2013 CRF450R supermoto conversion

[Guest Connader]

Everything is going according to plan, be patient and keep following up this thread.

There is a lot more to come

OEM - 128 CFM

Previous owner - 143 CFM

Long road to perfection - 187 CFM = + 46% vs OEM or + 31% vs previous owner.

Ah ok, got it!!

Nice work Mate!!!

[kab-pro]

Port welding.

I would gladly hear a specialist's opinion about the technology. In the picture below you can see an intake port that has been welded.

Does anyone know what tech is necessary to pull this off? Welding the intake ports on a 250cc head seems like a piece of art.

I know there is a guy in the US who does this but he will only present the results and never process.

Right now two options are considered:

- epoxy welding (also offered by the top tuning workshops);

- soldering with Techno-Weld rods.

After some consulting it was decided that we will try out the soldering option.

Conclusion:

The actual result was not as expected but I guess we were being too optimistic since soldering material on both floor and ceiling is a tough challenge. Props to the wiz who did it.

Regarding the realization process of this research - this was a long time mind boggle but the goals were of course the highest.

Plan A: this has to be done by a machine

Plan B: it will be done by hand using all sorts of models and comparison tools.

I have to admit - Techno-Weld usage was the last impulse to carry out Plan A.

The hardness of the soldering reaches ~ 93 by Brinel scale. When taking in consideration the thickness of the soldering layer, plan B is getting practically impossible.

Getting ready for Plan A.

CAM system is deployed to generate toolpath and couple it with cutting feeds, speeds and positioning.

The G-code produced is not always perfect so this is when we alter huge amounts of it by creating a secondary program in free-pascal.

After running the machine simulation we are finally looking at what could be the result of the first porting operation.

[kab-pro]

Tool and equipment.

We have finally come to the machining process. The tool type as seen below is not exactly right for such a job but it was the best option available at the moment given the resources we had. Traditionally lollipop mills are used but this cylindrical one is just enough to reach the furthest points of the port.

Next up: machining a jig to fix the head with the option to flip it around without losing positioning. Flipping will be necessary due to the machine type that simply does not support full tool travel in a single setting.

And of course the support spacers and bushings. Each spacer has it's own calibrated length that was calculated with FARO probe to avoid any human error.

These lengths are key to preserving the positioning when the head is flipped between operations. The one mistake that was made was to machine these spacers from aluminum but I will get into detail with this as we move on.

[Guest Connader]

Hi Andis,

Epoxy is the most common material used, as I heard. "JB Weld" is very popular here.

Here is a thread about porting, welding (using Epoxy) and how the finished product looks like (nothing what you don't know already, but anyway )

http://www.trx450r.org/forum/93-engine/290074-trx-porting-theory-tools-how-diy.html

The Thread-starter is the Guy who did my ports.

Keep on.....

[kab-pro]

Time to test the machine code and see how it will materialize itself. The first idea was to start with a wooden block, though it was decided to begin with a piece of plastic foam since damaging the tool and machine is fairly easy to do.

The amount of G-code is approx. 5 Mb which makes it physically impossible to inspect.

Let's start up the CNC and munch on the foam a bit, and instantly, you can see why testing is necessary. The first try could not have been more wrong than this.

But that all likes to change.

Still, some G-code errors pop up and must be accounted for in the CAM system so that we can continue with the wooden block.

The main reason for wood is to test how flipping the head around will work out and to see if the cutting speeds and feeds produce any smoke.

And soon enough, we get something like this. Looking good so far.

A few imperfections are still popping up, but those can be easily corrected.

[Throttle_Jockey]

Very nice! Testing is always a must. One day I asked my teacher how and who did the huge "mismilling" mark on the CNC machine table... Then he started to look a bit sad and disappointed and said "it was me - first try ever after we got the machine...."

I can already see it on eBay, "wooden head for crf, ported and ready to use..."

[dr.-hasenbein]

I can already see it on eBay, "wooden head for crf, ported and ready to use..."

Muuaahhaaa

[kab-pro]

CNC headworks:

Not much left to say here. Whats important has already been described earlier.

[video=youtube;leQWGJwXUFw]

In conclusion:

Mechanical processing:

- From intake side - 4 rough cuts + 1 finishing cut ~ 5 hours total.

- From valveseat side - 4 rough cuts + 1 finishing cut ~ 5 hours total.

Main factors influencing machining time:

- thickness of soldering layer

- soldering hardness

- tool type

Edited July 3, 2016 by kab-pro

[Nik]

Hi

This looks pretty nice! I have two small questions:

- Are you going to use new/bigger valves or will you use the original valves?

- Are you able to make a rough estimate how much time you spend for that "project" ? -> Time for calculating, constructing, making the machine code, testing, cutting , etc ...

[kab-pro]

I am going to use Xceldyne valves (OE size). At first the idea was to use bigger valves but as I had already decided to go with Xceldyne I had to stick with OE size since they didn't offer oversize. Nevertheless, I was able to achieve results very close to the oversize.

My estimate about the time spent on the project would be very rough since it was a "free time" project. Anyhow, it took a great deal of time, several months, spread across the whole year because I had to learn every single step from scratch. But right now, since all of this is no longer new to me, the time needed for another project like this would be significantly less.

[Nik]

I asked because I talked to someone yesterday, who is building a 2016 crf at the moment. He's making some changes to his head with a flow-bench too. I told him about your project and maybe he'll contact you for some more details.

He also told me that the valves are the biggest "problem" at the Japanese engine heads. He told me that the flow drastically decreases with installed valves. They act like a "restrictor". Bigger valves should fix those problems a bit

Did you measure your flow with or without valves?

[kab-pro]

The valves do the same on every head regardless of make. All the measurements were made to simulate real life scenario as close as possible. That includes the valves, combustion chamber and even the spark plug.. I did the simulations with oversize and stock size valves, and the difference was somewhat minimal.

I am looking forward to see his results with the 2016 head. And thank you for spreading a good word

[Nik]

Hmm okay... I'm not really into that subject so I don't have any personal experience. I also don't know how and what will influence the results. He just told me that your result (126 CFM for OEM head) is pretty high.

But if I interpret your results right it's the "opposite" of what he told me about the valves but nevertheless you both are doing pretty nice and interesting work

Interested in doing the same stuff at a kawasaki head?

[kab-pro]

I didn't actually test the OEM head since my port was already modified as I got it. The 128 CFM is a figure I found on the internet after some searching and the source seems to be legit as well. Maybe your guy was reffering to a different test pressure. As I got the info, it flows as follows (all numbers @ 0.400 inches valve lift):

- 116 CFM @ 10 inches of water

- 128 CFM @ 12 inches of water

- 195 CFM @ 28 inches of water

About the valves, I only experimented with the diameter, not the shape of the valves. Maybe that is why my results for stock and oversize valves were so similar.

What year is your kawasaki and is it a KX450F?

[Nik]

ah okay

as i mentioned before, I never worked with a flow bench or simulated sth. like this with a CAD programm so I don´t have any personal experience and of course I can´t tell excatly what method he uses.

And my kawasaki is a KX450F 2009. First year with EFI

[kab-pro]

Do you have any pictures from your cylinder head / intake port?

I might have to order a special tool for the job. I hope you understand that it is a delicate and time-consuming process.

Regards.

[Nik]

No I don't have any pictures of it at the moment

But if you want you can send me a msg with some more information

[kab-pro]

Digging thrpugh the treasure box to take the head to a local workshop for guide and seat installation.

And here it is as received back.

The guide ends are machined manually.

[dr.-hasenbein]

Definitly no need to grind the surface more smooth ?

edit:

wrong. why change for bronce seats ? because you can ?

Edited July 4, 2016 by dr.-hasenbein

[Ben]

That´s Beryllium right?

[dr.-hasenbein]

If ei rimämbär reit, Ei dond ßing ßo

[kab-pro]

Yes, that is beryllium bronze. It has proven to last forever when used together with titanium valves. That is actually a full Xceldyne valvetrain kit. A bulletproof setup.

About the surface roughness - I have left it as you see in the pictures. No smoothing. Smooth surfaces condensate and accumulate fuel droplets essentially ruining the mixture. Thus, it has been proven that the best aero and hydro-dynamic properties have been displayed by surfaces that are not smooth, for example a golf ball or shark skin. That is why I have left the surface wavy and textured. Simply out of curiosity.

[kab-pro]

When the head flow and cam profiles have been properly analyzed, we are finally ready to build a full mathematical model of the given engine. There are some powerful tools out there designed specifically for this task. They require a whole lot of different specs and evaluating those properly is the key to successfully generating a valid dyno graph. The most basic specs include bore, stroke, rod length and some of the more advanced specs include valvespring rate tables, head flow tables, cam files etc.

When the graph has been generated, it has to be compared to another valid source. Since we didn't make any runs by ourselves, there is a site called MotoUSA with plenty of different dyno sheets. Here is the comparison between what they have published and we have generated using ONLY a mathematical model.

2013 CRF450R stock

You can see the resemblence being quite believable. The slight differences that you are seeing are due to fuel mixture, spark curve and environmental conditions. And of course the allmighty error of math regarding such complex things, which to my surprise, is proving to be somewhat unimportant, since the goal of this virtual simulation is to evaluate the changes brought from every mod, not reaching absolute numbers.

Edited July 6, 2016 by kab-pro

[kab-pro]

Since now it is clear that this method works quite well and the stock baseline is ready, we can start figuring out what advantages are introduced with every upgrade.

First off, the manually ported cylinder head. You can see that it alone backfires in +11% of power. The highest curve is the condition of the bike as we received it, featuring manually ported head, reground cam and full HGS exhaust system. Backfire screen removed. Result = +15% of power with noteworthy increase in the high end of revs.

[dr.-hasenbein]

Youth Research

No *bling* *bling* Ti-Valve Pics this day ? Ooh... come on !