Synchro copters [beautilful]

embedded by Embedded Video

YouTube Direkt

Easy glider finally in the air

After long break, I’ve came back to RC. EZ glider has been lying unassembled for about half a year, and I’ve finally finished it. The test run wasn’t very smooth:

- I’ve threw glider without motor applied to see if it glides ok, it bumped nose to the ground. Nose heavy? I’ve measured COG 70mm from the leading edge, and it balanced perfect. On the second try I’ve switched on motor, and it went up real nice, but of course I had to compensate elevator trim.

- I haven’t mixed tail with ailerones, so it was kinda lazy on the turns. I still haven’t mastered coordinated turn, so I’ll fix it with the mix.

- I forgot to program the brake, so the folding propeller was not going to fold.

- Propeller adapter came off during the flight. It was fixed firmly before flight. Not sure what’s happened. My first folding propeller, maybe I’ve missed something.

Hopefully next time I’ll go with fpv gear and fixed plane.

embedded by Embedded Video

YouTube Direkt

Finally received GOPRO2!

Finally came my GOPRO2 from USA. New and improved. Better resolution, handles dark better, more pro. Haven’t installed it on a plane yet, but took it on my ride to work. It was pitch black, so you can evaluate how it works at night. It’s not a night camera, but even at such conditions it’s sensitivity is reasonable. I promisse to make more videos during the daylight, once I see it.

Video taken 1080p at 30 fps.
embedded by Embedded Video

YouTube Direkt

Converted DOSD power supply to linear

My gps unit was unable to lock stable signal. I’ve found out that interference source was DOSD swithing power supply unit. I’ve removed it, and inserted linear 3.3V voltage regulator (LM1117-3.3). Works like a charm. Stable satellite lock near window where was no at all.

Adjusting video level on DOSD

While lines
I’ve recently purchased DOSD. The first I wanted to fix was these white lines on the screen. They appear if black borders are enabled in the settings and signal is fed through RF video link. When DOSD is directly connected to the goggles or the computer, this problem is not noticeable.

As I’ve noticed on youtube clips this problem is not rare one. Few of my friends has bought Dragon OSD and had the same problem.

The problem is in the black level in the composite video signal

As you may notice on the scope view, very short bursrs of signal means white letters being drawn on the screen with black border. The lower amplitude is, the darker the view is. But it should stay above reference black level, which goes right after the sync signal. Here we see, that it goes under. That probably confuses video TX or maybe RX AGC. You may hide this problem by loading video signal with parallel resistors on the composite cable, but this might degrade your contrast, colors or even a synchronization. THIS IS NOT A MAXIMUM VIDEO LEVEL PROBLEM!!!

When I’ve looked at the DOSD PCB I’ve found that two signals comes from microcontroller. One of them draws white pixels, another is for black ones. White pixels right now probably is not the part of the problem, so I’ll draw only the black color part . This scetch shows how the parts are connected (guess only, based on observation):

Accidentally I’ve damaged onboard transistor, and tried to replace it with pin-to-pin compatible BC847 (100MHz). This general purpose NPN transistor seems too slow for this task, the image was very shady. So I’ve searched and found very good and kinda cheap transistor: BRF93A. It is able to work up to 6GHz (!). It worked really nice. Not sure what was the original part number, but text in the picture now is crystal clear.

To limit voltage being shorted to ground by transistor, I’ve inserted 100 Ohm resistor on the emiter, and it solved a problem. You can see in the video how I’ve tried adjustable resistor to find optimal blackness level in the picture. There was no slot for the load resistor, so I had to lift transistor and solder 100 Ohm resistor nearby. Not very elegant, but .. it works.

As you may see, I have great picture in the screen. With solid black border and no white shading ;) .

embedded by Embedded Video

YouTube Direkt

Hope this will help someone to improve the image!


Never tried, but maybe there is even easier way to solve this, by increasing marked resistor:

Finally reached 2 km range

I’ve tested my new Lawmate 1.2 GHz video rx. It seems it’s quite good receiver. Sensitivity was slightly better than sunsky rx, but selectivity is much better. It was quite immune to nearby 433 MHz LRS, contrary to sunsky vrx. Due to omni antennas signal wasn’t very strong, but good enough to do the flight.

Twinstar2 3S 2600 mAh, 8×4 APC. Inverted Vee antenna for 400mW @1280MHz sunsky video tx, omni stock antenna on lawmate video rx. Archos AV500 DVR. ChainLink 433MHz LRS stock antenna 200mW mode. Fatshark camera. Also fatshark 2.4 GHz 100mW video tx on ground station, to retransmit signal to fatshark goggles.

embedded by Embedded Video

YouTube Direkt

New ground station, new LRS

Finally I've purchased ChainLink LRS. It has switchable power between 0.1mW (range test, 0.2W and 0.5W). 12 Channel receiver. We've tested it in the city without ling of sight, signal faded in 2.4 km, which was very similar to popular DragonLink (DL). I've modified Smalltime OSD to show RSSI level. Problem was in SN777 camera which was very noisy. I've did modification to remove that noise (will be described later). I've did some changes to the ground station also. I've used photo camera tripod. That helped to reduce multipathing. I've attached 100mW 2.4 GHz fatshark transmitter near after 1.2 GHz receiver to retransmit the signal to the goggles. It gives more freedom around the GS. The main reason I've did this, to reduce CL influence to video receiver, because atleast 2 meters are recommended. embedded by Embedded Video

YouTube Direkt

Fatshark review

Recently I’ve bought FatShark 2010 Aviator edition video goggles. I’ve been using laptop as my primary navigation screen. I’ve started with EasyCap USB video adapter. It was very cheap (~7$), and very crappy product indeed. Then I’ve switched to Avermedia PCMCIA adapter. It was ok. The mayor problem is laptop screen itself. It’s very difficult to seen even there is moderate sun light outside. Even the laptop is in the dark car trunk, reflections are still very bad.

So I finnaly decided to get these goggles which I’ve heard so much about. Aviator edition is different from standart edition, because it has integrated 2.4Ghz or 5.8 GHz video receiver. Also it came with 100 mW transmitter. I didn’t tested it outside, but I don’t expect much from that power.

I had VR920 video goggles before, so I can compare them. First of all, the things I like:

+ Good field of view (FOV). It’s declared as 40ish degrees. It would be nice to have more, to get even more immersion, but it’s really enought to navigate and feel yourself in the plane. This is where it outperforms VR920.
+ Good contrast. My camera doesn’t produce good colors, so I can’t tell if there colors are good, but picture is quite good.
+ Really good sun shading. Foam around the eyes completely covers everything around, even with the bright sun outside, I’ve seen only screen. No silly caps or covers needed.

I know it’s not a big list, but it’s very important freatures in FPV goggles I think. Now what I didn’t like:
- The most important issue for me with this goggles is eyes fatigue even after very short usage. I must admit my eyes are quite bad, maybe there is something to do with it. But few people with good vision also confirmed this. I think there is something to do with focus distance and eye degree. The goggles are focused for 2 meters virtual distance. But eyes are looking straight forward, parallely into infinitive. Eyes quickly adapts to this and you can see fine in the goggles, but when you take it off, eyes needs to readjust and it feels sick for a few moments. Maybe this is something I need to get used to, but I didn’t like it at all!!!
- Lenses are very close to the eyes. Some people complained about it, personally I had no big problem with that. I guess it was needed to achieve such good FOV.
- Even you can change lenses distance to fit the distance between the eyes with two sliders in the front, it’s almost impossible to get clear picture in the whole screen. Sometimes left, sometimes right, or sometimes both sides are blurry. It’s not very big issue for me, because my osd shows main information near the center of the screen, but other people, could have some problems with osd like dragon OSD. They say that 2010 edition fixed this problem. I wonder how horrible it was before this edition?
- Battery charged came in US mains plug version. I need to buy aditional adapter to use it. I had to cut the cable, fir the T connector and use it with my primary battery charger.

Here are some pictures:

I like this label “inflatable shark not included”.

It’s not easy to take a picture of the view, because you need to be really close to the lens to see the whole picture.

This 100mW transmitter came with simple wire attached instead of antenna. I can live with that if it’s tuned to right frequency. I didn’t tested yet properly, but I’m sure inverted vee antenna will outperform this wire.

I hope this review will help you to decide if you want to purchase this unit. They are not great goggles. They are ok and do their job. Not sure if it’s possible to get anything better for such price though.

Neseniai nusipirkau FatShark 2010 Aviator edition video goggles. Seniau naudojau laptopa FPV skrydžiams. Pradėjau nuo EeasCap USB video adapterio. Gan prastas ir pigus produktas. Vėliau pakeičiau į Avermedia PCMCIA adapterį. Jis buvo geresnis, tačiau pagrindinė problema išliko dėl atspindžių ekrane. Labai sunku buvo įžiūrėti kažką netgi prie vidutinio apšviestumo lauke. Netgi įdėjus į tamsią bagažinę, iš lauko ateinanti šviesa labai trukdė.

Taigi nutariau nusipirkt šiuos akinius. Nemažai apie juos rašoma FPV forumuose. Aviator edition nuo paprastos skiriasi, tuo, kad integruotas 2.4GHz arba 5.8GHz imtuvas, ir pridedamas 100 mW siųstuvas. Dar neišmėginau siųstuvo lauko sąlygomis, tačiau iš 100 mW nesitikiu labai daug.

Prieš tai esu naudojęs VGA VR920 video akinius, taigi galiu palyginti su šiais. Iš
pradžių, kas man patiko:
+ Geras apžvalgos kampas (FOV). Deklaruojamas apie 40 laipsnių. Būtų gerai turėti daugiau, bet tiek tikrai pakanka, tam kad pasijustumėte tarsi lėktuve. Šiuo aspektu fatshark’ai lenkia VR920.
+ Geras kontrastas. Dabartinė kamera neduoda gerų, taigi negaliu pasakyti, kaip dėl jų, bet vaizdas gan neblogas.
+ Labai gera apsauga nuo saulės. Porolonas aplink akis visiškai uždengia viską aplink, ir netgi esant ryškiai saulei nieko nesimato išskyrus ekraną. Nereikia dėtis kepurių ar kitų dangalų.

Pliusų sąrašas neilgas, tačiau manau šitos sąvybės gan svarbios FPV. Kas man nepatiko:
- Labai pavargsta akys netgi po trumpo naudojimo. Keletas žmonių patvirtino, kad kažkas nėra gerai. Mano spėjimas, dėl ko taip yra – akiniai sufokusuoti virtualiam 2 metrų atstumui, bet akys žiūri tiesiai į tolį. Užsidėjus akinius greitai priprantama ir vaizdas normalus, tačiau nusiėmus, keletą akimirkų dar jaučiasi vaizdo dvigubėjimas ir nuovargis. Nors tai greitai praeina, bet klausimas ar labai sveika.
- Linzės labai arti akių. Man tai nekliuvo, bet yra žmonių kuriem tas gali maišyti. Tai turbūt buvo norint pasiekti tokį didelį FOV.
- Nors yra reguliatoriai kuriais galima keisti linzių atstumą, kad priderinti prie atstumo tarp akių, beveik niekad nepavyksta išgauti gero vaizdo abiejose akyse. Kartais neryškumas kairėje pusėje, kartais dešinėje, kartais abiejuose kraštuose. Man tai nelabai trukdo, nes mano OSD informaciją pateikia arčiau centro, bet su dragon OSD ar pan., gali kilti problemu. Gamintojas teigia kad ši problema pataisyta 2010 m. versijoje. Idomu kaip blogai buvo iki to.
- Baterijos pakrovėjas atėjo su US kištuku, taigi nelabai panaudojamas. Teko kirpti laidą ir prilituoti T jungtį, kad galėčiau krauti su pagrindiniu pakrovėju.

Keletas nuotraukėlių:

Patiko užrašas “inflatable shark not included”.

Mėginau nufotografuoti, kaip atrodo vaizdas iš vidaus, bet su standartiniu objektyvu praktiškai neįmanoma taip arti pridėti, kad matytųsi plačiau.

Pridedamas 100 mW siųstuvas su laidu vietoj antenos. Po testų bus galima pasakyti, ar ta antena dirba normaliai. Neabejoju kad Vee tipo antena dirbs geriau.

Tikiuosi ši apžvalga padės apsispręsti ar jum reikia šių akinių. Tai nėra puikūs akiniai. Tiesiog akiniai. Už tokią kainą geresnių šiai dienai gal ir nelabai pavyktų rasti.

Calibrating OpenPilot quadrocopter

embedded by Embedded Video

YouTube Direkt

FPV range calculator

Try this experimental range calculator:

New heights with new OSD, V-diversity and step-up

I’ve tested my new, custom made equipment:
- DC/DC setup-up 12V converter (custom PCB 1×2 cm made by myself;))
- Smalltim OSD
- 2.4 GHz video diversity.
I love it! EasyStar with 3 meter 35 MHz RX antenna.

embedded by Embedded Video

YouTube Direkt

2.4 GHz diversity controller

My new finished diversity controller based on airwave module. Soon to be expanded to work with 4 signals, video spliter, telemetry uplink to the PC. Expected price <100$.

520 TVL day/night camera

My current cmos camera was very bad quality, so I needed to change it. I’ve choose this one:



EasyStar with 4S1P

3S1P configuration was slightly underpowered so, I’ve added one more LiFePO4 cell to the setup. It still doesn’t have 90 deg. climb, but it’s pretty fast now. I think propeller 5.5×5x5 should be increased up to 6×4. Anyway 50% WOT time it get’s about 9 runtime. It’s really not much, but with 4C allowed charging current, battery will be charged in 15 minutes, and that’s not so bad.

embedded by Embedded Video

YouTube Direkt

Towards digital FPV

On RCGroups member announced, that he tested new digital video link, designed for FPV. “Rotoconcept Robotics” from Canada, has developed bidirectional 2.4 GHz link, which enables RC control and digital video transmission simultaneously. The details are unknown, but they are probably using OFDM modulation. Encoder and transmitter is about 12×13x4 cm. Right now it’s too heavey for models like EasyStar, but it was successfully used on quadrocopter. Right now company tries to make this device smaller. They promise not only digital video transmission, but also OSD and GPS functionality. Integrated computer gives a chance to use this system for FPV purposes. Despite that digtal video transmission get much criticism in the FPV arena, everybody agrees that in 5 years, this technology will be dominant due it’s high performance in quality and range. In this clip, you can see difference between 1W 2.4 GHz analog video transmitter (first part) and 50 mW digital video transmitter. The test was performed in 180 meters, non line of sight conditions. Analog video was constantly interfered, while digital link was consistent during all range.

embedded by Embedded Video

YouTube Direkt

embedded by Embedded Video

YouTube Direkt

Corona RP8D1 receiver RSSI mod – cheapest OSD ever

When flying FPV, it’s essential to have both video and RC links perfect. One of these missing will result grounding effect, when your plane will loose control and gravity force will pull it to the surface much faster than you usually do. It’s easy to notice weak video signal – colors are disappearing, frames are out of sync, view is “snowy”.

It’s much harder to notice RC control link failures. Especially if link is digital. Receivers usually has LED, which indicates link quality. Glowing green led means that everything is ok, blinking or red led means link failures. This led is not seen during the flight, because receiver is hidden somewhere deep in the plane where all the servo wires come in and it’s more easy to protect from external damage.

I’ve modded my Corona synthesyzed 35 MHz frequency receiver RP8D1, so that I could see link failures in video signal. I’ve desoldered the LED from the receiver and moved to the front of the plane. 1 mA current was enough to light up white led with moderate luminance. To make led omni directional I’ve sanded it’s surface. This mod is very basic. Going further we can take analog RSSI signal from the receiver, buffer it and show it on led’s row or put into OSD. The we could see signal strength expressed in percent, not only blinking. But right now I think I’ll be fine with that.

We need to take paper and plastic protection from the electronics, and foam which covers the microchips.

Receiver is a small part so you need to be careful. The magnifying glass can help you.


First launch of the Easy Star – not so easy

Finaly the day came when weather conditions allowed to test my new Easy Star floater. There was no wind at all, temperature below -8C. First of all I’ve tried to launch without the camera. To keep conter of the gravity where it supposed to be, I’ve mounted metal plate in the nose of the plane. It was very difficult to launch the model. If I slowly hand launch the model, it drops on the ground. If I through it faster, the nose climbs up, stalls and falls down. Giving a motor more thrust doesn’t help either. Motor pushes the model towards ground also. (I suspect that motor is mounted too high and this is not a behaviour by design). Once I’ve managed to lift off, I’ve realized that rudder control is inverted. So I had to come back. The rudder is to little for this model. When it slowly glides it’s almost impossible to control it. I need to give more thrust to control it, so it odd situation when you need to increase speed to land correctly. I’ll need to mod this thing. I’m not sure about the ailerons. Maybe there is no need for them when the rudder will work properly. Anyway, I was astonished how good this plane floats. Even travelling decent distance, it lost almost no height. Here is my flight:

embedded by Embedded Video

YouTube Direkt

Camera modding for servo-less aerial photography

To do aerial photos without servo motor, I need to electricaly connect push button on the camera to the receiver. Receiver outputs PWM signal, which, imho, by mistake is called PPM (RF modulation type). This PWM signal is 50Hz impulses, which width varies from 1 ms to 2 ms. It needs to be converted to logical 1 or 0. If impulse is > 1.5 ms, it should be treated as logical 1. There are plenty of schematics with 555 timer, or mircrocontroller. I’m going to use attiny 8 pin mircrocontroller, because it won’t require additional adjustment and is quite small. It can do autonomous periodic photo taking. I’ll write about this stage later (I’ll update this post or create new one).

If you want to take a picture, you need to press button half way (fist switch is activated). Then camera does autofocusing. When the picture is clear, you press button to the end and picture is taken. So I’ll do 1-2 second pause between pushes, which should give enough time for camera to focus. I’ve dissasembled my Canon SD600, and soldered wires to the button pins. I had to solder on plastic plate very carefully, because you never know how much heat resistant it is. I didn’t manage to solder wires to battery pins, because it’s deep inside in the camera under 2-3 PCB’s. So I’ll need to make battery shaped connector or fly with original battery.


Assembling Easy Star

Finally I’ve bought Easy Star! It’s the most important day in my FPV career :) . This model is a living legend, made by Multiplex. It quickly became most favorable model among modelists, because it has pusher motor, very strong ELAPOR material and nice look. There are many mods for this model: ailerons, bigger rudder, brushless motor etc. I’ve bought FC2830-9T 1290KV 56g “Outrunner” motor, which was recommended by salesman, but right now I think it was a bad decision. 5″ propeller rotates to slow to have any thrust. I hope tri-blade 6″ will make a difference. If the motor is in it’s default place, you can’t fit anything bigger than 6″. I’ve found in the forums, that 2000-2700 KV motors should be used in this model. It you ever decide do go brushles, choose something from that range. The time will show if I’ll be able to fly with 1290 KV. I’ve mounted my A123 LiFePO4 battery with separated cells, because I wanted to center the mass correctly. I’ll mount 150g camera in the front, so I wanted to move as much weight as I can to the back. Manual recommends to make CG exactly 37 mm from the front side of the wing.

Electric motors – Turborix 162W brushless outrunner

Our affiliate partners fro China R2HOBBIES sent us Turborix 162W motor. This RC shop has modest prices and good shipping offers. Packages are shipped by airmail, so it’s cheap and fast. (4-6$ to ship this motor). So I suggest to buy at this shop – it helps us! ;)

Old Hextronix was too weak to drag all the required equipment. This motor has twice weight, so it’s torque is better. Specifications:
rpm/v (kv) : 1070
Suggested Battery : 2-3 Cells Lithium Polymer Battery
Load Current (A) : < 15A
Recommended ESC : Turborix Advance 20A ESC
Suggested Propellers : 8 X 3.8″ – 11 X 4.7″
Weight : 60g (±5g)

In the photos you see the motor and included connectors with colored termoplastic. Quite nice setup setup for 8$ :) .


Here you see photos of preparation for second FPV flight with 500mW transmitter. Flight itself was horrible, it was windy, elevator broke up and plane was grounded. Once again, it requires rebuild almost for the scratch.

My first el cheapo FPV setup



When flying for the first time in FPV mode I was afraid that I’ll crash my equipment and loose some investments. Exactly that happened actually. So I didn’t want to buy any fancy expensive equipment. Something cheap, but upgradable. That’s why I’ve choose cheap camera with integrated 2.4 GHz 10mW transmitter standart 2.4 GHz receiver, USB video adapter and more powerful 2.4 GHz 500mW transmitter. I was sure that 10mW was not enough, but it was interesting to try also how 10mW works. It has much pros too (we’ll talk about it latter).

I had to mod 2 things: first of all receiver. It works from the wall plug (230V) by default. I’m pretty sure there is no such interface at the field where I’m flying. So I had to change it that it would work from my laptop USB port. I’ve seen some equipment on the eBay, which were actually USB receivers without external power supply. There is no secret that most of the are tend to be build on the same module, or atleast similar technology. So I’ve dissasmbled the receiver and slightly modified:

receiver and usb adapter

Receiver comes with 8V DC wall adapter which supplies the unit. Unfortunaly USB port provides only 5V. So we need to dig a little bit deeper. Let’s open the case and see what’s there.


NordStar autopilot – UAV system

Lately I’ve been working on autopilot system for UAV (unmanned arial vehicle). Main goal is to have system which backs you up during FPV flight in case video broadcast or control link is down. I’ve finished component search, almost finished with schematics and soon going to route the board. Not sure how big it will be, but I’ve tried to use small components, no redundancy and keep simple as possible. All aerial control systems are very similar, they usually have battery monitor, GPS, accelerometers/gyroscopes and some additional sensors. This is something like that but I hope superior in few ways from the ones which are available in the market. Let the specs talk for theirselves:

- 6-22V (2-5 cells supported) power supply input range
- Step-up voltage converter making 12V voltage for FPV camera
- Making 5V for video transmitter (no BEC required).
- Monitoring up to 50 A with 0.1A resolution
- 6 channel control/capture
- Receiver fault detection
- Roll & pitch gyros
- 3 axis accelerometer
- Altimeter
- Air speed sensor
- OSD (on screen display)
- Telemetry data overlay on broadcast signal (TBD)
- GSM modem port
- 2 buttons for calibration purposes
- Bidirectional alternate communication port (dedicated for ultra long range communications)
- Temperature sensor
- I2C interface for external digital device
- New generation ARM Cortex-M3 microcontoller running up to 72 MHz (1.25 DMIPS / 1MHz)

Once I have more information, I will update. I’m giving schematics overview. These are not final and will be supplemented.

Electric motors: HXT 24g 3000kv

Quick peek at the specifications:
Shaft diameter: 3 mm
Pull: 450 g
Current: 30 A
RPM/V: 3000 KV
MAX RPM: 21000
Recommended LiPo cell count: 2
Weight: 24 g.

I was learning to fly with HXT 24 gram motor from the start. I can’t even remember how many crashes it has seen. I had to repair plane ten times, I’ve changed about 6 propellers. But motor withstood almost everything. After last crash I’ve inspected how motor is, and there are very some bad damages to the wiring. Few wires was broken. It still works, but in my opinion it won’t be very reliable because it won’t be able to take max current as it was designed to. Some users has complained about weak protection for the wires, it seems they are absolutely right. This is the weakest side of the motor. Nothing could kill it, including overvoltage, glues and unmatched propellers, but small cut at the wiring made it useless for critical flights as FPV.



Radio control upgrade to 8 ch 35 Mhz WFLY (WFT08)

I’ve decided to change my 2,4 GHz radio control gear to 35 Mhz. I’ve choose WFLY radio. It’s the most expensive radio from the cheapest class :) . It still has all main functions which belongs to programmable radio.


A123 batteries instead of LiPo

Today I’ve received 3 A123 batteries. Specification of these batteries are unbelievable:

* Nominal voltage: 3.3V
* Nominal capacity: 2.3Ah
* Core cell weight: 70 grams
* Internal impedance: (1kHz AC) 8 mΩ typical
* Typical fast charge current: 10A to 3.6V CCCV
* 70A continuous discharge
* 120A, 10 sec pulse discharge
* Cycle life at 10C discharge, 100% DOD: over 1,000 cycles




The problem is that cell’s nominal voltage is 3.3V, and LiPo has 3.7V. Can we use it instead of LiPo? I had no time to mount these batteries into plane, but I’ve done some testings. First of all I’ve measured my Rhino 2S1P 7.4V Lipo 30C battery with Hextronix 3000 KV motor. Battery was slightly depleted, but motor was running quite ok. Measured current was 9.5A. Then I’ve measured A123 in 2S1P configuration. With same motor, at max rpm. The result was very similar, it was about 9,5A. I’ve connected 3 cells in series and connected to system. Motor was running insanely fast. I couldn’t measure the current, because my multimeter has 10 A limit. I didn’t try max RPMs. I don’t to damage my motor. Few users wrote in forums that they’ve damaged their motor with A123 3S1P configuration. In summary 2 A123 cells can replace somewhat depleted LiPo 7,4V battery :) . With 3 cells you can fly, but you should limit RPMs. When I have results how batteries work in the air, I’ll give you update.
Seems to be all right, just on high RPM engine starts to heat, ESC also.