DJI Mavic, Air and Mini Drones
Friendly, Helpful & Knowledgeable Community
Join Us Now

Mini frame mod for experienced builders

fGene

Well-Known Member
Joined
May 31, 2020
Messages
101
Reactions
166
Age
52
Location
Long Island, NY, USA
Hi everyone.

Let me describe the modification of Mavic Mini that I made during this winter, for which I already received requests from people who visited my introduction thread.

The main purpose of the mod was to make MM more resistant to wind, and consequentally less inclined to fly away in certain weather conditions. At this point, I think every Mini owner should be well aware of this problem. I believe I achieved some success in this regard, but the level of improvement is very hard to assess without a wind tunnel, where the wind is unidirectional, and the speed of it is adjustable and precisely known at any moment.

Initially, I thought I would be able to get the needed numbers for analysis by comparing the mean-maximum speed of the drone before and after the mod. Unfortunately, after many tries I didn't register any meaningful difference, the speed maxed in all configurations at around 29 mph. At this point I am inclined to believe DJI has capped the maximum speed of the Mini in software, relying on GPS data. Because of this, I cannot provide any hard, verifiable data which would back my claim of improvement. However, I have flown both original and modified Mini when UAV forecast was showing 15mph wind, and noticed this: while the maximum speed of the drone varied between 28 and 30 mph in both configurations (always drifting towards 29mph, as if PID-regulated), the modded Mini was significantly faster, up to +6 mph, when flying against the wind. Of course, these tests were months apart, taken at different temperatures, cloud cover etc., so please take them for what they are. They are not scientific.

Here is an attempt to summarize the pros and cons of this mod:

+ Better wind resistance -- improved aerodynamics and frame rigidity (hard to quantify, see above)
+ Longer flight time -- up to 32 min measured on stock battery, due to 12g weight reduction and improved airflow
+ Improved low light visibility -- 2 front-mounted 200mW LEDs
+ Retains foldability and compactness
+ Unique look (and bragging rights).

- Very hard and time consuming to do properly, even for an experienced modeler
- Voids the warranty
- Requires a CNC milling machine
- Potentially makes crashes more damaging to framebase by eliminating arm flexing (not tested)

Disclaimer: here I assume that people who would attempt this mod are quite familiar with tools and know how to disassemble the Mini without damaging it. And, in general, realize what they are getting themselves into. If anything goes wrong (a lot might!) I am not the one to blame.
If you are still interested, please read on...

Take a look at the Mini in its original form. It is a well-engineered machine as is, rigorously tested by DJI, and they obviously have much better means to ensure it performs up to their quality standards, which are quite high. Can we modelers possibly do better then them? I think yes, sometimes. If you have your Mini at hand, please unfold it and look it in the eye (pardon, camera). This is the side which faces the wind. In my opinion, it doesn't look very aerodynamic. The overall area against which the wind pushes the Mini away from you, the pilot, is too broad mostly because the arms are too thick. The problem is, they cannot be made thinner with the plastic DJI chose, because they would become too flexible. Too much flexing in the arms would severely impede flight characteristics, vibrations would increase and become unpredictable, and the IMU would have to work much harder to compensate for changing geometry. DJI could have used a different material for arms, but that, in turn, would increase either the weight of the drone or manufacturing costs to unacceptable levels.

Thankfully, we modelers are not restricted much by material considerations. So the first obvious area of improvement, to me, was the arms. To make the new ones, I used high quality Japan-made woven carbon fiber tubes 7mm in diameter. Not that I chose this specific size deliberately, they just happened to reside in one of my toolboxes since ancient times. They have a round cross-section which is by itself an advantage (aerodynamically speaking) over almost rectangular cross-section of Mini original arms, and they are also about 2 times thinner. Which means less resistance to both wind and prop wash, thus improving the efficiency of the propulsion system as well.

After the tubes were selected, I had to design motor mounts and pivots, and that was a nightmare. I can't tell how many times I had to revise and rebuild the whole thing from scratch. One of the photos I attached to this post, that shows some of the discarded parts which were all tested in flight and found to be inadequate in some way or the other, will give you an idea. I didn't include the parts which were discarded straight away because of some defect. Next I will try to describe some of the considerations and problems I encountered.

First of all, my material of choice for all plastic parts was nylon. It is very strong, retains its shape very well even when thin-walled, is relatively thermostable (motors could get hot enough to melt PVC, for instance) and most importantly, I had lots of nylon slabs laying around in my workshop. Unfortunately, it is also a very tricky material. It is milling-unfriendly, and sanding-unfriendly (and painting-unfriendly, as it turned out later). When cutting nylon with a CNC, the endmill has to be very sharp at all times, or the dimensions could end up way off. I only have a hobby-grade inexpensive 3-axis milling machine, and almost each part had to be made with several consequent setups (like this: cut a block with known dimensions, secure one side in angled vises, adjust angle, measure, cut, secure another side, adjust angle, measure, cut, secure third side, adjust angle, measure, cut, find the dimensions are off because the endmill dulled, discard, begin from scratch etc.)

Secondly, concerning pivots, which have to sit tight in sockets within Mini's original plastic frame and rotate freely without damaging the frame. Sometimes I think DJI deliberately attempted to make the frame un-moddable (which might be not far from the truth). No straight lines, odd angles, unexpected curvatures everywhere. In short, impossible to measure with a ruler or calipers. No matter how hard I tried to get the dimensions right, I always ended up using a Dremel and sandpaper afterwards. As I mentioned before, hardened nylon resists sanding very well, so it is quite a tedious and frustrating work. An unfortunate consequence of it, I can't provide any final dimensions here because, frankly, I don't know them myself. Please rely on photos which I tried to make as descriptive as possible.

Here is some info which concerns things not obvious from the photos, and some other things which are very obvious and are likely to raise a question.

The front motor mount holds antenna frame which doubles as landing support. The frame is secured in the groove of the mount by a single screw. That screw, if metallic, needs to be as far away from the actual antenna as possible. I made the frame from a piece of fiberglass, glued antenna inside with double-sided tape and put a piece of shrink wrap tube over it. It is important to use only fully dielectric and radio-transparent materials around antenna. Ideally, it should be either completely sealed from moisture, or the opposite, be open to fresh air enough so that moisture can evaporate fast. For the frame, fiberglass should be clean without any copper traces left, carbon fiber must be avoided altogether. I found that black heat shrink tube reduces the range (maybe because it has carbon particles as coloring?), hence mine are currently white. Performance over looks!

The front pivot. This one is particularly troublesome to get the dimensions right. I used the original pivot spring mechanism within my nylon frame. The mechanism is quite strong, so be careful that it does not damage the original frame when inserted or pulled off. Pay attention to the shape of the front portion of the pivot socket where the front arm rests in unfolded position. It is about 3 degrees off vertical, which might not be obvious at first glance. If the shape of the pivot doesn't match the socket in this position, the frame will get deformed by the pressure of the arm and you will end up with asymmetrical Mini (which will fly asymmetrically, too, if at all.)

For the rear pivot, I did not use the original spring-loaded mechanism because it provides only two stops 180 degrees apart, and my rear arms needed to rotate about 205 degrees from folded to unfolded position. Actually, I did not want to use them anyway because a pair of them weighs 5.5 grams! They absolutely don't need to be this heavy. So I devised my own, completely made of nylon, which are supported by a nylon clip running in a groove in the pivot cylinder. That nylon clip acts as a (pretty weak) spring, supporting the arms at takeoffs and landings. A tooth prevents the rear arms from rotating past high point. During flight, thrust from motors naturally prevents the rear arms from folding. My pivots weigh 1.4 g together with clips, thus saving more than 4 grams on rear pivots alone. Not good, DJI!

The rear motor mount. Oh, these angles... 22.5 degrees off motor rotating plane, if I remember correctly, same as the rear arm itself off the pivot.

I need to emphasize that all linear dimensions and angles need to exactly mirror each other on the left and right side, otherwise the drone will not fly straight and will become almost useless as a videographic tool. The CG of the drone (with the battery inserted and gimbal guard taken off!) should be right at the centre between all four motors.

The substance I used to glue mounts and pivots to carbon tubes is called "J-B Plastic Bonder High Strength Structural Adhesive". It has a 15 minutes cure time, enough to make sure the parts being glued are properly aligned, and it stays a bit flexible after curing (unlike most epoxies) so it will not break in any crash. Highly recommended.

As for the front LEDs, this is the easiest part. There are pockets on the frame where front obstacle avoidance sensors should have been (but are absent on the Mini), covered with a sticky black pieces of plastic. Remove them. Take a sticky LED strip, put a plastic piece you just removed on it so that it covers a single diode and use scissors to cut it from the strip, following the contour of the plastic piece. Repeat with another diode, using the second plastic piece. Solder two tiniest insulated wires you can find to each of them, then glue them to the frame exactly where you removed the black plastic pieces from. Solder them in parallel, so that you have only two wires to deal with -- it is easier and saves weight. These two wires go to the back to ESC cluster PCB, which on top has a free pad labeled "3V6". This is where LEDs are powered from. A free ground pad is also there, on the other side of the screw that supports the IMU. I don't think I have to explain the rest... It's pretty easy, really.

Finally, I would like to explain why I decided to rotate rear arms so far upward, compared to original Mini. The simple answer -- it flies a bit better. As to why, I believe the main reason is that in such configuration neither prop is in others propwash, which reduces turbulence and noise, and improves efficiency. When rear props are lower than the front ones, and the drone flies forward, the wash from the front propellers is shifted backwards by the wind and disturbs the airflow over the rear ones, making them less efficient and more noisy. This is noticeable only if you know what you are looking for, definitely nothing major (all Mavics have such configuration and fly well), but why not try to make good things even better if we can? Another advantage of my configuration over the original, is that it can take off where the original Mini can't: where there's short grass or something that might entangle the rear props if they are too close to the ground.

I can't cover everything about this mod in a single post (it is already too long!), but I will try to answer questions if there are any. Thanks.
 

Attachments

  • mm_assortment.jpg
    mm_assortment.jpg
    219.8 KB · Views: 696
  • mm_back.jpg
    mm_back.jpg
    65.5 KB · Views: 705
  • mm_folded_bottom.jpg
    mm_folded_bottom.jpg
    88 KB · Views: 683
  • mm_folded_front.jpg
    mm_folded_front.jpg
    62.7 KB · Views: 665
  • mm_folded_general.jpg
    mm_folded_general.jpg
    62.3 KB · Views: 653
  • mm_front.jpg
    mm_front.jpg
    67 KB · Views: 666
  • mm_front_motor_mount.jpg
    mm_front_motor_mount.jpg
    98.3 KB · Views: 659
  • mm_inflight_day.jpg
    mm_inflight_day.jpg
    75.4 KB · Views: 680
  • mm_inflight_night.jpg
    mm_inflight_night.jpg
    38.3 KB · Views: 676
  • mm_packed.jpg
    mm_packed.jpg
    170.5 KB · Views: 634
  • mm_rear_motor_mount.jpg
    mm_rear_motor_mount.jpg
    60.6 KB · Views: 597
  • mm_rear_pivot.jpg
    mm_rear_pivot.jpg
    66.9 KB · Views: 592
  • mm_rear_pivot2.jpg
    mm_rear_pivot2.jpg
    82.6 KB · Views: 593
  • mm_side.jpg
    mm_side.jpg
    71.7 KB · Views: 622
Impressive work, very nicely done. Do you think the nylon parts could be 3D printed? Nylon itself is 3D printable, although it's not always the easiest to get good results with. PETG is almost as strong, and pretty easy to print, so maybe that could work?

If the plastic parts were 3D printable I suspect it would make this much easier to do.
 
Impressive work, very nicely done. Do you think the nylon parts could be 3D printed? Nylon itself is 3D printable, although it's not always the easiest to get good results with. PETG is almost as strong, and pretty easy to print, so maybe that could work?

If the plastic parts were 3D printable I suspect it would make this much easier to do.

In general, CNC milling machines are orders of magnitude more precise than 3D printers, so I don't think it is a viable option for modelers. We are talking about very small parts, made with tolerances of about 0.05 mm or less. I have yet to see a 3D printer approaching this kind of precision.

Secondly, parts cut from bulk material are usually stronger and much more reliable than their printed equivalents of the same weight. They are completely solid, meaning, they lack microscopic air pockets which are usually present in printed parts. This is very important for flight safety.

Also, I am not sure that for motor mounts I would use any plastic that could soften and change shape at 100 C or less. Most filaments used in 3D printers are not thermostable. Nylon is one of the exceptions, but it is not printer-friendly. I don't know about PETG...
 
This is very interesting but I have a much simpler question - do you know what the washer is for from the original kit? I removed a rear rotation axis and it ended up on the bench - I don't know where it came from? About 5mm diameter - slight concave. It has a round centre and slightly flattened sides
 
  • Like
Reactions: LoudThunder
This is very interesting but I have a much simpler question - do you know what the washer is for from the original kit? I removed a rear rotation axis and it ended up on the bench - I don't know where it came from? About 5mm diameter - slight concave. It has a round centre and slightly flattened sides
How about a picture and which drone?
 
Attention to all people who are interested in buying the kit!

Guys, I feel flattered and overwhelmed simultaneously. Please understand that this mod took a lot of my time (looks like I failed to get this important point across in my initial post), and currently I am in no position to meet any kind of demand. With the current manufacturing process which I described, I simply can't justify spending that much of my time. Even if I am paid a minimum wage for every hour I spend building a kit, the final price will be at least several hundred US dollars, more than the Mini itself, which I simply can't ask of people. I need to develop a new, highly optimized manufacturing process to go down this route. Currently I am looking at my options and will update this thread if something changes. It would help to know how many people would actually buy such a kit, and how much they are prepared to pay for it, so please feel free to comment. Thanks for understanding.


I see that the rear props look kinda tilted when viewing from side- are the flight characteristics different from stock in any way?

Yes, this is intentional. The original frame also has it to somewhat lesser degree. Such configuration helps in separating thrust vectors, which leads to better overall stability. It also makes forward flight (main mode of operation) more efficient, as the energy is distributed more evenly between motors. On the downside, it slightly reduces efficiency at hovering, but it is almost negligible.

This is very interesting but I have a much simpler question - do you know what the washer is for from the original kit? I removed a rear rotation axis and it ended up on the bench - I don't know where it came from? About 5mm diameter - slight concave. It has a round centre and slightly flattened sides

I really have no idea. The original rear pivot mechanism cannot be disassembled, to my knowledge. It consists of a black flat plastic part which is secured by 3 screws to the frame, and a metal part which rotates on top of it. The latter has a spring mechanism inside and two "ears" with threaded holes to secure the arm.
 
Last edited:
Yes, this is intentional. The original frame also has it to somewhat lesser degree. Such configuration helps in separating thrust vectors, which leads to better overall stability. It also makes forward flight (main mode of operation) more efficient, as the energy is distributed more evenly between motors. On the downside, it slightly reduces efficiency at hovering, but it is almost negligible.

Awesome, yeah, I owned a Mini before and was aware of the tilt. might be a visual thing, but the rear props on your mini seemed a bit more tilted and was wondering if the flight was affected for the same reasons you mentioned. It is great that you got such a mod to work; personally I would be freaking out to attempt such a thing on my MA2 (I am a grad student, so money isn't something I make a lot. I wouldn't be able to forgive myself if I screw up).

You also mentioned that you are looking for a more efficient way of manufacturing this mod. If you are looking to mass-produce those individual components, I would say injection molding is the way to go. If you are looking to produce only a few, a desktop mill like the ones offered by Carbide3D might be your best bet. Carbide 3D

From personal experience, this machine cuts aluminum, let alone plastics. The accuracy is phenomenal too ( I think in the order of 0.5 thou, the TDS will give you more info).

Keep up the good work!
 
why not a mini on a new frame using only motors and electronics?

Because... Have you seen the original frame from inside? I simply can't make something as light and complicated, with means to house all the electronic components so that they work well. Also, we are talking diminishing returns here. The arms mod can offer significant improvements to flight characteristics, while a "full frame" mod --- not so much. I am mostly after performance improvements, the looks are secondary in my book. While I agree that the frame can be improved aerodynamically, it would be a major project in itself, and I simply don't have time for it at this point.
 
Last edited:
That is awesome. Bro you could seriously create this in bulk and really only do the work once. Take the parts you hand made and have them machine done as the person above suggested. Once each part is scanned and precise digital “blueprint “ for each part is Done you can put a chunk of raw material and let it make itself.. id be curious about different materials and you could expirment . Put the package together with an instructon manual and these would fly off the shelves. Question.. would you be able to revert back to stock arms if you wanted? Also what would be stopping you from upgrading the motors .. lol I literally can see a whole skew of variations.. different material options mabye even different lengths with possibly larger motors. Different colours.. you would definitely sell a lot more if it
Could made where it is possible tp revert
Back to factory stuff if you want... i wonder if it is possible to change the body shell for a lighter material and a slicker body lines because if you could get upgraded motors and shed enough weight to keep it 250grams.. a diy kit new motors with more power and new cool look that’s aerodynamic “laughing all the way to the bank “
 
That is awesome. Bro you could seriously create this in bulk and really only do the work once.

I wish it was so.

Take the parts you hand made and have them machine done as the person above suggested. Once each part is scanned and precise digital “blueprint “ for each part is Done you can put a chunk of raw material and let it make itself.. id be curious about different materials and you could expirment .

I already have all "blueprints", meaning CAD/CAM files. I did machine these parts, I wouldn't even consider doing it manually from scratch. The problem is, there are different kinds of machines, and to produce such parts on a scale you need some seriously expensive ones. Such as a 5-6 axis milling machine with automatic tool changer, or some advanced automatic injection molding machine. Neither of which I can afford. So one of my options is to find a manufacturer who has this kind of equipment and wouldn't charge much for low-volume production, but that's not an easy task either.

Question.. would you be able to revert back to stock arms if you wanted?

Yes.
 
  • Like
Reactions: Hauptmann
The round parts (motor clamps etc) indeed need 4 axes to machine in their current state. If you can change the design to something that requires only three axes, the desktop CNC mill (about $2.5k) will be more than what you need.

Or maybe even SLS 3DP technology?
 
  • Like
Reactions: Hauptmann
Lycus Tech Mavic Air 3 Case

DJI Drone Deals

New Threads

Forum statistics

Threads
130,865
Messages
1,557,273
Members
159,889
Latest member
MadMaurizius