transformer driving mosfet

[vincehhh]

I trying to use a mosfet (IRF730) as a high side switch controlled from a microcontroler.
The Isolation needed is 350v

I need to turn it on for around 8.5µs maybe 7 - 10. At first I looked at optocouplers but the ones available where too slow.

Transformers look the best bet as they can do the isolation but selecting or designing I don't know.

I thought it should be a high inductance primary so current is still rising at turn off.
But at turnoff there might be problems with ringing (oscillating due to L and winging capacitance).

For example using 12v and transformer primary L of 1mH this should give peak primary current of around 100mA. At turn off we get a -ve pulse to discharge gate , but I don't know if I get enough current Initially to charge gate and turn FET on.


The other option I have read about said low primary inductance and a current limiting resistor.

Any ideas welcome.

 

[dougy83]

Do you actually require isolation? Or just level shifting up to 350V?

 

[crutschow]

The transformer secondary current you get is a function of the primary current available from the driver and is not directly related to the inductance. The primary inductance only determines how much magnetizing current is generated during the pulse period. This current does not appear in the secondary during that time. But this inductive energy will generate a reverse (flyback) voltage and current when the primary is opened (but not if the primary is shorted).

To summarize, transformers can have two currents:

One is the magnetizing current in the primary winding, which is determined by the transformer inductance and the primary signal voltage and frequency characteristics. From that point of view, the transformer look like an inductor with multiple windings. For example, this inductance is used to transfer energy from primary to secondary in the operation of a flyback transformer.

The other is any load current which appears in both the primary and secondary windings when operating as a normal transformer, with relative magnitude determined by the transformer turns ratio. This current is unrelated directly to the magnetizing current.

 

[Leonard]

I need to turn it on for around 8.5µs maybe 7 - 10. At first I looked at optocouplers but the ones available where too slow.

Cause you have used a simple type of it, will be rather to used the Darlington type of opto-couple whose functions much faster.

 

[shortbus=]

With a high side n-mosfet you need to raise the gate voltage above the source voltage when turned on. Also unless your using a logic level mosfet you also need around 10 -12 volts to turn it on. Go with a gate driver like a IR2110 and you don't need a optocoupler(the gate driver isolates input from gate output). The IR 2110 is good for 600V. But you will still need the 10 - 12 volts for the gate.

 

[crutschow]

Cause you have used a simple type of it, will be rather to used the Darlington type of opto-couple whose functions much faster.

Darlington types are more sensitive, but are generally slower than non-Darlington couplers.

 

[crutschow]

The IR 2110 is good for 600V. But you will still need the 10 - 12 volts for the gate.

That's likely why he wants a transformer, since that can generate the gate voltage needed for a high-side N-MOSFET driver.

 

[vincehhh]

Thank you crutschow

I was thinking the wrong way round. I could see as the transformer drive was cut off I would get a good turn off pulse but turn on seemed uncertain.

Of cause it is done in reverse. transformer drive current is oriented to turn FET off and as ithe drive is cut we have max magnetic field falling to drive FET on.

At the instant of transformer drive being cut off.
Magnetic field is at MAX
magnetic field fall is fastest at this point because the secondary looks like it is shorted due to the load of gate charged in reverse.
once the gate is charged the secondary turns appear as open cct as gate is fully charged and gate voltage should stay at required voltage s magnetic field falls.

The transformer is the best option because it is simpler, cheaper than optocoupler. The optocouplers at this speed are much more expensive. The drawback with transformer is the possibility of oscillating due to the combination of the inductance and inter-winding capacitance, LC cct with low resistance, ideal foe resonance. Also there is the problem with saturating the transformer core. Suck and see

I just need to pull my finger out and make a simple test cct and have look at what happens with CRO.
use a 555 driving a FET or Bipolar to drive transformer with the test FET loaded and chech the FET voltage drops to low volt to indicate it is turned on properly.

Why did I not think of that earlier, I must be getting too old.


Thanks to crutschow

 

[Dvinchy]

Darlington types are more sensitive, but are generally slower than non-Darlington couplers.

Wrong opinion. Faster or slower are depending of resistance load between base and emitter. The more meaning of resistance, the faster function.

 

[shortbus=]

That's likely why he wants a transformer, since that can generate the gate voltage needed for a high-side N-MOSFET driver.

Maybe I'm thinking wrong here but, won't using a transformer still need a driver for high side? The transformer can raise the Microcontroler output to 10V but the gate still needs to stay 10V above the source voltage. And by adding the source voltage to the gate voltage and then into the transformer, won't that make the transformer primary voltage go higher and possibly blow out the microcontroller?

 

[crutschow]

Maybe I'm thinking wrong here but, won't using a transformer still need a driver for high side? The transformer can raise the Microcontroler output to 10V but the gate still needs to stay 10V above the source voltage. And by adding the source voltage to the gate voltage and then into the transformer, won't that make the transformer primary voltage go higher and possibly blow out the microcontroller?

You are not understanding how the transformer is connected. It is the driver for the high side. The transformer secondary is connected between the high-side transistor source and gate. Thus the gate-source voltage will be equal to the transformer secondary voltage, independent of what the transistor source to ground voltage is.

I don't understand your concern about blowing out the microcontroller. There is no voltage fed back from the secondary to the primary in this configuration. The common-mode voltage of the secondary winding will move up and down in response to the source voltage, but that has no significant effect on the primary voltage.

The advantage here of a transformer over an opto-coupler is that a transformer can transfer energy from the primary to the secondary to generate the gate voltage, and a standard opto-coupler cannot.

 

[crutschow]

Wrong opinion. Faster or slower are depending of resistance load between base and emitter. The more meaning of resistance, the faster function.

Well, I beg to differ with your opinion of my opinion.

If I understand your fractured English you are now discussing about resistance between base and emitter (whatever "meaning of Resistance" means). How does that pertain to whether it's a darlington or not?

 

[Leonard]

Well, I beg to differ with your opinion of my opinion.

If I understand your fractured English you are now discussing about resistance between base and emitter (whatever "meaning of Resistance" means). How does that pertain to whether it's a darlington or not?

Please take a more information about Darlington type here:

 

[shortbus=]

You are not understanding how the transformer is connected. It is the driver for the high side. The transformer secondary is connected between the high-side transistor source and gate. Thus the gate-source voltage will be equal to the transformer secondary voltage, independent of what the transistor source to ground voltage is.

I don't understand your concern about blowing out the microcontroller. There is no voltage fed back from the secondary to the primary in this configuration. The common-mode voltage of the secondary winding will move up and down in response to the source voltage, but that has no significant effect on the primary voltage.

The advantage here of a transformer over an opto-coupler is that a transformer can transfer energy from the primary to the secondary to generate the gate voltage, and a standard opto-coupler cannot.

Again I'm probably wrong but the way I understand it is you need 10 volt on the gate to get current/voltage to flow from drain to source on the mosfet. As the source voltage starts to flow why would you need a transformer to raise the voltage on the gate? The source voltage is already raising the amount needed.

I thought you were using the transformer to raise the voltage from the Micro to the needed 10V to turn on the gate. Then if you added the source and gate voltage together it would feed back into the Micro through the transformer.

It's also my understanding that a High Side mosfet source is not connected to ground but a load. Say if switching 20V from drain to source the gate volage would have to end up at 30V when mosfet is turned on, right? Unless a logic level mosfet is being used then you need 25V at the gate.

Are you reading/saying high side but thinking low side?

Also with out a gate driver the low amperage comeing from the Micro output is going to take a longer time to switch the mosfet on causing more heat in the mosfet.

 

[crutschow]

Please take a more information about Darlington type here:

And on the second page of the reference it says:

A variation on the standard optocoupler with a single​

output phototransistor is the type having a photo-
Darlington transistor pair in the output (Fig.7), such as the
6N138. As mentioned earlier this type of device gives a
much higher CTR and transfer gain, but with a significant​

penalty in terms of bandwidth.

A "significant penalty in terms of bandwidth" means it's slower than a standard opto coupler.

 

[crutschow]

Again I'm probably wrong but the way I understand it is you need 10 volt on the gate to get current/voltage to flow from drain to source on the mosfet. As the source voltage starts to flow why would you need a transformer to raise the voltage on the gate? The source voltage is already raising the amount needed.

I thought you were using the transformer to raise the voltage from the Micro to the needed 10V to turn on the gate. Then if you added the source and gate voltage together it would feed back into the Micro through the transformer.

It's also my understanding that a High Side mosfet source is not connected to ground but a load. Say if switching 20V from drain to source the gate voltage would have to end up at 30V when mosfet is turned on, right? Unless a logic level mosfet is being used then you need 25V at the gate.

Are you reading/saying high side but thinking low side?

Also with out a gate driver the low amperage coming from the Micro output is going to take a longer time to switch the mosfet on causing more heat in the mosfet.

You seem to have difficulty understanding the concept of using a transformer to drive a high-side transistor (no, I'm not confusing it with a low side-side connection).

Don't understand your comment:
Quote "I understand it is you need 10 volt on the gate to get current/voltage to flow from drain to source on the mosfet. As the source voltage starts to flow why would you need a transformer to raise the voltage on the gate? The source voltage is already raising the amount needed."
If you don't have the transformer how are you going to get the source current to flow?
You need 10V between gate and source for that, which the transformer supplies.

The transformer secondary is connected between the gate and source of the high-side transistor. The secondary is floating and so provides a voltage between the gate and source, independent of the source voltage above ground. Thus the transformer just has to generate 10V, not 10V plus the source voltage. So you don't need 30V from the transformer to provide 10V to the gate if the source is at 20V, you just need 10V.

And, as I previously stated there is no feedback from the source voltage through the transformer. The source voltage is common to both secondary windings and is thus not seen by the primary.

Your point about the Micro not having enough to drive a power MOSFET is valid. You may need a driver from the Micro to the transformer. But you still need the transformer to drive the high side transistor.

If you are still confused, perhaps you should read a tutorial about transformers.

 

[shortbus=]

Quote -"The transformer secondary is connected between the gate and source of the high-side transistor. The secondary is floating and so provides a voltage between the gate and source, independent of the source voltage above ground. Thus the transformer just has to generate 10V, not 10V plus the source voltage. So you don't need 30V from the transformer to provide 10V to the gate if the source is at 20V, you just need 10V."

All the other replies on this and other forums(even some by you) have stated that in a H/S N-fet,(non logic gate) the gate voltage must be kept 10V above the source voltage when on. If not the N-fet will be in the resistive/ohmic region of operation. Now your saying this isn't so?

Quote - "And, as I previously stated there is no feedback from the source voltage through the transformer. The source voltage is common to both secondary windings and is thus not seen by the primary"

So your saying that as the EMF of the secondary collapses it won't create a current in the primary? If the ratio of the transformer is 1:2 (5V on pri. to 10V on sec.) there will be 10V to the Micro terminal (in my example - 20V from source on sec. to 10V at pri.)

Also because all of these are DC voltages the output from the Micro will have to be a PWM signal lasting the of time the O/P wanted it to be on. Because the transformer won't work on straight DC. Plus for the transfomer to be viable it takes a lot more components than a high side gate driver IC.

 

[dougy83]

All the other replies on this and other forums(even some by you) have stated that in a H/S N-fet,(non logic gate) the gate voltage must be kept 10V above the source voltage when on. If not the N-fet will be in the resistive/ohmic region of operation. Now your saying this isn't so?

You seem to have missed the point that the transformer secondary is connected across the mosfet gate and source. Therefore, if there's 10V on the transformer secondary winding, there's 10V on the mosfet G-S; this is irrespective of the absolute voltage on any of the mosfet pins (source could be -100V or +100V, doesn't matter).

So your saying that as the EMF of the secondary collapses it won't create a current in the primary?

Why would the secondary be collapsing.

Also because all of these are DC voltages the output from the Micro will have to be a PWM signal lasting the of time the O/P wanted it to be on.

The micro output will be high for as long as you want the fet on (or vice versa if you reverse the transformer winding connection).

Because the transformer won't work on straight DC.

No, it won't. The thing is, driving a transformer with a square wave on its primary will get a squarewave on its secondary (limited be the inductor saturation current, the power supply or the voltage/resistance of the coil) - the reason is that the current through the pri will increase linearly (i.e. ramp) with a voltage applied (until the supply hits max current or the core saturates) which will give the derivative signal on the output (dV/dt of a ramp is a constant voltage).

Plus for the transfomer to be viable it takes a lot more components than a high side gate driver IC.

Yes, you need a buffer, a transformer and probably a resistor and capacitor.

 

[Ubergeek63]

Wrong opinion. Faster or slower are depending of resistance load between base and emitter. The more meaning of resistance, the faster function.

and you are wrong... darlingtons are NOTORIOUSLY SLOW, by an order of magnitude or more - you just can not flush the carriers out of the junctions fast enough. it all depends on what you are trying to do.

for gate isolation at 10KHz PWM and above there is no what you will not get a darlington to work, EVER!

Most people will not get a transformer to work either since practical circuits can not afford to reset the core. there are ways around this but most people do not know them and end up wondering why their switchers crash and burn ( every pulse that is not balanced pushes the core closer to saturation )

On the bright side there are affordable 10MHz optos now!

 

[dougy83]

Most people will not get a transformer to work either since practical circuits can not afford to reset the core. there are ways around this but most people do not know them and end up wondering why their switchers crash and burn ( every pulse that is not balanced pushes the core closer to saturation )

Isn't it a simple case of using a DC-blocking capacitor in series with the primary?? Seems both affordable and practical...