Not an audio circuit, but given the knowledge base here with large inductors, tubes, and high voltage, I am throwing out an idea for comments.
The attached is a circuit I am thinking about for fast demag of a large inductance. The AC source is continuously variable from 0-100V although it typically will run between 10 and 30. When the system is active, DC current flows through the inductor, and the IGBT must be switched on. When the system shuts off, AC power drops to zero, and the rectifier free wheels the inductance. At that point, polarity across the rectifier switches, and the IGBT must shut off quickly. This applies the nonlinear resistor in series with the inductor. L/R increases, and the magnetic energy is quickly dissipated vs simply allowing it to freewheel through the internal resistance.
The green circuit is not really subject to change. The red is what I am toying around with. I would like to keep things as passive as possible - no DC/DC converters or big charge pumps. A few of my current challenges:
a. At low voltages, I struggle to get the current sources up, as well as the problem of not having sufficient voltage to turn the IGBT on. Fortunately, switching frequency is extremely low, so I am not overly concerned with slow rise times. Fast fall time is a benefit.
b. Trying to keep watt dissipation low - the red circuit will end up being epoxy potted, with maybe one small heatsink sticking out.
c. Voltages can get pretty high (still working out actual maximums), so my transistor choices are limited.
d. Debating if the active Miller clamp is really necessary.
I probably have more thoughts as this develops, but wanted some extra eyes to put a sanity check on this.
The attached is a circuit I am thinking about for fast demag of a large inductance. The AC source is continuously variable from 0-100V although it typically will run between 10 and 30. When the system is active, DC current flows through the inductor, and the IGBT must be switched on. When the system shuts off, AC power drops to zero, and the rectifier free wheels the inductance. At that point, polarity across the rectifier switches, and the IGBT must shut off quickly. This applies the nonlinear resistor in series with the inductor. L/R increases, and the magnetic energy is quickly dissipated vs simply allowing it to freewheel through the internal resistance.
The green circuit is not really subject to change. The red is what I am toying around with. I would like to keep things as passive as possible - no DC/DC converters or big charge pumps. A few of my current challenges:
a. At low voltages, I struggle to get the current sources up, as well as the problem of not having sufficient voltage to turn the IGBT on. Fortunately, switching frequency is extremely low, so I am not overly concerned with slow rise times. Fast fall time is a benefit.
b. Trying to keep watt dissipation low - the red circuit will end up being epoxy potted, with maybe one small heatsink sticking out.
c. Voltages can get pretty high (still working out actual maximums), so my transistor choices are limited.
d. Debating if the active Miller clamp is really necessary.
I probably have more thoughts as this develops, but wanted some extra eyes to put a sanity check on this.
Attachments
Why wouldn't you use an antiparallel diode across the S and D of the IGBT? Demagnetization would take longer, but it is much simpler
Then, a suitable resistor in series with the diode? (I have no idea about the reverse voltage rating of the IGBT though)
I don't know if you are following what I am trying to accomplish. An antiparallel diode across the IGBT does nothing, current through the inductor is unidirectional. My bridge rectifier already performs free-wheeling action, which has insufficiently slow discharge for my needs. Although this might look like some type of PWM circuit, it is not. Just slow on/off action once per minute or so.
As noted in the OP, what is shown in green does not change. It remains as drawn.
I need to dissipate 32 kJ of stored energy very fast using a resistor such as these: https://metrosil.com/applications/generator-de-excitation/
The question is how to optimally drive the IGBT with limited parts. There is no auxiliary power supply available, and no remote control possible. All red control is based on what the green circuit gives me to work with.
As noted in the OP, what is shown in green does not change. It remains as drawn.
I need to dissipate 32 kJ of stored energy very fast using a resistor such as these: https://metrosil.com/applications/generator-de-excitation/
The question is how to optimally drive the IGBT with limited parts. There is no auxiliary power supply available, and no remote control possible. All red control is based on what the green circuit gives me to work with.
OK, I missed the point. Maybe a freewheeling diode in series with a resistor?
The only other option is to dissipate the energy in the IGBT using a suitable gate control, but for 32kJ, dissipation in a resistor is certainly preferable
The only other option is to dissipate the energy in the IGBT using a suitable gate control, but for 32kJ, dissipation in a resistor is certainly preferable
If you look closely, a "freewheeling diode in series with a resistor" is exactly what I already have. The bridge rectifier acts as the diode, and the nonlinear resistor is either in series OR gets bypassed by the IGBT when it is not wanted.
The thread is obviously not getting much traction, so I am putting it through LTspice modeling to try and get further. All I am asking for is help on the red portion of the circuit, not green.
The thread is obviously not getting much traction, so I am putting it through LTspice modeling to try and get further. All I am asking for is help on the red portion of the circuit, not green.
How did you calculate the resistor value? Is it the maximum practical value? If its a nonlinear resistor, is it increasing resistance as the inductor discharges?
Last edited:
Well, the resistor is green, which would imply I am not asking for any assistance with it. But I understand there might be some curiosity about how it works. The elements are either ZnO or SiC, which begin to conduct harder as voltage rises. Effectively, they decrease resistance with higher applied voltages. The website of post 5 has specifics.
With AC input source on (let's assume 30V) I have a bridge output of 42VDC, and the inductor is carrying lots of amps. Technically the source is three phase, but this is irrelevant to the thread. With 42V on the red gate circuit, I have a sufficiently strong gate voltage, and the IGBT is fully on. Resistor is therefore bypassed, running low watt dissipation. This is the normal mode of operation.
When the AC input source shuts off, the source of inductor current is interrupted - however, the inductor refuses to allow this, and becomes a source, changing polarity. This allows the inductor current to freewheel across the bridge rectifier. At this moment (polarity swap) my red circuit tries to discharge the gate and turns the IGBT off. Now, the resistor is in the discharge circuit with the same freewheeling action. The addition of this resistor will allow the high energy storage in the inductor to quickly dissipate, along with the benefit of fast discharge. The initial step response with this resistor (and highest current flow) will generate the the highest voltage emf within the inductor windings. I must select the resistor carefully to not exceed the turn/turn and winding/ground ratings of the inductor. Higher resistance is better for discharge time, but the higher this resistance the larger the emf developed in the inductor. There may also be additional limitations such as physical size and cost of this resistor, but the primary limitation of the design depends on the voltage rating of the coil.
This is the entire goal of the circuit - when I need to discharge the inductor, do it quickly. Without any external control or auxiliary supply voltage available to me.
Hopefully that completes the analysis of the green circuit, because that ain't gonna change.
The issues are what happens at low excitation voltage, where the gate circuit struggles to get up to 15V. With low coil current, some linear IGBT behavior and current sharing with the resistor might be of little consequence, but this is why I am trying to model it in spice. Then, how fast will my gate discharge with a unipolar gate supply (which is why the Miller clamp was installed). Are there better gate drivers to design? Again, I cannot change what is green, and I cannot use external controls (the red circuit will be permanently inaccessible).
With AC input source on (let's assume 30V) I have a bridge output of 42VDC, and the inductor is carrying lots of amps. Technically the source is three phase, but this is irrelevant to the thread. With 42V on the red gate circuit, I have a sufficiently strong gate voltage, and the IGBT is fully on. Resistor is therefore bypassed, running low watt dissipation. This is the normal mode of operation.
When the AC input source shuts off, the source of inductor current is interrupted - however, the inductor refuses to allow this, and becomes a source, changing polarity. This allows the inductor current to freewheel across the bridge rectifier. At this moment (polarity swap) my red circuit tries to discharge the gate and turns the IGBT off. Now, the resistor is in the discharge circuit with the same freewheeling action. The addition of this resistor will allow the high energy storage in the inductor to quickly dissipate, along with the benefit of fast discharge. The initial step response with this resistor (and highest current flow) will generate the the highest voltage emf within the inductor windings. I must select the resistor carefully to not exceed the turn/turn and winding/ground ratings of the inductor. Higher resistance is better for discharge time, but the higher this resistance the larger the emf developed in the inductor. There may also be additional limitations such as physical size and cost of this resistor, but the primary limitation of the design depends on the voltage rating of the coil.
This is the entire goal of the circuit - when I need to discharge the inductor, do it quickly. Without any external control or auxiliary supply voltage available to me.
Hopefully that completes the analysis of the green circuit, because that ain't gonna change.
The issues are what happens at low excitation voltage, where the gate circuit struggles to get up to 15V. With low coil current, some linear IGBT behavior and current sharing with the resistor might be of little consequence, but this is why I am trying to model it in spice. Then, how fast will my gate discharge with a unipolar gate supply (which is why the Miller clamp was installed). Are there better gate drivers to design? Again, I cannot change what is green, and I cannot use external controls (the red circuit will be permanently inaccessible).
I think I begin to grasp what you need: this circuit will start to quench the inductor as soon as the primary power is removed.
The Nmos is the IGBT obviously:
It is extremely simple, and voltage agnostic
The Nmos is the IGBT obviously:
It is extremely simple, and voltage agnostic
Looks like that will blow up at 150V excitation. Part of my challenge was working with the full range of possible voltage levels.
R1 would run at 10W continuous. I would even prefer to design everything to survive as much as 250V for robustness, so crank that up even more. If I increase R1 resistance, I have a long gate discharge time which is definitely a no go.
At light excitation (20V) I get 2 mA to charge up the gate.
Not just taking shots at your design, but I have considered options similar to that already.
R1 would run at 10W continuous. I would even prefer to design everything to survive as much as 250V for robustness, so crank that up even more. If I increase R1 resistance, I have a long gate discharge time which is definitely a no go.
At light excitation (20V) I get 2 mA to charge up the gate.
Not just taking shots at your design, but I have considered options similar to that already.
this is the most simplistic option. Resistors values can be adapted, and the circuit can be evolved towards anything. You simply need to to state your priorities: simplicity, low consumption, lowest quiescent consumption or even better.
You cannot have everything at the same time. Engineering is trade-off, chose yours
You cannot have everything at the same time. Engineering is trade-off, chose yours
Perhaps a suitable MOV could be additionally placed across the coil to constrain peak transient voltage, and work in collaboration with the existing discharge MOV. Have you worked through the peak energy in your MOV to see if the MOV will degrade, and likely how fast.
Perhaps use a relay to give you an isolated NC contact to connect in a controlled parallel dump across the inductor, or some other form of bypass R to augment the existing de-energisation of the inductor.
A few 100V CCS fets in series may be a better option to give you a defined current for control circuit from say 6V feed up to beyond a few hundred volts.
Perhaps use a relay to give you an isolated NC contact to connect in a controlled parallel dump across the inductor, or some other form of bypass R to augment the existing de-energisation of the inductor.
A few 100V CCS fets in series may be a better option to give you a defined current for control circuit from say 6V feed up to beyond a few hundred volts.