about the stock head is how evenly it flows. This is EXTREMELY critical because an engine can only make as much power as its weakest cylinder. The only way around that is to have one of each sensor for each cylinder (not something really feasible). So if all the cylinders flowed the same but one *cough* cylinder 4 *cough* flowed 20% less, then the rest of the cylinders will be forced to make 20% less power. That is why balance is paramount when it comes to airflow through an engine.

I've attached the results for all of my tests in case anyone is interested to see the whole numbers. Unfortunately I left the swirl tester off for all but one of the tests. I didn't want to run short on time for testing (which ended up happening anyway) but I was still curious about the numbers. I didn't have time to test port velocity and I might try and go back and do it but that might not be for a while.

Thanks again to @maisonvi; for loaning me the manis

Thanks for your time gents. Let me know if you have any questions.