Need advice on cheap SSRs

[Jwillis]

Interesting .I didn't mention China.

[wraper]

Interesting .I didn't mention China.

But that crap comes from China. If you buy it somewhere else, it's just Chinese crap being resold. What at least looks like a genuine Fotek according to the pictures sells for more than $20 for 25A SSR on ebay.

[Jwillis]

Marking things up doesn't make them genuine.

[wraper]

Marking things up doesn't make them genuine.

That is not marking things up. Every picture in cheaper listings show 100% counterfeits. Some which are expensive at least have pictures of how genuine Fotek should look like. I did not say they necessarily are. Although there are expensive listings with counterfeits on photos as well. Particularity they should bear "Taiwan made" written on them.

https://www.ul.com/newsroom/publicnotices/ul-warns-of-solid-state-relay-with-counterfeit-ul-recognition-mark-release-13pn-52/

Photo of the counterfeit relay


Photo of the relay authorized to bear the UL Mark

[Jwillis]

That is not marking things up. Every picture in cheaper listings show 100% counterfeits. Some which are expensive at least have pictures of how genuine Fotek should look like. I did not say they necessarily are. Although there are expensive listings with counterfeits on photos as well. Particularity they should bear "Taiwan made" written on them.
[/quote]

Direct from http://www.fotek.com.tw/Ecenter1.asp?classNo=4&class2_sn=16

No "Taiwan made" written on them.
Seems Fotek is counterfeiting Fotek.

[Peabody]

The Crydom SSR I ordered on Ebay arrived today.

https://www.ebay.com/itm/273522773981

It's exactly what is shown in the listing pictures, including the Crydom box and the 2006 date code.  It comes with the screw terminal hardware, but no heatsink screws.  Based on the packaging, I believe it is genuine new/old stock.

I had to order a heatsink separately, and it hasn't arrived yet.  So at this point I'm just testing it with a nightlight.  But it seems to work fine.

I have a question about the DC input drive.  It's supposed to work down to 3V, and it works fine at my 3.3V.  But that's driving a night light.  If I put 10 amps through it, will I have to increase the DC voltage, or will the 3.3V continue to work?

Compared to what new/new Crydoms cost, this seems to be a good deal.  That's assuming its age doesn't present any problems.  Seems like it shouldn't.

Also, I find that the metal plate on the back isn't galvanically connected to any of the four terminals, which means I won't have to worry about the heatsink electrocuting anyone.  I have grey thermal paste, which I believe is conductive, but if the plate is isolated that won't matter.

Anyway, I think this was a good buy so long as it keeps working.  But as of now he has zero left to sell.

Edit:  As of Dec 11, he's now showing more than 10 available.

[wraper]

Minimum firing voltage does not depend on load. As I said before, you don't need thermal paste. There already should be thermal interface applied.

If I put 10 amps through it, will I have to increase the DC voltage, or will the 3.3V continue to work?

Is it hard to open datasheet?

I find that the metal plate on the back isn't galvanically connected to any of the four terminals

Of course it's not. But I would highly recommend that heatsink/chassis are grounded.

[Peabody]

I wanted to report that when using the Crydom SSR and heatsink with my hot plate reflow controller, switching 1000 watts, the heat sink doesn't get warm at all.  I suspect that means it isn't really necessary.  The sequence only takes about 5 minutes, and I'd guess that's not enough to even make it warm up.  So I guess in hindsight the fake Fotek SSRs would probably work fine, but using the "40A" version in the hope of getting a real 10A.

[langwadt]

A cheap SSR can be made with high quality materials. A pair of back-to-back MOSFETs and a TLP3906 is all you need to implement a crude SSR.
You won't have fancy features like zero crossing detection (which can be implemented with even cheaper photo triacs and power triacs, but with higher on state loss).
And you won't have safety agency certificates. But technology wise, an $8 10A high quality SSR is more than feasible.

For instance, FCPF067N65S3 from DigiKey is $2.81 at 1kpcs, with 144mR maxim Rdson at 125C, and 2.7K/W thermal resistance.
Therefore, to design for 125C Tjmax and 40C baseplate temperature, the maximum power dissipation is 31.5W, or 14.8A.
To achieve 10A operation, you need two devices, connected back-to-back to block AC.
Totally you will need $5.62 on power FETs, plus $0.77 on gate driver, for a grant total of $6.39, then add cost for PCB, encapsulant and pins, say $7.5, is enough for a unit.
At full load current, it will dissipate less than 18.8W (Tc=40C, Tj=68C, Rdson_unit=1.4, Rdson_25c_max=67mR, three iterations) for both FETs.
You can also double the FETs for better efficiency and higher looser cooling requirements, and you still get ~$13.5 per module.

The cheapest Crydom 10A will cost you at least $30, and will have only a bit lower power dissipation as the $7.5 version.

another advantage to using FETs is that you can have near instant off so it can be made short circuit proof

[Zero999]

A cheap SSR can be made with high quality materials. A pair of back-to-back MOSFETs and a TLP3906 is all you need to implement a crude SSR.
You won't have fancy features like zero crossing detection (which can be implemented with even cheaper photo triacs and power triacs, but with higher on state loss).
And you won't have safety agency certificates. But technology wise, an $8 10A high quality SSR is more than feasible.

For instance, FCPF067N65S3 from DigiKey is $2.81 at 1kpcs, with 144mR maxim Rdson at 125C, and 2.7K/W thermal resistance.
Therefore, to design for 125C Tjmax and 40C baseplate temperature, the maximum power dissipation is 31.5W, or 14.8A.
To achieve 10A operation, you need two devices, connected back-to-back to block AC.
Totally you will need $5.62 on power FETs, plus $0.77 on gate driver, for a grant total of $6.39, then add cost for PCB, encapsulant and pins, say $7.5, is enough for a unit.
At full load current, it will dissipate less than 18.8W (Tc=40C, Tj=68C, Rdson_unit=1.4, Rdson_25c_max=67mR, three iterations) for both FETs.
You can also double the FETs for better efficiency and higher looser cooling requirements, and you still get ~$13.5 per module.

The cheapest Crydom 10A will cost you at least $30, and will have only a bit lower power dissipation as the $7.5 version.

As much as I like MOSFETs, they're still a lot more expensive than a TRIAC, especially if you're trying to get the power dissipation down at high currents. The Q4015R5TP I posted a link to earlier on in the thread, is much cheaper than those MOSFETs and will dissipate just over 10W, with an RMS current of 10A and can be purchased in single units for £1.71 from Mouser.

https://www2.mouser.com/datasheet/2/240/Littelfuse_Thyristor_Qxx15xx_Qxx16xHx_Datasheet.pd-775524.pdf
https://www.mouser.co.uk/ProductDetail/Littelfuse/Q4015R5TP?qs=sGAEpiMZZMvLfqGeKEshX8Jtl8j%252bpxKso7ZxoeqD3pz01KAeSOq5hw==

[wraper]

For instance, FCPF067N65S3 from DigiKey is $2.81 at 1kpcs, with 144mR maxim Rdson at 125C, and 2.7K/W thermal resistance.
Therefore, to design for 125C Tjmax and 40C baseplate temperature, the maximum power dissipation is 31.5W, or 14.8A.
To achieve 10A operation, you need two devices, connected back-to-back to block AC.
Totally you will need $5.62 on power FETs, plus $0.77 on gate driver, for a grant total of $6.39, then add cost for PCB, encapsulant and pins, say $7.5, is enough for a unit.
At full load current, it will dissipate less than 18.8W (Tc=40C, Tj=68C, Rdson_unit=1.4, Rdson_25c_max=67mR, three iterations) for both FETs.
You can also double the FETs for better efficiency and higher looser cooling requirements, and you still get ~$13.5 per module.

That's slightly worse efficiency than you would get with 25A triac like BTA24 + MOC306x. Much more expensive, no zero cross and won't have insulated tab. On top of that less surge current endurance.

[wraper]

A FET can turn off immediately, and has lower THD,

None of which are normally needed for SSR.

plus has intrinsic overload protection (positive Rdson vs temp).

How that would help in this case? It would only increase heat produced with temperature rise. The only good thing about it is that you can connect them in parallel.

[charlesrg]

Sorry for putting a message on an old topic, but looks like not many people are talking about those things lately, and most are going back to old threads to get info.
I want to share my experience:
I have some cheap SSRs that came with the sparkfun SSR relay board kit whole thing with 4 SSRs cost $40 USD. So I placed it into a heatsink and put a load of 1000W and after 21 minutes the heatsink reached 45 degrees celsius.

So I thought I could just buy one of latest Crydom SSRs and get much better results under the same controlled environment. I ordered 1 CWD2490 that cost me $90 USD.

Mounted the CRYDOM to the same heatsink and put the same 1000Watts load and after 10 minutes the heatsink was already at 45 degrees celsius.

So I'm very disappointed knowing that I paid so much for something that is so inferior.

[wraper]

So I thought I could just buy one of latest Crydom SSRs and get much better results under the same controlled environment. I ordered 1 CWD2490 that cost me $90 USD.
Mounted the CRYDOM to the same heatsink and put the same 1000Watts load and after 10 minutes the heatsink was already at 45 degrees celsius.
So I'm very disappointed knowing that I paid so much for something that is so inferior.

More like your expectations, measurements and conclusions are extremely flawed. Especially flawed is part where you estimate heating time to certain temperature instead of final stable temperature, it may just mean there is better heat transfer between pass element in SSR and heatsink or you did not give enough time for heatsink to fully cool down after first measurement, etc. First of all you need to ensure you are doing measurements properly which with precise temperature measurements is quite hard. Secondly power loss can be way more precisely estimated by voltage drop over SSR, not by some heatsink temperature measurement which can depend on many factors. Not to say there won't be a huge difference in heat dissipation since voltage drop over triac will be more or less similar unless triac is heavily undersized.

[Terry Bites]

Don't buy cheap, buy traceable so you can sue them when your eyebrows get singed.
Do you really need an SSR? Mechanical relays have astonishing life times (if you treat 'em right) and are much cheaper
 and they don't get hot. OK, a little warm sometimes.

Who are these THD lunatics?

[jesuscf]

Do you really need an SSR? Mechanical relays have astonishing life times (if you treat 'em right) and are much cheaper
 and they don't get hot. OK, a little warm sometimes.

One massive advantage of SSRs over relays is the precise control of the switching instant.  For instance, high power SSRs only switch when the AC voltage crosses zero.  That would be very hard, if not impossible to do with electro-mechanical relays.  Also, I doubt a relay would be cheaper than a Triac/Optocoupler pair...

[CatalinaWOW]

Do you really need an SSR? Mechanical relays have astonishing life times (if you treat 'em right) and are much cheaper
 and they don't get hot. OK, a little warm sometimes.

One massive advantage of SSRs over relays is the precise control of the switching instant.  For instance, high power SSRs only switch when the AC voltage crosses zero.  That would be very hard, if not impossible to do with electro-mechanical relays.  Also, I doubt a relay would be cheaper than a Triac/Optocoupler pair...

If dissipation is a problem use both.  Put the EM relay and SSR in parallel.  Use a small micro to close the SSR.  Once it closes, close the EM Relay. Then open the SSR.   Reverse on opening. Close the SSR, open EM relay, then the SSR. 

From a performance standpoint it is the best of both worlds.  Not the cheapest or most compact, but for some applications would be the way to go.

[charlesrg]

More like your expectations, measurements and conclusions are extremely flawed. Especially flawed is part where you estimate heating time to certain temperature instead of final stable temperature, it may just mean there is better heat transfer between pass element in SSR and heatsink or you did not give enough time for heatsink to fully cool down after first measurement, etc. First of all you need to ensure you are doing measurements properly which with precise temperature measurements is quite hard. Secondly power loss can be way more precisely estimated by voltage drop over SSR, not by some heatsink temperature measurement which can depend on many factors. Not to say there won't be a huge difference in heat dissipation since voltage drop over triac will be more or less similar unless triac is heavily undersized.

I agree with you that improper heat transfer could cause a skew on the tests. On my defense I've used thermal paste and exact the same Radiator for both tests. If the Internal heat transfer is not great than my results would not be valid.
In my defense I did multiple tests and ensured the initial temperature of the radiator was the same on each test for each SSR. I'm using an ESP32 to control the relay on/off and monitor the temperature each time with same sensor and same radiator.
How would  you suggest measuring voltage drop on the AC side, same way as on DC circuits ? (By measuring voltage before and after the SSR ?) If yes I will proceed with this test and post results.

Don't buy cheap, buy traceable so you can sue them when your eyebrows get singed.
Do you really need an SSR? Mechanical relays have astonishing life times (if you treat 'em right) and are much cheaper
 and they don't get hot. OK, a little warm sometimes.

Who are these THD lunatics?

SSRs will be installed near sleeping quarters and I don't want to hear click/clacks at night.

Do you really need an SSR? Mechanical relays have astonishing life times (if you treat 'em right) and are much cheaper
 and they don't get hot. OK, a little warm sometimes.

One massive advantage of SSRs over relays is the precise control of the switching instant.  For instance, high power SSRs only switch when the AC voltage crosses zero.  That would be very hard, if not impossible to do with electro-mechanical relays.  Also, I doubt a relay would be cheaper than a Triac/Optocoupler pair...

If dissipation is a problem use both.  Put the EM relay and SSR in parallel.  Use a small micro to close the SSR.  Once it closes, close the EM Relay. Then open the SSR.   Reverse on opening. Close the SSR, open EM relay, then the SSR. 

From a performance standpoint it is the best of both worlds.  Not the cheapest or most compact, but for some applications would be the way to go.

That will add complexity. I've about 8DC and 4AC things that I want to be able to turn on/off silently.  Lights, water heater, floor heater, fans are some of them.
Here is a picture of the project: https://imgur.com/a/CXYDmdZ
ESP32 connected to sensors will be the brains to turn on and off things. Example if batteries are at 90% turn on water heater so extra solar power gets turned into hot water.

[CatalinaWOW]

More like your expectations, measurements and conclusions are extremely flawed. Especially flawed is part where you estimate heating time to certain temperature instead of final stable temperature, it may just mean there is better heat transfer between pass element in SSR and heatsink or you did not give enough time for heatsink to fully cool down after first measurement, etc. First of all you need to ensure you are doing measurements properly which with precise temperature measurements is quite hard. Secondly power loss can be way more precisely estimated by voltage drop over SSR, not by some heatsink temperature measurement which can depend on many factors. Not to say there won't be a huge difference in heat dissipation since voltage drop over triac will be more or less similar unless triac is heavily undersized.

I agree with you that improper heat transfer could cause a skew on the tests. On my defense I've used thermal paste and exact the same Radiator for both tests. If the Internal heat transfer is not great than my results would not be valid.
In my defense I did multiple tests and ensured the initial temperature of the radiator was the same on each test for each SSR. I'm using an ESP32 to control the relay on/off and monitor the temperature each time with same sensor and same radiator.
How would  you suggest measuring voltage drop on the AC side, same way as on DC circuits ? (By measuring voltage before and after the SSR ?) If yes I will proceed with this test and post results.

Don't buy cheap, buy traceable so you can sue them when your eyebrows get singed.
Do you really need an SSR? Mechanical relays have astonishing life times (if you treat 'em right) and are much cheaper
 and they don't get hot. OK, a little warm sometimes.

Who are these THD lunatics?

SSRs will be installed near sleeping quarters and I don't want to hear click/clacks at night.

Do you really need an SSR? Mechanical relays have astonishing life times (if you treat 'em right) and are much cheaper
 and they don't get hot. OK, a little warm sometimes.

One massive advantage of SSRs over relays is the precise control of the switching instant.  For instance, high power SSRs only switch when the AC voltage crosses zero.  That would be very hard, if not impossible to do with electro-mechanical relays.  Also, I doubt a relay would be cheaper than a Triac/Optocoupler pair...

If dissipation is a problem use both.  Put the EM relay and SSR in parallel.  Use a small micro to close the SSR.  Once it closes, close the EM Relay. Then open the SSR.   Reverse on opening. Close the SSR, open EM relay, then the SSR. 

From a performance standpoint it is the best of both worlds.  Not the cheapest or most compact, but for some applications would be the way to go.

That will add complexity. I've about 8DC and 4AC things that I want to be able to turn on/off silently.  Lights, water heater, floor heater, fans are some of them.
Here is a picture of the project: https://imgur.com/a/CXYDmdZ
ESP32 connected to sensors will be the brains to turn on and off things. Example if batteries are at 90% turn on water heater so extra solar power gets turned into hot water.

Valid reasons.  And a real world example of requirements being added mid problem.  If your heat problem becomes bad enough, there are solutions to the noise problems.  Complexity is real, but it isn't more complex than programming a dot matrix LED display.

All engineering solutions are compromises between often conflicting requirements.

[wraper]

How would  you suggest measuring voltage drop on the AC side, same way as on DC circuits ? (By measuring voltage before and after the SSR ?) If yes I will proceed with this test and post results.

The easiest would be to use current limited DC source such as LAB PSU. Pass same DC current through both SSR and measure voltage drop across SSR terminals. AC measurement and load attached is fine too. Just make multiple measurements since mains voltage may change.

[langwadt]

How would  you suggest measuring voltage drop on the AC side, same way as on DC circuits ? (By measuring voltage before and after the SSR ?) If yes I will proceed with this test and post results.

The easiest would be to use current limited DC source such as LAB PSU. Pass same DC current through both SSR and measure voltage drop across SSR terminals. AC measurement and load attached is fine too. Just make multiple measurements since mains voltage may change.

or just measure across the SSR when it is on, but if it is both triacs I'd expect the drops to be very similar around ~1.5V

[wizard69]

Thanks for the responses.

If this project is ever completed, it will reside at a local makerspace. 

This immediately threw up red flags.   Your contraption has no business being in a public place if you are going to cut corners.   If anything happens you could be in a big hurt liability wise.

With that, I would not feel comfortable trying to build an SSR from scratch.  So it may be that a $45 SSR from Digikey is the only option.

You have lots of options.   This is sounding pretty defeatist to me.   You could consider a number of retailers with a web presence, or check out local suppliers to industry.   Here I'm talking electrical supply houses.

  But you know, it doesn't really make sense that an SSR should need to cost twice as much as the entire toaster oven.  It seems there should be some source of good stuff at prices substantially less than Crydoms.

This statement baffles me as even DigiKey handles many brands.   You don't have to buy Crydoms.

    You also need to realize what niche the likes of Crydom is selling into.   The cost of a SSR is trivial compared to the down time it would take to implement your own and verify that it is working.   workign on controls puts you on a clock and how fast that clock spins depends upon many factors.   In any event a down machine can cost a company anything from a trivial amount of money to tens of thousands per minute, the cost of a canned solution is trivial.

If Fotek is actually a legitimate brand, is there a place where you would be certain of getting their genuine stuff?  Are there any other low-cost (but not junk) SSR brands?

Never heard of Fotek until this thread cropped up.

As for brands; what do you need from me to convince you that you have dozens of brands to consider?

Now lets consider what you are building.   This oven is expected to go into a public space, thus it needs to be wired and constructed to be compliant with the NEC and UL.   Ideally you will have a separate over temp detector and a control switch (relay) for that.

Speaking of relays you actually have other choices beyond a SSR.    There are mechanical relays specifically made for thermal heating control.   You can still buy Mercury relays for a reasonable price.   https://www.wolfautomation.com/products/mercury-relays/?gclid=EAIaIQobChMI0PSBy7Oe7wIVC42GCh3N4gMwEAMYAyAAEgKwc_D_BwE. 

By the way for many use cases mercury relays are almost ideal.   They will literally last for many decades and are still popular in thermal control systems in industry.

[Peabody]

Well, this is what happens when old threads are replied to.  This project was completed three years ago in the form of a hot plate reflow device.  It was given to a subscription-only maker space populated by local EEs, and which has since closed down.  In the end, it didn't need an SSR at all.  The reflow process consisted of turning on and off the hot plate twice at specific intervals which roughly reproduce the suggested reflow curve for leaded paste, and that can be done manually using the hot plate power switch and a watch.

[CatalinaWOW]

Well, this is what happens when old threads are replied to.  This project was completed three years ago in the form of a hot plate reflow device.  It was given to a subscription-only maker space populated by local EEs, and which has since closed down.  In the end, it didn't need an SSR at all.  The reflow process consisted of turning on and off the hot plate twice at specific intervals which roughly reproduce the suggested reflow curve for leaded paste, and that can be done manually using the hot plate power switch and a watch.

While the late replies don't benefit the original project, they do help the community learn.  Not wasted time.

[charlesrg]

I'm happy that you guys concluded the reflow oven project. Sorry for hijacking the thread. I do find useful the pointing to Mercury relays, first time I've heard of them.

On my Crydom VS Cheap I've went back and this time instead of testing with temperature probe I used 2 multimeters and accounted for the +0.1V volt variation on the second multimeter.

I checked 5 times @425W load on CRYDOM vs Cheap and checked 5 times @925W load as well.

The results:

SSR	Load	Drop
Cheap NoName $5	425W	~0.8V
Crydom CWD2490 $90	425W	~0.9V
Cheap NoName $5	925W	~0.8V
Crydom CWD2490 $90	925W	~1V

I'm finding that paying more for an SSR has no benefit, this Crydom was supposed to handle the 925W @110V without a heatsink. Would arrow be selling counterfeit Crydoms ?