Winter Build 2019-2020: Top Flite P-51 (0.60 size)

[F106DeltaDart]

I had NO idea what to expect in terms of time or cost involved for its development let alone how much time it would take to get a 3D printed aluminum piece. Those unknowns are just part of the journey. I can tell you though, an off-the-shelf 4-blade rotor hub assembly is over $400 retail and my motor and assembly cost half that. The upshot too is that if I get down the road to testing and find that I cannot get things to perform as I would like, I have the ability to convert the motor back to normal and just roll with a 2-blade prop. No modifications to the airplane required.

Josh, I’m glad to see things moving on the build again. Definitely the right move to take the time in the planning stage to make sure things go together smoothly! I‘ve got the same kit on the shelf ready to go, so I’m extra interested to see how things work.

On a related note, scale helis have been my primary focus in the hobby for a while, and $400 is waaay more money than I typically see multiblade heads go for, especially in the 450 size. If I would have known about this before you spooled up the project, I probably would have just sent you a spare 450 4 blade head. They are usually in the $60-ish range. For any future projects, I put together a guide on where to find multiblade heads on another forum since they tend to be difficult to track down: https://www.hobbysquawk.com/forum/r...main-tail-rotor-heads-450-800-size#post282069.

Looking forward to seeing things come togethe!

Michael

 

[willsonman]

Josh, I’m glad to see things moving on the build again. Definitely the right move to take the time in the planning stage to make sure things go together smoothly! I‘ve got the same kit on the shelf ready to go, so I’m extra interested to see how things work.

On a related note, scale helis have been my primary focus in the hobby for a while, and $400 is waaay more money than I typically see multiblade heads go for, especially in the 450 size. If I would have known about this before you spooled up the project, I probably would have just sent you a spare 450 4 blade head. They are usually in the $60-ish range. For any future projects, I put together a guide on where to find multiblade heads on another forum since they tend to be difficult to track down: https://www.hobbysquawk.com/forum/r...main-tail-rotor-heads-450-800-size#post282069.

Looking forward to seeing things come togethe!

Michael

Yeah, I’m I’ve seen some of those... keep in mind that my $200 includes the cost of a $90 brushless motor. Also note that for my application I have to have the collective controlled by a control rod that must pass through the motor shaft. Given that I’m using this with a spinner, the collective control cannot interfere with the back plate of the spinner. Part of the issue with the off the shelf heads is being able to do all of that in one package.

 

[willsonman]

As I continue to work on other peripheral things, I spent about 5 minutes editing the Fusion360 file for the spinner. The idea here is that because aluminum spinners are stupid expensive and I've seen many other different methods used, my idea is to use a hybrid model of what I've run across.

I've vacuum pulled my own canopies before. It's much easier than you might think but can sometimes be somewhat annoying for various reasons. One of those is making the plug in the first place. In the past I've made the plug from foam and had to do a few layers of fiberglass to maintain the form and provide a sort of heat insulation from the foam. This of course requires the glass work but also the finish work with sanding, priming, sanding, and on and on and on.

I decided to try making my plug by 3D print. My thought here is that I will print in ABS for better strength as well as heat resistance. Instead of my usual meticulous sanding and acetone treatment of the printed part, I opted to use Bondo branded body spackle to fill the layers. I just smeared on a thin layer with a piece of scrap plywood and let it sit for a good hour. I came back and sanded for about 15 minutes (100-grit, then 220 and then 400) and this was my result.


Pretty pleased with that I shot a layer of primer and let it cure for a couple of hours and then another coat to cure overnight. A few notes about the plug. I filled in the part for strength reasons but left recesses for where the blades would protrude so I would have a reference of where to cut. The bottom of the entire part was extended a few mm to allow for the usual rounding of corners from a pull. This, of couse does not entirely factor any failed pulls as a result of webbing of the plastic. This area too is recessed so that I can either rough cut the part off the plug or have a reference line to cut the part once off the plug. The plastic I will use is PETG. I forget the thickness of the stock I have so I'll report that when I got onto pulling the part.

For those thinking ahead, attachment of the plastic spinner will be via screws on the edge of the spinner backplate. This is replicating the spinner type and attachment of the 1/4 scale P-40 build that @wilmracer did a few years back and works well. The benefit of this approach is 2-fold. 1) attachment with the screws allows me to make slots in the spinner so that I can slide the spinner on and off the backplate without having to entirely remove the screws. 2) being able to have 2 spinners. One painted red for flying and one left clear for display of the guts of the blade pitching apparatus. It is a teaching/sharing opportunity that I cannot miss out on.

 

[rockyboy]

I hadn't thought about how freaking cool a clear spinner would be on this! Very cool!

 

[Tench745]

I hadn't thought about how freaking cool a clear spinner would be on this! Very cool!

Same. I am excited to see how this turns out. Do some of the larger spinners ever need a bulkhead in the spinner to register the front of the prop shaft like on full scale aircraft?

 

[The Hangar]

I hadn't thought about how freaking cool a clear spinner would be on this! Very cool!

Yeah, it’ll be the icing on the cake!

 

[willsonman]

Let's go over the transmitter programming for the pressure sensor. I filmed the process and grabbed screenshots so I'll do my best to guide you through this.

First, you need to navigate to your telemetry page and find the sensor by selecting "Discover new sensors"


In my instance, the new sensor came up as "A3" and this is after I've already done subsequent steps so the value is misleading... more on that in a bit.


Selecting the sensor takes you into a screen with finer details. There is no "psi" unit so that is left with a unit-less figure. The first thing you will want to do is change your offset to essentially "zero out" your pressure sensor. Then you begin adjusting your ratio to correct the scaling of the value so that you are accurate in your value readout. This value is in the upper left corner. In this image, 11.94 is my value and I'll explain more on this in a bit.


So now we can apply pressure to the system and using the gauge from the pump, which is not calibrated, my target threshold of where I expect the air to not be able to lower the gear to down and locked. This is around 60psi. So, with the pressure gauge at around this point (~62 here) I start manipulating my ratio until the readout roughly matches. Here the radio reads 65.54 so pretty close.


I work with a lot of calibrated instruments in my industry and there are so many little aspects of measurements that it will really make your head spin. Pressure is one of those measurements that is even harder to do since there are a number of environmental factors that contribute to this measurement. Let's take a moment to cover this in a bit more detail so we understand some values and discrepancies.

First, there is the value of regular atmospheric pressure. There is a blanket statement that at all times you are experiencing 1atm (1 atmosphere) of pressure at all times. However, this is at sea level and should be ~14.69psi there. However, the altitude of where I live is about 300 feet above sea level and mathematically translates to 14.54psi. Further still is factoring weather. Low pressure systems will influence your pressure reading and generally we will be flying on high pressure kinds of days.

But all of this shouldn't matter right since the pressure gauge from the pump is experiencing those same effects? Well, note earlier when I noted that the pressure gauge was not calibrated. Everything that we do in this system is relative to some external factor or force. If we were to try and get THE most accurate readout, we would be chasing our tail. To that end, let's go over a few things from the above images.

The base line readout is not going to be 0psi because this pressure sensor reads ABSOLUTE pressure, which includes the current atmospheric experience. That is why the baseline is around 12psi... obviously from the above discussion this was on a stormy day when a low pressure system was passing though. Probably not 2.5psi lower than normal but again... nothing here is calibrated and everything is relative. Also not the !3psi difference between my readout on the radio and the readout on the pump gauge. When I initially setup the sensor THEY DID MATCH within 1psi. Weather changes everything.

All of that combines to force you into a perspective where making things exact is next to impossible and we are just going with relativistic approaches.

The last part here is the failsafe for the landing gear. Navigate to the logic switch page and we will select A<X where A = A3 (telemetry value of the pressure) and X = 60psi (our target for our low pressure failsafe) and the action we want to take is switch SG as down (landing gear switch in the down position). Even though the physical switch will be in the up position, this will override that physical switch. From here we can integrate both the main gear, tail gear, and the inner main gear doors to sequence during this failsafe operation. This can be done any number of ways in OpenTX so I'll not cover all of that.


I hope this helps clarify several points for you guys.

 

[rockyboy]

Perfect - that helps connect the idea to the menu steps for sure

On the calibration fun, I see some of the challenges there - and completely agree that a relative approach is best for the application. But to geek out for a little bit, the sensor should only be taking readings from inside the closed Robart air system, right? If local atmospheric conditions were influencing the pressure inside the retract system, wouldn't that mean you've got a leak?

 

[willsonman]

Maybe. The differences are going to be relatively minor from location to location. The other factor is that the system will be open during travel and when the airplane is stored. So, wherever she is put together to fly, she will be open to the atmosphere at that point. The plan is to have the pressure canister and control valve in the fuselage. Some quick connects (luer lock) on the pneumatic tubes makes the connections easy during assembly. I did this on my Corsair and it has proven to be very good.

 

[rockyboy]

I wonder.... maybe the sensor measurement is not an absolute measure like a caliper - if it's measuring the relative difference between the pressure on each side of the sensor then local atmospheric pressure will certainly have an impact on the readings.

Feel like building a hyperbaric chamber so we can get to the bottom of this?

 

[willsonman]

Feel like building a hyperbaric chamber so we can get to the bottom of this?

...




no

 

[rockyboy]

...




no

 

[willsonman]

LOL, I'm glad you got the humor in that post. Some would not. At any rate... I'm basically summing up everything you read here but with much more in-depth visuals, which always help, in my video on all of this custom sensor business. It's long, but I lay out literally everything I learned from this project so that you can do it too. Sure, short videos are nice for "entertainment" value. I do not consider myself so much as an entertainer as I do an educator.

 

[willsonman]

Moving on to more vacuum forming, the raw sheets I use are 0.020" thick. I purchased my PETG stash back in 2016 from a company that no longer offers this thickness. I still would recommend that you stick to this thickness for a number of reasons. The first one is that it is widely accepted as the best compromise between durability and work-ability. By work-ability I mean the ability to form something well without too many issues. Webbing at corners is the primary one. You can avoid this by going to a thinner sheet but then you run into durability issues. Like touching the plastic that is rounded can cause an irreversible indentation. If the sheet is too thick then vertical webs can form at the corners or even rounded sides depending on the evenness of the heat applied to the sheet. Generally, PETG starts to soften at around 85 degrees Celsius (185 F). For the purposes of our use, we want the softening of the plastic to be a bit more gradual so that the heating will be as even as possible. The problem is that if you go too slowly, you will start to bake the plastic which looks like tiny bubbles forming inside the sheet. It's unmistakable once you complete your pull. For this reason, I never recommend setting your oven temperature over 2 times the melting point of the plastic you intend to use. So, 300 - 350 F is a good place to be. 300 you risk the melt being too slow and you end up either waiting too long and you start to get baking or at 350 your sheet melts faster than you anticipate and your pull is awful because you do not get a good seal from the plastic being too soft. More on a proper seal later.

I'll refrain from including actual links to plastic sheets as there are a number of considerations here. First, in a year's time when someone looks back at this post, the product may be unavailable or the link is broken. Second, you need to consider the physical size of the forming pane you intend to use. Third, the forming pane size is determined by the amount of heating capacity you have... in other words, your oven size. Some folks go so far as to build their own vacuum table that has a built-in heating element. I'm not that fancy. I have a 12"x12" box made with scrap wood. It's a good size for me to physically handle (remember this entire frame coming out of the oven is HOT) and accommodates the general size of parts I'm willing to pull. Larger parts are far more difficult as are ones with more detail.

That said, I have a vacuum box with a frame made of scrap 1/2"x3" boards as sides with a 3/8" particle board bottom and the top is pegboard. I assembled the bottom and sides with screws, then applied some leftover bathroom sealant (caulk) in the inside corners, then applied more sealant on the top lip and placed the top peg board on. I flipped the whole thing upside down and let it cure out with some weight on top. I used some foam rubber around the perimeter of the pegboard that is picked up usually in the area of the hardware store where window seals can be found. Sticky on one side so it self-adheres. A hole is drilled in the side of the box just large enough to get my shop vacuum hose to snugly fit into it. Next I made a frame out of more 1/2"x3" boards. Some folks like to make fancy aluminum surrounds that clamp the PETG onto the frame. I use a staple gun through the sheet. Yeah, it's annoying to dig them all out once the part is done but frankly, it works and I do not pull parts that often to justify making something overly complex.

 

[willsonman]

SO, I've been slowly making progress on the actual build at this point. The risk you take when buying a kit second hand is that there are missing pieces. I'm telling NOW MATTER WHO you buy it from there is the possibility that a piece or 17 may be missing. Be prepared to improvise a bit. The tail end of this airplane is no exception. Half of the H-stab ribs were entirely missing. Thankfully I could just trace the ribs from scrap balsa and sand them to shape. It's very important to match the rib standoffs very carefully otherwise you may incur some warping or twist.


From there it was just a matter of following the instructions to frame things up.


Sheeting was also simple and per the instructions. A quick tip on this. When gluing sheets together, I use a pane of glass to join them. It is a nice a flat surface and the CA peels off fairly easy compared to a cutting mat. If it is hesitant its also nice to be able to use a razor underneath to persuade the balsa to release. Getting good seams on these sheets is crucial to getting a nice finish so make sure you use a sanding block to smooth out the seam before you install it. I am following scale outlines here so note the notches for the elevator tabs.


What you do NOT see in the above picture is that the tip blocks were ALSO missing from the kit. I grabbed more scrap from my pile and made my own. I was able to make one block by removing a bit at a time with 100-grit sandpaper on a block until it matched the plan profile. I then traced it onto another block and made them match. After attaching these to the tips of the stab I used a razor plane to roughly match the airfoil profile and then used a standing block to taper and do final shaping. I used a separate 220-grit sandpaper block to do a final sand down of the entire part. There are only a couple of minor places that need touchup prior to glass work.

 

[TooJung2Die]

It took me a second to understand the odd shape of the ribs before assembly. Are those "legs" to stand the ribs on so you don't have to spend a lot of time carefully shimming the leading and trailing edges?

 

[willsonman]

It took me a second to understand the odd shape of the ribs before assembly. Are those "legs" to stand the ribs on so you don't have to spend a lot of time carefully shimming the leading and trailing edges?

Yes, a lot of the more advanced kits utilize these tabs so that as you build on a flat board, there is built-in washout, wash-in, or nothing at all. You build the frame out first and then add the top skin. Removing the build from the board after that you cut away the build tabs and then final shape the leading and trailing edges before applying the bottom skin. From there it's business as usual to do final LE and TE shaping.

 

[willsonman]

Josh,
Which SBUS to PWM converter are you using? I've used a couple purchased from Amazon that came from China, and they work well, but I found that the signal is "compressed," for lack of a better term. The issue I've seen is this: A servo connected directly to the PWM output of the receiver (I use FrSky) gets the full -100/100 travel. The same servo connected to the same channel, but via the SBUS converter only yields about -80/80 travel. Have you seen this as well? What is your workaround?

I love the builds! Keep them coming!
--Brian

Brian,

I've finally gotten some time to dig into this issue a bit. The short answer is... The RMILEC converter is the one you want... maybe. Let me break this down into something more comprehensive but digestible.

So, I went after this question with the basic approach of angle. Plug a servo in on the receiver and then on the converter. Compare the angle and you'll get an answer. I was wrong. Immediately I noticed that the SPEED at which the servo changed was notably higher on the converter. It was much more precise in terms of its movement. When I moved my stick slower the servo reacted better on the converter. I thought that this must be somehow due to using an old servo, so I broke out two brand new servos for this build and decided to directly compare them side-by-side. It showed the exact same thing. I even went so far as to take a slow-motion video on my phone to look closer at the reaction time. Decoder was for sure reacting faster with more precision.

It did confirm that there was a sharp difference in servo throw.

I was able to get my hands on the HiTec HFP-25 digital servo programmer and tester. Nice thing is you can put it on your receiver and see what PWM pulse (measured in micro seconds [µs]) it is spitting out. I'll report the data here:

Model	Source	Center (µs)	High (µs)	Low (µs)
Top Flite P-51	RX8R Pro	1500	2011	987
Top Flite P-51	QwinOut Converter	1498	1907	1088
Hangar 9 P-47	X8R	1495	1698	1271
Hangar 9 P-47	RMILEC V3	1493	1696	1270

I opted to NOT pull out all 13 channels of the P-47 and just note that I have some trim (center not being at 1500) and rates (High and Low not being very high or low) applied for this model. The point is that we are talking about a difference of just a couple of µs here on the RMILEC V3.

Now if you look at the data on the P-51 setup we are at full 100% throws and no trims (yet). The centering does not seem too bad but then a the high and low we seem to be off by about 100µs. That can certainly account for the discrepancy.

Moving forward this may not pose an issue as I generally prefer to NOT use the extremes of my servo throws. Further still, I will note that the throw angle coming off the RX8R Pro was definitely greater than 30-degrees off center. The decoder was right at 60 degrees end-to-end. The industry standard for servo deflection (non retract servo) is 60 degrees of movement. This means that generally you are getting more throw by default when using an FrSky radio? Sure, the PWM pulse says so. Is that great for a longer life of your servo? Probably not would be my guess. Is there a discrepancy between the SBus and PWM signals? Yes, in this particular instance. I cannot confirm that this will happen on EVERY receiver with EVERY radio.

Some food for thought here. I'd love to know anyone's thoughts and see how our taste buds align.

 

[PsyBorg]

Two thoughts on this...

First.. Does the converter use a higher resolution for instance 2048 vs 1024? It would still travel the same total range set by end points but would have double the steps to get there thus giving a noticeably more smooth movement specially on digital servos.

Frsky sets the default ppm range at 1000us to 2000us +- the tolerances. Anything open tx does this I believe. Where Spectrum and Flysky gear only use ~80% of that as their defaults. I assume this is to make sure servos dont slam and get stripped out or locked because they can at the 1000-2000 range go past physical limits of some servos.

The 1000us to 2000us is more for newer gear like flight controllers to use their full range and give best resolution. This is why so many people on non open tx gear struggle with flight controller set up and super twitchy controls. A lot of the times the range is too small to get between flight controller minimums to unset the arm flags thus locking them out or shortening the resolution thus making it twitchy.

Like you said for servos and such you should set your own end points and centers for your needs to get the best resolution but not over drive a servo out of the box. I also think this is one of the reasons radios are now using scripts to choose between multi rotor, fixed wing, and heli set ups One of the things in the scripts is that variation in control range default settings. Max for FC's but not so for fixed wing and heli.

 

[willsonman]

I could be wrong but I believe that the input and output are what determines how crucial the 2048 vs 1024 debate may be. For example: my Horus gimbals are actually 4096 bit. For those who are unaware of this situation, this means that the extent of the range of movement for one axis of the gimbal is divided up by 4096 steps. This makes control VERY precise. It is a feature that currently only exists on high-end radios like Hitec, Multiplex, and top line models from Futaba and Spektrum. I'll note that some people do not even notice this kind of resolution but if you are a person who does fly precision maneuvers on only ONE airplane, you get so familiar with an airplane that this level of precision really does become noticeable.

This level of high-precision input all becomes an entirely worthless point if you do not have the output to match. Unless you are using some digital servos that extend beyond the average hobby budget, you really will not see much difference. Analog servos are VERY limited to their precision and when you get smaller servos, the problem gets worse because of physical gear tooth engagement. Smaller servos with high precision get very expensive fast because of this limitation and you go to the coreless direct-drive stuff. Essentially you end up with a miniature stepper motor that uses a digital signal.

The point of mentioning all of this is that your initial point about resolution, I think, is not entirely relevant because the signal automatically encodes the precision of the input, transmits it, and then spits it back out. The fact that the SBus converter is simply an additional relay plays in when it simply converts the digital SBus signal to an PWM signal. The interpretation of that PWM signal is done by the servo itself.

So, all of that to say this is a point-to-point system. I will say that I believe that this converter issue may have a filter on the output that does not exist on the other. This filter would limit the range of the signal a bit but the extent of that is TBD. Could I get more travel from endpoint adjustment? Not sure. Would I get the SAME range if I used another protocol? Not sure. It's all speculative so it's best to just accept the KNOWN parts of the system and work with the limitations you have unless you are so inclined to investigate.