Solid-state relay holding/minimum current

[nostril]

I'm building a switched mains outlet to control my sound system based on the presence of 5V on the TV's USB socket. Since USB outlets are fairly weak, I want to avoid relays, so I've been looking at TRIAC control and off-the-shelf solid state relay modules. Using a generic mains outlet on the device invites me to build more than one and reuse the design with appliances other than my sound system, so I've fallen down the rabbit hole of designing the device to be reasonably accommodating of unknown mains loads. I'm looking at random turn-on SSRs capable of 25A. Sufficient headroom for inrushes on 10A loads, potentially inductive. I also want it to be capable of switching fairly light loads such as the sound system which draws approx 30 mA idle, mainly down to transformer magnetising current, power supply quiescent, etc.

I've so far found most off-the-shelf SSRs in this range have minimum current ratings 100 mA and above. A few to mention:

• Panasonic AQA421VL ~$30: 100 mA min
• Carlo Gavazzi RF1B23L25 ~$40: 150 mA min
• TE 240D25R ~$50: 100 mA min

A few stand out from the others in performance and price:

• Sensata 84134210, ~$70: 50 mA min -- almost usable on paper for my 30 mA load
• Crouzet 84137210N, ~$60: 5 mA min -- seems ideal. Confirmed by Crouzet engineers not to be a typo

I'm confused why that particular Crouzet part stands out on minimum current. I couldn't (yet) swallow its $60 price, so I began designing my own board and realised I think the combination of an opto-TRIAC triggering a larger 25A-capable TRIAC (see attached circuit scraped from Vishay App Note 34*) may allow a very low holding current. As I understand it, the opto-TRIAC takes the load current at the beginning of a trigger event until enough current flows on the larger TRIAC's gate, at which point it takes over. This means for loads too light to ever trigger the larger TRIAC, the opto-TRIAC will conduct all load current, and its holding current would apply.
My main questions are:

• I'm comfortable with mains, and realise I need transient and fault protection, snubbers etc. Am I crazy to build my own circuit here, should I just swallow the cost of an OTS hockey puck SSR or three?
• If I build my own, am I missing anything in thinking I can operate it with the opto-TRIAC conducting full load current for light loads, given the current and dissipation are within its limits?
• Is the Crouzet's specified 5 mA holding current unique, or are other manufacturers simply specifying the main switching element's holding current as their SSR's minimum to avoid complication/liability for what's probably an uncommon corner case? Maybe they're intentionally crippling their performance to avoid false triggering or lockup?

Any help or hints appreciated
* https://www.vishay.com/docs/84780/appnote34.pdf

[wraper]

You talk about 25A capable SSR but what load do you actually expect? For example, to provide constant 16A capability, you need to somehow dissipate about 25W of power. Use of SSR's generally don't make sense unless power needs to be switched in frequent manner. I'd rather go for a sensitive relay.

I'm looking at random turn-on SSRs capable of 25A.

For what?

[moffy]

 If you are really concerned about discontinuous AC just use two suitably rated MOSFETS back to back driven by a photovoltaic opto: https://www.vishay.com/docs/83469/vom1271.pdf

[PCB.Wiz]

Is the Crouzet's specified 5 mA holding current unique, or are other manufacturers simply specifying the main switching element's holding current as their SSR's minimum to avoid complication/liability for what's probably an uncommon corner case? Maybe they're intentionally crippling their performance to avoid false triggering or lockup?

It may be that the components in the gate branch, are not rated high enough to be the only power path.
ie they expect to only have brief gate trigger pulses.

I have used a generic 10A SSR block to slave some low power LED strips to a rangehood light circuit.
Worked very well, even tho it was at the 'bottom end' of current specs.

You could get a unit and try it?

[Whales]

Also checkout "latching relays".  You only have to give them a pulse to switch them.

[Someone]

Also checkout "latching relays".  You only have to give them a pulse to switch them.

For an application like the OP is describing where the delay in transitioning between on and off can be so slow the high peak power requirement can be worked around. Various “power harvesting” regulators can limit the input current to a significant capacitor, then some logic to apply the pulses (and reverse polarity if single coil). So a fair bit more complex design than a triac or mosfet switching but easier to understand/prove.

[PCB.Wiz]

As I understand it, the opto-TRIAC takes the load current at the beginning of a trigger event until enough current flows on the larger TRIAC's gate, at which point it takes over. This means for loads too light to ever trigger the larger TRIAC, the opto-TRIAC will conduct all load current, and its holding current would apply.

I pulled a couple of triacs and tested them, in the sub-latch region. ie with valid GATE drive but not enough T2 current to latch.
They do work as a transistor, and they do conduct T1-T2, so long as gate current is present.
eg Flows as low as 1-2-3mA are possible, on a 16A triac with 12mA DC gate drive. (well under the latching or holding current)
 That eases the demands on the combination, as even very light loads will share and reduce the power demands of the driver.

You could confirm that with some tests.

[moffy]

Holding current varies with temperature, gets lower as the triac heats and larger as it gets colder. This is a good app note about it: https://www.st.com/resource/en/application_note/an302-thyristors-and-triacs-holding-current--an-important-parameter-stmicroelectronics.pdf

[neillnz]

Another thing to consider is the leakage current. 

The datasheet for the the Crouzet part has a spec of 5 mA for the random version. 

Will your load cope with that? Or will it slowly charge a capacitor, reach a threashold, attempt to power on for a brief moment discharging the capacitor as it "blips".  This is the sort of thing that can happen with an LED light, the driver charges up with the leakage current and once it gets to a high enough voltage the driver arrempts to start and the LED brefly flashes as teh capacitor discharges.

There are SSRs with very low leakage, such as CPC1998J with 100 uA.

A snubber will add to the leakage.

 

[tooki]

I'm building a switched mains outlet to control my sound system based on the presence of 5V on the TV's USB socket. Since USB outlets are fairly weak, I want to avoid relays, so I've been looking at TRIAC control and off-the-shelf solid state relay modules. Using a generic mains outlet on the device invites me to build more than one and reuse the design with appliances other than my sound system, so I've fallen down the rabbit hole of designing the device to be reasonably accommodating of unknown mains loads. I'm looking at random turn-on SSRs capable of 25A. Sufficient headroom for inrushes on 10A loads, potentially inductive. I also want it to be capable of switching fairly light loads such as the sound system which draws approx 30 mA idle, mainly down to transformer magnetising current, power supply quiescent, etc.

I've so far found most off-the-shelf SSRs in this range have minimum current ratings 100 mA and above. A few to mention:

• Panasonic AQA421VL ~$30: 100 mA min
• Carlo Gavazzi RF1B23L25 ~$40: 150 mA min
• TE 240D25R ~$50: 100 mA min

A few stand out from the others in performance and price:

• Sensata 84134210, ~$70: 50 mA min -- almost usable on paper for my 30 mA load
• Crouzet 84137210N, ~$60: 5 mA min -- seems ideal. Confirmed by Crouzet engineers not to be a typo

I'm confused why that particular Crouzet part stands out on minimum current. I couldn't (yet) swallow its $60 price, so I began designing my own board and realised I think the combination of an opto-TRIAC triggering a larger 25A-capable TRIAC (see attached circuit scraped from Vishay App Note 34*) may allow a very low holding current. As I understand it, the opto-TRIAC takes the load current at the beginning of a trigger event until enough current flows on the larger TRIAC's gate, at which point it takes over. This means for loads too light to ever trigger the larger TRIAC, the opto-TRIAC will conduct all load current, and its holding current would apply.

The 500mA of a regular USB port would be more than enough to trigger and hold a relay. With the relay, you could either switch the load directly, or use it to switch a beefier contactor to switch the load. (For example, a 5V relay to switch a 230V contactor.)

Why would you want a random-turn-on SSR instead of a zero-crossing one? I’d only use one for a known, large inductive load.

[wizard69]

Electrical mechanical relays are still sold for a reason, it is hard to imagine why one would go to the trouble of an electronic circuit to turn random devices on an of,

[tooki]

Electrical mechanical relays are still sold for a reason, it is hard to imagine why one would go to the trouble of an electronic circuit to turn random devices on an of,

Well it has all to do with how frequently it gets turned on and off! You certainly wouldn’t want to use a mechanical relay to switch a heating element under PID control with a cycle time of 1s or less!

[floobydust]

I use an SSR to power up my audio system, for many years.
Junkbox parts like Idec RSSDN-25A SSR; dual-SCR 50mA Minimum Holding Current rated, and 25mA max. off-state due to the internal snubber I think.
The main audio amplifier has a 1.2kVA power transformer so there can be huge inrush currents because the SSR is zero-cross switching but it is rated 300A 1/2 cycle. It just causes the house wiring to rattle  and shake sometimes during power on.

SSR heat dissipation is roughly 1W per amp. You will need a heatsink.
Stay away from the honey pot temptation of cheap Fotek SSR's out there, most are junk. Sure you could build an SSR if HV spacings/isolation at the opto-isolator were decent and you have room.

OP I think your 30mA minimum current is not right. Even tiny power transformers have higher magentization current. It's a highly distorted, non-sinusoidal waveform that non true-RMS multimeters can misread and give a bad reading. You might check that again.

[nostril]

You talk about 25A capable SSR but what load do you actually expect? For example, to provide constant 16A capability, you need to somehow dissipate about 25W of power. Use of SSR's generally don't make sense unless power needs to be switched in frequent manner. I'd rather go for a sensitive relay.

I'm looking at random turn-on SSRs capable of 25A.

For what?

Fair point about dissipation. Standard outlets here are limited to 10 A continuous, so about 16 W to dissipate (still quite a bit). At and above 25A, the SSRs begin to have instantaneous current ratings that line up with medium to large transformers' inrush currents. As for random-turn on, perhaps I should say "instantaneous turn-on" -- switching on the zero-crossing guarantees starting inductive loads at the worst possible time, every time. Random turn-on would avoid this many times, as I understand it.

If you are really concerned about discontinuous AC just use two suitably rated MOSFETS back to back driven by a photovoltaic opto: https://www.vishay.com/docs/83469/vom1271.pdf

Thanks for the tips! I wasn't aware of this architecture of solid state relay, but it will be good reading. I had also read AN302 front to back already 
Thanks @PCB.Wiz for the two real-world examples. Even if I don't go with a SSR/TRIAC approach I'll have to stock my parts bin with some and have a play.

Another thing to consider is the leakage current. 
...
A snubber will add to the leakage.

I'll have to measure the amps' performance on leakage-type currents (one has a transformer, one has SMPS). I have an AC power supply that can limit down to a few mA. I hadn't considered snubber leakage, either. That will carry across to relays too, if they get snubbed to alleviate contact arcing.

The 500mA of a regular USB port would be more than enough to trigger and hold a relay.

You've found where I went wrong! I had a 100 mA current limit in my head, but I now see this is not true for USB-2.0, which a TV's port is definitely going to be (at least). A relay may become the better option then.

Electrical mechanical relays are still sold for a reason, it is hard to imagine why one would go to the trouble of an electronic circuit to turn random devices on an of,

My main two concerns would be higher power consumption on the controlling side, and long-term unreliability of contacts. As with semiconductor switches, they must be correctly de-rated for reactive loads.

Additional thanks to @floobydust for lived experience in the same situation. It sounds like I should re-measure the magnetising current. I was reading it off a generic mains power meter until now.

I have some further homework to do, it would seem. Thank you all so far for the helpful information and experience offered.

[nostril]

You've found where I went wrong! I had a 100 mA current limit in my head, but I now see this is not true for USB-2.0, which a TV's port is definitely going to be (at least). A relay may become the better option then.

I see where I got this 100 mA from, and it would appear to hold true if I stick to the letter of the standard. 100 mA limit until enumeration, at which point it lifts to 500 mA. I won't likely be being enumerated.

[PCB.Wiz]

The 500mA of a regular USB port would be more than enough to trigger and hold a relay. A relay may become the better option then.
....
I see where I got this 100 mA from, and it would appear to hold true if I stick to the letter of the standard. 100 mA limit until enumeration, at which point it lifts to 500 mA. I won't likely be being enumerated.

Even 100mA is ok, with modern low power coil relays.
I see at lcsc even up to 25A relays with < 100mA (500mW) ratings)

https://www.lcsc.com/product-detail/C48579779.html
360mW and specs a 75% of 5V pull-in rating, so just needs a low RDS FET drive.

[Seekonk]

Use a capacitor on the output as a load.  This will supply sufficient current to operate the SSR and keep the voltage on the output low enough.

[wraper]

You've found where I went wrong! I had a 100 mA current limit in my head, but I now see this is not true for USB-2.0, which a TV's port is definitely going to be (at least). A relay may become the better option then.

I see where I got this 100 mA from, and it would appear to hold true if I stick to the letter of the standard. 100 mA limit until enumeration, at which point it lifts to 500 mA. I won't likely be being enumerated.

No USB host is built to care about that. Not to say many smartphones will happily draw 1A or even more if connected to the device with charging only cable that don't have data lines.

[MKRD]

1) You say "potentially inductive", etc, so why not a zero-crossing type SSR???
2) You say that you can't afford a $60 SSR which meets your needs, but have you budgeted for a heatsink, and a safety enclosure to contain the dangerous voltages inside, while cooling down the SSR? Have you budgeted for machining and labor? You will multiply your BOM by several times to get to "product" cost.
I did exactly this for work, it was a sturdy metal box, outlet on outside, 120V cord going inside, SSR inside, heatsink on outside, several fuses added, and it took me two days to put all parts together and build it. If I factored in my salary for two days, it cost us a shocking amount to build such a box, but I was lightly occupied in labor that week, so who cares.
In the end, you realize it, someone else had already made such a product already, for a far cheaper price, and it's all potted etc, and much safer.

[Zero999]

1) You say "potentially inductive", etc, so why not a zero-crossing type SSR???

Because an SSR is possibly the worst thing to use for an inductive load. Switching at 0V, is the worst thing to do. It pretty much guarantees the highest inrush current and risk of core saturation.

Inductive loads are best switched when the voltage is as close as possible to the peak of the sine wave.

[MKRD]

https://youtu.be/8DNl4cfENBk?si=wjwFLv9dPpwQsUax&t=4502

[Zero999]

https://youtu.be/8DNl4cfENBk?si=wjwFLv9dPpwQsUax&t=4502

That video is blocked in the UK.

Just to clarify: I mean zero crossing is the worst type of switching for inductive loads, not SSRs in general, which are suitable.

[MKRD]

I don't go on the internet without a VPN and a dozen+ script,ad,malware-blocking browser extensions.
Dark Star the old 1974 sci-fi movie, the "I must think on this further" by the star-bomb scene.

[nostril]

Thanks again for the input, all. I've done some thinking and searching, and found a discontinued product I will likely imitate -- the "Power Switch Tail". It has been "reborn" in the form of a kit available from a Belgian outfit: https://shop.mchobby.be/en/relay-modules/1289-power-switch-tail-240v-ac-3232100012899.html for €31.40, with full schematics in the product description for free.
It uses an electromechanical relay with its coil driven by an opto-TRIAC. For my purposes, this gives the best all-round feature set:

• A low cost <$5 relay which could have a high coil power for all I care
• Draws only a few mA from the input -- not a material concern for my original use case, but allows the design to be reused with microcontrollers, etc, without an extra drive circuit
• Allows a range of voltages to be used as the trigger -- again, a benefit for reuse with unknown setups in future

I intend to deviate from their design and add a fuse, add input and output LEDs, drive the control-side LEDs with current-limiting diodes, and at leave some footprints for a snubber network across the relay contacts.

Best