You guys may need to use a sweep with multiple guide curves to define the shape and trajectory. There are a couple of different ways to skin this one.
My shop primarily deals in the design of molds for the optics industry as well as anything that could possibly be created in a plastic injection mold. I've modeled some very complex forms over the years but never a cylinder head port. Would need accurate port CAD data to actually construct one in the first place. That and the time, which I don't have.
I'm still suggesting digitizing, then work from that. Don't try to construct a -free hand- port model for machining. We don't know where the water pasages are.
As a rule of thumb, anytime a loft fails to produce a complex shape, it's time to go to sweeps with guide curves. That and some surface integration should get the desired end product.
If anyone wants to send me CAD data of a port, I will look it over and try to help a little better.
A sweep along the flow axis of a port is rarley useful in modeling ports. The reason is that the definition of a swept surface can only include 3 guide strings maximum. 3 guide strings are only enough control to define one corner radius for only part of the length of the port. The second problem with swept surfaces is that they do not have continuity control to adjacent surfaces.
A parametric model for a SBC type port will have more than 100 features and more than 200 parameters that adjust port shape.
We are talking about very different levels of model complexity here.
I suggest you try to model something that even looks like a real port before giving more advice on something you know nothing about.
I'm still suggesting digitizing, then work from that. Don't try to construct a -free hand- port model for machining. We don't know where the water pasages are.
What advantage does not developing in CAD have in this regard?
Any chance you could do a basic "walk through" of doing a port like
this in Surfcam since several of us here are using Surfcam?
I think a friend of mine has SurfCAM but it might be a few releases old. Let me see what he has and what modeling tools it has. What versions are those of you with interest using, (just to be sure we can share files)?
Here's a suggestion, model a real port succesfully before misguiding others.
Schmidt,
I fail to see where trying to HELP people can be equated with "misguiding". Could you please give specific examples of where I tried to do this?
First, I stated very clearly that I had never modeled an actual intake port. I suggested a couple of different approaches I thought might work. If they didn't, then how much was actually lost in the process? I have learned a number of different CAD/CAM modeling packages over the years and most of that learning has come via TRYING to do something. If I had an actual port mold or something concrete to take dimensions from I'd do what I could to help them along further. Regardless of your opinion, we all have one, I'm confident I could model a port given the time and resources. I could do it in Solidworks with some trouble and I could definitely do it in ProE WildFire with parametrics. I don't get paid to sit around and model theoretical engine components all day long but I envy those who do.
Glad you finally decided to add something constructive to the discussion.
I fail to see where trying to HELP people can be equated with "misguiding". Could you please give specific examples of where I tried to do this?
#2
I'm confident I could model a port given the time and resources. I could do it in Solidworks with some trouble and I could definitely do it in ProE WildFire with parametrics.
Do #2 and you will learn the answer to #1, you are so off base on nearly everything you post that I don't believe I could educate you with arguments. You really have to learn this lesson by doing.
Any chance you could do a basic "walk through" of doing a port like this in Surfcam since several of us here are using Surfcam?
I spent the evening with SurfCAM 2001, sorry but there is no practical way to model ports with this software. I can see why people would like it for work where there wasn't much need for experimentation but it is just a mismatch for modeling ports.
Maybe there is a College or University near you that teaches a high-end CAD system that allows lab time to work on your own stuff? Don't waste your time with any of the mid-range stuff, they just aren't there yet.
Some people have done some impressive work with Rhino3D.
I think it only costs around 1K but I don't know for sure. I think you can also get a demo download.
I wouldn't have the patience or time for this but you could probably import some points, model a tube shape and drag the surface poles around until you got the shape you wanted. This might be the best way to figure out how you want to do this.
I still think the most enlightening thing for all would be to get a scan of the ports and chamber of a head most all of us know like LS1 head and I will show the steps in UGNX, you will probably see an analog to the software you use or see a reason why you need a functionality that you don't have.
First off, .. guys, please let's keep this as a friendly debate, . . .
while you guys don't agree, you both have posted very valuable information,
and the "back & forth" discussion between you has lead to the bulk of
information posted in the thread.
So, .. please don't turn this sour, .. you're "disagreement" is the key
to this thread, .. so keep it friendly, .. we are all professionals here, let's
continue to act like it.
OK, .. so back to the currently scheduled program, ..
Jon, .. yes I've had trouble with Surfcam, .. and everyone I know
using it for the most part is just reverse engineering. Reverse engineering
is my primary goal right now, .. but to be able to manipulate the scanned port
shapes is really important.
I have the demo copy of Rhino, .. and have found it fairly easy to learn
and get around in, .. I have not however worked out how to model a
port properly. I've got to get a port scan of something I'm willing to post
here, .. but am just too busy for the next few weeks.
Does anyone have a basic port scan, .. small block chebby or something?
One thing I can see happening, .. say I have a port I really like, .. but
it's for a 400" engine and a 2.200" valve, .. I need a port for a 350" engine
with a 2.125" valve, .. I'd want to take the exact same port, .. and "shrink"
it down for the smaller valve & engine combo while keeping the same basic shape.
ok, .. I'm off to the races today, .. don't know if I'll have net at the hotel, ..
I road race a Formula car if you're interested, ..
Cheers,
Curtis
Race Flow Development
Simultaneous 5-axis CNC Porting
http://www.raceflowdevelopment.com
I took CAD/3D and Advanced ProE after completing my degree Schmidt. I appreciate your concern for my education though.
Curtis,
I am all for civil debate but I refuse to deal with people who are disrespectful of other opinions so please excuse my demeanor. I think egos get in the way of civil debates far too often on internet message boards. And it's always easy spotting the guy who has the "biggest".
Before I go any further I should make a correction to an earlier post referring to 3 as a maximum number of guide curves. Simply not true but I think we're well past that now.
Rhino 3D is a good surfacer. If you can get your hands on a demo its well worth the time. I mentioned Geometry Works 3D in an earlier post. For those of you who like the intuitive nature of Solidworks, you'll also like GW3D's integration. ShapeWorks and SurfaceWorks are also fully associative surfacing packages with Solidworks interoperability. I haven't used ShapeWorks, but SurfaceWorks is a nice package. However, my preferance still leans towards GW3D.
Interestingly enough, I did find a case study on GeoMagic's website that might interest some of you. Seems that GeoMag is used by Richard Childress Racing in their cylinder head development program. The steps outlined in the article are basically the same as the ones I detailed earlier. I have to wonder how much time RCR has devoted to other means of creating ports and why they are still doing it this way? With virtually unlimited resources at their disposal, this is a serious question.
You will also see that RCR is using GeoMagic as a means of analyzing port scans to improve them. The polygonal ("quilting" mentioned earlier) to NURBS surface creation. As mentioned, the surfaces are hardly set in stone.
Schmidt,
I want to know if you build your port in your software will it make all the G-codes so it can be machined? If you are doing it on a 5 axis machining center do you have to draw the cutting tool you are using so that the software know where its limits are? How does it know to rotate the head so that the side of the tool (cutter) does not crash into the side of the port. I will have more question letter I am sure.
Schmidt,
I want to know if you build your port in your software will it make all the G-codes so it can be machined?
It isn't as easy as "it make" but you have touched on an important advantage of using an integrated software. By this I mean that the same model created by work in the Design part of the application can be cut in the machining part of the application without any export translation/imprt. The key advantage of this is that when changes are made to the model the CAM program can be updated to cut the new design with minimal effort, sometimes a few clicks but on 4 and five axis work if the changes are big you may want to make changes about the geometry that drives how the cutter follows the port.
Five axis machining is much too big of a subject to explain here but it isn't something that is done automatically (at least no very well). There are softwares out there that that try to automate both CAd and CAM tasks, in simple cases they can do very well sometimes. It would be a mistake to extrapolate that a software that can do simple work automatically will be better than another software that takes a more detailed interactive approach. A lot of people make this mistake when evaluating CAD CAM for difficult projects, they see the mid-range software and how simple it seems and think it means more than it does, that ussually lasts for about a week or two when they reach a capability limit and find they have wasted time and money on a dead-end effort.
I have to wonder how much time RCR has devoted to other means of creating ports and why they are still doing it this way? With virtually unlimited resources at their disposal, this is a serious question.
First, before you believe what you read in a press releases take a close look at the port on the web site, copy it and blow it up. Is that a real port? Does something look funny around the guide boss? "0.0001 accuracy" would you like to buy a bridge?
The answer to your question is easy, US motorsports industry trails aerospace manufacturing technology by about 10~15 years, the talent/experience pool in the US motorsports industry hasn't been developed yet. Also the motorsports industry doesn't pay well enough to attract people from aerospace. This is why I call it an opportunity for those willing to put out the effort and produce something of unique value.
The question you should be asking is "why Toyota, Honda, Ilmor, Cosworth etc. don't use RE to develop ports"
Another thing I think is being over looked here, even if you do intend to use a reverse engineering software you will want to make so many refinements to the ports as scanned that I think it would be a short time before you realized you were in fact developing ports in CAD with tools that weren't meant for design work.
SchmidtMotorWorks wrote:First, before you believe what you read in a press releases take a close look at the port on the web site, copy it and blow it up. Is that a real port? Does something look funny around the guide boss? "0.0001 accuracy" would you like to buy a bridge?
I don't believe the port in the pictures is really the issue here is it?
We don't know all the details behind the screen captures. Was Raindrop allowed to take actual port screenshots from RCR or did they try to recreate something for the press release? Who's to know for sure.
The point is irrelevent. Raindrop could be sued for false advertisement if they were intentionally misrepresenting their involvement with RCR.
The answer to your question is easy, US motorsports industry trails aerospace manufacturing technology by about 10~15 years, the talent/experience pool in the US motorsports industry hasn't been developed yet. Also the motorsports industry doesn't pay well enough to attract people from aerospace. This is why I call it an opportunity for those willing to put out the effort and produce something of unique value.
Hmm, I don't totally agree. Many of the motorsports divisions have picked up engineers from aerospace and other high-tech industry.
If we're going to draw comparisons with Indy/F1/etc vs a street/race head for the masses. Well I just don't see that. An Ilmor cylinder port is designed with no compromises. The entire cylinder head is designed for a specific purpose. If this analogy was a good one then we might as well design a model of a port, chamber, coolant passage, ENTIRE cylinder head and cast our own.
I still say that whether or not the port was entirely CAD developed is a tertiary importance. Many multi-billion dollar aircraft have been designed with CAD and some of them weren't any good. Who's to say that a port will be?
I won't be the first to fly a new jet on its maiden voyage and I wouldn't run out to buy a cylinder head that boasts 100% CAD creation either. If the results are there, then maybe. Until then, you have a very expensive item targeted at a market that is driven by results. I see a big chance of failure there.
