Negative discharge effect

[nelsonrochaa]

The problem is, i have no good tools to measure. I measured it by connecting the coil and a potentiometer in series to a 12 V AC wall adapter, it was really 14.4 V. Then i adjusted the potentiometer so that the voltage on that was half of it, 7.2 V AC. Disconnected the potentiometer, measured its resistance and calculated the inductance by the 2. method there http://www.wikihow.com/Measure-Inductance , when changing the resistance instead of frequency. And with that method i got the inductance 6.6 H.

But then i tried to check that method and measure a known inductance. I had only a small known inductor though, and i don't have so small potentiometer, so i had to use small resistors. I also had to use a 100 ohms resistor to restrict the current of the adapter, to not to burn the adapter. This 100 ohms resistor went so terribly hot that i thought i can use it as a soldering iron. I measured the voltage of the inductor and a combination of resistors in series, and choose resistors so that their voltage was half of that voltage. I got that way 80 mH, but what was written on the inductor was 8,2 mH , that is, 8, what looked like a short vertical line, and 2. I thought that means 8.2 mH but i'm not entirely sure that the thing in between there was a dot, so maybe it was 82 mH.

The core is the core of the deflection yoke of a CRT monitor, and there is 900 turns of a 26 gauge magnet wire on it. I calculated the inductance, and got only 0.16 H. Assuming that the relative permeability is 100, but for such ferrite core it is likely much higher.

So this is what i could get with my primitive tools. As much as i remember,  the potentiometer was 2.2 k when the total voltage was 14.4 V, and there was 7.2 V on the potentiometer. Frequency should be 50 Hz. I measured in the 200 V AC range of my multimeter. So this is how it is, with primitive tools it is not possible to get too good results.

Hi ayeaye,
I make a similar tests and all my mosfet and igbt burn after some time working.
I think that BEMF will fry any mosfet in a common configuration like you illustrate because the gate is not isolated.The diode protection of mosfet in the peak of collapse will not work because the static is so much higher
that diode will let flow the current backward and didn't make the job.
I try the commutation with relay and see what happens. https://www.youtube.com/watch?v=pf_qUlwSZl0
https://www.youtube.com/watch?v=DfxEAQNOjp0.

Good tests

[ayeaye]

Yes mosfet has a body diode in parallel to it, which always conducts positive current from source to drain, even when the mosfet is closed. I tested it, connecting positive probe of the multimeter to the source and negative to the drain when the mosfet was closed, in the diode range, and it conducted. This has no importance though for that circuit, because the diode in the circuit is opposite to that diode, and thus when the mosfet is closed, nothing can go through the circuit for that reason. As i said earlier, it is only important for that circuit that the body diode is a Zener diode.

But the real reason why i thought about negative discharge, was because without it, because of the diode, the circuit is disconnected for the forward current in the coil, and this is the only thing that can be caused by switching. And without switched forward current, there cannot be back-emf, and when there is no back emf, then nothing can go to the charged capacitor. So to say in short what i mean, is that negative discharge is the only way how the circuit can be closed (connected) for the forward current. And for the opposite current the main circuit is only closed when the back-emf goes through the mosfet's Zener diode, as we talked.

[TinselKoala]

Circuit elements are not perfect. It seems to me that you are modelling your components in your mind as "perfect". But, for example, the diode you specify is very slow, it has a reverse recovery time of 2 microseconds.
This means it will be ineffective in blocking fast spikes in the reverse biased direction.  Allow me to suggest that you compare the system's performance with different diodes. Try the UF4007 for example and see if your capacitor voltages are the same as when you are using your present diodes. CRT monitors have some fast HV diodes in them, check the internet for the diode data sheets for the part numbers that you find in your scavenging.

The mosfet is not a perfect switch either. For gate voltages near the threshold (usually around 4 volts) the mosfet will be operating in a "linear conductance region" where its on-state resistance is linearly related to the charge on the gate. It is more like a variable resistor than a switch when the gate charge is in this range. So if you have a gate drive that is low voltage and slowly changing, or has only the ability to deliver small currents (filling the gate capacitance slowly) the mosfet is no longer going to be switching cleanly.

Furthermore if there is no way for charge to _leave_ the gate then the mosfet will not turn _off_ cleanly. This is one reason that you sometimes see mosfet gates being driven with AC signals: the reversed (negative) voltage sucks charge out of the gate and turns the mosfet off faster than simply bringing the gate voltage to zero.  Also "pulldown" resistors may be incorporated from gate to source, to allow the charge to leave the gate when the mosfet is supposed to be off.

I speak of course of N-channel mosfets; P-channel are the same but polarities are reversed.

You can make a good inductance meter with an Arduino (or other microprocessor system) and a few other components. You do not need the LCD display, the Arduino can report its data over the serial (usb) line and display on your computer monitor.

http://www.youtube.com/watch?v=S6N8ys8FiA4

http://www.youtube.com/watch?v=SCxypoN8-xc

[ayeaye]

Talking about microcontroller. My microcontroller board is kl25z, and this is a 32 bit arm microcontroller. This means that its logical 1 voltage is 3.2 volts, different from 5 volts in microcontroller boards such as arduino uno. But i tested it, switching the mosfet on and off, and this 3.2 volts switches the mosfet fully on, so that on the resistor in series there is a voltage almost equal to the source voltage.

And yes i controlled the microcontroller, that is generating pulses, through usb, using the minicom terminal emulator.

About pulses, i think that what matters is the exact length of the pulses, not frequency, frequency is important only because when the frequency is higher, the things are changing faster. But i may be wrong about that particular thing.

[TinselKoala]

Talking about microcontroller. My microcontroller board is kl25z, and this is a 32 bit arm microcontroller. This means that its logical 1 voltage is 3.2 volts, different from 5 volts in microcontroller boards such as arduino uno. But i tested it, switching the mosfet on and off, and this 3.2 volts switches the mosfet fully on, so that on the resistor in series there is a voltage almost equal to the source voltage.

And yes i controlled the microcontroller, that is generating pulses, through usb, using the minicom terminal emulator.

About pulses, i think that what matters is the exact length of the pulses, not frequency, frequency is important only because when the frequency is higher, the things are changing faster. But i may be wrong about that particular thing.

No, at 3.2 volts the IRF630  mosfet is not turning fully on, especially if you are sourcing the gate current directly from the microprocessor. The gate _threshold_ voltage of the IRF630 is between 2 and 4 volts but it will not be fully on until the gate voltage is near 8 volts. See the graphs below, taken from the Vishay data sheet.

And no, frequency is always important, since the mosfet has a finite switching time, the reactances vary with frequency, the rise time of the drive pulses probably varies with frequency, etc etc.

You really need an oscilloscope monitoring the mosfet drain voltage to see what you need to see in this experiment.