SBD Dauntless, from scratch

[Witold Jaworski]

To build the model from scratch you need a good reference. Initially I decided to use for this purpose detailed scale plans from the monograph published by KAGERO in 2007 (Authors: Krzysztof Janowicz, Andre Zbiegniewski, ISBN: 978-83-60445-25-9). It contains SBD Dauntless plans in scale 1:48, traced by Krzysztof Lukasik. They are quite detailed (up to the rivets on the aircraft skin). Apparently these drawings were made using Corel Draw or similar software. I scanned these drawings to do the basic verification. During this phase I did not find any flaws:

All the key locations of the fuselage are in the same place in the side view and the top view. The proportions of the length and the span of the top view is correct: 0.787. (This ratio comes from 996/1266. According the dimensions specified in this monograph, the length of the SBD-3 fuselage was 996 cm [32' 8"], while the wing span of all the Dauntless versions was 1266 cm [41' 6"]). This good impression disappeared, when I compared side views of two different Dauntless versions: SBD-3 and SBD-5:

According the monograph data, SBD-5 fuselage was 4 inches (about 10.1 cm) longer than SBD-3. (The SBD-5 and SBD-6 fuselage was 33' long. Most probably it has slightly different engine cowling and the propeller. The airframe after the firewall was the same in all Dauntless versions). However, in this monograph they have the same length!

Maybe the textual data contains an error? In such a situation I try to find an "official", archival drawing of the aircraft. They do not show many details, but contain the key dimensions. I have found on the Internet a BuAer Navy drawing of the SBD-5, from 1944. From the front view you can read the precise wing span: 41' and 6 5/16". From the side view you can read the exact length: 33' and 1/8".

This BuAer drawing isn't an ideal source: it does not contain such details as panel seams. You can also find here some manual errors, made by its draftsman. While the aspect ratio of the top view matches the span and length specified in the dimensions, the actual fuselage length on the side view is somewhat shorter. (The positions of the wing and horizontal tailplane match in the side and top view match each other. It seems that the part of the vertical tail contour was shifted). On the other hand, the BuAer top view is a little bit asymmetric, and the firewall line is moved forward a little.

The good news is that the wing and the tailplane arrangement on the KAGERO plans and the BuAer drawing match each other:

Then I compared the side views of these drawings (I marked the correct fuselage length measured on the BuAer top view in red):

The differences of the side views are overwhelming: this is not only the engine cowling but also the cockpit canopy, the fin, and the tailwheel. (In general: none of these drawings shows the correct tail).

Thus, I can conclude: never trust the scale plans! I need a better reference, to fix these drawings.

In the next week I will show how I verify these drawings using photos.

Note: for editing these images I use two free, Open Source programs: GIMP (it is similar to Adobe Photoshop) and Inkscape (it is similar to Corel Draw or Adobe Illustrator). You can find more about them in this e-book.

 

[vikingBerserker]

Very nice!

 

[Wurger]

 

[Lucky13]

Very nice indeed!

 

[chris brown]

Very cool! Should be a fun build to watch.

 

[Wayne Little]

Interesting..!

 

[Witold Jaworski]

Before I start a new model, I collect its photos — as many as I can find, everywhere: in the books, magazines, on the Internet. Some of these photos are high-quality, detailed photos of restored airplanes. One of them is this a high-resolution photo from the web page of Chino Planes of Fame Air Museum:

This is a special photo: it was made from a long distance using a "telescope" lens, which minimized the perspective barrel distortion. The airplane on this picture lowered its right wing, so its bottom parts are slightly shifted downward, but except this area it is a perfect reference!

I placed this photo in Inkscape, and set it horizontally (along the canopy frames). Then I mirrored it, for the comparison with the left side view:

In Dauntless there are two long parallel lines that were perpendicular to the fuselage centerline: the trailing edge of the wing center section, and elevator leading edge. On this photo they are also parallel (more or less). This is the proof that we can neglect the perspective (barrel) distortion.
Now let's compare the BuAer drawing (see the previous post) with the reference photo:

In the first post I mentioned that this BuAer side view is too short. To make a fair comparison, I marked on this drawing the proper fuselage length (as on the BuAer top view). As you can see, the drawing matches the reference photo quite well!

It fits the photo even better when you correct the tail contour (so it matches the fuselage length in the BuAer top view):

It seems that the BuAer drawing from 1944 matches the contour of the real aircraft quite well. In fact, it is much better than the contours of the detailed KAGERO drawings from 2007 (see previous post), which most probably are based on the drawings made by previous authors:

I think that these KAGERO plans "accumulated" many decades of various errors. Do not be surprised: before the 1990 it was practically impossible to make such a "photo verification" like this one. Even today authors are used to redrawing earlier plans. They seldom compare their work with the real photos in the manner shown above.

Concluding: there is no good reference among the existing Dauntless drawings: the BuAer lacks details, while the KAGERO plans contain too many deviations. The plans from other authors have similar errors (I will not elaborate about it here).

It seems that I have to crate my own drawings!

In the next post I will refer the progress of this work (I hope that I will show you the corrected side view).

 

[Wurger]

Great !!! However I hope you took into consideration the angle the picture was taken with. I mean the shot isn't the full side view at the kite.

 

[Shinpachi]

I agree that commercial drawings are not reliable no matter how detailed they may look as they are copies of copies in almost cases.
You are doing good job, Witold Jaworski.

 

[Witold Jaworski]

Thank you for the comments! @Wurger: yes, this airplane is deflected by 1.4 deg and rolled by 5.5 deg. I did a compensation for this deviation.

Below you can see the first version of the side view (click the drawing to see the high-resolution version - 1900px wide):

This is not an ultimate drawing: I suppose that it will be updated during my work, following the new findings about the airframe shape and/or details. The dotted lines mark the rivet seams, but size and spacing of these dots does not match the real rivets. I prepare these plans to build a model: that's why I removed the outer wing section and horizontal tailplane. For these parts the most important drawing is the top view. To build them, on the side view I need the precise contours of their key sections (i.e. their airfoils as well as the incidence angles and spar locations). I draw three profiles: first of the wing root, then the root of the outer wing section, and then the wing tip. Two different sources specifies different wing tip profiles: NACA-2409 (Performance Test Report, 1942) or NACA-2407 (BuAer drawing, 1944). However, the bottom contour of the NACA-2407 seems to be a little concave. Because I did not observe such an effect on the photos, I decided to use the thicker airfoil of NACA-2409. I still have to verify this detail when I build the wing. The airfoils of the tailplane were specified nowhere. I copied its root airfoil from a photo.

While drawing the side view you still have to think "in 3D". That's why you can see around this silhouette some auxiliary sketches: the front view of the engine cowling, and the contours of the center wing section. I draw the latter element just to mark the exact position of the first rib of the wing. It was hidden inside the fuselage. Note that this airfoil was a specific modification of the NACA-2415 shape: the part of the wing that houses the main landing gear was reshaped. In the effect, the leading edge of the center wing section has a small downward inclination.

On the next week I will present side views of two earlier Dauntless versions: the SBD-2 and SBD-3. I will discuss where are the differences in the length of the SBD-5 and SBD-3, 4, as well as the mystery of the "missing 7 inches" of the SBD-2.

Within two weeks I will present the corrected/verified top view.

 

[Wurger]

@Wurger: yes, this airplane is deflected by 1.4 deg and rolled by 5.5 deg. I did a compensation for this deviation.

Great ! Looking forward to the verified top view, Witold.

 

[Witold Jaworski]

Thank you, Wurger!

In this post I will show you how do I create Dauntless side views. First I used the "semi-orthogonal" photo of the SBD-5 as the reference to draw the side view of this version. This is the most important picture, because it provides reliable "general reference":

Then I used many other photos and sketched fragments of the other views to complete the side view details (note the multiple guide lines on these pictures):

Note the large B/W photo that I used to verify the shape of the propeller hub and engine cowling. I could not compare it with other areas — the cockpit canopy, for example — because its perspective (barrel) distortion was too intense.
However, when the barrel distortion is moderate, we can revert it! See for example this side photo of another Dauntless version: the SBD-3:

First I identified the undeformed fragment of the fuselage (in this case — around the firewall and the windscreen) then fitted this part of the photo into the drawing. Then, comparing the lengths of this photo and the side view, I concluded that it has a moderate barrel distortion. From the history of this design I know that the SBD-3 and SBD-5 had different engine cowlings. The other parts of their airframes had the same shape. This means that I could use the existing SBD-5 drawing from the firewall to the fin as the reference for the unwrapping process of this SBD-3. Then I unwrapped this photo using the GIMP. (Speaking more precisely – its "Lens Distortion" image filter. You can find all the details of this process in this book).

Note that this operation "flattens" only the airplane contour that lies on the symmetry plane of the fuselage. All protruding elements, like wing and horizontal tailplane remain deformed. But it's OK, I need this just this contour. While drawing, I will compensate the small remaining deformation of the bulkhead lines. (For example, the leading edge of the NACA cowling from this photo should be a straight line, but it is a very flat ellipse).
Of course, it is always better to prepare more than one of such "flattened" pictures, to minimize inevitable errors:

NOTE: these two SBD-3 photos depict training aircraft, without the telescope gunsight. Such a gunsight was protruding through the windscreen in the combat airplanes.
​

On both SBD-3 photos you can see that the engine cowling is somewhat shorter than in the SBD-5. In fact, in the specifications you can find that these versions had different overall length:
- SBD-5: 33' 1​/8​";
- SBD-3: 32' 8 " (in some sources I also saw 32' 8 4​/5​")
However, on the scale plans authors attribute this difference to the longer propeller hub of the SBD-5 (It used different propeller: Hamilton standard hydromatic). Others did not bother about the different lengths of the SBD variants, and draw the profiles of all Dauntless versions alike.

Following the findings on the unwrapped photos I analyzed many other archival pictures. Below you can see the conclusion:

It seems that in the SBD-5 the engine, together with the NACA cowling, was moved slightly forward. All other bulkheads remain in the same places. This modification shifted forward the center of gravity. I suppose that this correction improved some handling characteristics that changed after the doubling of the rear guns. (The second gun in the rear was introduced in 1942 to the SBD-3 as the "field" modification. It shifted the cg backward).

In my next post I will finish the side view matter, delivering you the complete side views of the SBD-3, SBD-2 and SBD-1 variants. I will also write about other non-existent length difference, which you can find in the books about the SBD Dauntless.

 

[Wurger]

 

[Lucky13]

8)

 

[Witold Jaworski]

In addition to the side view of the SBD-5 presented in one of my previous posts, I have prepared side views of the earlier Dauntless versions: SBD-2 and SBD-3:

(Here are the links to the high-resolution profile images of: SBD-2, SBD-3).

When you look into Dauntless specifications, you will find that all its models have the same span, but they often differ from each other in the length. This is a typical case, because the wing geometry determines the aircraft behavior. Thus, once "debugged" in the prototype (the stall characteristics etc.) it remains unaltered between subsequent versions. The fuselage shape is not so important, so it is often modified. In the effect, the length of the airplane often vary between subsequent versions.

In the previous post I described how the photos confirmed the different length of the SBD-5 (33' 1​/8​") and the SBD-3 (32' 8"), listed in their specifications. The reason was the different engine mount, modified in the SBD-5. The same sources specify the length of the SBD-2 as 32' 2". This is something strange, because I cannot find any evidence of this significant, 6 inch difference between SBD-3 and SBD-2 on the photos!

The SBD-3 was a "quick and dirty" adaptation of the SBD-2 to the recognized requirements of the modern war. Douglas added armor plates to the pilot and gunner seats, self-sealing fuel tanks (reducing their capacity), doubled the rear guns. All the key elements of the design: the airframe and the engine, remained the same. Where is there the modification that changed the overall of length of the SBD-3 by 6"!?

I started to look for the sources of this information (the subsequent publications copy their specification data from the earlier ones, up to an ultimate source document). Ultimately it seems that it comes from the BuAer Performance Data Reports. There are two of such documents, created in 1942: one for the SBD-2 and one for the SBD-3. On their last pages you can find the measured airplane dimensions. The difference is there: LENGTH, LEVEL: 32'-8" in the SBD-3 report, and LENGTH, LEVEL: 32'-2" in the SBD-2 report. (Unfortunately, they did not specified the length on wheels for the SBD-2, so there is no double-check). Note that all other dimensions are the same. I speculated that the reason of these differences lies in the propeller spinner: it was often removed. If the tested SBD-3 had this spinner, and the SBD-2 didn't — what was the eventual difference? I tried to check this option, but it shortens the fuselage length by less than 4".

What's more interesting: the only survived SBD-2 is owned by the National Navy Aviation Museum in Pensacola. On their web page the owner specifies the length of this airplane as 32' 8" — the same as the SBD-3! Thinking further about it, I noticed the manual corrections of typing errors in other SBD performance data reports. So I have following hypothesis:

• The SBD-2 and SBD-3 had the same length: 32' 8", as specified by the owner of the restored SBD-2 (NAM in Pensacola);
• The typist of the BuAer Performance Data Report made a mistake (most probably —deciphering the handwritten measure results he/she read "2" instead "8"). The authors of the first publications about SBD Dauntless used this source, and the others used their publications. So the initial error was multiplied;

Thus I assumed that the SBD-2 length specified in Performance Data report is wrong. Basically it was the same as the SBD-3. It also applies to the SBD-1:

(Here is the link to high resolution profile image of the SBD-1). The only external difference between the SBD-1 and SBD-2 is the larger air scoop on the top of the engine cowling.

Conclusion from this little investigation (in fact, it took me a few days): do not trust blindly the specified width and wing span of a historical airplane! When you compare the different sources you will find that sometimes these figures are different. Always try to verify available data. The wing span is less error-prone because it usually does not vary between subsequent versions. Remember that the photos are always the ultimate evidence.

In the next post I will present the updated/verified Dauntless top view.

 

[Wurger]

 

[Witold Jaworski]

This Monday I finally got the "Instructions for the Erection and Maintenance of the Model SBD-6 Airplane" book – more than 600 pages about the Dauntless, published by Douglas in March 1944. Because of the lengthy title, I will refer to this book as the "SBD Maintenance Manual" or the "Douglas manual". In spite that it describes the last produced version, it is also usable for the earlier models: as I mentioned in one of the previous posts, the SBD-1 airframe behind the firewall differs only in a few details (the double gun mount, gunsight type, lack of the YAGI antennas) from the SBD-6.

Inside you can find the SBD-6 general arrangement drawings, as well as the stations diagram:

Here are the links to the high-resolution versions: side view (cropped from the page), top and front view, stations diagram. As you can see these Douglas diagrams contain more dimensions than the BuAer drawings. Their chains on the side view allow for verification of the wing location, as well as the wing and tailplane incidence angles. They also allow you to determine the basic "trapeze" around the rudder and the fin. From the front view you can also read the dihedral angle of the outer wing panels (9⁰ 19').


The dimensions from the top view allowed me to draw the basic trapezes around the wing and tailplane, as well as to determine locations of the aileron and elevator hinge axes. This information, combined with dimensions from the side view, allows for determining the precise location of the firewall, wings and empennage. I used them to verify my scale plans. Sometimes they just confirmed what I determined before (for example — locations of the wing or the last bulkhead). Sometimes they revealed the errors I made. I will write more about it in the next post. So here is the current, updated version of my drawings:

Because of the formatting issues I had to split this image into two parts:

(Click here to get these drawings as a single, high-resolution image). Note that I draw the outer wing panel without its dihedral (it is much easier to build its model using such a "flat" reference). Thus when you check proportions of this top view, its span/length ratio is somewhat greater than the expected value of 41' 6" / 33'. What is interesting, the dimensions on the general arrangement drawing indicate that the "official" wing span does not include the size of the running lights:

To obtain the "physical" wing span value you have to add 1.5" to each wing. I used similar convention when I matched the fuselage contour against its dimension (33' 1​/8​"). These dimension lines are more obscured on the side view, but for the matching purposes I skipped the length of the running light cover protruding from the tip of the tail (1").

In general, after all these updates I feel more confidence in my drawings. I know which elements come from the explicit dimensions of the general arrangement diagram, which from the photos, and which are based on other drawings or just on an assumption. The only larger element that I was not able to verify is the fuselage width (i.e. its contour in the top view). It is copied from the Douglas drawing. I was able just to verify it at the 9th​ bulkhead (station 140). I have a photo of this bulkhead from one of the Dauntless restorations, so I am sure that it fits properly into the fuselage contour on both views: the side view and the top view. However, I did not verify in any way the curved contour of the tail on the top view.

Frankly speaking, after this experience I am really glad that I am doing such a "slow start" to the modeling by preparing these drawings. It forced me to think twice (or even more times than twice) about every part of this airplane, resulting in better understanding of various nuances of its geometry. Sometimes I had to deliberate over a single line (like the gap between the elevator and stabilizer) for a whole day, watching and comparing hundreds of photos. In the effect I had to move a few lines around it on the plans. It was not a big deal. However, if I already started to build the model, adaptation of such findings would require a lot of work!

In next post I will tell you more on how I used the explicit dimensions from the Douglas drawings. They allowed me to find some flaws in my plans. Description of this case will give you an insight into the errors that you can make using the photos.

 

[Wurger]

 

[Lucky13]

Very nice work!

 

[Witold Jaworski]

@Wurger, @Lucky13, thank you!

The ultimate shape of airplane from post #17 resulted from matching my initial drawings against the Douglas general arrangement diagram. I couldn't do it before, because this diagram comes from the Dauntless maintenance manual, which I received in previous week.

In this post I will show you how I do such a matching using the diagram shown below:

When you use such a drawing, you can follow the general rules of the technical drawings. In particular:

• The ultimate contour of the depicted object is on the outer side of the drawing lines;
• When the shape on the drawing differs from the result of its explicit dimensions, the result of these dimensions prevails;

So, starting from the thrust line (i.e. the propeller axis) and from the firewall (the base of all dimensions), we can use the dimensions from this diagram to determine the wing chord position (points A and B on picture below):

We can read from the side view dimensions that A (the rib tip) is located 20.38" from the thrust line and 9" from the firewall. The end of this rib is located 2.5⁰ lower, and the chord length of this rib is 115.12" (this dimension you can read from the top view). This determines location of point B.

I used the scale of my drawing (3.02 px/in) to convert the dimensions listed above into drawing units. Then I used guide lines to find these points of the wing chord on my plans:

Fortunately points A, B on my plans occurred very close to wing leading and trailing edges.

You can use dimensions from the general arrangement drawings to sketch the basic trapeze around the wing (in the top view) as well as around the fin, rudder, and horizontal tailplane. These trapezes allow you to determine the basic shape and proportions of the airplane. I will show this method on the example of the fin and the rudder. Figure below shows their dimensions on the original drawing:

Using these dimensions you can draw the basic trapezoid around the rudder and fin. You can also locate the chord of the horizontal tailplane as we did for the wing.

When I mapped these elements onto my plans, they revealed a serious flaw in my drawings:

The whole tailplane seems to be shifted downward, and the rudder hinge is moved left! However, if the wing chord fits to the dimensions, then most probably this is the result of a random rotation. I have quickly verified this hypothesis using the reference photo:

When I set the pivot point of this transformation above the wing see figure below), it was enough to rotate this photo by 0.27⁰ to fit the rudder and fin into given contour:

It seems that I made mistake at the very beginning, trying to set this photo horizontally (in post #7). I estimated it using cockpit edge (as in figure above), because the better candidate for such a reference — the seam running on the side view along the reference line — is not visible on this picture. It seems that this fragment was too short for precise estimation of the horizontal direction. What's more, I did not know at the beginning that in the top view this edge is not parallel to the fuselage centerline. Because the depicted airplane is slightly inclined toward the photo, I had to estimate location of this edge as the line lying between two cockpit edges visible on this picture. The BuAer drawing (copied from the Douglas general Arrangement Diagram) would help, if its draftsman did not made additional errors around the tail and empennage (see post #7).

Of course, the drawings that I published in post #17 did not contain any of the flaws that I have found here. I fixed all of them before. I just wanted to show you in this post what kind of errors you can do using a photo reference.

Conclusion: always try to find a general arrangement diagram of the airplane and use its explicit dimensions to verify your drawings. They often allow you to fix severe flaws in the geometry of the depicted aircraft!