jdl

  • ***
  • 172
GepRC Mark4 7" long-range quad
« on: May 22, 2020, 08:47:19 pm »
Hi,

This is my latest LibrePilot build: 7" long-range quad with autonomous capabilities.

Made it during the lockdown. I had luck that all neccesary parts had arrived till March 2020.


Build details:

Code: [Select]
Frame: GepRC Mark4 7"
FC Revolution with 433MHz OpLink for radio control and telemetry
ESC: 4-in-1 T-motor F55A PROII Blheli32, protocol used: DSHOT1200
Motors: BrotherHobby Avenger V2 2507 1500kv
Props: DALProp Cyclone 7 x 5.6 x 3
Battery: 4-6S, 2250 - 2700mAh, 65-130C. Currently flying with Turnigy Nano-tech 4S 2250mAh 65-130C
VTx: Eachine TX1200
FPV Cam: Runcam Micro Eagle
OSD: Micro MinimOSD
GPS: Beitian BN-200
AuxMag: GY-273 (HMC5883L)
HD Cam: Mobuis Mini (waiting for ages for an already ordered Runcam5 Orange...)
Extras: DIY tBeacon locator, sensor for ambient air temperature (connected to OSD)
LibrePilot version: Next r.735 with few tiny mods

AUW w/o battery:   489g.


The frame is well designed and manufactured, lot of care in details. However, it is not a spacious one and placing all the modules inside it required patience and precision, as well as the making of some additional details.

All cables are carefully twisted, Auxmag board location is just below the FPV camera. No disturbance from motor currents, tested up to 96A current bursts during flight, no MAG alarms. RTB, Position Hold, VR, Auto-land, etc. work reliably. Of course, RTB is most important for long-range, it's set as a fail-safe mode in case of radio control loss.

I'm using a tiny Matek 5V BEC to power the Revolution FC, GPS, Auxmag, HD camera and tBeacon. Max. load measured is 0.7A @ 5V, not a problem for the BEC.

The T-Motor ESC has its own onboard BEC providing 10V. I'm using it as a power supply for the VTx with some good protection against voltage spikes during motors braking).

The VTx itself outputs 5V for the OSD and FPV cam. I've added there an extra LC filter (100uH, 470uF).

The DIY tBeacon (Blue v4.3) locator https://tbeacon.org/?page=15&language=en  is connected to the GPS cable and is located at the back of the frame. It weights only 6 grams together with its battery and antenna. Btw, I've equipped all my quads and wings with these modules. Just for peace of mind! :)

The UHF antenna is a DIY monopole. Not the best choice for long-range, I agree.
The carbon-fibre frame is connectd to GND (the shielding of the antennas coax cables). So the entire frame acts as a counter-poise to the 1/4 wavelength antenna element. I've used a Network Analyser to individually tune this monopole antenna, mounted on the frame, and set its center frequency to 435MHz.
So far, flights up to 2-3 kilometers away show RSSI and Link quality results similar to my older quads that use coil-loaded dipoles (individually tuned, too). At some later stage I may consider to make a coil-loaded dipole for this quad, if the monopole behaves unsatisfactory.

The UHF antenna and the VTx antenna are mounted on DIY steel stands. I've found nice elastic 0.5mm steel plates from old computer cases and have used them to dremell, file and drill the stands.

The GPS was initially mounted entirely (centered) in the back compartment of the frame. However, it turned out this location is not good, it had great difficulties locating sats (maybe a Faradey cage effect?!). So I had to move it slightly otuside of the frame, now it works outstandingly. Last flight I've noticed it was using 27  satelites with differential corrections (DGPS).

GPS is set to 57600bps, UBX, 5Hz update rate, GPS + GLONASS + GALILEO, ENGOS for ranging + corrections.

Autotune produced good PID values, I decided to leave QuickSmooth to 0 to have smoother HD video.

This quad is noticeably faster and more efficient then my older ZMR250s wth 2204 2300kv motors.

Hovering current is 6A (24% throttle) with a fresh 4S battery. The cruise speed is about 60km/h (@ 8-10A) - 65km/h (@ 12A) - 75km/h (@ 16A) (4S battery) and needs just a blip  on the throttle to reach 100km/h. So far the max speed I've recorded with it was 139km/h (while climbing, not diving). I guess it can score 170km/h or even more, with 6S battery.


Here are some close-ups showing the location of the modules:













































And a sample from in-flight video (unedited, unstabilized, nothing special):



« Last Edit: May 23, 2020, 01:22:08 pm by jdl »

karla

  • *****
  • 599
Re: GepRC Mark4 7" long-range quad
« Reply #1 on: May 24, 2020, 02:00:21 am »
Really nice build jdl!

jdl

  • ***
  • 172
Re: GepRC Mark4 7" long-range quad
« Reply #2 on: May 24, 2020, 09:23:57 am »
Thanks karla!
I decided to share it because some ideas in the design of this particular build turned out to be successful and may be useful to others.

karla

  • *****
  • 599
Re: GepRC Mark4 7" long-range quad
« Reply #3 on: May 24, 2020, 11:58:38 am »
Yes very useful with all details you provided.
The 433MHz OpLink radio, is that operating 0.1 or 1 watts?
If over 0.1 then how did you solve that?

jdl

  • ***
  • 172
Re: GepRC Mark4 7" long-range quad
« Reply #4 on: May 24, 2020, 12:52:47 pm »
0.1W from onboard RFM22B modem (integrated in Revolution FC).

1W power from the radio control TX (relay box). Modified OnLink Mini (its RFM22B replaced with 1W RFM23BPS on separate higher voltage supply):

https://forum.librepilot.org/index.php?topic=4705.msg31701#msg31701







Btw, I'd further improved the relay station. It'd got a modified RFM23BPS module that outputs 1.3W / (3.5W in "turbo" mode). Modifications are partially inspired by "Narpat" mod:

http://www.itluxembourg.lu/site/2-5w-narpat-mod/
http://www.itluxembourg.lu/site/modding-the-orangerx-1w-modules-for-full-power/

However I believe he had made some wrong assumptions that harm the performance (my opinion, and I had to state I'm not an RF engineer).

I don't mind to share the details, just have to dig my archives and find what I've done. I don't clearly remember  all the details...

It is nice to have a 3.5W TX power for radio control but I've used only few times for few minutes for the last years. Also, in "turbo" mode RFM23BPS operates very close to the absolute maximum ratings of its RF Power MOSFET and requires using of very well matched antenna. I've also placed an RF ferrite choke on the feeder line as an extra protection. Worsening the SWR by simply touching the antenna by hand can cause the famous RFM23BPS "smart" reset or even damage the RF MOSFET.

1W is pretty enough for up to 20km range flights with good omni antennas in clean (non-urban) environment.

Re: GepRC Mark4 7" long-range quad
« Reply #5 on: May 24, 2020, 07:56:36 pm »
That is a really great looking build.  Your lockdown time was well spent.  :)

Wow, antenna testing on a network analyzer.  I have an RF Explorer with separate signal generator, but I haven't bought a directional coupler yet.  I guess I need to get several directional couplers, one for each band, rather than continuing to look for something that covers close to 0-6GHz since I additionally want to be able to do 1.3GHz and 5.8GHz.

Might I suggest a Nagoya NA-24 airborne antenna (8 grams) (you would need a 90 degree SMA elbow too) if you are interested in range testing.  About USD $15 plus the buyer's country VAT.
https://www.ebay.com/itm/New-Genuine-Nagoya-NA-24-SMA-Male-144-430Mhz-2-15dB-10W-Flexible-Antenna/122292796574
I have one and have tested it, but most of my long range stuff is fixed wing and I use Retevis RHD-771 (heavy 34g) on those.

I'd like to caution those that see 3.5W and say "I want one of those" (I must say that I want one of those too :) ), you should first work on getting good antennas.  0.00125W (the lowest OpLink setting) will get you hundreds of meters with good omnidirectional antennas and several km with the unmodified full power of 0.1W.  These are from my own actual testing of stock antennas that have not been specially tuned.  As I recall, some original testing at OpenPilot had it working at 0.1W at 7km with omnis, but I would not plan on that kind of range.

The little stubby antennas are especially bad, but really anything less than about 38cm or 15 inches is not the best antenna unless it has a load choke (or a counterpoise / ground plane and correct mounting for a 1/4 wave antenna).  I might be tempted to test a 1/4 wave mounted pointing down.  I'm not sure whether there is a virtual 1/4 wave antenna signal lobe below the ground plane.  Isn't there effectively a null straight out to the side or is that only for real earth ground planes?
« Last Edit: May 24, 2020, 08:17:46 pm by TheOtherCliff »

jdl

  • ***
  • 172
Re: GepRC Mark4 7" long-range quad
« Reply #6 on: May 26, 2020, 02:52:17 pm »
That is a really great looking build.  Your lockdown time was well spent.  :)

Thanks Cliff! I'm also very pleased with the result.


Wow, antenna testing on a network analyzer.  I have an RF Explorer with separate signal generator, but I haven't bought a directional coupler yet.  I guess I need to get several directional couplers, one for each band, rather than continuing to look for something that covers close to 0-6GHz since I additionally want to be able to do 1.3GHz and 5.8GHz.

I'm afraid I've said something misleading. I'm not speaking of a real (industrial grade) Network Analyser. No chance to get my hands on one of these...

The "poor man's" VNA I'm using is a cheap spectrum analyser with onboard tracking source (LTDZ_35-4400M),
RF bridge (https://www.ebay.com/itm/332052527822 or https://www.ebay.com/itm/332640254339),
MAC Technology C2045-20 Directional Coupler 2-8GHZ 20dB,
and the nice VMA Simple Spectrum Analyser software: https://vma-satellite.blogspot.com/2017/03/vma-simple-spectrum-analyser-download.html

The LTDZ module is far from perfect, definitely inferior to RF Explorer. It is good enough, however, for testing and tuning antennas in the range 100-2500MHz.

Btw, there is currently an affordable octave band direction coupler at very descent price (ebay):
MAC Technology SMA Directional Coupler CA3205-20F (4-8GHz): https://www.ebay.com/itm/263097636864
Spec sheet: http://www.mactechnology.com/pdf/2008-4.pdf


Might I suggest a Nagoya NA-24 airborne antenna (8 grams) (you would need a 90 degree SMA elbow too) if you are interested in range testing.  About USD $15 plus the buyer's country VAT.
https://www.ebay.com/itm/New-Genuine-Nagoya-NA-24-SMA-Male-144-430Mhz-2-15dB-10W-Flexible-Antenna/122292796574
I have one and have tested it, but most of my long range stuff is fixed wing and I use Retevis RHD-771 (heavy 34g) on those.

Here are the measurement of the Retevis RHD-771 I'm using with the 1.3/3.5W Oplink relay station. The orange line is the measured directivity of the RF bridge.



I have three of these antennas, and this is the best matched one. Other two have their center frequences slightly lower.


I'll try to get one of these Nagoya NA-24 to test on a wing. Quads just cannot go that far (due to battery limitations) to justify usage of superior quality onboard antennas.

Here are the dimensions of the coil-loaded dipoles I make for my quads. They are easy to build and proved to be robust enough for long-range / freestyle (no crashing, of course :)). I've found that if made precise enough, there is no need to additionally fine-tune them on VNA.












I'd like to caution those that see 3.5W and say "I want one of those" (I must say that I want one of those too :) ), you should first work on getting good antennas.  0.00125W (the lowest OpLink setting) will get you hundreds of meters with good omnidirectional antennas and several km with the unmodified full power of 0.1W.  These are from my own actual testing of stock antennas that have not been specially tuned.  As I recall, some original testing at OpenPilot had it working at 0.1W at 7km with omnis, but I would not plan on that kind of range.

Once I've flown to over 11km with 0.1W ground (control) oplink and omnis but this was on the edge...

On the other side, telemetry from the onboard 0.1W RFM22B (Revo built-in) keeps coming from over 15km (some minor packet loss starts at 13+km) and omnis. Onboard antenna, despite tiny 0.1W TX power, benefits a lot from being high above ground :)

Haven't tested longer distances yet. A directional Moxon or Yagi on the ground relay station is a logical move then...


The little stubby antennas are especially bad, but really anything less than about 38cm or 15 inches is not the best antenna unless it has a load choke (or a counterpoise / ground plane and correct mounting for a 1/4 wave antenna).  I might be tempted to test a 1/4 wave mounted pointing down.  I'm not sure whether there is a virtual 1/4 wave antenna signal lobe below the ground plane.  Isn't there effectively a null straight out to the side or is that only for real earth ground planes?

I've read that being high above ground, quarter wave dipole has a radiation pattern similar to the one of the half-wave dipole.
https://ham.stackexchange.com/questions/6325/does-mounting-a-ground-plane-antenna-upside-down-change-the-takeoff-angle

It might worth trying with monopole mounted upside down... Still, I doubt there would be a serious improvement. The quad constantly changes its attitude during the flight and so is the antenna, it does not stay in the optimal vertical position all the time...
« Last Edit: May 26, 2020, 02:57:58 pm by jdl »

jdl

  • ***
  • 172
Oplink "Turbo" Mod
« Reply #7 on: May 26, 2020, 04:04:32 pm »
Oplink / RFM23BPS mods:

1. RFM23BPS needs adding a secondary R25 transistor parallel to existing one. It is 2SC3356 SOT-23. Get it from Aliexpress or from another damaged RFM23BPS. This will raise the output power to from 1W to 1.3-1.4W.

2. Cut the PCB trace at the position marked with yellow, add an extra 100nF 0805 SMD capacitor as pointed and solder 28-30AWG wire to it (as on the image) to feed the higher (6-10V) voltage to the RFM23BPS.










3. I recommend using a shielding enclosure: https://www.aliexpress.com/item/32787641224.html



4. Follow the schematic to build the power supply for the modified RFM23BPS.





Oplink transmits quite efficiently, short bursts, thus the energy stored into the caps is sufficient to achieve the high output power AND the RFM23BPS stays cool. No extra cooling is necessary. LDOs are not stressed, either. I hadn't tested the modified RFM23BPS with OpenLRS and do not plan to!



Parts:

LDOs:
ROHM - BAJ0BC0FP-E2 - Fixed LDO Voltage Regulator, 3V to 16V, 300mV Dropout, 10Vout, 1Aout, TO-252-3
ROHM - BA06CC0FP-E2 - Fixed LDO Voltage Regulator, 4V to 25V, 300mV Dropout, 6Vout, 1Aout, TO-252-3

Capacitors:
220uF / 25V LOW ESR
330uF / 25V LOW ESR
1000uF / 6.3V LOW ESR
330nF - ceramic

Diodes:
1N4007 or even better: 1N5817G (has lower Vf)

Ferrite choke:
LAIRD - 28B0375-100 - Ferrite Core, 154 ohm, 14.48 mm Length, 5.08 mm ID, 9.53 mm OD
I found it in Farnell. As I recall, it's the same as ones IBCrazy used to put on his antennas.









I do strongly advise to use/put the choke on the feeder line! It helps to protect in some extent the RF Power MOSFET from reflected power if bad antenna is used, especially when operating at full power.

Make sure your TX antenna is good! Retevis RHD-771 is my favorite!





NEVER run the modified module without antenna! Avoid touching the antenna while operating at full power! Avoid placing a massive objects (like your body) near the antenna while operating! All this degrades the SWR of the antenna and overloads unnecessary the RFM23BPS module.

Last words: Don't use the "turbo" mode if not absolutely necessary! Have it on a knob / switch for emergency occasions only! The 1.3W is pretty enough!
« Last Edit: May 26, 2020, 04:09:55 pm by jdl »

Re: GepRC Mark4 7" long-range quad
« Reply #8 on: May 27, 2020, 06:22:53 am »
I also have an OpLink with an RFM23BP on it, but the RFM23BP is stock.

As we know, the RFM23BP mod introduces a 2nd power supply, so this OpLink now has two, 5V from USB regulated down to 3.3V and the new power supply that is regulated down to 5 or 6 or ?? volts.

A small circuit addition that I find handy is a diode from the 5V USB power to the new RFM23BP power.  This allows the RFM23BP to be powered from the 5V USB if you don't have your new, higher voltage supply connected.  When powered from USB alone, the output is about 0.4W, and I generally don't need more.

A bit of information that I gathered when doing this project is that the RFM23BP uses a different type of output stage than the RFM22B, so the 23 doesn't throttle well.  The reduced power settings are undefined in the 23.  According to the information I read, it's probably best to leave the power setting at max and reduce the RFM23BP voltage to reduce the RF power.

jdl

  • ***
  • 172
Re: GepRC Mark4 7" long-range quad
« Reply #9 on: May 27, 2020, 09:22:49 am »
As we know, the RFM23BP mod introduces a 2nd power supply, so this OpLink now has two, 5V from USB regulated down to 3.3V and the new power supply that is regulated down to 5 or 6 or ?? volts.

6V Linear Voltage Regulator with an adequate (330uF) low ESR capacitor on output and a schottky diode between the output and RFM23BPS Vcc pad. This way an unmodified RFM23BPS is powered with 5.6-5.7V that is safe and enough for 1W TX power.

A small circuit addition that I find handy is a diode from the 5V USB power to the new RFM23BP power.  This allows the RFM23BP to be powered from the 5V USB if you don't have your new, higher voltage supply connected.  When powered from USB alone, the output is about 0.4W, and I generally don't need more.

The diode you mention is presented as D4 (schottky diode) on the hand-drawn schematic. It is soldered to the input pin of the Oplink onboard 3.3V LDO (internally connected to 5V USB).

With the additional R25 preamplifier (RF driver) transistor soldered, I've measured that TX output of the modified OpLink raises from 0.4W to 0.55-0.6W when powered from the 5V USB only.

Even if you use a completely unmodified RFM23BPS module, it's still a good idea to have a serious capacitor (C5) (330-1000uF) between Vcc and Gnd pads of the RFM23BPS module.

A bit of information that I gathered when doing this project is that the RFM23BP uses a different type of output stage than the RFM22B, so the 23 doesn't throttle well.  The reduced power settings are undefined in the 23.  According to the information I read, it's probably best to leave the power setting at max and reduce the RFM23BP voltage to reduce the RF power.

+1.