My name is Berca Iulian, I am from Romania I followed the Politechnical University of Bucharest, Faculty of Electrical Engineering and i like to build things like electric cars , scooters, bicycles, quadcopters, brushless motors and inverters (ESC or motor controllers)

I am very passionate of electronics, now i’m working to improve the brushless controller and i look to build new motors and vehicles.

I`m willing to collaborate for different projects  or working in a creative company, to develop and test new technologies.

My youtube channel is on

If you have any question about this constructions you can send me an e-mail at

I`m designing in Eagle-cad : brushless controller circuits, dc motor controller circuits,protection circuits,etc.
I`m constantly improving my design for a robust brushless controller, with all protection needed (overcurrent, over-temperature,fault output, torque control) also ramp acceleration and deceleration and adapting timing angle.
Now i also work on a new design of a large 48 pole 60kW direct drive brushless motor.

I`m now building new laboratory and i would appreciate any small donations for new shop materials or equipment to improve designs, and make new interesting and innovative projects.

Experiment replication to observe the thrust in EmDrive device.

The device uses a magnetron to produce microwaves which are directed into a metallic, fully enclosed conically tapered high Q resonant cavity with a greater area at one end of the device, and a dielectric resonator in front of the narrower end. The inventor claims that the device generates a directional thrust toward the narrow end of the tapered cavity. The device (engine) requires an electrical power source to produce its reflecting internal microwaves but does not have any moving parts or require any reaction mass as fuel. If proven to work as claimed, this technology could be used to propel vehicles intended for all forms of travel including ground travel, marine travel, sub-marine travel, airflight and spaceflight.

EmDrive is a device invented by Roger Shawyer in 1999 and replicate with success by a team from China led by Yang Juan and a team from Nasa this year. The device was also tested in vacuum with the thrust is sill present so the air convection or other possible air movement is ruled out.

I will replicate the experiment and try to observe the thrust.

Materials for the drive.

  • Cooper sheet 0.3mm (initially intended 0.6mm)
  • transformer from a microwave oven (power ~ 800-1200W)
  • Magnetron from microwave oven: Anode cathode voltage ~ 4kV , and 3-4V @13A for the filament. Frequency 2,45Ghz.
  • solder, 4mm screws, pcb

Test equipment: Current measurement, voltage measurement, temperature measurement, micro-gram scale.




Frustrum 3D modeling in Autocad Inventor

emdrive homemade2

Connection diagram for the magnetron. Warning: charged capacitors can kill very easy. Always discharge the capacitor by putting 100kOhm resistor on the ends and also to the external case for your safety. After discharge put the ends in short circuit and wait for a couple of seconds to be absolute sure that there is no voltage.

emdriveAlmost all the materials arrived today:


Emdrive building

emdrive (4)emdrive (22)Today i will make the frustum to see the results.

emdrive (3) WP_20150513_00_22_34_ProToday after work, i will finish the setup and i will connect the frustum to a plate and suspend in the air from  4 nylon wires.

 After i power on  the temperature of the magnetron increased to 60 degree celsius.(140F) in around 5-6 seconds. If the magnetron does not have any load the temperature should rapidly increase i think, even though the microwave oven will not burn (overheat) if left on without any food inside.

In this paper : they say that the thrust comes after 20 seconds you power the magnetron. But in 20 seconds the magnetron will be very hot without proper cooling ( or maybe because the magnetron has no load)

I still do not know if the waveguide in a microwave oven plays other role than just feed the microwave in to the cavity.

Another think i want to test is to try to reduce the current on the filament with a separate power supply in hope that i will decrease the power in  “search”of some thrust.

In a few days i will receive 2 plates of  PCB single-sided, and i will try them instead of cooper ends.


Today i did the first test with the setup suspended in a pendulum. The power was applied  for 40 seconds. No thrust resulted 🙁

i will post the video soon.

Tomorrow i will move the magnetron above the middle of the frustum, to the small end. After power on for 40 sec temperature was 85 degree Celsius.

Next step will be adjusting the current of the filament and maybe the frequency by adding 2 separate coils over the magnets with adjustable current, to try to change the oscillating  frequency.



emdrive 008


Test No. 2

Modifications: magnetron moved to the smaller side

Coil added to the magnetron to try to adjust the frequency.

Still no visible thrust in the pendulum

WP_20150514_22_34_09_Pro WP_20150514_23_19_48_Pro WP_20150515_01_41_24_Pro



Test No 3, Thrust observed.

As you notice in the movie the weight of the foam on the frustum is 10.2gr and real weight is 3.58gr so the lever ratio is 1:2.894 in this way the real thrust was 0.508gr.

New tests will be done with the coil to see any change in thrust.

I will modify again the frustum and add adjustable length so resonance adjustment.

Emdrive test


I did not have time time to do any new setup.  Frequency counter has arrived, and i measured 2463Mhz.  . By changing the current in the coil around the magnet i will change the magnetic field created, overimpose to that of the magnet. With this change in magnetic field, the frequency output should change. I hope i can change enough to find the resonance frequency of the frustum and hope for higher thrust. The other method of finding the resonance is to adjust the length of the cavity. This can be done with a movable  plate and a screw . I can make that out of PCB.

frequency counter

frequency counter01




Hi, guys, i`m still alive . Sorry if i did not post anything this days. I noticed some guys think i died , relay strange. I do not have a Tweeter account by the way.

I flipped the cone in the original setup and i have the thrust downwards (scale goes positive). Unfortunately the thrust downwards is around 7 times smaller. difference on the scale is only 0.20 grams and is consistent with the power on and off .

Temperature discussion: during the tests the temperature of the frustum does not change to much maybe 1-2 degree.

The biggest change in the temperature is on thee fins of the magnetron. Can be as high as 80 degree Celsius. Definitely the air is going upward from the fins. (What is this meaning the the change in weight?)  The tests  shows that after power off the frustum weight is continue to decrease. up to – 0.30 grams at least. How we can explain this ?

Spring discussion:

Andy P. said something interesting : “When comparing the different thrusts, you will also have to take into account that in test 3.1 the thruster has to fight against the upward force of the spring onto which it is attached. This will lower the observed weight change on the scale, but does not necessarily mean the thrust is lower.”


This fight “against the spring” is real or not ? Lets assume this: you put 1kg on a scale and push TARE button. When you remove the weight it should not indicate -1Kg if the fight against the spring was real. Inside the scale is also a “spring” to keep the weight the the test should be the same.


I`m working the modify the cone now. Test No 4 will be with new setup.

Because i do not have a cooler to the magnetron i can not put a servomotor to continuously adjust the cavity length because  magnetron will heat fast. So i will need to manually adjust the length for each test to  observe the scale and then let the magnetron to cool ant test again.


First i will adjust in bigger steps 1cm smaller for each test.  then i will see witch one has the most thrust. After i will go around that value from mm to mm with the screw.

I do not own the picture with the adjustable setup. is from NasaSpaceflight forum.

emdrive 1b

Emdrive 4

emdrive 11

Emdrive 4

emdrive 10



emdrive 12

I hope others can make their own EmDrive and have positive results.


After successfully building a brushless controller i have decided also to build my own brushless motor. Auto-cad Inventor 3D cad software was used by me to design 3D model of the motor.

Before start to design something you need to know the RPM needed,torque needed,running voltage, max Amps. This formula is for calculating the torque if you have the power and the speed. motor Torque foruma  After reading the gear ratio on  Opel Agila it resulted that i need a speed of 4000RPM to reach 73.4Km/h in 3-rd gear ( i do not need more than that in a city). I decided to make a reverse outrunner motor design because is more easy to cool the stator if is outside. If the sator is inside is more difficult to cool down. The draw-back is that you loose torque because you have a smaller diameter of the rotor. I opted for a 48 slots (teeth) and 40 magnets design and i will have 142.5Nm of torque. You can check on to see all the motor slots and poles combinations.

  • Nominal Power: 45KW
  • Nominal Voltage: 230v
  • Nominal current:200A
  • Winding configuration : Delta
  • RPM: 2600
  • Torque 165Nm
  • Construction : 48 slot, 40Magnets Neodimium
  • Lamination Grade M330
  • Cooling : Glicol
  • Weight: 17Kg
  • Cooper weight ~ 2.7kg


  • Can be : delta or Star ( WYE)
  • Delta connection will give you higher power per cooper amount, higher RPM, higher current, lower phase voltage.
  • STAR will give you lower RPM, higher torque (1,73 more than dela) , higher the voltate, lower the current.
  • Can be: concentrated or fractional slot type.
  • If there are concentrated can be: LRK, distributed LRK etc.
  • A good winding scheme calculator can be found here.

winding scheme calculator




electric motor winding calculator

How to choose the magnets ?

After you choose the slots and poles, you need to choose the magnets. This is not very easy task because you need hi temperature magnets and are not so cheap and easy to find for your needed size. I bought custom made magnets from a Chinese website.

  •  The temperature rating for Neodymium is only a guide value. The actual temperature where magnet start to loose strength is size, shape and magnetic circuit dependent. If you have a magnet attached to a piece of steel  i will demagnetize at higher magnetic flux than in a free space.  On the other hand demagnetization temperature will be  lower if you subject the magnet to a strong opposite magnetic field such in a motor.
  • If the thickness of the magnet is bigger and  you will need a bigger magnetic field to start demagnetize it.

Temperature classification for neodymium magnets, N stays from Neo from .

  • N42           ≤80℃                                      magnet
  • N42M      ≤100℃
  • N42H       ≤120℃
  • N42SH     ≤150℃
  • N42UH    ≤180℃
  • N42EH     ≤200℃
  • N42VH    ≤230℃

Neodymium magnets needs a coating otherwise they will rust in contact with air.

The coatings can be: Nickel, Zinc, Phosphate, Epoxy, Gold and others.


Mathematical magnetic flux density analysis.

Next step is to run mathematical analysis in magnetic field to see if i have some areas with saturated magnetic field. We want to avoid core saturation. For this i used Finite Element Method Magnetics Tool A Windows finite element solver for 2D and axisymmetric magnetic, electrostatic, heat flow, and current flow problems with graphical pre- and post-processors brushless simulation FEMM motor simulationWe can observe that i have areas in pink color with to much magnetic flux, above 2Teslas, so i need to increase the thickness of the tooth to stay under 2 Teslas because the saturation of the laminations.

Multiple factors can have a huge difference in motor performance and efficiency:

This factors can be:

1. Maximum working frequency (depending on RPM and no. of poles). The frequency is calculated by next formula:

f= rps (motor rotation per second) x (nr. of poles/2). no.of poles is equal with no. of magnets.

Example for 1000Rpm: the rps will be 1000rpm/60s = 16,66 then  f=16,22 x 40poles/2 will result in: f= 333.2Hz

Because the losses in the core lamination are increasing  with increase (non linear) in frequency we want to have a frequency as low as possible for max motor RPM.  For exampe for lamination grade M330-50 the loses at 50Hz and 1Tesla are 1,29W/kg but 132W/kg at 1000Hz.

2. Proper combination between slots and poles count.

3. Material properties and thickness of stator and rotor laminations.

4. Air gap thickness.

5. Magnets grade.

6. Current density.

7. Slot fill factor.

8. Cogging torque. A summary of techniques used for reducing cogging torque:

  • Skewing stator stack or magnets
  • Using fractional slots per pole
  • Modulating drive current waveform
  • Optimizing the magnet pole arc or width
Brushless motor simulation

Brushless motor simulation 45KW 0.2mm lamination and 0.75mm airgap


motor simulation

Brushless motor simulation results


  • The final motor without the caps.
  • The motor with the caps and without the ball bearings.
  • I have also an YouTube video to present the 3D model.

Design stage Brushless motor60Kw




Brushless motor 50Kw


Close up view of the coils, magnets.







Brushless 40 poles 48 slotes 60kw


Air gap 0.75mm







Custom Magnets received  and tested by heating up to 120 Celsius to see if  any drop in magnetic field occurs.

Custom mamgnets brushless motor

Custom cut, N35UH grade, Phosphate coating magnets

Because the laser shop in Romania only had M330-50 grade, i was forced to use this material with higher losses for high RPM (frequency)


brushless motor design

Motor laminations  Grade M330-50 and other elements.

Drill to embed the screw in material.

Brusless motor construction

Drilling process

Brushless motor Rotor

Rotor assembly

Brushless motor homemade

Rotor with epoxy and kevlar to keep the magnets from flying from centrifugal force.

I used 14 strands of cooper 0.5mm in parallel.

48 poles brushless motor

Brushless motor winding

Winding the motor



brushless motor, car motor. Electric car brushless motor.

Motor finished, ready for testing.

brushless motor contruction, brushless motor water cooling

Motor Water Cooling

brushless motor homemade

Brusless motor ready for testing under load.







In the last 6 months i`m working at my electric car conversion. The car was in good shape and was a good candidate for the conversion. 
After i found a place where to work to the car (a garage) i took out the ICE and i begun to measure all the interior spaces to see if the motor will fit.
Today (27.07.2013) i did the first test on the electric motor to see if the motor fits ok inside and join well the gear box. 




After i removed the engine i start taking measurements for creating  the mounting plates.




The car with the engine removed. Now i have space to work.



This motor is a Permanent Magnet Dc motor with 100Nm and 1900RPM total power will be limited to ~ 20Kw depending on how i will manage to remove the heat from the  motor.

  I also received the Batteries:
  Model : A123 Systems.
  Capacity : 20Ah
  Nominal voltage 3.3v
  Continuos Discarge 600A
  Puls discharge 1200A
  Total number of cells in series : 72

 As the DC motor Controllers are to expensive for my budget, and i have  experience in electronics i designed my own pwm controller using Eagle  Cad program.

Specs: Max voltage: 400V
            Max current: 200A
Adjustable current limitation feature,
Undervoltage lockout, thermal shut down, curent sense.




New 30 A123 Systems cells arrived.

The mounting plate for 24 cells in series

1234_cr4The schematic and PCB where designed in Eagle cad.

electric car PWM controller




This is the power module formed by 10 mosfets 500V 32A in parralel.

Now new Toyota Prius first generation IGBT module arrived from ebay.

Toyota Prius invertor igbt module


The internal schematic, Each igbt inside has 27nF gate capacitance.

toyota prius  invertor

Opel agila electric

electric car conversion




















Homemade Qadrocopter

Posted: 13th November 2012 by iulian207 in Homemade stuff
Tags: ,

Motors: 12v Dc Brushless 1000Kv, 20Amp
ESC Mystery 30A Blue series
Batt: 3 cell 2300mAh Li-Fe A123 Systems.
Home Made Contoller Board Atmega48, 3xMurata Gyroscops.

First you need to buy the materials:

  1.   Tow cubic meter of extruded polystyrene from insulation materials.
  2.   6 pcs  wooden planks
  3.   6 pcs threaded screw
  4.   3 meters of fiberglass mash.
  5.    Polyurethane foam
  6.    Some hot wire machine to make a trench in polystyrene



Now the actual construction.

How to make a pontoon boat step by step

How to make a pontoon boat step by step

pontoon boat build




How to make a pontoon boat step by step


First mount of the Boat to see if everything is ok,

And it was ok 😀

Going to the lake site to test the boat.







Electrical motor 32lbs of trust.









I just finished a new controller for my electric scooter project.

The current is limited to 50Amp but it can sustain at least 80Amp.

It is also based on Mc33035 IC and it is capable to decode signal from the hall sensors.



Now i`m also building a new Go kart controller.I will use the same logic board as the scooter.

The board was designed in Eagle Cad:

The Brushless motor Controller Schematic: (command module)


I used photo resist PCB to make the circuit :




After that i used  Ferric Chloride etchant to remove all the unnecessary cooper.

And the result was :

Here is a picture of my Homemade Electric Go Kart:





Today i received the charger and i put it to the test.
I am very satisfied about the capability of the charger.


ICharger 1010B+ insite board




This is the best cheapest DC motor speed controller circuit that you can find on internet.

In the past i tried wit NE555 and other circuits but the results were every time in shorted mosfet`s 😀 and not stoppable GO KART (not very good thing when you do not have a big red kill switch).

In in the following i will  present my DC Motor speed controller capable of adjusting speed (PWM)  form 0 – 100% and the frequency form  ~ 400Hz to 3kHz, based on LM339 comparator.

The power supply is from 14-30volts,  expandable to practical any value with little modification.

From R15 VR 10k you can adjust the speed from 0 -100%

From R14 VR 100k you can adjust the frequency.

If the jumper JP1 is shorted you can adjust the PWM frequency from 400Hz  to 3kHz. If jumper is open Freq is fixed at 100Hz.

The circuit is designed in Eagle cad 6.2



You can use almost any Channel N mosfet`s you want. The fets will be mounted on a radiator if the current is  higher than 2-5 amp.

It is possible to increase the voltage supply to any value if you separate the power to the logic circuit from power to the load and mosfet`s

This is my second prototype of dc motor controller


Depending on Rds ON value of the mosfet`s you will need a smaller or bigger radiator.

The wires will be at least 12 AWG for a 30 -35  amp load.

For any questions you can  ask me any time via my e-mail found on about menu.

Success with the circuit.

This is the real life testing of the circuit.




High quality improved PWM controller based on MC33035 IC.

Schematic :





Eagle 6.1 design.



I used as mosfet driver the TC4452 IC with is capable of 12Amp output.

I used this schematic in conjunction with power stage formed by 10 mosfets in parralel with all gate connected via 10pcs  20 ohm resistor to the output of the IC driver.

For current sense circuit i used allegro sensor ACS758

50-200A current sensor IC

The Allegro CA and CB package current sensor ICs are fully integrated current sensor solutions. They contain the primary conductor, concentrating ferromagnetic core and the analog output Hall-effect linear in a single IC package. The conductor resistance is a typical of 100 µΩ for ultra low power loss when sensing current up to 200 A. These sensors are automotive grade devices that can take the heat and deliver highly accurate open loop current sensing in the most harsh applications environments.

The Allegro medium current devices are much smaller than bulky current transformers and have the added advantage of sensing both AC and DC currents. The package design also provides galvanic isolation to 3000 VRMS and can be used in many line side applications.


Controller Parameters and Features

  1. Features:
  • – Uses analogical chip with no software inside.
  • – Work only with sensored brushless motors.
  • – Speed adjustable via a potentiometer
  • – Adjustable acceleration deceleration
  • – Loop Control
  • – Fackword / Forward
  • – Dynamic breaking
  • – Over-current sense from external  shunt resistor 100mV threshold level.
  • – Overheat protection.
  • – Undervoltage protection.
  • – Fully Accessible Error Amplifier for Closed Loop Servo Applications
  • – Adjustable PWM frequency
  • – 6.25 V Reference Capable of Supplying Hall Sensor Power

I used Eagle Cad to make the schematic and the board.


New version of the schematic and it is easy to understand :

brushless controller schematic



This is another version with more mosfets in parallel and different drivers.

I used only N-channel mosfets in junction with  IR2110 half bridge mosfet driver.

You also need a inversor gate for the Top Drives (4049)

This is the newest version with 4 mosfet in parallel per switch  IR4110 but doesn`t matter what mosfet you use as the volgate and current is good for you.

top driver_cr

The pin 3 JP8  goes  to the Board 2 (Comand module) at JP20 pin 1

The pin 1 JP8 goes to Board 2 JP20 pin 2

The Pin 3 JP9 goes to the Board 2 at JP20 pin 3

The pin1  JP9 goes to the Board 2 JP20 at pin 4

The pin 3 JP10 goes to the Board 2 JP20 pin 5

The pin 1 JP10 goes to the Board 2 jp20 pin 6

The pin 1 gnd of the JP5, JP6. JP7  can be left in the air. because gnd is common.





Brushless controller schematic mc33025

In the upper part you can se the current sensor Allegro ACS758 200A.

You can also see in the left the bottom module next to it the driver+ top module.

In the right upper corner command module and in lower corner a dc-dc converter module from ebay.

to power the cmomand module and the driver board.

You can put any mosfets channel N you need. Best are with internal resistance as low as possible and higher current.

Board 3 mc33035 Brushless controller







This diagram was draw by a website visitor by name “Bill Catalena” from my specification.

Brushless controller schematic mc33035

This is the 3-rd Board with bottom mosfets.

The U,V,W need to be connected to the U,V,W to the top part of the mosfets.

The Pin 1 of the JP1 goes to the pin 2 JP5 from the Board 1

The pin 2 of the JP1 goes to the pin 2 JP6 from the Board 1

The pin 3 of the JP1 goes to the pin 2 JP7 from the Board 1

Ground pin is connected from the power supply of 48v




bottom mosfets mc33035 driver



Toyota Prius IGBT module

This toyota prius igbt module inverter was from a scrap and i took apart the driver board. now i-m building my own diver board.igbt module toyota prius

In order to be able so start the motor  you need to put a floating dc power supply to the IR2110

I have tested some of 1200V 600A FZ600R12KE3 IGBT module and the input capacitance was ~ 55nF. The time rinse obtained with avago IC ACPL-P343 was 1,2uS at 12Khz, not so good  if you want the switching losses low. the datasheet:

In this video i used 12 Mosfets irf3205z and IR2110 driverInfineon IGBT







Now the big thing : Testing the Electric Go Kart to measure the results