Component Experiments

Simple Test of Optocoupler and IGBT

Simple test of new FGA50S110P and 6N136:



Original gEDA schematic

const int sensorPin = A0;
const int ledPin = 13;
const int sensorFactor = 5;
const int sensorLimit = 1024 >> sensorFactor;

void setup()
{
  pinMode (ledPin, OUTPUT);
}

void loop()
{  
  int sensorValue = analogRead (sensorPin) >> sensorFactor;

  digitalWrite (ledPin, HIGH);

  delay (sensorValue);

  digitalWrite (ledPin, LOW);

  delay (sensorLimit-sensorValue);
}


Using Current Amplifier for Gate


Original gEDA schematic

const int sensorPin = A0;
const int ledPin = 13;
const int sensorFactor = 5;
const int sensorLimit = 1024 >> sensorFactor;

const int TEMPLIMIT = 700;
const int TEMPSAFE = 600;

int sensorValue = 0;
int OverTemp = 0;

void setup()
{
  pinMode (ledPin, OUTPUT);

  Serial.begin (115200);
}

void loop()
{  
  int TempSensor = analogRead (A1);

  Serial.print (TempSensor);
  Serial.print ('\t');

  sensorValue = analogRead (sensorPin) >> sensorFactor;

  Serial.print (sensorValue);
  
  if (TempSensor < TEMPLIMIT && (!(OverTemp) || TempSensor < TEMPSAFE))
  {  
      digitalWrite (ledPin, HIGH);

      delay (sensorValue);

      digitalWrite (ledPin, LOW);

      delay (sensorLimit-sensorValue);
      
      OverTemp = 0;
  }
  else
  {
      digitalWrite (ledPin, LOW);
      
      OverTemp = 1;
  }
  
  Serial.print ('\n');
}


const int sensorPin = A0;
const int ledPin = 13;
const int sensorFactor = 5;
const int sensorLimit = 1024 >> sensorFactor;

const int TEMPLIMIT = 800;
const int TEMPHIGH = 750;
const int TEMPSAFE = 750;

int sensorValue = 0;
int OverTemp = 0;

void setup()
{
  pinMode (ledPin, OUTPUT);

  Serial.begin (115200);
}

void loop()
{  
  int TempSensor = analogRead (A1);

  Serial.print (TempSensor);
  Serial.print ('\t');

  sensorValue = analogRead (sensorPin) >> sensorFactor;
  
  if (TempSensor > TEMPHIGH)
  {
      sensorValue -= ((TempSensor - TEMPHIGH) >> 2);
      
      if (sensorValue < 0)
          OverTemp = 1;          
  }

  Serial.print (sensorValue);
  
  if (TempSensor < TEMPLIMIT && (!(OverTemp) || TempSensor < TEMPSAFE))
  {
      digitalWrite (ledPin, HIGH);

      delay (sensorValue);

      digitalWrite (ledPin, LOW);

      delay (sensorLimit-sensorValue);
      
      OverTemp = 0;
  }
  else
  {
      digitalWrite (ledPin, LOW);
      
      OverTemp = 1;
  }
  
  Serial.print ('\n');
}


Original gEDA schematic

  if (TempSensor < TEMPLIMIT && (!(OverTemp) || TempSensor < TEMPSAFE))
  {
      digitalWrite (ledPin, HIGH);

      delayMicroseconds (sensorValue);

      digitalWrite (ledPin, LOW);

      delayMicroseconds (sensorLimit-sensorValue);
      
      OverTemp = 0;
  }
  else
  {
      digitalWrite (ledPin, LOW);
      
      OverTemp = 1;
  }

Using Higher Gain Current Amplifier


Original gEDA schematic




Original gEDA schematic

const int TEMPLIMIT = 800;

const int CONTROLINPUT = A0;
const int TEMPINPUT    = A1;
const int MOTOROUTPUT  = 13;

const int TEMPHIGH    = 800;
const int TEMPSAFE    = 700;

int ControlInput = 0;
int TempInput = 0;
int MotorOutput = 0;
int TempHigh = 0;

void setup()
{
  Serial.begin(115200);
}

void loop()
{
  ControlInput = analogRead(CONTROLINPUT);
  TempInput = analogRead(TEMPINPUT);

  MotorOutput = map(ControlInput, 0, 1023, 0, 255);
  
  if (TempHigh)
  {
      if (TempInput < TEMPSAFE)
          TempHigh = 0;
          
      analogWrite(MOTOROUTPUT, 0);
  }
  else
  {
      if (TempInput > TEMPHIGH)
          TempHigh = 1;
          
      analogWrite(MOTOROUTPUT, MotorOutput);
  }

  Serial.print("ControlInput = " );
  Serial.print(ControlInput);
  Serial.print("\tTempInput = " );
  Serial.print(TempInput);
  Serial.print("\t MotorOutput = ");
  Serial.println(MotorOutput);

  delay(100);
}

ControlInput = 291	TempInput = 770	 MotorOutput = 72
ControlInput = 292	TempInput = 770	 MotorOutput = 72
ControlInput = 286	TempInput = 770	 MotorOutput = 71
ControlInput = 291	TempInput = 777	 MotorOutput = 72
ControlInput = 288	TempInput = 769	 MotorOutput = 71
ControlInput = 290	TempInput = 770	 MotorOutput = 72
ControlInput = 290	TempInput = 770	 MotorOutput = 72
ControlInput = 288	TempInput = 770	 MotorOutput = 71
ControlInput = 290	TempInput = 769	 MotorOutput = 72
ControlInput = 290	TempInput = 769	 MotorOutput = 72
ControlInput = 290	TempInput = 769	 MotorOutput = 72
ControlInput = 291	TempInput = 769	 MotorOutput = 72
ControlInput = 290	TempInput = 770	 MotorOutput = 72
ControlInput = 290	TempInput = 769	 MotorOutput = 72
ControlInput = 293	TempInput = 769	 MotorOutput = 73
ControlInput = 290	TempInput = 769	 MotorOutput = 72
ControlInput = 291	TempInput = 768	 MotorOutput = 72
ControlInput = 291	TempInput = 768	 MotorOutput = 72
ControlInput = 290	TempInput = 769	 MotorOutput = 72
ControlInput = 288	TempInput = 768	 MotorOutput = 71
ControlInput = 290	TempInput = 769	 MotorOutput = 72
ControlInput = 288	TempInput = 770	 MotorOutput = 71
ControlInput = 290	TempInput = 769	 MotorOutput = 72





Higher Speed Optocouplers

This is a test of the 6N136 in inverting mode.

Original gEDA schematic

The test is done both without then with the biasing resistor (R3)

Red without R3, Orange with R3

TIME:200nS/div, Blue:TP1(2vDC/div), Red:TP2+R3(2vDC/div), Orange:TP2-R3(2vDC/div).


TIME:200nS/div, Blue:TP1(2vDC/div), Red:TP2+R3(2vDC/div), Orange:TP2-R3(2vDC/div).

The effect of the biasing resistor is to speed up the switch off at the cost of switch one time.
In the circuit a 10k resistor added means the rise and fall times are now roughly equal at 200nS


Original gEDA schematic

Add the trinary circuit, also using a DC-DC converter for high side gate supply

Original gEDA schematic


TIME:200nS/div, Blue:TP1(2vDC/div), Red:TP5(20vDC/div), Magenta:TP3(20vDC/div), Cyan:TP4(20vDC/div).


TIME:200nS/div, Blue:TP1(2vDC/div), Red:TP5(20vDC/div), Magenta:TP3(20vDC/div), Cyan:TP4(20vDC/div).


Using Two NPN IGBTs in Push-Pull


Original gEDA schematic




Direct Coupling of the MCU


Original gEDA schematic




Original gEDA schematic

Thermistor Reading

Thermister transfer plots for different lowside loads and linear approximations:

Original GNumeric Spreadsheet
const int CONTROLINPUT = A0;
const int TEMPINPUT    = A1;
const int MOTOROUTPUT  = 11;
const int MOTOROFF     = 255;

const int TEMPHIGH    = 450;
const int TEMPSAFE    = 370;

int ControlInput = 0;
int TempInput = 0;
int TempInC = 0;
int MotorOutput = 0;
int TempHigh = 0;

void setup()
{
  Serial.begin(115200);
  Serial.println("Grahams Motor Test, "__DATE__" ("__TIME__")");
}

void loop()
{
  ControlInput = analogRead(CONTROLINPUT);
  TempInput = analogRead(TEMPINPUT);
  
// For 1k lowside: 45C < Temp ~ +-1C < 112C
  TempInC = (TempInput * 10 / 76) + 21;
  
  if (TempInput < TEMPSAFE)
      TempHigh = 0;
  
  if (TempInput > TEMPHIGH)
      TempHigh = 1;

  MotorOutput = ControlInput >> 2;
          
  if (TempHigh)
      analogWrite(MOTOROUTPUT, MOTOROFF);
  else
      analogWrite(MOTOROUTPUT, MotorOutput);

  Serial.print("ControlInput = " );
  Serial.print(ControlInput);
  Serial.print("\tTempInput = " );
  Serial.print(TempInput);
  Serial.print("(");
  Serial.print(TempInC);
  Serial.print("C)\tMotorOutput = ");
  Serial.println(MotorOutput);

  delay(100);
}
Grahams Motor Test, Apr 20 2016 (12:12:09)
ControlInput = 892	TempInput = 165(42C)	MotorOutput = 223
ControlInput = 891	TempInput = 165(42C)	MotorOutput = 222
ControlInput = 891	TempInput = 166(42C)	MotorOutput = 222
ControlInput = 892	TempInput = 165(42C)	MotorOutput = 223
ControlInput = 891	TempInput = 165(42C)	MotorOutput = 222
ControlInput = 892	TempInput = 162(42C)	MotorOutput = 223
ControlInput = 891	TempInput = 165(42C)	MotorOutput = 222
ControlInput = 891	TempInput = 165(42C)	MotorOutput = 222
ControlInput = 891	TempInput = 165(42C)	MotorOutput = 222
ControlInput = 892	TempInput = 164(42C)	MotorOutput = 223
ControlInput = 891	TempInput = 165(42C)	MotorOutput = 222
ControlInput = 892	TempInput = 164(42C)	MotorOutput = 223
ControlInput = 891	TempInput = 160(42C)	MotorOutput = 222
ControlInput = 891	TempInput = 164(42C)	MotorOutput = 222
ControlInput = 891	TempInput = 164(42C)	MotorOutput = 222
ControlInput = 891	TempInput = 166(42C)	MotorOutput = 222
ControlInput = 891	TempInput = 164(42C)	MotorOutput = 222
ControlInput = 891	TempInput = 165(42C)	MotorOutput = 222
ControlInput = 891	TempInput = 155(41C)	MotorOutput = 222
ControlInput = 891	TempInput = 165(42C)	MotorOutput = 222
ControlInput = 892	TempInput = 164(42C)	MotorOutput = 223
ControlInput = 892	TempInput = 165(42C)	MotorOutput = 223
ControlInput = 891	TempInput = 165(42C)	MotorOutput = 222



Rise time about 10uS:

Fall time about 10uS:

Someone from the Arduino Community has coded a conversion algorithm for the thermistor setup used above [link].
A comparison graph for different low-side resistors against the ND03N00103J is here:

This verifies the use of the 1k low-side (cyan) and means we can now refine the temperature readings.

High Side


Original gEDA schematic




Low Side


Original gEDA schematic




Original gEDA schematic

Gate charging with a current source


Original gEDA schematic




Original gEDA schematic

Rise time in the order of 10uS:

Fall time in the order of 1uS:

Adding current amplifier for gate


Original gEDA schematic




Original gEDA schematic

Original inkscape drawing




Rise time is about 4uS:

Fall time is about 500nS:

Continued on Controller Software, Modelling and Tuning and Field Control


Latest version of DueTest.cpp:



MicroChip Driver IC

Sourced a new driver chip made by Microchip:



Constructed a simple test to see if it works.


Original gEDA schematic

Seems to work quite well.
Below is a test as 62.5kHz.


TIME:2uS/div, Blue:TP1(2vDC/div), Red:TP2(4vDC/div), Green:TP3(10vCD/div).

..and tested with 320A IGBT (IXGN320N60A3)


TIME:2uS/div, Blue:TP1(2vDC/div), Red:TP2(4vDC/div), Green:TP3(10vDC/div).

From these test it seems the driver chip is working in the nanosecond switching range with the low power motor switching within a microsecond too.

Short switch testing shows we can have sub-microsecond PWM mark and space.


TIME:500nS/div, Blue:TP1(2vDC/div), Red:TP2(4vDC/div), Green:TP3(10vDC/div).


TIME:500nS/div, Blue:TP1(2vDC/div), Red:TP2(4vDC/div), Green:TP3(10vDC/div).

Seems to work OK until a low frequency PWM is used.
This show the switch to space at 30Hz:


TIME:200nS/div, Blue:TP1(2vDC/div), Red:TP2(10vDC/div), Green:TP3(40vDC/div), Orange:Gate(10vDC/div).

At higher frequencies (4kHz) the signal is much better:


TIME:200nS/div, Blue:TP1(2vDC/div), Red:TP2(10vDC/div), Green:TP3(40vDC/div), Orange:Gate(10vDC/div).

Unfortunately, this voltage spike (+30v/-50v, above) on the gate is beyond the specification of the IGBT (+-20v), which causes it to blow.

Adding a Schottky parallel diode (D3) makes quite a difference.
Also gate resistor of 11R and a gate discharge diode were added so the turn on times were extended.


Original gEDA schematic


TIME:200nS/div, Blue:TP1(2vDC/div), Orange:TP2(10vDC/div), Red:TP3(10vDC/div), Green:TP4(40vDC/div).

It seems D3 is just accounting for the long wires connecting the gate circuit.



So these were shortened and the result was also better (without D3).




Original gEDA schematic


TIME:200nS/div, Blue:TP1(2vDC/div), Orange:TP2(10vDC/div), Red:TP3(10vDC/div), Green:TP4(40vDC/div).

And now we have success with the high power IGBT.


Original gEDA schematic



This is minimum mark:


TIME:500nS/div, Blue:TP1(2vDC/div), Orange:TP2(10vDC/div), Red:TP3(10vDC/div), Green:TP4(200vDC/div).

This is with more (about 5%) mark:


TIME:500nS/div, Blue:TP1(2vDC/div), Orange:TP2(10vDC/div), Red:TP3(10vDC/div), Green:TP4(200vDC/div).

It's still not perfect as the gate is being taken up to +50vDC (on a 12vDC motor) for about 10nS with higher mark, but at least the IGBT is working stably.

Testing at 30Hz PWM frequency on the milk float motor on 4 x 12v/80Ah batteries



Minimum mark:

TIME:500nS/div, Blue:TP1(2vDC/div), Red:TP4(200vDC/div).

Spinning up:

TIME:500nS/div, Blue:TP1(2vDC/div), Red:TP4(200vDC/div).

This is probably cleaner as the the milk float motor is series wound and has no permanent magnets.

Peak current during testing 92.9 amps:


The main cause of this has to be RF feedback as it's affecting the MCU output signal.
Adding some shielding will probably stop this.


Trinary 3-Phase

Now the basic power testing shows it'll work, the testing moves to the patent 3-phase trinary controller.

UnoBLDC.cpp

static const char Trinary[][TRITS+1] = {
    "++-",
    "+ -",
    "+--",
    "+- ",
    "+-+",
    " -+",
    "--+",
    "- +",
    "-++",
    "-+ ",
    "-+-",
    " +-" 
};


Original gEDA schematic


TIME:500uS/div, Blue:TP1(1vDC/div), Brown:TP2(1vDC/div), Green:TP3(1vDC/div).

The trimmers were set to arbitrarily offset the open-circuit states so we can see them.
Also the mark and space were just short and equal so we can see the switching.

Just to see the transitions of a single channel (blue trace), from high to off then low, off and back to high

TIME:50uS/div, Blue(2vDC/div), Red(2vDC/div), Green(10vDC/div), Yellow(10vDC/div).
Off to low:

TIME:200nS/div, Blue(2vDC/div), Red(2vDC/div), Green(10vDC/div), Yellow(10vDC/div).
Low to off:

TIME:200nS/div, Blue(2vDC/div), Red(2vDC/div), Green(10vDC/div), Yellow(10vDC/div).
Off to high:

TIME:200nS/div, Blue(2vDC/div), Red(2vDC/div), Green(10vDC/div), Yellow(10vDC/div).
High to off:

TIME:200nS/div, Blue(2vDC/div), Red(2vDC/div), Green(10vDC/div), Yellow(10vDC/div).

Simple PWM testing

Using a single channel:


Original gEDA schematic


TIME:500uS/div, Blue:TP1(2vDC/div), Brown:TP2(2vDC/div).

TIME:100nS/div, Blue:TP1(2vDC/div), Brown:TP2(2vDC/div).

TIME:100nS/div, Blue:TP1(2vDC/div), Brown:TP2(2vDC/div).

IGBT testing:


Original gEDA schematic


TIME:1mS/div, Blue:TP1(2vDC/div), Brown:TP2(4vDC/div).

TIME:1uS/div, Blue:TP1(2vDC/div), Brown:TP2(4vDC/div).

TIME:1uS/div, Blue:TP1(2vDC/div), Brown:TP2(4vDC/div).

Low side:


Original gEDA schematic


TIME:100uS/div, Blue:TP1(2vDC/div), Red:TP2(1vDC/div), Green:TP4(10vDC/div), Yellow:TP5(10vDC/div).

TIME:1uS/div, Blue:TP1(2vDC/div), Red:TP2(1vDC/div), Green:TP4(10vDC/div), Yellow:TP5(10vDC/div).

TIME:1uS/div, Blue:TP1(2vDC/div), Red:TP2(1vDC/div), Green:TP4(10vDC/div), Yellow:TP5(10vDC/div).


Original Inkscape diagram

High side:


Original gEDA schematic


Optocoupled Gate Driver

NOTE: The HCPL-3120 gate driver will not work under 15vDC.
These tests were done at 18vDC which seemed to be the minimum to achieve stability.


Original gEDA schematic

For the test, R1 was tuned to give the shortest but cleanest transition on the IGBT gate.
This was 13R4.


TIME:200nS/div, Blue:TP1(2vDC/div), Red:TP2(4vDC/div).

TIME:200nS/div, Blue:TP1(2vDC/div), Red:TP2(4vDC/div).

TIME:20uS/div, Blue:TP1(2vDC/div), Red:TP2(4vDC/div).

Transition times are in the order of 200nS, with a delay about the same so fast for an optocoupled part.
The noise would be improved by using shorter connections.


Original gEDA schematic

Op Amp for Trinary

Due to it's 84MHz CPU speed the Arduino Due is a better choice for an SVM controller.


Original gEDA schematic

      MOV r4, REG_PIOC_SODR  
      MOV r5, REG_PIOC_CODR  
      MOV r6, REG_PIOC_OER   
      MOV r7, REG_PIOC_ODR   
      MOV r8, REG_PMC_PCER0  
      
      MOV r0, #0xFF             
      LSL r0, #2                 
      
      MOV r9, #0x2000           
      STR r9, [r8]                ; PMC to enable Parallel I/O Controller C (PIOC, PID13)
      
  L1: STR r0, [r4]                ; HIGH
      STR r0, [r6]                ; output
      NOP                       
      NOP                       
      NOP                       
      NOP                      
     
      NOP                       
      STR r0, [r7]                ; OFF (input)
      NOP                       
      NOP                       
      NOP                      
      
      STR r0, [r5]                ; LOW
      STR r0, [r6]                ; output
      NOP                       
      NOP                       
      NOP                    

      NOP                       
      STR r0, [r7]                ; OFF (input)
      
      B L1                        ; repeat


TIME:20nS/div, Blue:TP1(1vDC/div).

Here you can see the waveform running at about 14MHz.

Using the MicroChip MCP6002 which is rated at 1MHz, and at a lower frequency of about 3kHz it can be see that the theory is good,
but it's just far too slow to be useful.

TIME:100uS/div, Blue:TP1(1vDC/div), Red:TP2(2vDC/div), Green:TP3(2vDC/div).


TIME:5uS/div, Blue:TP1(1vDC/div), Red:TP2(2vDC/div), Green:TP3(2vDC/div).

Transition times are in the region of 5uS.

A 15MHz TI RC4560 was also tested which has better transition times in the region of 1uS.


TIME:500nS/div, Blue:TP1(1vDC/div), Red:TP2(2vDC/div), Green:TP3(2vDC/div).

...but the part seemed to have a problem running from a single rail supply


TIME:5uS/div, Blue:TP1(1vDC/div), Red:TP2(2vDC/div), Green:TP3(2vDC/div).

Low side seems to switch OK, but the high side was also switching for low side.

Reverting to Bipolar Design


Original gEDA schematic

Checking the bipolar transition times at approx 1MHz.


TIME:200nS/div, Blue:TP1(1vDC/div), Red:TP2(1vDC/div), Green:TP3(1vDC/div).

So here transition times are in the region of 100nS.


Original gEDA schematic

Adding the SN74LS14 Schmitt Trigger Inverters brings transition times to 50nS with around 100nS delay.


TIME:200nS/div, Blue:TP1(2vDC/div), Red:TP3(2vDC/div), Green:TP4(2vDC/div), Yellow:TP5(2vDC/div).

All these times are an order of magnitude shorter than the 1uS needed, so this will work well.

The output is a bit noisy due to the wiring, but the Schmitt Trigger inverters seem fine with 3v3 logic input.
At 3kHz the 3v3AC trinary to twin 5vDC binary is clear.


TIME:100uS/div, Blue:TP1(1vAC/div), Green:TP4(2vDC/div), Yellow:TP5(2vDC/div).


Electromagnets

Since a new patent motor is being designed this needs some research into inductors and magnetic circuitry.

Some preliminary research shows that the most powerful magnets for the same volume are electromagnets.
Also the copper in an electromagnet is much cheaper and easier to use than cobalt or neodymium, and far more environmentally friendly.

It is possible, and the subject of more research, that a motor using an electromagnetic exciter
would be less efficient than one made from permanent magnets.

	Volts	Current	Ohms		
air	0.822	3	0.274		-0.002
	1.14	4.2	0.271		0.001
	1.52	5.6	0.271	0.272	0.001
bar	0.833	3.2	0.260		0.007
	1.17	4.3	0.272		-0.005
	1.54	5.7	0.270	0.268	-0.003
					
air	5.1	4.8	1.063		0.016
	4.3	3.9	1.103		-0.024
	3.45	3.2	1.078		0.001
	2.55	2.4	1.063		0.016
	1.85	1.7	1.088	1.079	-0.009
bar	1.85	1.7	1.088		0.044
	2.6	2.3	1.130		0.002
	3.49	3.1	1.126		0.006
	4.37	3.8	1.150		-0.018
	5.25	4.5	1.167	1.132	-0.034
Original GNumeric spreadsheet
CoilTest.ino


Original gEDA schematic



Field Strength of a Planar Solenoid

Total volume is 55mm x 1.12mm = 61.6mm2 (not subtracting the central hole)

So the only way to work this out is by treating each loop separately.
From Magnetic Field of Current Loop (HyperPhysics / Georgia State University)

B = μ0 x current / ( 2 x radius )

radius B length
5.00 mm 75.398 µT 31.42 mm
5.53 mm 68.205 µT 34.73 mm
6.11 mm 61.697 µT 38.39 mm
6.75 mm 55.811 µT 42.44 mm
7.47 mm 50.486 µT 46.92 mm
8.25 mm 45.669 µT 51.87 mm
9.13 mm 41.312 µT 57.34 mm
10.09 mm 37.370 µT 63.38 mm
11.15 mm 33.805 µT 70.07 mm
12.33 mm 30.580 µT 77.46 mm
13.63 mm 27.662 µT 85.63 mm
15.07 mm 25.023 µT 94.66 mm
16.65 mm 22.636 µT 104.65 mm
18.41 mm 20.476 µT 115.68 mm
20.35 mm 18.522 µT 127.88 mm
22.50 mm 16.755 µT 141.37 mm
Total 631.408 µT 1.184 m



Total field strength is 0.1mT per Amp.

So this is the field strength at the centre of the magnet, but in a motor the rotor would be close to all the windings, including the outer loops.
This means the effect on a large flat plate would be more that something requiring the field being focused at the centre.

Also given that the idea is to create a field vector, rather than some localised field at the magnet, the graph above is not portraying the true strength.