ElectroxX
Newbie level 6
- Joined
- Dec 21, 2011
- Messages
- 12
- Helped
- 0
- Reputation
- 0
- Reaction score
- 0
- Trophy points
- 1,281
- Activity points
- 1,421
I'm making a controller for a DC generator. The generator in question is connected to a windturbine so is has highly varying levels of input voltage and current.
I have an idea about how to do it and would like to discuss it.
The basic idea is to use the DC machine's inductance and not any external coils. The tricky thing here is that the DC machine's inductance is merged together with it's voltage source.
The current concept is using a buck-boost type circuit for the control:
The double blocking FETS are like the blocking diode in a regular boost converter when the setup is runned as boost, and as the main FET when runned as a buck converter.
1. As the coil and the voltage sources are bundled together in the DC machine, I can't freewheel the energy stored in the inductance with a diode when the circuit is runned as a buck. The solution I have is to dump the energy into a snubber capacitor.
However, when running the circuit as a boost this snubber capacitor would absorb the whole voltage spike and ruin it's functionality.
Therefore I have a selectable snubber that can be turned off when running the circuit as a boost. Is this a good idea?
2. How to accomplish the control of the transistors?
There are one signal for each transistor cluster, and also a fourth digital signal B that indicate wether the circuit is runned as a boost or buck. Q2 and Q3 is controlled by the same signal Q2. Each transistor can be controlled by a PWM signal Qx_PWM. The whole thing is simply described in boolean logics by
Q2 = B AND NOT Q1 OR Q2_PWM
Q1 = Q1_PWM
Q4 = NOT B
Should I buy logic gates and in that case which ones? Or build my own? A PLD feels like overkill for this. I'm using an Atmega 328 for PWM. Probably two PWM channels on the same clock could be configured to work inversely in relation to each other. Would that be a good option?
I want active rectifying and not schottkys as this is a low voltage- high current application.
Typical inductor current is 30A and input voltage around 30.
I have an idea about how to do it and would like to discuss it.
The basic idea is to use the DC machine's inductance and not any external coils. The tricky thing here is that the DC machine's inductance is merged together with it's voltage source.
The current concept is using a buck-boost type circuit for the control:
The double blocking FETS are like the blocking diode in a regular boost converter when the setup is runned as boost, and as the main FET when runned as a buck converter.
1. As the coil and the voltage sources are bundled together in the DC machine, I can't freewheel the energy stored in the inductance with a diode when the circuit is runned as a buck. The solution I have is to dump the energy into a snubber capacitor.
However, when running the circuit as a boost this snubber capacitor would absorb the whole voltage spike and ruin it's functionality.
Therefore I have a selectable snubber that can be turned off when running the circuit as a boost. Is this a good idea?
2. How to accomplish the control of the transistors?
There are one signal for each transistor cluster, and also a fourth digital signal B that indicate wether the circuit is runned as a boost or buck. Q2 and Q3 is controlled by the same signal Q2. Each transistor can be controlled by a PWM signal Qx_PWM. The whole thing is simply described in boolean logics by
Q2 = B AND NOT Q1 OR Q2_PWM
Q1 = Q1_PWM
Q4 = NOT B
Should I buy logic gates and in that case which ones? Or build my own? A PLD feels like overkill for this. I'm using an Atmega 328 for PWM. Probably two PWM channels on the same clock could be configured to work inversely in relation to each other. Would that be a good option?
I want active rectifying and not schottkys as this is a low voltage- high current application.
Typical inductor current is 30A and input voltage around 30.