Primitive Archer
Main Discussion Area => Bows => Topic started by: Aussie Yeoman on December 13, 2025, 04:09:14 pm
-
Hi all,
I'd like to get into deflex reflex bows, having made longbows and flatbows (straight and perry-reflexed) pretty much exclusively for almost three decades, with one or two recurves in the mix. I've been teaching bowmaking classes for about 15 years.
So I'm not new to bow making by any means, but only to this particular style. I've tried once or twice, but it didn't end well. I think the problem was in having the belly slat too thick, and/or trying to go for too much reflex, in the glue up.
I'd like to hear from the experienced, what common mistakes beginners can make when constructing and tillering this kind of bow. Maybe even some of the uncommon but important to know mistakes too.
Cheers!
-
R?D is still my favorite, but not as extreme as shown in the diagram. I have built quite a few like those pictured, and I have had a very high failure rate. My biggest mistake is getting a hinge in the middle of the limb where it just suddenly appears. I believe my biggest mistake on those is trying to go for too much reflex. If the tips are level with the back, they are not too bad to tiller out, but they are a nightmare with a few inches of reflex. My favorite R?D design is more like an R&D with a semi-recurve. I use a very similar form to the one Gary Anderson used. Hard to beat.
-
Thanks Badger, always good to hear from you.
Had you meant to attach a picture or link to one?
-
I was referring to the photos of bows in the other thread about bow shapes. I find that if the limb is straight at brace, I am in pretty good shape, but if it is still showing the dip from the deflex, it is under extremely high tension on the string and will twist itself into a knot.
-
Ah! I see. You mean #6 here? http://www.primitivearcher.com/smf/index.php/topic,73421.0.html
Or perhaps Post #13 in the same thread?
The D/R bows I've seen that are done well do seem to look pretty much dead straight when braced. I did some analysis of a couple and at full draw the bent limb looks to have the same degree and shape of bend as a straight bow with a ~7 inch brace. Doing that bit of graphics was pretty informative, because an elliptical bow at brace has very little bend at the inner limb. Which is something I probably didn't do well the time I tried.
-
Here's one of the two designs I'm currently playing with. The tips are essentially dead-even with the back of the handle at rest, the limbs pretty straight at brace.
To get this profile, it needs to have an amount of reflex equal to about 80 mm (3 1/4") of reflex on a straight bow. Which I assume would need maybe five inches of deflection during glue up, to account for spring-back.
-
Biggest problem is over stressing the limb. A limb will always have a specific amount it can bend no mater its profile. Make sure the bend radius never is exceeded.
-
Biggest problem is over stressing the limb. A limb will always have a specific amount it can bend no mater its profile. Make sure the bend radius never is exceeded.
A few forumites have mentioned already how difficult it is to not create a hinge when tillering a R/D design. How would you go about actually making sure the bend radius is never exceeded?
-
You make sure the reflex is even thought the bending portion of the limb. Whe you bend the limb, you make sure the bend stays even as it straightens out, eventually becoming a straight line as the limb flexes flat. After its a straight line, you make sure it bends evenly into a perfect curve as you pull it through towards full draw. You set yourself up for success by evenly reflexing the bow. Start even, finish even.
-
So to drop down on that further, how do you ensure that? Tiller only with a scraper? Some other stratagem?
-
I think the biggest mistake is having too much reflex (tips bent too far behind the back). The second mistake is not using the VirtualBow program to model the bow and find the correct taper rate for a specific front profile.
In my opinion, and what I prefer to do, is keep the reflex at about 0–50 mm—no more in normal bow. If I want to make a flight bow or a bow just for speed, and I can trust my materials, then more.
I always make deflex–reflex bows as laminates, because then the correct taper rate for evenly distributed stress is easy to achieve. Usually, a taper of 0.006 or 0.008 works well. I have a good taper sled, so the tiller is almost always spot on. If the tiller needs correction, a scraper works well and is almost the only tool I use.
-
Start even, finish even.
knowing what bend to eyeball for is not as intuitive with a R/D as it is with a straight bow.
Removing wood to the correct tapers will produce the desired bend.
-
Thanks folks. This is good material so far.
When reflexing laminations so close to finished dimensions, what do you expect the spring back to be off the caul? As in, if you want to have the tips deflected 3" from their deflexed position, do you glue the limbs in 6" of reflex? More? Less?
-
When reflexing laminations so close to finished dimensions, what do you expect the spring back to be off the caul? As in, if you want to have the tips deflected 3" from their deflexed position, do you glue the limbs in 6" of reflex? More? Less?
The more laminates used, the less springback there is. I prefer to use four laminates; with that configuration, the springback is about 10 mm (1/2") at most. With three laminates there is slightly more, but it has never been a problem or something I would take into account. For a 45–50# bow, the total thickness is about 15 mm. With four laminates, each laminate is under 4 mm thick, so they are relatively thin. Set is another matter: if you want the finished bow to have a side profile similar to the model, you should take that into account.
-
I'm planning on using two lams - belly and backing. At least to start. Bamboo and an Australian hardwood, called Ironbark. It's heavier than water and stiffer than the timbers you folks think are stuff, like bulletwood or ipe.
According to virtual bow, the thickness overall at the start of the working limb will be about 11 mm, and just over 5 at the tip.
-
heat shape the lams to the forms before gluup? I know the bamboo will be easy,
-
Here's one bow I'm working on in VirtualBow. It's a takedown with the riser omitted.
Brace: 150 mm
Draw: 711 mm
Handle-
length: 175 mm
setback: 0
Angle: -7.5
Profile: Spiral
Length: 750
R start: 3000
R end: 2500
Width-
0//40
12.5//39.9
50//39.9
75//31.4
100//10
Thickness-
0//12.05
100//5.75
The material is Ironbark. With a stiffness of 24.212 GPa, and a max working stress of 167 MPa.
-
And here's another I'm still tooling around with. This time a 1-piece.
Brace: 150 mm
Draw: 711 mm
Handle-
length: 0
setback: 0
Angle: -7
Profile: Spiral
Length: 838
R start: 4000
R end: 2250
Width-
0//30
2//30
4//30.7
6//31.8
8//34.4
10//37.6
12.5//39.8
50//36
75//28
100//10
Thickness-
0//25
2.5//25
5//24
7.5//17.2
10//12.5565
12.5//11.7125
15//11.556
100//6.68576
Same material as above
-
So, I've collated some commentary and for future reference the key things to keep in mind when making a D/R bow are:
- The laminations must be pretty accurate to begin with (difficult if using natural bamboo, but we must try what we can, eh?)
- Avoid having uneven reflex. Smooth curves at the start = smooth curves at the end. This would be a manifestation of the above point being gotten right, in addition to having a good form/caul.
- At brace, the limbs should be actually or close to pretty straight
- Avoid putting in too much reflex. Having the tips around even with, or just in front of, the back of the handle helps to avoid nasty problems during tillering
- Go slowly through tillering - this done to avoid putting too much bend somewhere in particular, which can happen very quickly with a bow like this. Consider using a scraper of some sort. The accuracy of the lams to begin with helps reduce the amount of bulk stock removal, making the scraper effective.
- Know exactly what it's meant to look like at full draw before you start. This done by sketching, modelling or emulating another bow. This helps prevent hinges or overshooting the limb's curvature.
Are those the main points or have I missed something vital?
-
- Know exactly what it's meant to look like at full draw before you start. This done by sketching, modelling or emulating another bow. This helps prevent hinges or overshooting the limb's curvature.
Are those the main points or have I missed something vital?
with virtual bow you should be able to see what the bow should look like at any stage of the draw, eliminating the need to compare it to a different bow
-
How hard is it to download the Virtual Bow program and what do you use to open/use it?
-
https://www.virtualbow.org/
It's an independent program that you download and install. It's really easy to install and quite easy to use, with a fairly shallow and short learning curve.
-
https://www.virtualbow.org/
It's an independent program that you download and install. It's really easy to install and quite easy to use, with a fairly shallow and short learning curve.
I have win 11 and get a warning about it "not being commonly downloaded. Be sure that you trust the link" It did download something but the file is called "unconfirmed download" and it won't open...
-
Several questions about the programs. Do they allow for some set at given stresses? Does increased set affect efficiency or FDC? If so, by how much? Are there any allowances made for working part of the limb length? Efficiency is lost in working limb length and increases as the working limb shortens, even though strain figures might go up.
-
https://www.virtualbow.org/
It's an independent program that you download and install. It's really easy to install and quite easy to use, with a fairly shallow and short learning curve.
I have win 11 and get a warning about it "not being commonly downloaded. Be sure that you trust the link" It did download something but the file is called "unconfirmed download" and it won't open...
does your file name end with .crdownload? if so try again
-
Several questions about the programs. Do they allow for some set at given stresses? Does increased set affect efficiency or FDC? If so, by how much? Are there any allowances made for working part of the limb length? Efficiency is lost in working limb length and increases as the working limb shortens, even though strain figures might go up.
set is not taken into account with the program. It is up to you to choose a MOE value for the materiel used, and review your design to see that stresses remains below the level you feel set will occur at.
-
https://www.virtualbow.org/
It's an independent program that you download and install. It's really easy to install and quite easy to use, with a fairly shallow and short learning curve.
I have win 11 and get a warning about it "not being commonly downloaded. Be sure that you trust the link" It did download something but the file is called "unconfirmed download" and it won't open...
does your file name end with .crdownload? if so try again
Yes... I have hit 'download 5 or 6 times and the file is the same every time. "Unconfirmed(random number).crdownload"
Thank you for helping...
-
Nah, unfortunately it doesn't account for set. Alan Case's spreadhseet allows a bow to be 'pre-stressed' to account for set.
If One happens to have bend test data, then One can predict how much set the bow will take at a given stress level. That's what I've done with my D/R bows above, and anticipate they'll take between 3/4 and 1" set once shot in.
Several questions about the programs. Do they allow for some set at given stresses? Does increased set affect efficiency or FDC? If so, by how much? Are there any allowances made for working part of the limb length? Efficiency is lost in working limb length and increases as the working limb shortens, even though strain figures might go up.
-
https://www.virtualbow.org/
It's an independent program that you download and install. It's really easy to install and quite easy to use, with a fairly shallow and short learning curve.
I have win 11 and get a warning about it "not being commonly downloaded. Be sure that you trust the link" It did download something but the file is called "unconfirmed download" and it won't open...
does your file name end with .crdownload? if so try again
Yes... I have hit 'download 5 or 6 times and the file is the same every time. "Unconfirmed(random number).crdownload"
Thank you for helping...
windows 11 may be too helpful for its own good
try checking these settings
https://www.supportyourtech.com/tech/how-to-allow-downloads-on-windows-11-a-step-by-step-guide/
-
Several questions about the programs. Do they allow for some set at given stresses? Does increased set affect efficiency or FDC? If so, by how much? Are there any allowances made for working part of the limb length? Efficiency is lost in working limb length and increases as the working limb shortens, even though strain figures might go up.
No, it is not possible to define set as we understand it. However, there is a “damping” parameter, but it only describes some general energy losses in a bow. Set relates to material properties, and it will be included in the future, so you cannot currently take set into account. Of course, you can interpret strain values: if the compression strain is high, the bow will take set.
Regarding the working part of the limb – yes, of course. You simply define how the bow tapers (just like in my Straight Bow example) and simulate it. I am not sure that efficiency is lost with a longer working limb length; I do not believe it is the determining parameter. Efficiency relates mainly to moving limb mass and hysteresis, and working limb length affects these factors.
Nah, unfortunately it doesn't account for set. Alan Case's spreadhseet allows a bow to be 'pre-stressed' to account for set.
If One happens to have bend test data, then One can predict how much set the bow will take at a given stress level. That's what I've done with my D/R bows above, and anticipate they'll take between 3/4 and 1" set once shot in.
Alan Case’s SuperTiller program (in Excel) does not take set into account. The “pre-stress” feature is purely for pre-stressing, similar to Perry-reflexing a bow. For example, in VirtualBow a reflexed bow has zero stress when unbraced, whereas in SuperTiller you can introduce pre-stress so that the bow has a certain amount of stress even when not braced. As a result, at full draw this kind of bow is more highly stressed than a bow with no pre-stress. It is a nice feature, and hopefully we will have it in VirtualBow in the future as well.
Compression strain values are very useful when predicting set. Wood will take a permanent set when the strain value exceeds about 0.4–0.6% in compression, depending on the wood species. In tension, wood can tolerate on average about 0.9–1.0% strain, but the limit is much lower in compression. Of course, there is an important difference: if the maximum strain is exceeded in tension, the bow breaks, whereas if the bow takes set, it is not yet broken. I am not sure how much compression strain is required to cause chrysals, and I am sure it varies greatly depending on the wood species. However, if you want to estimate set in VirtualBow, I would use 0.4–0.6% compression strain as a guiding range – beyond that, the bow will take set.