Gating a MOSFET in a flyback inverter
2:45 PM
The switching frequency is 100kHz. The mosfet is connected on the primary side of flyback transformer. It is a low side mosfet and hence the gating signal should be provided with respect to the ground (PV panel negative). The gating signal however, appears to be a ramp with a sudden drop to zero , just like a triangular waveform. The possible issues as pointed out by experts contacted on various forums are as follows
- The Mosfet gate can be viewed as a capacitor of ~15 pF or thereabouts. You say nothing about your circuit, so we don't know. Your drive current is insufficient to charge this capacitor quickly, so the result is similar to an RC delay. It is actually a ZC delay. The 6n137 seems designed to pull down, not up. Is this a P channel Mosfet? If not the 6n137 seems a poor choice. Go figure. If you want us to figure, more details are needed. All your current into the ~15pF is needed instantaneously on the gate if you want a square wave. The ideal switching circuit has a LV rail with current behind it, and switching is done by a fast device, e.g. a bipolar. I have also seen pulse transformers driven by oscillators making a supply to switch bipolars when required. Is this switch high side or low side? (Answer by Declan Moriarty)
- Mostly agree with Declan (and, especially, more details are needed!). But, depending on the transistor you're driving, its gate capacitance may be as much as several nF (as opposed to pF). The limited current capability of the 6N137 (along with the related pullup current capability) may be far too low to obtain square wave gate drive of your transistor. Your mention of a ramped gate voltage tends to confirm this. Investigate the gate drive requirements for your transistor in detail. "Gate charge" datasheet plots can be very helpful for this.Additional comment: to summarize, you may need a much higher current capability gate drive circuit (both sourcing and sinking) in order to obtain reasonably fast switching and gate drive waveforms. (Answer by Robert Scott)
- You can use 6N137 together with driver. The UCC2732x is a very popular IC. It provides 9A peak current and is capable to operate in quite high frequency range. I often use it for low-power converters. There is a modification with 2-fet output cascade.(Answer by Roman Gorbunov)
- I assume the 6N137 has an open collector output. What's the value of the pull-up resistor? If the gate resistor is too large - or maybe missing! - you will have a large time constant together with the gate capacitance and that could result in a ramp voltage at turn on. What MOSFET do you use? What does the falling slope of Vgs look like (at turn off)? If you see a ramp here as well it appears your optocoupler is not capable of driving the gate for some reason. Well, if the turn off seems OK, when looking at the gate voltage, I would say the optocoupler actually seems to be able to sink the current as you extract the charge from the gate.
Assuming you apply 7V, which is the permissible voltage, via the pull-up resistor, to Vo of the 6N137, you should use a pull-up resistor of no less than 140ohm in order not to exceed the maximum sink current of 50mA. The capacitance of the gate, which indeed is complex to define (and it is dynamic, too), should be around 3nF anyway, according to datasheet. The time constant of the gate voltage at turn on would thus equal: 140ohm*3nF=0.5us. Mind you, this number is only a rough estimation but will give you a hint of what to expect. Would this turn on time be fast enough for your application as regards switching losses? Should your ramp be much different from 0.5us I strongly suspect that your pull-up resistor is much larger than 140ohm. What resistance do you actually use?
Another aspect of the gate drive is that the Rds-on depends on the gate voltage. If there is a need to minimize the conduction losses you would benefit from using a gate voltage higher than 7V, but less than 20V.
Although swapping to a proper gate driver very well could solve the problem, I think you should try to find out exactly why your circuit fails before you do that. This way you will really learn something and you will better know how to design the next time. On my reply about circuit pull up resistor value and about circuit configuration OK, so this explains at least part of the problem. Good work! It is possible that using a proper gate driver actually would solve the problem, just as Bruce Long suggested. On the other hand, without knowing more about your design and the waveforms, I wouldn't exclude alternative reasons for slow transitions. The Flyback inverter is very much depending on the properties of the secondary side components as well as the transformer. Furthermore, the freewheeling diode is placed on the secondary side of the transformer which makes the flyback design different from many other designs, I think. And as for all switch mode circuitry you must be beware of any stray reactive elements.
There are numerous application notes available on the web that address the principles of gate drive of MOSFETs and IGBTs. Go and search at the web sites of the major semiconductor manufacturers! I suggest you pick a few of those app notes and then analyze how your design behaves compared to the guidelines. Try to identify and compare the various waveforms of voltage and current. For instance how do the gate voltage and drain voltage interact in time? Is the layout OK? Any inductive or resistive elements in the source path results in negative feedback, which in turn slows down the transitions.
Sorry, I had just sent my comment when I saw your remark about the hot resistor. Well, if you use 140ohm, have a voltage of 7V, and if we assume the voltage is applied continuously, the power loss equals 7*7/140=0.35W. At any duty cycle less than 100% the power decreases, naturally. If you use a fairly large resistor, like 1/2 Watt size, you should be OK.
I will certainly try to give you further support if I can. Please keep the forum updated on your further attempts!(Answer by Lars Thomasson) - I think Lars is giving you good information. I checked the 6N137 data sheet. It is a logic level, open collector device with an output absolute maximum sink current of 50 mA and an output absolute voltage rating of 7v. You did not tell us what the MOSFET part number is but I could easily imagine situations where the 6N137 would develop marginal gate drive.
Solution Choose another way to provide the MOSFET gate drive choose another optocoupler with better output ratings.Add an emitter follower to the opto-coupler output maybe a npn pnp emitter coupler pair so you have low impedance and high current drive for both upwards and downwards transistions.Use a gate driver integrated circuit instead of an optocoupler. Keep in mind that 7 v is the absolute maximum voltage for the internal logic gate inside the optocoupler. It is possible to run the optocoupler vcc on 5V and use a higher voltage on the open collector output and a pull up resistor sized to allow the opto-coupler to sink 50 mA minus an appropriate safety margin as once again 50 mA is the absolute max sink current allowed. In principal the opto-coupler transistor burns out if the current increases to 50.001 mA
Lars has again given you excellent advice for you to work until you understand why the mosfet gate drive waveform is something other than you expect it to be.(Answer by Bruce Long) - If you are not too limited for space try using two or three resistors in parallel. Also, there are several manufactures of the 6N137, which one are you using? What are you using as a controller for the flyback? What is the input voltage to the transformer? I am concerned about a MOSFET with a 100V rating as it will see spikes that are significantly higher when it switches. (Answer by Wayne Parades)
- Without access to a schematic we can only guess, however I wonder whether it is correctly understood, that you drive the MOSFET directly from the 6N137? If so, then most likely the 6N137's maximum output current is way too small to produce a fast rising gate voltage, due to the gate capacitance of the MOSFET..(Answer by Klaus Bahner )
1 comments
Mosfet working pdf , you can Download it and Read Online.
ReplyDelete