rmeyers
Well-Known Member
Not wanting to hijack the original thread I thought that I'd start a new one for discussing the relative advantages and disadvantages of billet construction of VW aircraft heads.
This discussion pertains only to heads for a single cylinder. A single cylinder head will be far easier to CNC manufacture and should have a greater market.
This thread was started with all due respect for Hot Wings, someone who has given a great deal of time, thought and energy to this project. Please understand that if I do not agree with some of the positions stated in the original thread it is with that respect that I state my opinion.
That said, to what I perceive as advantages to billet construction. First the list of advantages and then my reasoning below the list.
1. Arbitrary fin depth, thickness and spacing.
2. Better alloys to chose from.
3. Raw material cost should be about equal.
4. Speed of production. For the numbers spoken of in the original thread, 100's per year, CNC manufacturing will be much faster.
5. Cost. Billet production will cost less in the given numbers.
Reasoning:
1. In post #15 in the original thread it was stated "There is really no practical way to get the kind of fin spacing needed with CNC.*", the asterisk said "*The Jabiru is a much simpler head.". I didn't understand why the author felt that way and then it occurred to me that maybe the author was thinking that fins needed to be milled. That would be correct, there is no practical way to mill the fins at the needed pitch. What I would do, and have a fair amount of experience in, is saw the fins in (on the mill) with a slitting saw. Slitting saws are available in almost any diameter and thickness you can imagine. A 4.0 inch DIA saw on a standard arbor can cut a slot approx 1.25 inches deep. A 5.0 in saw can cut approx 1.75 in deep. Assuming that the design is created with a fairly even fin depth over a linear length (up and down the cylinder head) the saws can be ganged and 4 or 5 saws put on one arbor with maybe another arbor with 2 or 3 saws and possibly a third with 1 blade. The only downside is that the saws can only cut an inside curve greater than the radius of the blade. Of course there is no limit on how sharp an outside curve can be cut. ---> confession here, I have actually used Harbor Freight saw blades before for quick and dirty AL cutting, my shame. As most airplane guys know, several of the big radial engines used fins that were gang sawn with specially made band saws, this is just the modern CNC version.
A .125 thick blade and a .080 thick fin will give you 5 fins in 1.025 in linear height.
Bottom line, you can get a lot of fins in a little space. Far more than can possibly be cast.
2. Self explanatory. Nearly any of the wrought alloys are stronger and have better heat resistance than any of the common casting alloys. I would choose 2024 or 7075. Not as much difference in the cost as opposed to 6061 as you might imagine.
3. Not knowing the exact dimensions of Hot Wings' head design I made a couple of guesses. A piece of billet will probably run $75 to $110. (although with the current state of supply chain affairs who knows?)
Casting material costs will vary widely. If lost wax or lost foam your yield will probably be in the 50% to 60% range, if dry sand your yield will be on the order of 25%. A thing to consider is that the gates, runners and risers cannot be repeatably thrown back into the foundry. With every cast there are metallurgical changes as alloying elements come out of solution. After being recycled once or twice you will not have the alloy that you think you have. Large foundries test and adjust constantly to ensure the elemental analysis. You will be forced to use more AL than it might appear at first glance. I think that it will come out about the same. I could be wrong and casting may have noticeably lower material cost. Requires more detailed analysis.
4.& 5. Speed of production and cost are tied together so I will address them as one. Time is money and money is time even for a semi-hobby.
Post Work;
Simply put, casting has a great deal of post work to make a completed part that a CNC'd part does not;
a. Heat treating; the CNC part does not require heat treating, the cast part probably will.
b. Post production machining operations; The CNC part should need none, the cast part will need many. The cast part will need fettling and probably a fair amount of hand cleanup. The cast part will need automotive machining for the valve seat pocket and the valve guide bore and possibly the rocker shaft bore. Without special tooling these are not home shop operations. The cast part will need to have the deck surface machined to a fairly fine finish. Spark plug thread inserts will need to be threaded. This can be done by hand but is slow.
Tooling; this is potentially the big one. On a simple 4 axis VMC there will be 2 to 4 fixtures needed, depending on the port placement in the head. These will be relatively simple indexing fixtures so the raw material can be indexed in a different orientation for further machining. Surprisingly enough, there really isn't all that much high precision machine work on a head. Mostly the stuff that I described in part 'b' above.
Tooling for the cast part includes not only the machining tooling but the casting tooling. There's a bunch of that. Copes, drags, core boxes, runners, gates, risers, etc. And don't forget the patterns, whether disposable or hard patterns.
Left and right hand heads; For a cast part, basically double all of the things above. For the CNC part a mirrored model and modified CAM program, with the caveat that I don't know if the head bolt pattern and valve rocker geometry is mirrorable. Still, modifying the model is a lot easier than making all new patterns.
Running cost; For the CNC part the recurring costs are for the material and the CNC time itself. How much machining time? Good question and important. Several variables will affect the final cost. If the head is designed with CNC machining in mind cost will be much less. Can't go into all that entails right here, this post is long enough, but for example, if the head is designed so that all of the machining fixtures can be placed in the VMC before the start of the run, all the operator has to do when one machining sequence is finished, is to open the door, unclamp the part, move it to the next fixture position, clamp it, close the door and hit run. This greatly reduces the cost. Spindle time usually runs (at least before covid) $75 - $110/hour. My best guess, and it is only a guess at this point, is 1 to 2 hours. Port design, shape and length will be the slowest part, you can't push those spindly little spherical cutters too hard.
For the cast part recurring costs are for the materials, the energy costs of the foundry, and the automotive machining. I would be surprised if the automotive machining was less than $200.
Intangibles - Annoyance factor; For the CNC part you call the metal supply house and order the material, have it shipped to the machine shop and finally pick up the finished pieces. Maybe take the material from the supplier to the machine shop yourself to save some.
For the cast part there are a bunch of things to do, make cores, make molds, cast, all of the hand work mentioned above, transporting the semi-finished parts to the automotive machine shop.
I've stopped at the point where both kinds of head are in a comparable state. There is still the installation of the valve seats and valve guides, the valve job itself and installation of the valvetrain components.
Bad parts; Once the initial bugs are worked out, and there will be bugs, the reject rate for CNC parts is effectively zero. Not so for the cast parts. Casting the fins in the density that is needed is not for the faint of heart. There will be failed casting, some unknown percentage, constantly.
That's all that I can think of off the top of my head. I'm sure the knowledgeable people here will come up with all sorts of things that I have not listed. I look forward to the discussion.
This discussion pertains only to heads for a single cylinder. A single cylinder head will be far easier to CNC manufacture and should have a greater market.
This thread was started with all due respect for Hot Wings, someone who has given a great deal of time, thought and energy to this project. Please understand that if I do not agree with some of the positions stated in the original thread it is with that respect that I state my opinion.
That said, to what I perceive as advantages to billet construction. First the list of advantages and then my reasoning below the list.
1. Arbitrary fin depth, thickness and spacing.
2. Better alloys to chose from.
3. Raw material cost should be about equal.
4. Speed of production. For the numbers spoken of in the original thread, 100's per year, CNC manufacturing will be much faster.
5. Cost. Billet production will cost less in the given numbers.
Reasoning:
1. In post #15 in the original thread it was stated "There is really no practical way to get the kind of fin spacing needed with CNC.*", the asterisk said "*The Jabiru is a much simpler head.". I didn't understand why the author felt that way and then it occurred to me that maybe the author was thinking that fins needed to be milled. That would be correct, there is no practical way to mill the fins at the needed pitch. What I would do, and have a fair amount of experience in, is saw the fins in (on the mill) with a slitting saw. Slitting saws are available in almost any diameter and thickness you can imagine. A 4.0 inch DIA saw on a standard arbor can cut a slot approx 1.25 inches deep. A 5.0 in saw can cut approx 1.75 in deep. Assuming that the design is created with a fairly even fin depth over a linear length (up and down the cylinder head) the saws can be ganged and 4 or 5 saws put on one arbor with maybe another arbor with 2 or 3 saws and possibly a third with 1 blade. The only downside is that the saws can only cut an inside curve greater than the radius of the blade. Of course there is no limit on how sharp an outside curve can be cut. ---> confession here, I have actually used Harbor Freight saw blades before for quick and dirty AL cutting, my shame. As most airplane guys know, several of the big radial engines used fins that were gang sawn with specially made band saws, this is just the modern CNC version.
A .125 thick blade and a .080 thick fin will give you 5 fins in 1.025 in linear height.
Bottom line, you can get a lot of fins in a little space. Far more than can possibly be cast.
2. Self explanatory. Nearly any of the wrought alloys are stronger and have better heat resistance than any of the common casting alloys. I would choose 2024 or 7075. Not as much difference in the cost as opposed to 6061 as you might imagine.
3. Not knowing the exact dimensions of Hot Wings' head design I made a couple of guesses. A piece of billet will probably run $75 to $110. (although with the current state of supply chain affairs who knows?)
Casting material costs will vary widely. If lost wax or lost foam your yield will probably be in the 50% to 60% range, if dry sand your yield will be on the order of 25%. A thing to consider is that the gates, runners and risers cannot be repeatably thrown back into the foundry. With every cast there are metallurgical changes as alloying elements come out of solution. After being recycled once or twice you will not have the alloy that you think you have. Large foundries test and adjust constantly to ensure the elemental analysis. You will be forced to use more AL than it might appear at first glance. I think that it will come out about the same. I could be wrong and casting may have noticeably lower material cost. Requires more detailed analysis.
4.& 5. Speed of production and cost are tied together so I will address them as one. Time is money and money is time even for a semi-hobby.
Post Work;
Simply put, casting has a great deal of post work to make a completed part that a CNC'd part does not;
a. Heat treating; the CNC part does not require heat treating, the cast part probably will.
b. Post production machining operations; The CNC part should need none, the cast part will need many. The cast part will need fettling and probably a fair amount of hand cleanup. The cast part will need automotive machining for the valve seat pocket and the valve guide bore and possibly the rocker shaft bore. Without special tooling these are not home shop operations. The cast part will need to have the deck surface machined to a fairly fine finish. Spark plug thread inserts will need to be threaded. This can be done by hand but is slow.
Tooling; this is potentially the big one. On a simple 4 axis VMC there will be 2 to 4 fixtures needed, depending on the port placement in the head. These will be relatively simple indexing fixtures so the raw material can be indexed in a different orientation for further machining. Surprisingly enough, there really isn't all that much high precision machine work on a head. Mostly the stuff that I described in part 'b' above.
Tooling for the cast part includes not only the machining tooling but the casting tooling. There's a bunch of that. Copes, drags, core boxes, runners, gates, risers, etc. And don't forget the patterns, whether disposable or hard patterns.
Left and right hand heads; For a cast part, basically double all of the things above. For the CNC part a mirrored model and modified CAM program, with the caveat that I don't know if the head bolt pattern and valve rocker geometry is mirrorable. Still, modifying the model is a lot easier than making all new patterns.
Running cost; For the CNC part the recurring costs are for the material and the CNC time itself. How much machining time? Good question and important. Several variables will affect the final cost. If the head is designed with CNC machining in mind cost will be much less. Can't go into all that entails right here, this post is long enough, but for example, if the head is designed so that all of the machining fixtures can be placed in the VMC before the start of the run, all the operator has to do when one machining sequence is finished, is to open the door, unclamp the part, move it to the next fixture position, clamp it, close the door and hit run. This greatly reduces the cost. Spindle time usually runs (at least before covid) $75 - $110/hour. My best guess, and it is only a guess at this point, is 1 to 2 hours. Port design, shape and length will be the slowest part, you can't push those spindly little spherical cutters too hard.
For the cast part recurring costs are for the materials, the energy costs of the foundry, and the automotive machining. I would be surprised if the automotive machining was less than $200.
Intangibles - Annoyance factor; For the CNC part you call the metal supply house and order the material, have it shipped to the machine shop and finally pick up the finished pieces. Maybe take the material from the supplier to the machine shop yourself to save some.
For the cast part there are a bunch of things to do, make cores, make molds, cast, all of the hand work mentioned above, transporting the semi-finished parts to the automotive machine shop.
I've stopped at the point where both kinds of head are in a comparable state. There is still the installation of the valve seats and valve guides, the valve job itself and installation of the valvetrain components.
Bad parts; Once the initial bugs are worked out, and there will be bugs, the reject rate for CNC parts is effectively zero. Not so for the cast parts. Casting the fins in the density that is needed is not for the faint of heart. There will be failed casting, some unknown percentage, constantly.
That's all that I can think of off the top of my head. I'm sure the knowledgeable people here will come up with all sorts of things that I have not listed. I look forward to the discussion.
