B-26 3D Model using Martin's original blueprints (1 Viewer)

Ad: This forum contains affiliate links to products on Amazon and eBay. More information in Terms and rules

Maty12

Senior Airman
311
319
Nov 6, 2019
Hi all,

I've already posted about my B-26 research efforts elsewhere in the forum, but I figured it was about time to make a thread for my model itself since it's starting to take shape. For those who haven't seen that thread, I spent the last six months or so scanning microfilm rolls I purchased from the Smithsonian. I have yet to finish categorizing them fully, but have simplified that process for the future (mainly, going through each individual drawing to note all of its details as I was originally doing is not important for modeling purposes, all that is needed is placing the drawings into the folders that correspond to the subassemblies the drawings correspond to).

Types of drawings:
There are more types, but these are the ones relevant to the model in its current phase.
  • Assembly- Drawing of a part with multiple views and dimensions.
  • Installation- Same as Assembly but with less or no dimensions (much less useful). These are instructions on attaching a part to the aircraft and are sometimes the same drawing with only the text modified.
  • Installation & Assembly- Usually a combination of both in the same drawing, sometimes identical to an installation drawing.
  • Casting/Molding- Similar to Assembly in utility, full of dimensions. Meant to be used by factory workers to make an individual part out of metal.
  • Frame/Rib Assembly- Assembly drawing of one specific frame or rib of the aircraft. Full of dimensions, as well as instructions on were bolts and rivets go and how to bend she sheets of metal.
  • Ordinates- Tables of number coordinates with a guide drawing, meant to represent the outline of specific parts or large assemblies. Numbers are rounded to the nearest thousand of an inch where practical (such as the pilot's enclosure), or to the nearest hundredth of an inch where the table is too large (such as for the overall fuselage and engine nacelles). The rows of an ordinates table form the outline of a cross section at a specific distanced measured either from the most forward point of the assembly, from its horizontal centerline, from its vertical centerline or an external reference.
  • Contours- Meant to go with the Ordinates drawings. The contour drawings overlay the several cross sections formed by the ordinates over each other, lining them up.
  • Skin- Diagrams for the skin of the aircraft, common for complex shapes such as the tips of the wings and stabilizers. These are the only drawings that measure the dimensions to the outside of the skin of the aircraft. Dimensions for the internal frames can be obtained by removing the skin thickness, which is always stated in the diagram.
Company-specific Terminology:
Martin had specific company names for parts of the B-26, which sometimes match industry-wide terms. There is no official name for some parts as long as it gets the idea across. A rear spar is sometimes called that and sometimes called an aft spar.
  • Fin- Refers to the vertical stabilizer.
  • Stabilizer- Refers exclusively to the horizontal stabilizer.
  • Empennage- Refers to the combination of the fin and stabilizer, sometimes including the rudder and elevator.
  • Venturi- Refers to the intake for the oil cooler.
  • Venturi Cowl- Refers to the section of the engine nacelle that houses the oil tank, oil cooler and exhaust stacks.
  • Engine Hood- Refers to the engine cowling. It includes the Carburetor Intakes and the Venturi.
  • Navigator's Turret- Refers to the astrodome assembly, which does not have any armament.
  • Turret- Refers to both powered and unpowered gun positions.
The model's current state:
The model is currently a collection of different assemblies combined as best I have been able to based on mounting points and information from the general assembly/arrangement drawings. The individual assemblies' positions may change as the model progresses and more data is found. Because the thrust line and fuselage centerline are at an angle to each other and their precise centerpoint isn't marked on any blueprints, all distances in the model are approximate, specifically to within thousands of an inch. Data for the model comes from:
  • Ordinates tables for the fuselage (Station 33 through 658 1/2, missing tail turret and nosecone), the pilot's enclosure, the nosecone itself, the engine nacelle and cowlings, the propeller spinners, and the fillet around the fin and stabilizer.
  • The skin drawings for the tips of the wing, fin and stabilizer (of which the wing has the least useable information).
  • The theoretical outline of the wings and stabilizers based on general assembly/arrangement drawings.
  • The vertical stabilizer, made from the diagrams for the individual ribs and spar and aligned using the aforementioned assembly drawings.
B-26 22-06-25 Front Quarter.png
B-26 22-06-25 Front.png
B-26 22-06-25 Top.png
B-26 22-06-25 Right.png


The drawings are in varied stages of legibility, and there is nothing either I or the Smithsonian could do about that, it's the state the original rolls were in. Some are blurry, but worse of all some are just very badly faded, to the point where barely any dimensions can be worked out. Fortunately for some of these, there are also contour drawings that can be used in conjunction with the ordinates. If an ordinate isn't on the line, then I have read at least one digit wrong, or (in the fully legible drawings) the draftsperson has made a copying error. These copying errors range from the copier mistaking similar-looking numbers, switching digits around, copying a line twice without noticing, or sometimes just writing down a number whose origin cannot be traced. Because this is a 3D model, ordinates in the wrong place can also be easy to spot even without contour drawings if a single ordinate is very far from any of the ordinates that were next to them on the original table.

The fin:
The fin is the most complete assembly so far, and the only one where I've modeled the ribs and the spar. I had a brief scare at first because my spar was not lining up with any of the ribs... only to then read the drawing's description again and realize what I had actually built was the Stabilizer's fwd spar. Even with that fixed though, the fin has caused me plenty of headaches. The fin's theoretical stations are parallel to the fuselage centerline, and are measured from 0 to 145.5. The fin's theoretical root is at station 34. That's where the straightforwardness ends. The angle for the fin's spar is not mentioned anywhere in the diagrams and it is not fully vertical. Many of the ribs do not mark the distance where they attach to the spar in their diagrams either, and because of all the angles involved, any values mentioned are rounded. The lowest rib in the fin is the only one that gives attachment information and an angle. It is angled 1d 30' 33" from horizontal, because the top of the fuselage is angled upwards towards the rear in order to accommodate the tail gunner. The other ribs provide no information.

B-26 22-06-25 Fin.png


The assembly drawing for the entire fin and the frame drawing for the spar give other dimensions that lead to the calculation that the spar is angled 1d 21' 58" from vertical. However, the spar's drawing also states that most of the ribs meet the spar at an angle of 88d 30', the complement of which is 1d 30'. I don't really know what to do with that information. Angling the spar 1d 30' from vertical results discrepancies in the magnitude of quarter inches in parts where the distances are provided to the thousand of an inch, and I don't think anyone would ever approximate 1d 21" 58" as 1d 30", when the logical approximation is 1d 22". 1d 30' is also not an approximation of the 1d 30' 33" measurement from the bottom rib, as this again results in discrepancies that are too large given the precision of other measurements, and would also be the incorrect way to approximate. My current hypothesis is that for whatever reason Martin decided to angle most of the ribs 88d 30' from the spar, which itself is 1d 21' 58" from vertical, meaning the ribs are almost parallel to the fuselage centerline & stations, but not quite, and almost perpendicular to the spar but not quite. I don't know why they would ever do this and have yet to finish testing the results of this hypothesis. In the current model they are parallel to the fuselage centerline and stations.

Extra tidbits about the fin:
  • The fin only has one spar. I find this very unusual, especially given that the stabilizer has two, but there genuinely is no forward spar near the leading edge. The leading edge attaches directly to the skin and ribs (which have angled sections that I have yet to model.
  • The bottom rib of the fin is the only one that includes the skin thickness in its measurements, as it is exactly .032 of an inch longer and wider than the theoretical fin in a section where the skin is .032 inches thick.
  • Most drawings of the fin in my possession list stations both for the short tail and long tail models of the aircraft. Station 145.5 on the B-26 is Station 131.5 on the B-26B1, and so on. The difference is always exactly 14 inches, which is odd given that the B-26B1's tail was 20 inches taller. What gives? Well, when the fin was lengthened Martin decided to change the reference datum from the fuselage centerline to the theoretical root of the fin, Station 34. 145.5 +20 -34 =131.5.
  • The fin is one of the only parts of the aircraft to not have round-ish numbers for its stations, the fuselage, wings and stabilizer all use whole numbers, or down to 1/64 of an inch as needed, while some of the fin's stations are 75.854 or 102.927. Why is that? Because of the rudder's hinges, which are exactly 15" and 69" from the top of the theoretical hinge line, respectively. The hingeline which is angled 3.08d degrees from vertical, resulting in hinges at station 130.0035... and 75.8559... . The remaining stations are merely the result of dividing the interval between 34 and 75.854 and the interval from 75.854 to 130.000 evenly into increments of 7.95868..." and 9.024602..." respectively. I do not know why 75.8559 was rounded to 75.854 instead of 75.856.

What's next:
I plan on modeling the stabilizer differently, because it occurs to me that using measurements that are rounded to the thousand of an inch that then almost match the theoretical airfoil makes less sense than modelling the airfoil and then cutting slices of it and doublechecking the numbers. The wing and all control surfaces will also follow a similar approach. The landing gear has individual drawings with basically all the info necessary to model them, but the main gear's drawing for some reason had only one mostly illegible copy within the 5,513 drawings that I scanned. The nose gear's drawing is more legible, and it has at least two other duplicates, one of which is fully legible, so it can be modeled with ease.

I will detail my process of modeling the pilot's enclosure in a future reply and will share more progress as it happens.
Best,
-Matt
 
Nice work, but drawing in Blender, O man, how's that possible. I try it several times to work with Blender , but after opening of the main screen, I really didn't know were to start. It's not really an intituative(??) program, but it looks really very nice. Are you plant to use the drawings for a printer or ...
 
Nice work, but drawing in Blender, O man, how's that possible. I try it several times to work with Blender , but after opening of the main screen, I really didn't know were to start. It's not really an intituative(??) program, but it looks really very nice. Are you plant to use the drawings for a printer or ...
Blender can be a bit tricky to pick up. My approach is to just add a plane, merge all the vertices in the center and then extrude from there to get the shape I want usually. This project is not for printing (it has no curves, it's all polygons), I plan on trying to get it into MSFS at some point.
 
Is this project capable of recreating a clear new set of blueprints to replace the faded originals? And is this data something that could be used by a model maker? Hint, hint, hint.
 
Is this project capable of recreating a clear new set of blueprints to replace the faded originals? And is this data something that could be used by a model maker? Hint, hint, hint.
I sure could use a squadron of 3d printed B-26/B-26A at 1:300 or 1:285
Lea and Rabaul need some bombing.
I have considered 3D printing, though I'm modeling with polygons instead of CAD so I don't know how well that works out. With regards to recreating the blueprints, I would have to double check the terms of purchase for the microfilm. I think I'm allowed to do whatever as long as I don't sell them, and that they're not liable if someone decides to build something using said blueprints and crashes. I'll look into the matter once I'm back from Europe. A difficult feature of using these blueprints is that they are all incomplete, you need data from 5 or 10 different diagrams to model a single part definitively. This might become easier once I'm able to finish sorting the drawings into the proper folders. I have so far moved some components from one of the rolls, but this will take weeks.

Greg, the B-26/B-26A tailgun has unfortunately proved one of the parts with the least data so far. I am committed to getting that area as accurate as I can though. There is a lot of data on the wing flaps and engine nacelles/cowling though, which I consider some of the other most iconic parts of the early Marauders.
 
Yep, tailside is only important at a woman :)

I have the same issue with tail details of a DH Mosquito. Just send an email to NewZealand, hope get some answers
Why I'm afraid you're mistaken, Bert, for you're missing half of the buffet on that front! :p That being said, I tend to focus my attention on one particular lovely chap's tailside, as he has been indispensable on the mathematical side of things on this project since I don't have the brains for it myself. Just don't take him to a book store, for the same reason he shouldn't take me to a model store. There'd be no inventory left on the shelves and no money in our wallets!

Best of luck on the Mosquito!
 
Great!
If you get stuck in Blender - feel free to ask me, I will help.

I have considered 3D printing, though I'm modeling with polygons instead of CAD so I don't know how well that works out (...)
You do not need a CAD NURBs model to create a 3D print. The basic input format for these devices (STL) describes everything as solids, which boundaries are made from thousands of small triangular faces. The key idea for 3D printing is that these solids have to be "watertight" i.e, all edges of their faces have to be connected to adjacent faces. The mesh models made in Blender, 3D Max and similar programs are OK for 3D printing, as long as you create each printed element as "watertight" solid. [Exception: analyzing a RC P-47 3D set (a data file), sold by a company from Czechia for self-printing, I found that they modeled the thinnest elements, like the wing skin, as simple one-sided surface built from faces. In the model their thickness was zero, but it seems that the 3D printer recreated them as the thinnest possible surface. I cannot explain such an approach, but it seems that it was made by someone well familiarized with the practical nuances of this process]. As in every physical process, you also have to take care for additional supporting structures which you remove from the final model. There are also other details - for example, the head movement across printed layer should not concentrate around a small area, because this can re-melt the cooling plastic filament. Modellers often split their 3D-printed model into parts, which they manually glue (this allows to build larger models than the 3D printer workspace). In practice, you always have to adapt a general 3D model for the printing, and it can require significant amount of work.

(...) With regards to recreating the blueprints, I would have to double check the terms of purchase for the microfilm. I think I'm allowed to do whatever as long as I don't sell them, and that they're not liable if someone decides to build something using said blueprints and crashes. (...)
A few years ago I bought from Smithsonian microfilms fo the SBD/A-24, for the same purposes. Assuming that the text of their standard agreement I had to enclose to this order is the same, it allows the buyer to use these microfilms just to her/his personal purposes. It seems that Smithsonian extends these restrictions to the scans of these microfilms. (For example, when I wanted to donate my SBD scans to AirCorps Library, they explained me that it would violate the original Smithsonian agreement). Of course, Smithsonian cannot claim any right to an ordinates table rewritten (and fixed - the added value!) in Excel, or to a 3D model based on these blueprints, because these are separate works whch used these blueprints as the input data.
 
Last edited:
Great!
If you get stuck in Blender - feel free to ask me, I will help.


You do not need a CAD NURBs model to create a 3D print. The basic input format for these devices (STL) describes everything as solids, which boundaries are made from thousands of small triangular faces. The key idea for 3D printing is that these solids have to be "watertight" i.e, all edges of their faces have to be connected to adjacent faces. The mesh models made in Blender, 3D Max and similar programs are OK for 3D printing, as long as you create each printed element as "watertight" solid. [Exception: analyzing a RC P-47 3D set (a data file), sold by a company from Czechia for self-printing, I found that they modeled the thinnest elements, like the wing skin, as simple one-sided surface built from faces. In the model their thickness was zero, but it seems that the 3D printer recreated them as the thinnest possible surface. I cannot explain such an approach, but it seems that it was made by someone well familiarized with the practical nuances of this process]. As in every physical process, you also have to take care for additional supporting structures which you remove from the final model. There are also other details - for example, the head movement across printed layer should not concentrate around a small area, because this can re-melt the cooling plastic filament. Modellers often split their 3D-printed model into parts, which they manually glue (this allows to build larger models than the 3D printer workspace). In practice, you always have to adapt a general 3D model for the printing, and it can require significant amount of work.


A few years ago I bought from Smithsonian microfilms fo the SBD/A-24, for the same purposes. Assuming that the text of their standard agreement I had to enclose to this order is the same, it allows the buyer to use these microfilms just to her/his personal purposes. It seems that Smithsonian extends these restrictions to the scans of these microfilms. (For example, when I wanted to donate my SBD scans to AirCorps Library, they explained me that it would violate the original Smithsonian agreement). Of course, Smithsonian cannot claim any right to an ordinates table rewritten (and fixed - the added value!) in Excel, or to a 3D model based on these blueprints, because these are separate works whch used these blueprints as the input data.
Thank you, Witold, that makes sense. I don't have 3D printing plans anytime soon (nor do I own a printer), but it's good to know in case I ever get one. I do also have some friends who own some, they just happen to be a few thousand miles away.

It is indeed the same standard Smithsonian agreement. I was originally sending the scans back to them, but stopped receiving responses. I was however told that these might help with the restoration of Flak Bait, so that's neat. It's good to know they cannot claim rights to my blueprints or my model.

I will post more updates once I return from Europe,
-Matt
 

Users who are viewing this thread

Back