SSR vs mosfet with drivers, why is SSR so slow?

[JohanHoltby]

Hi!
I have been looking in to solid state relays and noticed that they are very slow compared to optoisolated  gate drivers for mosfets.
Why is this? I would like to have a "SSR" switching at 1mhz. This is a safety switch which should be turned of when an electric level is outside the allowed rates.

[BravoV]

I would like to have a "SSR" switching at 1mhz.

1 mhz ? or 1 Mhz ?

[T3sl4co1l]

You really wouldn't want that, actually -- at 1MHz, you'll get all sorts of nasty peak voltages and currents from the unavoidable reactances (inductance and capacitance) present in the surrounding wiring.  And coming so frequently that the thing will destroy itself.  If it survives the first cycle, even.  You really need extra circuitry around the switching device to limit that, and that circuitry tends to make it special-purpose.  So that's basically why we still have bare transistors around, for the most part.

Note there are basically two kinds of SSRs: AC-only (thyristor based), and AC/DC (MOSFET(s) + photovoltaic driver).  Thyristors (namely TRIAC or SCRs) turn on reasonably fast (fractional microsecond, which can be tricky concerning peak currents through capacitors), but stay on until current drops to zero, so can only be used on AC.  The PV driver delivers very little current (microamperes), so turns on the MOSFETs very slowly (several milliseconds).  The switching loss can therefore be pretty high, but that just means you don't want to turn it on and off very frequently (which is usually fine for such applications).

The MOS SSRs are also fairly useless at delivering fault current, so, won't clear a fuse before blowing up themselves.  Thyristors can survive such rough treatment, just barely, if the fuse blows quickly.

Tim

[Zero999]

An isolated MOSFET driver requires power on the secondary side permanently, whether it's on or off. A photovoltaically coupled MOSFET gets its power from an LED, on the primary side. As mentioned above, the current generated by the photocell is tiny, so the MOSFET takes a long time to switch.

A solid state relay should never be trusted in a safety critical application. It can easily fail short circuit, causing electrocution or a motor to run causing a serious injury, when it should be off.

[David Hess]

Internally SSRs use a light source and photovoltaic cell to create the voltage to turn on the MOSFETs.  Unlike an optocoupler, the voltage is sufficient to fully enhance the gate so 5 to 15 volts.  The current available in low so it takes time to charge the gate capacitance of a large device and turn it on.  For the same reason, the gate to source shunt resistance is high and takes time to remove charge to shut it off.

So switching times are typically in the 100s of microseconds to milliseconds.

[JohanHoltby]

Hi
It's the photovoltaically version I was intended. The 1Mhz is only to be able to turn of fast enough to not damage any equipment it will not be switching on and off multiple times.  The safety is only for equipment so that the DUT in the test system don't damage the test equipment. So do any one know of a way to switch faster? The signal should be effected as little as possible my be negative in current direction since it's a generic module.

[sourcecharge]

Hi!
I have been looking in to solid state relays and noticed that they are very slow compared to optoisolated  gate drivers for mosfets.
Why is this? I would like to have a "SSR" switching at 1mhz. This is a safety switch which should be turned of when an electric level is outside the allowed rates.

How much current and voltage are you looking to switch, and what R(d)on is your maximum?

[JohanHoltby]

About 0.5-10A. Will have two modules, one low current and one high current.
R(d)on should probably no exceed 1 ohm but the lower the better.

It's always a giva and take during designing for components.

[sourcecharge]

If you are switching ground on and off I may have something you might be interested in...

[JohanHoltby]

Sadly I don't know what I'm switching . It's a generic test module so it can be serial data, a 2 volt signa to an other pin on the DUT or maybe a supply current.

[langwadt]

I've used back-to-back fets controlled by an isopower icoupler

[T3sl4co1l]

10A but serial data or "2V signal to another pin"?

Are you sure you want all this functionality in a single device?

Such a description might actually be possible with GaN transistors these days, but it's not going to be cheap.

There is good reason why there are many kinds of optos and switches available today.  It's not practical to use one for everything.

Tim

[sourcecharge]

Sadly I don't know what I'm switching . It's a generic test module so it can be serial data, a 2 volt signa to an other pin on the DUT or maybe a supply current.

Well, I was going to suggest back to back mosfets too, but I think you could possibly damage the ICs, DUTs, or PS if the cutoff safety module grounds out the input to the test equipment.

My suggestion would be using zeners as + and - voltage bleeders, with a super quick burn fuse in series instead.  Not sure if the fuse would burn in 1us though...

I've used back-to-back fets controlled by an isopower icoupler

Can this be used without grounding one side of the back to back mosfets?

[langwadt]

Sadly I don't know what I'm switching . It's a generic test module so it can be serial data, a 2 volt signa to an other pin on the DUT or maybe a supply current.

Well, I was going to suggest back to back mosfets too, but I think you could possibly damage the ICs, DUTs, or PS if the cutoff safety module grounds out the input to the test equipment.

My suggestion would be using zeners as + and - voltage bleeders, with a super quick burn fuse in series instead.  Not sure if the fuse would burn in 1us though...

I've used back-to-back fets controlled by an isopower icoupler

Can this be used without grounding one side of the back to back mosfets?

everything on the output side is floating, that's what the icoupler is for.

here's a schematic of someone doing something similar,

https://m.eet.com/media/1179930/f1l.jpg 

[Richard Crowley]

Remember that most SSRs have built-in zero-crossing circuits which turn on only during the period when the mains waveform passes through zero volts.  That would completely circumvent an kind of phase-angle triggering.  You must be careful to get a SSR that allows phase triggering.  I just assume that all SSRs are zero-crossing unless proved otherwise.

[JohanHoltby]

I understand that there is challenge in making something work for every thing and that there will be paid a price in cost. The back to back mosfets is a way that I believe in most. I will have some more freedom but will increase PCB area since I will have more ICs. The zero crossing is something to be aware of so I totally agree and I will keep a sharp eye for it if I go with SSRs instead of using mosefets and gate drivers. The gate drivers however might have the same "feature" to watch out for.

[langwadt]

Remember that most SSRs have built-in zero-crossing circuits which turn on only during the period when the mains waveform passes through zero volts.  That would completely circumvent an kind of phase-angle triggering.  You must be careful to get a SSR that allows phase triggering.  I just assume that all SSRs are zero-crossing unless proved otherwise.

there's  "SSRs" and there's SSRs, the "SSRs" are triacs and is only a relay in that they can replace a relay in some specific AC applications

[JohanHoltby]

Hahah Every thing got a name thees days. What I intended what the plural form of SSR. Will keep that in mid though!