CAL Multirotor Blog

I have a quite a few different ESCs lying around.  Most ESCs have something called a "BEC" that converts the battery voltage to a steady 5V for the use by the controller, most of the radio equipment, the GPS, and by the ESC's themselves.

Some are BECs are listed as analog, some are listed as digital, some are listed as "opto".  What's the difference?

The "Opto" designation was first used with ESC that had an opto-isolator on their control input.  An opto isolator can allow one device to control another - even if their GND potential is different.  That can be important if the primary (main power feed) wires to the ESC are long or are of too fine a gauge to handle the currents required and have a high voltage drop (> 1V).  I have not seen a problem with this in quadcopters. If you happen have that much voltage drop in your power feed lines (at full throttle), you are wasting an awful lot of power and you should rewire immediately.

The ESCs that I have purchased recently that have been called OPTO do NOT have an opto-isolator on board. To that manufacturer, (and probably most others), OPTO simply means that it doesn't have a BEC at all.  A quad with 4 of these ESCs needs some other 12V -> 5V converter somewhere simply to fly at all.

Some of the ESCs have an ANALOG BEC.  The one I'm looking at right now claims 3Amps output.  Do I believe that -NO!.  A linear voltage regulator has an input current equal to the output current even though the voltages are different.

So, if an ANALOG ESC has an input voltage of 12V and an output voltage of 5V, it has a "drop" of 12V - 5V = 7V.  If you take the output current times that drop, you get the power dissipation of the regulator in the ESC.  If the output current is 1A, the power dissipation is 7 Watts!!  That is more than those little regulators can handle.

The problem is much worse if we use a 4 cell battery (16V).  Now the voltage drop is 11V and a 1A load causes a dissipation of 11Watts.  This ESC will melt (or the regulator will most certainly fail.

If a digital or switching ESC is used, the regulator drops the voltage with an efficiency of around 85%, so if the output current is 1A, the output power is 1A X 5V = 5 Watts X (1- 0.85) = .75 Watts.  This is a level that can be sustained.  The best part is that when we increase the input voltage from 12V to 16V, the output stays the same, and the efficiency is nearly the same, so the dissipation will STILL be around 0.75W.  

I can't stand having a pile of junk lying in a box waiting for repair.  I have to drop everything and fix the broken item.

So, I fixed my little 6" (actually 6.5") prop quad (picture below)

And I'm nearly finished fixing the 16" prop "Santa Cruz Project" test bed. I may get a chance to try it out this afternoon.

Tomorrow at 8:00 AM, I'm hosting a MEETUP at Cataldi Park in San Jose.  I'm expecting at least 15 attendees (could be up to 30).

Leave a comment if you need directions.

Building a multirotor "from scratch" is a daunting prospect.  I would recommend that you first buy a kit from a a company such as Hobby King. 

But, If you are going to design and build your own machine from scratch, you will need some tools and supplies. The list can be intimidating, but you would be surprised how most of the stuff listed below can be used for a lot of other household and hobby tasks as well. Although you can get by with less than the extensive list below,having everything listed will make your job a lot easier and save time waiting for deliveries from China.

Two soldering irons (or a temperature-controlled one with interchangeable tips).  One should be high-power with a large tip (for 10-12GA power leads). The other should be 35W or so, with a fine tip, suitable for PCB work.

Solder

Two wire cutters (dikes). One should be small for cutting 18-30Ga wire, the other should be large enough to easily handle 10GA.

Small needle-nose pliers. Most of the ones you find will be too big. I like the ones that are about 4" long overall.

3mm socket head capscrews of all lengths. 25 of each size - 5mm (thread length), 8mm, 10mm, 15mm, 25mm, 30mm, 40mm.

Lots and lots of 3mm nylok hex nuts

7/32" nut-driver (works fine with 3mm nuts)

2.5mm hex wrenches - buy 10.  You will use them for nearly EVERYTHING.

Small metric hex wrench set (so that you will have all the sizes that aren't 2.5mm).

English hex wrench set with .050 as the smallest size (believe it or not, some Chinese equipment is built with English-sized setscrews).

3/8" and 1/2" aluminum angle and channel stock (Home Depot or McMaster-Carr).

Hacksaw with fine teeth.

Dremel tool with assorted abrasives/cutters.

Electric drill, preferably one that can be mounted in one of those cheap drill-press stands.

One of those cheap drill-press stands.

Small center punch

XActo knife

Metric drill set. (or a very complete English set).  Make certain that the smallest bit is no larger than 2mm (.079")

Some very limp 12GA and 20GA wire in red and black (silicone insulation)

3.5mm "bullet" connectors, male and female.

2 pin JST connectors is both genders.

MT60 connectors in both genders.

Shrink tubing in all sizes.

Heat gun (air).

Hot-melt glue gun.

Digital calipers.

1/4" Tapered reamer

Voltmeter.  A cheap one ($10) will do.

Lots of small tie-wraps

Double-sided foam tape

Adjustable square.

Some .050" or .060" aluminum sheet 

Some .063" bare fiberglass sheet (bare PCB material).

A 3D printer is not necessary but very useful

Patience!

It should be noted that I have everything listed above except the last item, and I'm still hurting for a few tools, like a sheet metal shear and a metal bending brake. I can borrow those for the time being, but I'll eventually need to buy my own. 

I'm just going to have to learn the patience part on my own.

.

When wiring up a multirotor, you often find that there is a need to splice some very heavy gauge wires together (like connecting 4 ESCs to a single wire to go to the battery). 

If you twist all those wires together to solder them, you have a big mess.  The wires are so large that you need to strip at least 1" of insulation off the wires just to get enough bare wire to make one "wrap".  And one wrap is barely enough to get them to hold together. And if you move them, they fall apart. Not to mention that the resultant connection looks like a knot and is as big as a marble.

So what is the solution?  Buy a spool of solid 28Ga or 30Ga Kynar-insulated solid wire - also known as "wire wrap wire" and strip off the insulation from a piece 5" or 6" long.  Put the ends your heavy gauge wires together - just lay them side-by-side.  Don't twist them together.  Now take the 30Ga wire and wrap it tightly around the bundle to hold everything together.  If it makes things easier, wrap your 30Ga wire several times around just one wire.  Now lay the second wire next to it and wrap both wires. Add a 3rd and wrap and  a 4th and wrap ... Solder.  

Now you have a neat, well-soldered connection.

It is not uncommon to find yourself needing to splice several small wires together and put heat-shrink tubing around them for insulation.

Many times, you need to splice some wires that are short.  If you cut your heat-shrink tubing long enough to cover the splice you may find that - when placed on the wire,  you can't get one end very far from the splice that needs to be soldered.

I find this happens when I want to shorten and connect several ESC power-feed wires together.  I cut the heat-shrink tubing 1" or so long, and slide it over the wire.  But the wire itself is only 1.5" long.  Now, when I solder the wire, the heat flows through the heavy gauge copper wire and heats the heatshrink tubing to the point where it shrinks. 

So what is the solution?  Get a pair of medical hemostats.  You might need some larger ones for the larger gauge wire, but clamp the wire with the hemostats between where you are soldering and the end of the heat-shrink tubing.  The hemostats will act as a heatsink and prevent the tubing from shrinking before you want it to.

I wanted to repeat my "wiring tip #2" again, and with a bit more information.

The technique works REALLY well!

Today, I had to make a wiring harness that connected a 12GA battery wire to (quan) 4  14GA wires (ESCs).

The technique below mentions 12GA and 14GA wires, but it will work for any wires, small or large.

Strip about 3/8" insulation off the ends of all wires and twist each one tight. I recommend that you do not 'tin' them, and do not twist the several ends together at this point.

Strip about 4" of the insulation off some 30Ga KYNAR insulated wire (solid wire commonly used for wire wrapping).

Hold the 30GA wire (the part that still has insulation) in your hand and wrap 4 turns of the bare 30GA  around the 12GA wire. Then take one of the 14GA wires and lay the stripped end next to the stripped end of the 12GA wire (but in the opposite direction - overlapping only the stripped sections.  It may help to use a medium-sized hemostat to help hold them together (but it can be done without a hemostat). Now wrap the remaining 30GA wire around both the 12G and the 14GA wires.

Strip some more 30GA wire.  Wrap 3-4 turns around the splice and add a second 14GA wire.  Now wrap the 30GA wire around all 3 wires.  Do the same for each wire you have. 

Now solder.  Because the wires were not previously tinned, they will take a lot of solder. I recommended that you do not pre-tin the wires because the solder adds bulk, and will prevent the wires from fitting tightly together, resulting in a larger joint.

I think you will find that this technique works remarkably well.

I'm finally getting back on track with the Santa Cruz project.  The frame needs to be very strong, yet very light. I have settled on the "Box - X" design, that is, the frame is a normal "X", but the arms of the "X" are fastened together with small diameter carbon fiber tubing.

By connecting the ends of the arms, the diagonals (the "X" part) can be much thinner while still maintaining overall strength and rigidity.

The craft will need a "landing gear", so I plan on extending short legs down from the end of each "X", and to add strength to those - as well as further preventing twisting of the arms, I'm adding a very thin carbon fiber "X" between all the legs.  Since these "arms"are - in effect - trusses, they can be thin while being strong.

I'm planning on the main "Box X" part (3 pieces to make the "X", and 4 to make the "Box") from 6mm round carbon-fiber tubing. The legs (approx 6" long)  will also be made 6mm carbon fiber tubing, while the "X's" that connect the legs will be made from 3mm carbon fiber rod (solid).  4 more pieces of 3mm carbon fiber rod will connect the bottom of each leg back to the center platform of the copter. These additional supports are needed to handle the 5-6 lb weight of the batteries, which will hang under the center of the craft. Carbon-fiber disks glued and screwed to the carbon fiber tubing will act as motor mounts and a few 3-D printed plastic pieces will aid in holding things in place. An acquaintance who is very familiar with (real) aircraft frames and carbon fiber fabrication and has a machine shop has tentatively agreed to provide help.  I'll need it!

At least three of the 4 batteries that power the craft are mounted vertically in "chutes" made from carbon fiber angle stock.  A nylon string (fishing line?) connected across the chute under each battery holds them in place. A 26GA nichrome wire is twisted around each nylon string and the ends of the wires brought to the battery controller. The fact that the batteries are mounted vertically explains the quite long length of the legs.

The battery controller consists of a Microchip PIC 18F2321 uC.  4 outputs are routed to 4 N-FETs acting as a level shifters.  The drain of each of the N-FETs is connect to a 60A,40V P-FET held in the "OFF" state with a 10K resistor. Each battery is connected to the Source of a separate FET. The drains of all FETs are connected in parallel and feed power to the craft's main power rail. 

4 channels of 10-bit A/D (in the 18F2321 uC) are each connected to a separate battery. When initiated by a trigger signal, the uC brings the first output high which turns on FET #1 and battery #1 powers the craft.  When battery #1 voltage drops below a set potential, the 18F2321 brings output #2 high.  This has two effects:  It turns on the FET connecting battery #2 to the main power rail, and it also feeds power to the nichrome wire wrapped around the nylon string.  Within a second or two, the heat generated by current flowing through the nichrome wire melts the  nylon string and battery #1 is no longer held in place.  It drops down the (short) chute by means of gravity and a short loop of wire allows the battery to gain a bit of momentum before it "yanks" on the 3.5mm bullet connectors that connect it (electrically) to the craft.  The disturbance from the "yank" is small enough, and of short duration, so the craft simply shudders and continues running on battery #2.  This process repeats for battery #2 and battery #3.  Battery #4, of course does not need to be disconnected!

I'm laying out the PCB for the above controller right now.

I have the basic airframe designed in AutoCad, and a friend is converting the design to SolidWorks.  SolidWorks has a FEA 'add on' that will show strengths and stresses.  I will use that to refine the airframe.

I need to join a "10 step" program - MULTIROTORS ANONYMOUS.

I am in the process of building a small (9" prop size) folding quad, but I wanted to try out some ideas. 

So I built this 10" prop foldable model as a "test bed".  The diagonal is a full 500 mm, and as you can see, it folds down to shoebox size.

It has 

6 channel reciever,

KEDA 28-15  1050KV motors

10" APC props

CC3D controller

Cheap HobbyWing 40A ESCs

3 Cell 3850 mAH battery

The frame has a few plastic parts from THINGIVERSE (3D printed by me) and arms made from 3/8" X 1/2" AL channel with 6mm square AL tubing (in the channel) for extra stiffness.  The body is simply 2 pieces of 3/8 AL channel with "expanded AL (.060") forming the top and bottom of the body.  The legs are short pieces of 6mm square AL tubing.  The whole thing (with battery) weighs 3.2lbs. 

I have had it a few feet off the ground and it appears that it wants to fly.  I'll find out in earnest tomorrow.

When building small model aircraft, it is often necessary to provide an insulation barrier between a PCB and other components or between a PCB and a metal frame. That insulation barrier must be thin, light and puncture resistant.

And sometimes you must prevent a nearby screw from cutting through heat-shrink tubing or wiring insulation

So what is the best material?

I have found that old credit cards or old gift cards or stolen hotel key-cards are absolutely excellent. They are thin, lightweight and very puncture-resistant.

They are readily obtainable and can be cut with scissors or drilled, and most (flexible) glues will stick to them as well. If you look at my aircraft, you will probably see a Mileage Plus card under the flight controller, or as protection for the ESC wires.

When I check out of a hotel, I always have two new fresh cards to use for my "builds".

I wanted to explain my concept of the frame for the Santa Cruz Project in more detail.

In the picture below, the RED frame members are 6mm round carbon fiber tubing

The BLUE frame members are 3mm round carbon fiber rods (not hollow)

The LIGHT BLUE frame members are also 3mm carbon fiber rods.  They are shown in a different color because they extend from the ends of the "legs" back up toward the center of the craft..

The square in the center of the top view is the electronics platform ( I realize now that I'm going to have to make it a bit bigger).

The yellow in the side view are representations of the 4 batteries in their "chutes" made from thin carbon-fiber angle stock.

Not shown are:

The motor mounts made from carbon fiber.

The ESCs

The wiring tie-wrapped to the arms carrying power to the motors.

The 3-D printed parts that form the joints at the junctures of the carbon fiber pieces.

A test model cobbled together from 1/8 and 1/4" wooden dowels shows that the basic design of the airframe is very strong and stiff.Carbon fiber is stronger and stiffer than hardwood.