Mosfet AC SSR

[OC71]

Hi everyone,
I would like to make a 10A AC SSR using anti-series mosfets.
Will the attached schematic do the job and if so will it survive without
a heat sink? ---it looks too simple.

[wraper]

Voltage rating of the Mosfets used is too low (200V).

[Zero999]

Voltage rating of the Mosfets used is too low (200V).

What makes you say that? The peak voltage here is only 170V.

EDIT:
To answer the original question. It depends on the ambient temperature. The thermal resistance is 40^oC/W and the power dissipated by each MOSFET is 970mW which gives a temperature rise of 38.8^oC and the MOSFET has a maximum temperature rating of 175^oC but you don't want to go near it.

[wraper]

Voltage rating of the Mosfets used is too low (200V).

What makes you say that? The peak voltage here is only 170V.

Take into account mains voltage tolerance too. With a tiniest transient, peak voltage will easily exceed 200V. Note that there is RMS, not peak on the graph.

[OC71]

Is the rest of the circuit OK? Can it be done this easily? it looks too simple.

[eneuro]

Can it be done this easily? it looks too simple.

It depends what you plan to do with this 1.2kW heater 
Look at page 2 in vom1271t datasheet where at 30mA IF you have ONLY Isc= 45uA (0.045 mA) , so you need estimate its maximum switching speed.
Even for  ON/OFF probably I'd add also AC mains zero crossing detection and while this  vom1271t has such weak gate drive current controll mosfets switching in zero crossing.

But ok. I didn't used this  vom1271t while sometimes difficult to source, however I used 800V-1000V AC mosfets switch but with optoisolated gate driver PSU and cheap $1 TLP350F  2.5Amax, which means 2.5A/0.000047A~50000 times biger AC mosfets switch gate driver current 

I do not like switching 1kW with such low gate current like 47uA without zero crossing detection, but maybe someone else used this low current vom1271t sucessfully, so it could be nice to see how it works.
Test it and tell us if those quite nice (and expensive) 8mOhm 200V mosfets survived given switching frequency 

[dom0]

Uhm... has this any specific sense? A COTS SSR will be cheaper and likely more reliable... the big ones are almost all CAT III and very tough.

[eneuro]

For heater triac BTA16 a few resistors and MOC3041..MOC3043 with built in zero crossing and... bigger radiator to ensure good heat dispation that is all for resistive heater

[OC71]

Thanks for the feedback everyone.
I would like to use mosfets for the experience.
I cannot find a mosfet >200V with a RDSon low enough to operate without a heatsink.
@eneuro, -- I think I will play safe and use the driver you mentioned.

Driver:
http://www.ebay.ca/itm/191554680124?_trksid=p2055119.m1438.l2649&ssPageName=STRK%3AMEBIDX%3AIT

Mosfet:
http://www.aliexpress.com/item/IRFP4668PBF/1641388922.html

[wraper]

I would suggest to not get mosfets and optocouplers from China. There could be anything inferior hiding inside the part.

[eneuro]

@eneuro, -- I think I will play safe and use the driver you mentioned.

There are plenty of them-I choose TLP351F while it is usable for other projects too, but you will need optoisolated PSU, however in tests you can use 12V battery with lets say 27R you will get around 0.5A gate current.

It looks like that ebay folks earns a lot on those TLP351F 3$ for this thing? 
TME is not the cheapiest, but they have is for less than $1 - $0.80 when you buy 10 
http://www.tme.eu/en/katalog/?idp=1&search=tlp351&cleanParameters=1
Also if you buy there 10 those mosfets there IRFP4668PBF than each is around $6, so 2 times cheaper 
http://www.tme.eu/en/katalog/#search=IRFP4668PBF&s_field=accuracy&s_order=DESC

Shipment in TME costs me less than $4, so usually for prototypes looking chips in TME.eu first 

UPS: Didn't notice than those China IRF's is 5 pieces...
Mised also this: Estimated Delivery Time: 15-34 days (ships out within 5 business days)   
I've got things from TME.eu usually 2nd day after order 

[OC71]

@ eneuro. TME shipping cost to Canada is 17.90 EUR plus duty.
Why would I need an opto isolated supply for the driver. Isn't the input isolated
already?

[Zero999]

Voltage rating of the Mosfets used is too low (200V).

What makes you say that? The peak voltage here is only 170V.

Take into account mains voltage tolerance too. With a tiniest transient, peak voltage will easily exceed 200V. Note that there is RMS, not peak on the graph.

It's not practical to use such high voltage MOSFETs. Open up any mains powered switched mode power supply and you'll find the MOSFETs will be rated to 400V in the 230V world and 200V in the 120V world. If it's an industrial unit you may find IGBTs rated to 600V but nothing like the voltages shown in that graph.

If there's a high voltage spike, the MOSFETs will simply go into avalanche breakdown and will not be harmed, as long as the energy in the spike is not too great. If it's an issue, there are this things called  metal-oxide varistors, which can be connected in parallel with the MOSFETs to will absorb some of the spike and provide some degree of protection.

[eneuro]

Why would I need an opto isolated supply for the driver. Isn't the input isolated
already?

I meant galvanic insulated/floating power supply for AC mosfets switch gate driving, but as I wrote battery is good starting point to test this thing.
You can use even 9V battery with lets say 20W mains load (small light bulb), while it should switch on those bulky mosfets with decent RDSON, but for higher loads 12V or sometimes 15V even better to lower down mosfets RDSON.

http://forum.allaboutcircuits.com/threads/floating-power-supply.10345/

A floating power supply can aso be imagined as a battery.

[Zero999]

Rather than using an opto-isolator, you could inductively or capacitively couple the control signal to the MOSFETs.


The pulse transformer needs to be rated to give adequate insulation between the primary and secondary and the capacitors need to be Y1 rated. The value isn't important but should be low enough to provide a high impedance to 60Hz so the leakage current is kept to a save level: 100pf should be more than enough.

[OC71]

Thanks everyone.
@ Hero999
 -- What is the purpose of R1?
 -- Does it matter whether RL is on the Hot or Neutral side?

[Zero999]

1) R1 discharges the gate-source capacitance when the control signal is switched off. Without it, the MOSFET will take a long time to turn off or may never turn off.

2) No. Why would it? The MOSFETs just act like a switch in series with the load. Generally the phase conductor should be switched for safety reasons but in this case MOSFETs should not be relied upon to safely isolate the load from the supply.

[OC71]

@ Hero999
Thanks for your input.
Why 200KHz control? Why not 60Hz mains frequency?

[dom0]

- Smaller pulse transformer (at 200 kHz you can use "standard issue" pulse xformers for SMPS)
- Cgs + R1 give a RC time constant for turn off. At 60 Hz this time constant must be unreasonably long, which leads to very slow turn off speeds and eventual thermal issues for the FETs. With a 200 kHz control signal the time constant can be much shorter.

I still don't get why you guys try to build your own SSR 

[Zero999]

- Smaller pulse transformer (at 200 kHz you can use "standard issue" pulse xformers for SMPS)
- Cgs + R1 give a RC time constant for turn off. At 60 Hz this time constant must be unreasonably long, which leads to very slow turn off speeds and eventual thermal issues for the FETs. With a 200 kHz control signal the time constant can be much shorter.

That's true for the pulse transformer version but what about the one with the Y1 capacitors?

@ Hero999
Thanks for your input.
Why 200KHz control? Why not 60Hz mains frequency?

Think about it.

What do the capacitors do?

How does the impedance a capacitor vary with frequency?

I still don't get why you guys try to build your own SSR 

As a learning exercise.

[langwadt]

- Smaller pulse transformer (at 200 kHz you can use "standard issue" pulse xformers for SMPS)
- Cgs + R1 give a RC time constant for turn off. At 60 Hz this time constant must be unreasonably long, which leads to very slow turn off speeds and eventual thermal issues for the FETs. With a 200 kHz control signal the time constant can be much shorter.

I still don't get why you guys try to build your own SSR 

Most SSRs use TRIACs, you can't turn off a TRIAC until the next zero crossing and most cannot handle a half
cycle into  a dead short and a fuse is useless for protecting a TRIAC

With FETs you can turn off in microseconds so you can make it practically bullet-proof

I have seen a few "random off" SSRs but they were ridiculously expensive

[dom0]

Most SSRs will be used for ohmic loads where fusing is not really an issue. Generally speaking short circuits are not a really an issue for larger SSRs (25+ A) in household circuits, since short circuits currents are generally rather low (200-400 A). Smaller SSRs like for example a Sharp S202S02 have 10 ms peak currents of something like 75 A, so you need to pay a bit of attention here.
Fusing SSRs for inductive loads is much harder and yes, it might even be impossible to find a fuse that provides the necessary protection in some cases. In those cases I'd just go for the next larger SSR and be done with it, or consider wholly different approaches (e.g. using a motor driver instead of a SSR to switch a motor, or using classic relays if switching is infrequent etc.)

[eneuro]

Rather than using an opto-isolator, you could inductively or capacitively couple the control signal to the MOSFETs.

Whet you have 200kHz and toroid or small transformer than you can easy make a few mA 12V-15V PSU and power up TLP351F, than another few mA for optoisolator diode and you have really fast ON/OFF switching 
In some projects use TLP351F with +/-12V (yep it is possible make such PSU for this optoisolator) for faster turning OFF with negative voltage rather than by 0V 

When you use 220k resistor on those two AC mosfets with this IRF http://www.tme.eu/en/Document/7f58240ac39dafede53e055f51f59c8f/irfp4668pbf.pdf  total gate charge: 2*Qg= 2*162nC= 324 nC,
C=Q/V , for 12V we have: C= 27nF , so you get RC ~6ms (http://www.referencedesigner.com/rfcal/cal_05.php ) and when we multiply this by lets say optimistic 2x RC ~14% of 12V ~1.6V (probably below gate threshold voltage for this IRF), than we have >10ms for AC SSR switch OFF without high current gate driver using bloody 220k resistor 
I use 100k gate resistor, but only as additional pull down with TLP351F or even 0.8Amax BC337/BC327 totem pole discrete driver.

[Zero999]

Rather than using an opto-isolator, you could inductively or capacitively couple the control signal to the MOSFETs.

Whet you have 200kHz and toroid or small transformer than you can easy make a few mA 12V-15V PSU and power up TLP351F, than another few mA for optoisolator diode and you have really fast ON/OFF switching 
In some projects use TLP351F with +/-12V (yep it is possible make such PSU for this optoisolator) for faster turning OFF with negative voltage rather than by 0V 

When you use 220k resistor on those two AC mosfets with this IRF http://www.tme.eu/en/Document/7f58240ac39dafede53e055f51f59c8f/irfp4668pbf.pdf  total gate charge: 2*Qg= 2*162nC= 324 nC,
C=Q/V , for 12V we have: C= 27nF , so you get RC ~6ms (http://www.referencedesigner.com/rfcal/cal_05.php ) and when we multiply this by lets say optimistic 2x RC ~14% of 12V ~1.6V (probably below gate threshold voltage for this IRF), than we have >10ms for AC SSR switch OFF without high current gate driver using bloody 220k resistor 
I use 100k gate resistor, but only as additional pull down with TLP351F or even 0.8Amax BC337/BC327 totem pole discrete driver.

I agree it would be slow but no worse than a optically coupled MOSFET. There is no reason why the gate resistor couldn't be decreased if switching speed is an issue.

It's good this has been brought up - I need a high speed (or not so slow) SSR for another project but it's not mains. I've just tried the capacitively coupled MOSFET SSR on my bench but used 15k for R1 and increased the frequency to 1MHz. I used the old 2N7000 (yes I know it's not suitable for mains and has a low gate capacitance) and was able to switch the MOSFETs at 17kHz. The main issue I can see with higher frequencies is RFI but that can be solved with capacitors and ferrite beads.

I agree the ultimate solution to switching speed is to use a transformer or capacitively coupled isolated power supply to power a proper MOSFET driver.

[OC71]

Thanks for the info everyone. I am currently trying to soak it in.
Meanwhile, with my limited knowledge is it possible we have gone full
circle from the VOM1271T  driver which has integrated turn-off
circuitry.

[Zero999]

I think eneuro is talking about using a separate isolated power supply and the TLP351F to drive the MOSFETs.
http://www.toshiba.com/taec/components2/Datasheet_Sync/201410/DST_TLP351F-TDE_EN_22634.pdf

This will enable a very high switching speed at the expense of more complexity.

How often do you intend to switch the relay on/off? The VOM1271T  driver will be far too slow for any kind of power control.

Steady state wise, I think it'll be fine without a heat sink providing it's not in a hot environment. Why do you feel the need to avoid a heat sink? Is there not room in the enclosure for one?

[eneuro]

How often do you intend to switch the relay on/off? The VOM1271T  driver will be far too slow for any kind of power control.

Just trying to estimate how slow it could be with those mosfets choosen by @OC71 
We need such amount of charge AC mosfets switch gates from typical values for irfp4668pbf :
dQ= 2*Qg= 2*162nC= 324 nC
At 30mA VOM1271T says about 47uA gate current, while I=dQ/dt -> dt= dQ/I, so dt= 324nC/47uA ~7 ms  At 50Hz mains we have 20ms period, so it looks like full switch on could take 30% of 50H z period time ? 
Could someone verify this? maybe I've made some mistake there?

There is also something else VOM1271T open circuit voltage at 30mA iF is maximum ~9V, so we can not get Vgs even 10V? 

I was thinking it could be usable when we add zero crossing, but 7ms turn on and Vgs<10V, I do not know, I do not like it, so that is why I dismissed it.

[OC71]

@ Hero999,
It is for on/off control of a heater (I think it is called bang bang).
 --There will be no more than 30 per minute. Could that be considered steady state?
 --If I use the TLP351F,will I need to isolate its power supply?
 --I am trying to do it without a heat sink mainly just to see if it is possible.

@ eneuro,
With the switching speed you have calculated, can I use the VOM1271 for the
30 per minute on/ off control?

Is this graph appropriate for determining the suitability of a driver, or am I
Wrong on that ( by the way, the labels for the 10ms and 1ms curves appear to be
reversed)

[Zero999]

I think it'll be Ok.

When the MOSFETs turn on, at the peak mains voltage, the voltage across them will not be the full 170V. The load will also drop some of the voltage too. They will also not always turn on/off when the mains voltage is at it's peak. Sometimes the voltage will be nearer zero. It will come down to chance.

Yes, if the TLP351 is used, it must have an isolated power supply with the same level of separation as you would the mains, as the negative rail is connected to the MOSFETs which are connected to the mains.

[OC71]

Will this schematic provide the necessary isolation?

[Zero999]

Will this schematic provide the necessary isolation?

Yes, as long as you treat everything connected directly to the secondary of that transformer with the same precautions as mains.

[OC71]

Thanks hero.

[eneuro]

Will this schematic ...

Remember that gate driver output might need resistor to limit this current, especially when you have no any voltage regulator on driver PSU-TLP351F has Vmax 27V if good remember, IRFs gate +/-30V, so... have you ever looked at sample Mosfets gate driving circuits?

[OC71]

@ eneuro,
Can you please clarify, I do not understand what you are saying?

[Zero999]

No I don't understand either.

The MOSFET gate voltage will be below the peak AC voltage on the secondary of the transformer which should be below 20V for a 12VAC transformer.

[eneuro]

The MOSFET gate voltage will be below the peak AC voltage on the secondary of the transformer which should be below 20V for a 12VAC transformer.

Maybe power consumption in TLP351F will be high enougth to keep this gate driver "PSU" below peak AC voltage in the case of mains primary transients, but rather common thing have low current voltage regulator in PSU and gate resistors to limit gate current to levels safe for driver itself
I'm not sure which is RDSON of TLP351F internal mosfets, so adding gate resistor assuming it is quite low below 10 ohm, etc.....

[Zero999]

The MOSFET gate voltage will be below the peak AC voltage on the secondary of the transformer which should be below 20V for a 12VAC transformer.

Maybe power consumption in TLP351F will be high enougth to keep this gate driver "PSU" below peak AC voltage in the case of mains primary transients, but rather common thing have low current voltage regulator in PSU and gate resistors to limit gate current to levels safe for driver itself
I'm not sure which is RDSON of TLP351F internal mosfets, so adding gate resistor assuming it is quite low below 10 ohm, etc.....

The rectifier diodes will keep the voltage below the AC peak regardless.

And how do you know what the peak AC voltage is when he didn't even specify the secondary voltage rating of the transformer? I made the assumption it's something sensible such as 9V or 12V which shouldn't exceed the MOSFET's gate rating.