Hi there!
I've used days searching and reading forums for finding a good solution to switch AC with mosfets in totem, source-to-source. TRIAC is not an option, as I want to switch of in other times than phase-over(PWM, reverse phase dimmer, etc). SSR are overkill, and to expensive.

Like many others, I stranded upon the problem with the floating source, and found solutions with a photovoltaic optocoupler. But, the photovoltaic optocouplers available from my supplier are pretty expensive.

So, I made a circuit with LTspice, using a rectifier and a optocoupler with transistor output. That way, I can feed the mosfet gate with a positive/negative pulse.

The cheap optocouplers usally have about 10+us on/off time, so that will be a limitation with high frequency switching(for example 100khz PWM).

This seems to work, but I wondered if any of you had real life experince with this type of setup?

The application will be fed with 12VAC. The load will be max 50-100W resistive. The models used are random, and I will adjust that to my application(Rds on, max current, max gate-source/drain-source voltage, type of optocoupler, etc).

 

And, the controlling pulse will be 5V from a micro controller.

 

The simple way to do this is by connecting two N-channel mosfets in series as shown.

This works because fets are actually bi-directional. A positive voltage on the gate (with respect to the source) switches the channel to a low resistance regardless of the direction of current in the channel. So if you choose a fet with a sufficiently low on resistance, the body diode on the reversed fet will not conduct. This can make a very low voltage drop AC switch.

I have used this circuit to switch 24VAC 250W halogen lamps with high efficiency.

Be careful though: Make sure that the voltage on the gate is either zero or sufficient to switch full-on. Even during power up. Or your fet will be quickly destroyed. I know from experience!

Another complication: Your drive circuit will probably have to be isolated from your micro-controller. As the sources are probably at some unknown AC voltage. A small floating power supply or DC-DC converter might have to be used. Unless your complete micro-controller circuit is attached to the sources.

 

Thanks for sharing your experience.

As you say, the complication lays in ground level, and floating mosfet source.

The way I planned to do it, was to set one of the AC phases as common ground(the circuit will be totally isolated from touching), and have +/- DC floating from that level. That way, the sources at the mosfets will float to a minus voltage when the body diode of the upper mosfet is forwarded. As you state, this will potentially burn the mosfets, if the gate is not at a lower voltage.

A problem with this, might be that gate-source voltage gets to high. When not conduction, we have potentially -17V(minus body-diode drop) amplitude plus the +/-17 voltage.

Do you have any schematics how you made it work with isolated DC on the micro-controller?

 

Actually, as my simulation shows, the sources are pulled against ground level when load is above the mosfet totem and gate rising from -16V. Maybe the optocouplers rise-time is enough to get a smooth transistion, and not blow the mosfets because of the +-20V gate-source voltage limitation.

 

There are many ways of doing this but consider that

• The MOSFET must always be fully OFF or fully ON
• The gate has a high capacitance, so to switch the transistor fast the driver must be able to sink and source high currents.
• There needs to be an under-voltage lockout so that the transistor can never be partially on when the supply voltage is low.

I suggest using a mosfet driver IC. It's all on the chip. An example driver circuit is attached. I have used an opto-coupler and a small DC-DC converter. Gate current is only needed when switching so only a small supply current is needed. It is possible to get the supply current from the AC supply itself using diodes and resistors.

 

Thanks. One step further to prototype

 

The simple way to do this is by connecting two N-channel mosfets in series as shown.

This works because fets are actually bi-directional. A positive voltage on the gate (with respect to the source) switches the channel to a low resistance regardless of the direction of current in the channel. So if you choose a fet with a sufficiently low on resistance, the body diode on the reversed fet will not conduct. This can make a very low voltage drop AC switch.

I have used this circuit to switch 24VAC 250W halogen lamps with high efficiency.

Be careful though: Make sure that the voltage on the gate is either zero or sufficient to switch full-on. Even during power up. Or your fet will be quickly destroyed. I know from experience!

Another complication: Your drive circuit will probably have to be isolated from your micro-controller. As the sources are probably at some unknown AC voltage. A small floating power supply or DC-DC converter might have to be used. Unless your complete micro-controller circuit is attached to the sources.

JD, thanks for this useful information. I am not a technical person. I am going to abuse by asking to give me more details in the way to connect two N channel mosfets to switch AC current. In the diagram that you posted where is the input AC and how can I get the driving voltage and current for the mosfet gate. I am sorry about my ignorance but I can not figure out how is that done when it alternates. Thanks for the help.

 

The two MOSFETS are placed in series on the hot lead. DO NOT under any circumstances connect the MOSFETS across the hot and neutral, in parallel with the load. This is similar to the mistake of trying to measure current with an ammeter by placing the leads across a resistor.

The driving voltage for the gates comes from the DC-DC converter that produces a +12V output to the MC34151 gate-driver(pin 6). The input to the DC-DC converter was not specified.

 

Below is a simulation of an SSR circuit using a 4N25 opto isolator and two back-to-back N-MOSFETs to control the load. The switching looks good at a 25kHz control frequency and 60Hz AC power frequency. The input is isolated from the output (Rsim is just to avoid a floating node error in the simulation). Power for the MOSFET drive circuits is generated using a half-wave rectifier from the 12Vac supply.

You may have to add a driver if you want to drive the opto from a micro.

 

Below is a simulation of an SSR circuit using a 4N25 opto isolator and two back-to-back N-MOSFETs to control the load. The switching looks good at a 25kHz control frequency and 60Hz AC power frequency. The input is isolated from the output (Rsim is just to avoid a floating node error in the simulation). Power is generated for the MOSFET drive circuits using a half-wave rectifier from the 12Vac supply.

You may have to add a driver if you want to drive the opto from a micro.

View attachment 75649

Excellent circuit crutshow! could you please share the LTspice file for that circuit?

 

I added the LTspice .asc file to post #10.

 

I am trying to switch two-phase 250Vac/100Hz AC with two pairs of NMOSFET (FCA36N60NF, one pair for each line). I use Vishay VO1263AAC Photovoltatic MOSFET driver to drive MOSFET gates. The setup has no problem in switching DC voltage but not AC.

Below are the problems I noticed:
1. After connecting two NMOSFET Gate terminals together and Source terminals together. Capacitance around 8.7nF can be measured between two drain terminals of these two FETs. I guess AC signal can go straight through drain terminals whenever Vgs is on or off.

Any comments?

Thanks,

The simple way to do this is by connecting two N-channel mosfets in series as shown.

This works because fets are actually bi-directional. A positive voltage on the gate (with respect to the source) switches the channel to a low resistance regardless of the direction of current in the channel. So if you choose a fet with a sufficiently low on resistance, the body diode on the reversed fet will not conduct. This can make a very low voltage drop AC switch.

I have used this circuit to switch 24VAC 250W halogen lamps with high efficiency.

Be careful though: Make sure that the voltage on the gate is either zero or sufficient to switch full-on. Even during power up. Or your fet will be quickly destroyed. I know from experience!

Another complication: Your drive circuit will probably have to be isolated from your micro-controller. As the sources are probably at some unknown AC voltage. A small floating power supply or DC-DC converter might have to be used. Unless your complete micro-controller circuit is attached to the sources.

 

I am trying to switch two-phase 250Vac/100Hz AC with two pairs of NMOSFET (FCA36N60NF, one pair for each line). I use Vishay VO1263AAC Photovoltatic MOSFET driver to drive MOSFET gates. The setup has no problem in switching DC voltage but not AC.

Below are the problems I noticed:
1. After connecting two NMOSFET Gate terminals together and Source terminals together. Capacitance around 8.7nF can be measured between two drain terminals of these two FETs. I guess AC signal can go straight through drain terminals whenever Vgs is on or off.

Any comments?

Thanks,

This thread is more than a year old.
I suggest you start a new thread with this question, and most importantly, post a schematic of what you're trying to do.

 

There are many ways of doing this but consider that

• The MOSFET must always be fully OFF or fully ON
• The gate has a high capacitance, so to switch the transistor fast the driver must be able to sink and source high currents.
• There needs to be an under-voltage lockout so that the transistor can never be partially on when the supply voltage is low.

I suggest using a mosfet driver IC. It's all on the chip. An example driver circuit is attached. I have used an opto-coupler and a small DC-DC converter. Gate current is only needed when switching so only a small supply current is needed. It is possible to get the supply current from the AC supply itself using diodes and resistors.

Have replicated your circuit – works perfectly fine up to 28V AC!!! I'm using the 12V second (switching PSU) DC source, the control current consumption does not exceed 20-30mA with STF10NM60N with the gate capacitance 540pF and total gate charge 19nC. Did not try mains AC switching, but conceptually nothing prevents that. Will update with mains AC switching results (have currently all breadboard setup – not safe for mains AC projects).