Radio controled torpedo experience

[jlcrl]

Hello,
I thought cool to have a model torpedo capable to follow the target for my 2m type VIIc
So I imagined the following constraints and tried to realize them:
- small size torpedo length 17.5 cm diameter 15.5 cm
- radio controlled since common transmitters (PPM signal)
- going straight and fast enough
- realized with cheap components
- main components 3D printed

The first prototype was made with a scratch build 72mhz receiver circuit
You can see detailed steps of the construction here (we are a non profit organization of RC sub modelers in south of France)
https://www.ladsm.fr/viewtopic.php?f=5&t=21
The main constraint was the antenna size. The simulated electromagnetic (EM) gain was only 3% (1.83m required for 1/4 wave length)
Also this prototype was using a common propeller, no contra rotating system.
Radio circuit was made around a TK83361M with -99dBm sensitivity and motor and rudder using a Silabs C8051F300 microcontroler
A magnet servo was done with 325 rounds of 7/100mm enameled copper wire
The patch antenna is a large and top curved trace acting as a top loaded capacitor to decrease reactance, making the antenna matching network possible with current commercial inductors values.
The antenna circuit tuning was made with a network analyzer (on circuit and in air calibration) and I got a -30dB resonance point so 99% of signal passing through.
Anyhow remember that range refers to the realized gain = K*EMgain*MatchingNetworkGain - AirWaterInterfaceLoss - AmbientNoise so the result is not so good
The test in waters shows that
1) the magnet servo works quite well for torpedo rudder
2) the mall pager motor (6mm direct drive) was not powerfull enough
3) ejection system was done successfully into a PTFE tube (the torpedo body is coated and sanded 4 times with a car painting filler until smooth as a mirror and then I spray PTFE grease on it)
this reduced dramatically friction forces and torpedo gently eject at 1/2 throttle
4) 72 mhz RC was not really reliable due to the reduced antenna size giving poor gain

Thinking to solve this I measured the wavelength of 72 mhz RF wave and it was only 43.6 cm, giving 32% EM gain on RF simulation.
Conclusion is that the antenna tuning should be made with the VNA probe and antenna into water and not in the air.

I choose to let this version to make a try in 433 Mhz
A -117 dB sensitivity Silabs chip was chosed, commercialized on a tiny circuit called HC12 capable to up 2 km range in air.
Successfull underwater experiments have been done already with it.

Parts achieved successfully until now
- 433 mhz rc Tx may be connected on any Tx school catch (Vbat,GND,PPM out)
- up to 8 torpedoes per model, one active at a time, configurable on conventional or multi switch channels
- arduino code mini and mega 2560 compatible
- 100 frequency channels authorizing different people to launch torpedoes together

Rx side
HC-12 + C8051F300 circuit proportional control of motor and rudder (magnet servo)

Contra rotative gearbox using easy to find pinions and a 6mm drone powerfull pager motor with 1/5 micro planetary gearbox

Final mounting is actually in progress and I hope to make a test & video in may










magnet servo (first version)

[coryhenry]

That is an excellent project, Thanks for sharing all the details!!

[Monahan Steam Models]

Nice work!

[trout]

Excellent beginnings.

[Slats]

Wow - wonderful work

[jlcrl]

Some useful links:

2 km range test (DIY TECH BROS):

https://www.youtube.com/watch?v=awOPJK5He28

submarine HC-12 433 MHz in a lake
https://www.youtube.com/watch?v=oLuW4o6 ... e=youtu.be

dive test -1.5m in a swimming pool

https://www.youtube.com/watch?v=T84hvI7wjPE

https://www.youtube.com/watch?v=oLuW4o6 ... e=youtu.be

technical articles about 433 MHz transmission into water:

https://www.researchgate.net/publication/258496191_Electromagnetic_Wave_Propagation_into_Fresh_Water

https://iopscience.iop.org/article/10.1088/1742-6596/450/1/012048

++ Performance of RF underwater communications operating at 433 MHz and 2.4 GHz

https://ieeexplore.ieee.org/document/8646491

This last link is closed to what we hope to do in configuration 1: air to water
HC-12 is like a wireless UART (serial) transmission, sending 1 byte at a time
the software packet sent for the torpedo is like this (each item is 1 byte = 8 bits)
StartByte RudderByte MotorByte ActivationByte CRCByte EndByte so 6 bytes = 6*8 = 48 bits

default configuration is Ch1 9600 bauds 100mW
see the HC-12 datasheet for configuration settings: http://avrproject.ru/112/rf_hc12/201...C-12_v2.3B.pdf

48/9600 gives 5 ms delay to send a packet = 0.83 ms/byte
As the data refresh is locked on the Tx PPM so 22.5 ms/frame we have enough time to send our packet each 22.5 ms = 44.4 Hertz
In real I measured 1 to 1.1 ms delay per byte. This is possibly due to the preamble. The preamble is 3 or 4 dummy bytes sent to enable the Rx to adjust its gain.
If the Tx is narrow to the Rx it will set the gain low to avoid saturation for example.This preamble is possibly included into the HC-12 software.
So coming back to the last link you can see that the test was done into a swimming pool, deep is some tens centimeters but not specified
156 packets/minute sent by the 433 mhz RF transceiver
At 10 m packet error was 10/156 = 6% and 31/156=20% at 30 m
The torpedo is not so deep and will be between surface and -10 cm
If the packet loss is too high, we have the possibility to decrease the baudrate to 2400 or 1200 bps (for high range)
At 9600 bauds HC-12 Rx sensitivity is -111 dBm (decibels milliWatt) and -124 dBm at 1200 bps (!)
If you look carefully the picture the Rx for the test is the low cost SF1000A with -100 dBm sensitivity only
48 bits/1200 bps gives 40 ms delay per packet so 25 Hz so data refreshed 25 times/s
This means we have a large margin. (Theoretically the range doubles every 6 dB)
Additional improvement is to use a + 3.3 dBi gain antenna for the Tx instead of the 0 dBi original one as we can connect any antenna on HC-12 through the FLU connector.

[jlcrl]

Thanks for encouragements. I like the contra rotating system done by Monaham Steel model. Excellent work on the lathe.
Height of the crown gear is actually 2.5mm from plate to top of teeth.
Possible to decrease it to 2mm by milling a POM crown gear with the CNC. Program is done. Waiting for micro drills straight and ball noose head,
all of them 0.2 mm diameter

[jlcrl]

EJECTION SYSTEM

3-4 layers of car paint filler, water paper nr 1000 sand each time
then spray the PTFE grease: it makes a thin PTFE film on torpedo body surface
Launching tube may be PTFE od 20 id 18mm or 3D printed with 0.3mm PTFE sheet inside.
Drill holes to prevent suction effect, letting water flux free circulation.
Doing all this and the torpedo gently eject from RC command 50% throttle without anything else.
The test done without grease. Problem with the grease is the solvent dissolving the filler layer.
I think to increase number of holes to decrease contact surface. Small contact surface = smaller friction forces.
Somebody in my club also realized a successful ejection with a 4 brass rods cage instead of a tube.

[jlcrl]

HOW TO CUSTOMIZE A GEARED PAGER MOTOR
You can find these small geared pager motors on robotic or Asian websites (inexpensive)
Gearbox are 1/5 to 1/250 6mm outer diameter
The original motor is very weak, we sacrify it and keep the gearbox.
Remove it by sliding the sharp head of a cutter under the 4 tabs.
Print the new motor holder as explained. Work also with 10mm motors.
( As the torpedo motor holder is screwed and removable, a 10mm motor may also be installed in place of the 6mm )




bevel the axis like this using a silicon carbide grinding wheel, axis locked with this tool

[jlcrl]

the U96 and the enemy

[Monahan Steam Models]

Good work you’re doing. I used the same planetary gear box as you are using currently during my experiments last year developing the 1/48 scale counter rotating propeller drive for my electric torpedoes. I found that with only 1 planetary gear stage, the rpm was too slow. This was using a longer more powerful 6mm motor. I eventually decided not to use a gear reduction at all and instead test different length & rpm 6mm motors to get the performance I was seeking.

Regards,
Nick

[jlcrl]

Thanks for the return Nick.
Yes the motor/battery/weight combination is a problem to solve. With RC version as additional constraint we cannot put anything metallic under the antenna.
So no ground plane, no lead, no wire, no battery under antenna. I spent 2 weeks of antenna EM simulations and VNA measurements before to start.
We have to consider the far field looking like an apple on a wire or coil antenna, this is an image of antenna gain when in resonance.
Far field represent resonance of antenna and all the things closed to it. With the battery under, I measured around 60 mhz forced resonance that is the self resonance of
battery dielectric and the antenna impedance is spoiled. Anyhow putting the battery under the circuit ground plane and (-) connected to ground don't modify antenna & mixer input impedance
but increase the gain. So to balance the torpedo I 3D print a PLA container half tube shape, glue it under antenna and fill it with aquarium quartz sand.

For the battery I will try the 300 mah 2g 3mm thin( but suspect it to be only 1C 2C discharge rate) and 6mm motors with 1:5 1:2.5 1:2 gearbox ratio or without gearbox.
This in may when they lift the confinement here as I cannot access to pools actually.
Also I will try the exLipoRC 551140 battery 150 mah 15C, source of many 150-175 mah drone batteries and its 220 mah declination 551148
Changing circuit with the servo esc controler circuit under the HC-12 makes space enough to place the battery, keeping 7-8g positive thrust on the weight quote.

Did you test a brushless motorisation ?
I think to have a combination working and plan to open a new thread for this
Jean Louis

[Monahan Steam Models]

Did you test a brushless motorisation ?
I think to have a combination working and plan to open a new thread for this
Jean Louis

Hi Jean,

I did not try brushless as my main work was with coreless brushed. The torpedoes I designed were based on 1/48 scaled down exact G7E types. So O.D. was limited to roughly 11mm to be scale and the I. D. was 10mm. No RC. Just point and shoot. They work great, run straight and range is controlled by a timer currently. The videos I posted show how they run when launched.

regards,
nick

[jlcrl]

I mounted the gearbox and take time to remove all hard points. Friction forces were tested on a laboratory power supply at 3.7 V with the geared 1:5 D=6mm 15mm length drone motor.
This one delivers 0.08A@3.7V without anything. The main hard point was found at the junction gearbox/axis and I made something loose to remove it.
Without the stern tube result was 0.25A@3.7V for 1:5 gearbox + contra rotating pinions. The PTFE stern tube was not as good as expected due to mechanical deformations giving 0.45A
I carefully machined an alu stern tube 7mm length, tested different oils (PTFE, lathe neat cutting oil and WD-40). WD-40 gives the best result with 0.31 A all mounted.
Depending on the results into water, direct drive an low gear ratio may be tested also.
About the battery the 2g 150 maH 25C E-flite may be inserted under the circuit but this one has to be modified. I first test low rate batteries 200 300 mah. I unmounted their protection circuit and the smd marking was F8205A and DW01A. Obviously they use a mosfet Gate -Source voltage threshold circuit to monitor the voltage, preventing it to fall under lipo dangerous limit. It is built so that when the battery voltage drops too low, the GS threshold is reached, cutting the battery connection (all also all circuits connected to the battery !)

[jlcrl]

New circuit has been milled, soldered and tested. Width is now 0.65mm (1.4 original), making the insertion of a powerful 25C 150 mah battery possible
Different pinion and adapter milling has been tested successfully, making custom size gears possible.
Motor and magnet servo radio control test OK
Rudder done.
I have to make the coil and think to start in-water tests next week