The Adapt-a-Damper - Open Source Project


 
glad you guys are interested in what I'm working on/sharing......thanks for testing it out to beggs, and its fun sharing a 'hobby-type' project.....I'm new at this forum at it seems not everyone is "on board" with this whole sharing of ideas....

I haven't got around to "the fan" part my projects yet (i just use an auber fan currently)....but I like your latest design....will use it with my all-in-one project.....
 
Hey Will. So I wanted to get some real numbers out there for you because there was a big concern of switching from the "smiley face" to the linear design. I downloaded your STL files from Thingiverse and duplicated the dimensions in AutoCAD so that I could calculate the errors associated with the "smiley face" design that your damper (and another one out there) has. Below is a photo showing the associated percent errors with running the "smiley face" opening. You'll notice the highlighted areas are where the pit should be running. This is also the area where the largest percent errors take place. A mean percent error in this range is an astounding 47.9%! This is also without adding resolution error from the servo.

So then the big concern with switching to a linear damper is that you have to split up the intake to two different holes. Doing so reduces the resolution of the servo operation. The reduction in resolution is a factor of the difference in range. The "smiley face" rotates from full open to full close in approx 141.7 degrees. The two hole linear design rotates from full open to full close in 88 degrees. Therfore the resolution is reduced by (141.7-88)/141.7*100 = 37.9%. So already, hey look at that, it's better than you thought.

So then the big question is, well what is the actual resolution of a MG90 or SG90 servo? According to TowerPro their MG90 and SG90 servos have a deadband of 1us. Now I've seen another spec around the internet that says it's as high as 5us, and I will address that. So then to get the deadband to resolution you have to multiply the full range that the servo is capable of by the deadband and divide by 1000. That means that with 1us deadband your resolution is 0.18 degrees. IF the servo has a deadband of 5us then your resolution is 0.9 degrees.

So what does that mean when you apply this error to a linear damper? Well I downloaded your linear damper and drew that up in AutoCAD to find out. I calculated that a 5% closer on your damper requires a rotation of 4.4 degrees. With a 1us resolution with a resolution reduction the 5% closure is 4.65 degrees. So when that damper is rotated 4.65 degrees your percent error is 0.3%. Surprised aren't you? Let's say for whatever reason the deadband is actually 5us. That means a 5% closure on your damper would require (with reduction and error) 5.64 degrees. What's the associated percent error with that? 1.7% would be the answer.

So how did this "smiley face" design come about if it's so inaccurate? I have no idea. I can only presume that it was designed this way because nobody really took the time to actually think about what they were making. Which is odd, because it has been purported many times over by some designers that there was a lot of thought put into the design. Hmmf.

Here's that table with the "smiley face" errors:

fLeguCZ.png
 
Last edited:
Holy Batwater, Steve! I actually understood that. What an awesome proof of your original hypothesis. The "smiley face" has officially turned into a frown. I'm thinking, in design, it's a lot easier for people to make a half circle than two arced circles centered around a point of rotation, thus, the smiley face. The best cooks I've ever had were with Tom Kole's Barrel Damper design, which gets me thinking. What is better? fan before damper, or damper before fan?
 
WBegg said:
The best cooks I've ever had were with Tom Kole's Barrel Damper design, which gets me thinking. What is better? fan before damper, or damper before fan?
Click to expand...

I feel bad saying this, but I've never really looked hard at Tom's design. There is a lot more marketing on this forum for the other design. I'll download Tom's and see what I think. I for sure see a lot of excellent graphs coming from cooks that used his design though.
 
If you need a blast for high heat, 100% open will always = 100% open. Other than that, don't overthink it too much. Using Tom's offset rotary damper, it's usually at 30-50% on my Kamado Joe.

It's not like we're designing a fuel rail and injectors for a race car!
 
Steve_M said:
If you need a blast for high heat, 100% open will always = 100% open. Other than that, don't overthink it too much. Using Tom's offset rotary damper, it's usually at 30-50% on my Kamado Joe.

It's not like we're designing a fuel rail and injectors for a race car!
Click to expand...

True Steve, but my HADD (Hobby ADD) implores me to make things more perfect than they need to be. If I can get it to within 1 degree, then I'll want to get it to within 0.5 degrees.

Regarding Tom's design, I've used the Barrel Servo extensively. It just works. I run in open/closed only so it makes sense, but I'm trying to think of those that don't have the same setup. Why not start with the math as Steve described.
 
Well, the half moon design at 100% open is going to have more airflow compared to the linear opening, so it's got that going for it.
 
My last cook with Tom's Barrel Damper design before I started messing with this rotary adapt-a-damper crap. Although a bit long on the print, it works.

KjrKRD7.jpg
 
SteveCK said:
Actually, his linear design has 60% more cross sectional area
Click to expand...

60% more to what? Do what mining engineers do, and re-do the math until you reach a number I'm happy with. The Steel-a-Damper has 891 mm2 roughly. The two hole has 1300 total.
 
SteveCK said:
I feel bad saying this, but I've never really looked hard at Tom's design. There is a lot more marketing on this forum for the other design. I'll download Tom's and see what I think. I for sure see a lot of excellent graphs coming from cooks that used his design though.
Click to expand...

The barrel still suffers from some non-linearity when it comes to servo control. The reason it never really mattered is that I found out early in my testing of the damper that running the servo in on/off mode gave the most impressive results.

As you've elegantly demonstrated, there can be large errors with the smiley valve depending upon where you are in the output range but I'm not sure if it makes a big difference in terms of controlling a smoker that has relatively large lag times. Like I've said before, if someone can print 2 valves that are identical except for the valve pattern and show me a difference, I'll be a believer. You could also test whether the resolution of the servo actually matters by testing the 2 vs 1 hole valves. I suspect that the sometimes erratic nature of the servos will lead to much larger errors than the deviations from linearity that are in your table.
 
Last edited:
..how about instead of a smiley face, you use a 'smirky' face.....where the width of the opening is different along the arc to compensate for the errors.....you would lose total opening area but should not have any errors......

and of course this is all super-over-kill......that's what makes it so interesting ;) The best solution is to do it in software.....'under an advanced setup tab I must add'.....
 
Tom Kole said:
I suspect that the sometimes erratic nature of the servos will lead to much larger errors than the deviations from linearity that are in your table.
Click to expand...

I'm going to print a large calibrated disk, and test a few servos and see what the real world resolution is on them. I did a quick calc to see what the error would be with the Adapt-a-damper linear design with a 20um servo deadband (which I would hope is a just a crappy servo) and still the percent error associated with it is only just over 6% - still much much better than the smiley. I'll have some real world data very soon. But yes, you are correct that the difference isn't going to be anything huge. The reason why the smiley design does work is because the HeaterMeter does all the leg work for the error. That's fine and all, but there is also nothing wrong with making a finely tuned machine.
 
Last edited:
DavidNP said:
..how about instead of a smiley face, you use a 'smirky' face.....where the width of the opening is different along the arc to compensate for the errors.....you would lose total opening area but should not have any errors......

and of course this is all super-over-kill......that's what makes it so interesting ;) The best solution is to do it in software.....'under an advanced setup tab I must add'.....
Click to expand...

I actually spent a good deal of time thinking this over while I was designing the linear intake for the MicroDamper. Once I started digging into it I found that designing a precise intake that works radially is just insane to design. I would have needed to have taken much higher mathematics classes than I did in engineering school to come up with the correct formula to make this work, then I would have to actually model it in 3D. Way above my pay scale and knowledge, plus just what I mentioned above the PID system of the heatermeter takes care of those small errors.
 
Will, I've been using your first design from here: http://www.thingiverse.com/thing:1177115 In fact those pictures are my damper. I like the simple no box design since I have no thermocouple and I did not want to jumper all those wires either.

It seems though that the smaller fan does struggle regulating the temps on my 22" WSM when it is below freezing out, even with a wind screen. So before I try the 15CFM version, any chance you will be updating that design to a two hole version as well?
 
Aristotle Allen said:
Will, I've been using your first design from here: http://www.thingiverse.com/thing:1177115 In fact those pictures are my damper. I like the simple no box design since I have no thermocouple and I did not want to jumper all those wires either.

It seems though that the smaller fan does struggle regulating the temps on my 22" WSM when it is below freezing out, even with a wind screen. So before I try the 15CFM version, any chance you will be updating that design to a two hole version as well?
Click to expand...

Now updated with the 2 hole version included in the files. http://www.thingiverse.com/thing:1272011
 
I like the idea of achieving more of a linear output by using a different-shaped intake hole so I was going to print one of these up. I know the idea is that you can swap out the ethernet plug box with something else but wouldn't it make a lot more sense to have the body and plug box be the same printed part, with a cover on the plug box?

Then you don't have to worry about the dovetail of the two pieces fitting together right, removing the plug box to wire it, and most importantly in my mind not having to fill the whole box with support to print it in that orientation. The lid can just slide on top that piece, and optionally add a screw if it isn't secure enough. The plug box can also be smaller because opening the lid will get you right to the ethernet jack wiring so no extra "working space" is required around it.
 
Bryan Mayland said:
I like the idea of achieving more of a linear output by using a different-shaped intake hole so I was going to print one of these up. I know the idea is that you can swap out the ethernet plug box with something else but wouldn't it make a lot more sense to have the body and plug box be the same printed part, with a cover on the plug box?

Then you don't have to worry about the dovetail of the two pieces fitting together right, removing the plug box to wire it, and most importantly in my mind not having to fill the whole box with support to print it in that orientation. The lid can just slide on top that piece, and optionally add a screw if it isn't secure enough. The plug box can also be smaller because opening the lid will get you right to the ethernet jack wiring so no extra "working space" is required around it.
Click to expand...

Thanks Bryan.

When I was originally designing, I was going to go just this route, but wanted it make it "swappable", so I went with the current design. If the box is integrated, one will have to print a complete lower if they have a different box layout. I will agree that printing the box in the optimal strength orientation produces alot of overkill as far as support material and time.

SteveCK had a good idea, and that was to print the "lid" or flat onto the main body, and the case can still slide on with the dovetail, but could be printed without any support materials (other than those needed for the keystone hole), as there would be one full open side.

As far as the dovetail, maybe I'm taking it for granted because I have a very well calibrated printer, and have had no issues with fit, but maybe I need to start thinking about those that aren't as well calibrated.

I guess there's more than one way to skin a cat, so I guess I'll be skinning cats these next few days. I'll try to get some designs for both ideas.
 
Out of curiosity, I wanted to test the air flow efficiency of the "smiley" vs "linear designs. I made an adapter to attach an anemometer (flow meter). Here is the setup.

u8TrYX5.jpg


I tested each setup in 5% increments, with the same fan, same servo, and the same Heatermeter settings (servo to follow the fan, not open/close only).

The results were as I suspected, as outlined by StevCK's posting above. The numbers show quite a marked difference in the normal operating range (5% to 40%), which would imply that a linear system should run more efficiently in this range. Of course, there are no benefits to one or the other if running the Servo open/close only, but with the recent options given in the new snapshot, and the soon to be released V13 firmware, I again see the Linear (2 hole) design a better option.

APbMFnB.jpg


Cheers,
Will
 
You must log in or register to reply here.

 

Back
Top