Homemade Solid State Relay for 12 Volt Automotive Application

[David Hess]

I have been making various improvements to my old 2002 GMC Sonoma pickup.  It came with two 12 volt accessory sockets however like the cigarette lighter socket they are always powered, which means manually connecting/disconnecting my GPS, Bluetooth to FM converter, phone, and battery bank.

So to improve the situation I am going to make the accessory sockets switchable.  Originally I was going to tie them to the accessory/power from the key through a relay, but I think I will just add a manual switch.  The problem then becomes how to switch potentially 10+ amps.  I decided that a solid state relay would be best, however MOSFET output solid state relays with suitable ratings are incredibly expensive, so I am making my own using the TLP591 photovoltaic optocoupler and a pair of IRFP7530PBF power MOSFETs wired in anti-series.

Using the optocoupler removes some potential failure modes involving ground currents and provides the boost voltage needed to use n-channel MOSFETs for high side switching.  Using two MOSFETs in anti-series allows reverse blocking when off which also removes some failure modes.  The large MOSFET is selected based on keeping power dissipation low at high currents.

I also considered using a Linear Technology high side driver which would have allowed for fault protection, but decided that the photovoltaic optocoupler would be better.

Any thoughts?

[langwadt]

what's wrong with a relay? it is easy and reliable

[thm_w]

Is an appropriately rated high current switch too large or too expensive?
https://www.digikey.com/en/products/detail/cw-industries/GRB234A2NABA1/3190606

edit: if you could find a trusted source there are the fotek SSRs, no idea if the genuine ones are reasonably priced though

https://www.fotek.com.tw/en-gb/product-category/144
https://www.aliexpress.us/item/3256802316870663.html
https://www.aliexpress.us/item/3256805242722323.html (fake?)

[coromonadalix]

sound complicated for nothing, you have  ssr   with 5 to 30 vdc in for the isolated input control, and the main contacts between 20 30 40 60 amps  of your model choice

[langwadt]

sound complicated for nothing, you have  ssr   with 5 to 30 vdc in for the isolated input control, and the main contacts between 20 30 40 60 amps  of your model choice

that work with DC?

[David Hess]

what's wrong with a relay? it is easy and reliable

Current draw would be higher and relays suffer from contact welding, and there would not be any price advantage.  A mechanical switch would be cheaper but still suffers from contact welding and an SSR can be soft starting.

I also wanted something like an illuminated safety switch and they do not come with high current ratings, so  relay would have been needed anyway.  I am actually spending more for the switch than for the parts for the solid state relay.

Is an appropriately rated high current switch too large or too expensive?

I searched for a suitable SSR module however almost none of them give a voltage drop or on resistance specification, and many DC modules were bipolar transistors with zero crossing control.  It was difficult or even impossible to determine which were actual MOSFET outputs.  The ones which might work well were incredibly expensive.

[DarkMode]

Hi David, what kind of Amps are you wanting to draw from the power socket?
Typically a car electrical socket is rated for about 10 Amps and at 12V that's about 120 Watts of power.

Mosfets could fit your bill, and yes, they are expensive, but just as expensive as suitable switch.

I was looking at Dave's project, Soft Switch. I was also planning on powering a few safety lights.

[floobydust]

Cranking is the biggest problem for these switches.
Relays can chatter and cut out, and a mosfet expecting 10V drive gets much less which can kill it if it's supplying load.
Or the day you plug in an air compressor and expect 15A. Toyota has a hilarious 2A cig lighter ability screw that 20ga wire.

You need a charge-pump for sufficient gate-drive, or the PV optocoupler.
There are many caravan "battery isolator" circuits same application but higher current, or "ideal diode" similar circuit on Ali but hard to find on/off control and surely fake specs.

Silicon Chip Magazine "Battery Isolator" July 2019 uses two IRFS7434-7P N-ch 40V 240A 0.70mohm D2PAK 7 pin. LM339 for LVCO and charge pump for gate drive. I'd post sch but not sure if that causes trouble.
Smaller IRFS7434PbF D2PAK or TO-262 195A 1.8mohm at 6V drive, 50A.

But for both C_iss ~11nF is high and I think would switch slow with the PV drive at say 15-40uA. I would simulate it and see how long it takes to switch them or go with smaller mosfets.
Opto-couplers are not popular in the automotive industry due to cost and few (PV) have AEC-Q101.
Vishay VOM1271 SOP-4 or TLP3906 and TLX9905/6 has AEC-Q101, they have internal discharge circuit.

I use an ATTiny 85 for control so it basically follows IGN after a 5-sec on and off delay (I used to drive a stick and sometimes stall it, don't want the NAV rebooting)  0.12mA quiescent.
Or press a button: single run for 3 hrs. Ign key not required. Press again: off. For charging a phone or people in the car needing power/radio when I go into a store.
Long-term average voltage used for a LVCO to avoid a dead battery. I have to do a redesign, it's all point to point wiring for years now.

[jonpaul]

use a properly rated contactor not a SSR or relay.

See epay

Up to 100s of A

Jon

[David Hess]

Hi David, what kind of Amps are you wanting to draw from the power socket?
Typically a car electrical socket is rated for about 10 Amps and at 12V that's about 120 Watts of power.

The existing circuit is protected by a 10 amp fuse and powers 1 cigarette lighter and 2 accessory sockets.  I know I blew the fuse once long ago, but do not remember what the load was.  The largest load I have is my power station which can draw about 10 amps.  My charger which can connect through the circuit provides 10 amps.

Mosfets could fit your bill, and yes, they are expensive, but just as expensive as suitable switch.

I think I am getting a lot more capability for the price by making a solid state relay from parts.  I do not understand why the suitable SSRs that I found are so incredibly expensive.

Cranking is the biggest problem for these switches.
Relays can chatter and cut out, and a mosfet expecting 10V drive gets much less which can kill it if it's supplying load.
Or the day you plug in an air compressor and expect 15A. Toyota has a hilarious 2A cig lighter ability screw that 20ga wire.

You need a charge-pump for sufficient gate-drive, or the PV optocoupler.

That convinces me that I am on the right track.  A photovoltaic optocoupler or gate driver will be largely immune to low voltage conditions during starting.

But for both C_iss ~11nF is high and I think would switch slow with the PV drive at say 15-40uA. I would simulate it and see how long it takes to switch them or go with smaller mosfets.

I am not adverse to simulating it, however it only takes a couple minutes to calculate with pen and paper.

Opto-couplers are not popular in the automotive industry due to cost and few (PV) have AEC-Q101.

That explains why they do not use them, but I can use whatever I like, and it is not like I am placing it into the engine compartment.

use a properly rated contactor not a SSR or relay.

I just occurred to me that there are suitable relays; the fuse block in the engine compartment is full of them for controlling various loads, and automotive relays are relative inexpensive.

[jfiresto]

Have you looked at the Hella 931774031 automotive SSR? It has an active low ("negative pulse", ground to activate) input. Unfortunately, it has roughly doubled in price since I first used one.

[langwadt]

what's wrong with a relay? it is easy and reliable

Current draw would be higher and relays suffer from contact welding, and there would not be any price advantage.  A mechanical switch would be cheaper but still suffers from contact welding and an SSR can be soft starting.

If that was a big issue the car wouldn't work, there is a lot of relays used in car

Is an appropriately rated high current switch too large or too expensive?

I searched for a suitable SSR module however almost none of them give a voltage drop or on resistance specification, and many DC modules were bipolar transistors with zero crossing control.  It was difficult or even impossible to determine which were actual MOSFET outputs.  The ones which might work well were incredibly expensive.

then use something like this, https://www.digikey.com/en/products/detail/infineon-technologies/BTS6143DAUMA1/1281799

[DavidAlfa]

It's just 12-14V DC, you don't need anything fancy, a simple P-FET will work great.
Most cars turn auxiliary systems off while cranking, even the radio.
Voltage going down during cranking is les than a issue, you had to be idiot to be sucking 20amps while also cranking lol.

You could take the crank signal so it disables automatically, this is what I would do.

[David Hess]

Have you looked at the Hella 931774031 automotive SSR? It has an active low ("negative pulse", ground to activate) input. Unfortunately, it has roughly doubled in price since I first used one.

That probably would have worked, although specifications are lacking.

Current draw would be higher and relays suffer from contact welding, and there would not be any price advantage.  A mechanical switch would be cheaper but still suffers from contact welding and an SSR can be soft starting.

If that was a big issue the car wouldn't work, there is a lot of relays used in car

Car relays can rely on operating with a known load.

I searched for a suitable SSR module however almost none of them give a voltage drop or on resistance specification, and many DC modules were bipolar transistors with zero crossing control.  It was difficult or even impossible to determine which were actual MOSFET outputs.  The ones which might work well were incredibly expensive.

then use something like this, https://www.digikey.com/en/products/detail/infineon-technologies/BTS6143DAUMA1/1281799

I looked for integrated protected switches but none had a high enough current.  I also considered using a load protected switch driver from Linear Technology which includes a charge pump, however it would have been more expensive and more complicated than the photovoltaic optocoupler.  Ultimately I decided that derating the power transistor and relying on the fuse would work better with sensitive loads.

It's just 12-14V DC, you don't need anything fancy, a simple P-FET will work great.

I started with that design but p-channel parts are so much more expensive, and I had already decided to use a photovoltaic optocoupler or a charge pump switch driver so there was no reason not to use an n-channel part.

Most cars turn auxiliary systems off while cranking, even the radio.

I will have to check that.  I do not know if mine does.  I was originally thinking of taking the accessory signal from the wiring harness where it goes to the radio since it is close by.  I will have to see if I can find detailed documentation for the signals going into the radio.

Voltage going down during cranking is les than a issue, you had to be idiot to be sucking 20amps while also cranking lol.

I agree, but mistakes happen.

The accessory outputs are continuously powered now and it has not been a problem with any of my loads or the charger.

You could take the crank signal so it disables automatically, this is what I would do.

Do you mean the signal to the starter relay?  The wiring diagrams for my pickup are not as thorough as I would like.

[jfiresto]

Have you looked at the Hella 931774031 automotive SSR? It has an active low ("negative pulse", ground to activate) input. Unfortunately, it has roughly doubled in price since I first used one.

That probably would have worked, although specifications are lacking.

I could probably dig up a data sheet somewhere if that would dispense with the subjunctive.

IIRC, it has something like a 10~15mA input current, to wet a mechanical switch, and is protected against surges, to drive incandescent bulbs. It seems to have no problem PWM switching a pair of 12V 55W halogen bulbs having a perhaps 70A inrush current. I do not know its load dump rating, but I would assume it is protected against the usual catalog of automotive electrical horrors.

[Siwastaja]

Photovoltaic gate drivers are slow with large Qg, so SOA can become a problem as the MOSFET is spending too much time in linear region.

Robust, short-circuit protected semiconductor switching in presence of small impedances like batteries is surprisingly difficult, which is why integrated solutions exist. They also have the benefit of being able to directly monitor die temperature, not accessible in discrete design.

[DavidAlfa]

P-ch so much expensive? Search properly and you'll find it's not always true.

Going to china makers, you can get AOD403 (-30V, 70A) for peanuts.
If your car makes 30V you'll have way more serious problems than a burned mosfet.

Or you can get a IRF4905S (-55V, 74A) from a local distributor for ~60 cents.

[David Hess]

P-ch so much expensive? Search properly and you'll find it's not always true.

Going to china makers, you can get AOD403 (-30V, 70A) for peanuts.
If your car makes 30V you'll have way more serious problems than a burned mosfet.

Automotive designs should be capable of handling negative voltages, and up to 60 volt spikes.

Or you can get a IRF4905S (-55V, 74A) from a local distributor for ~60 cents.

I selected the IRFP7530PBF based mostly on performance per price.  The closest p-channel equivalent is slightly more than twice the price, so not unreasonable, but with the charge pump gate driver included in a protection IC or a photovoltaic gate driver, there is no reason to use a more expensive p-channel part.

Photovoltaic gate drivers are slow with large Qg, so SOA can become a problem as the MOSFET is spending too much time in linear region.

I agree.  I will have to keep an eye on that but I think the current and power derating of the power MOSFET that I selected will be able to handle it.  I had always intended to use 2 photovoltaic optocouplers in parallel to double the drive current; I ordered 4 optocouplers and 4 MOSFETs to make 2 SSRs.

[DavidAlfa]

Automotive designs should be capable of handling negative voltages, and up to 60 volt spikes.

Try connecting the battery backwards and tell me how many $$$ it was! 
Ever wondered what would happen to your gps, bluetooth and all that if you fed reverse power or 60V to them?
It's just a power socket and DIY application.

[David Hess]

Automotive designs should be capable of handling negative voltages, and up to 60 volt spikes.

Try connecting the battery backwards and tell me how many $$$ it was! 
Ever wondered what would happen to your gps, bluetooth and all that if you fed reverse power or 60V to them?
It's just a power socket and DIY application.

-15 to +60 volts is just a common standard to handle load dumps, reverse battery, and various other calamities.  Everything except my power bank and charger connect via 12 volt to USB converters which hopefully would handle anything abnormal without destroying their loads.  The charger and battery bank handle up to at least 24 volts and I know the charger handles reverse voltage.

I brought this pickup out of storage last year after it had sat in the garage since 2014.  Back in 2014 I had to cut the battery cables to get the battery out because the side posts had corroded.  Last year I spliced new cables in and installed a top post battery.  GMC of course makes all of their battery cables black, so I had to decipher which were positive.  Luckily I got it correct.

[floobydust]

Adding SC schematic. I don't think the TVS are useful, the body diode avalanches well enough, and TVS can short and burn up esp. 15V ones.

IRFP7530PBF mosfet C_iss 13.7nF, two are 27.4nF which is a lot for the PV opto. If I get time I will try a LTSpice sim and see how slow that is, as far as SOA limits.
First rev I used a single P-ch BMS3003 but backfeed can happen which I don't like.

[David Hess]

Here are photographs of the construction.  Continuous current is limited by power dissipation to about 50 amps which is plenty; this could be considerably improved with heat sinking, but I chose such large MOSFETs just to avoid heat sinking.

The 1N270s form a bridge rectifier so the drive polarity to the optocouplers does not matter.  I only used 1N270s because they were immediately available; I would have preferred 1N4448s.

I will measure the switching time later today.

Update:

Turn-off: 10 milliseconds
Turn-off Delay: 34 milliseconds
Turn-on: 1.0 milliseconds
Turn-on Delay: 1.3 milliseconds

[Scrts]

You should look into high-side switches. Something like this one: https://www.ti.com/product/TPS27SA08

Automotive companies use these for multiple applications. We're using them for camera power control on driver assistance systems for the last 15 years...

[Doctorandus_P]

There are Ic's that combine some LED's with an integrated solar cell, and these are specifically designed to make a solid state relay in combination with a MOSfet. I'm not sure of the exact name, but now you know they exist, you should be able to find them easily yourself.

[David Hess]

You should look into high-side switches. Something like this one: https://www.ti.com/product/TPS27SA08

Automotive companies use these for multiple applications. We're using them for camera power control on driver assistance systems for the last 15 years...

I think I need to be able to switch more than 10 amps so all of the integrated high side switches that I checked were marginal at best.

I also looked at using a fault protected high side switch driver with a big external MOSFET.  This definitely would have worked but was more complex and there was some question about the fault protection causing more problems than it solves.

I decided to rely on the existing fuse for fault protection as it has done so for years without problems, but I did need to make sure that the solid state relay will not limit the current preventing the fuse from blowing.

There are Ic's that combine some LED's with an integrated solar cell, and these are specifically designed to make a solid state relay in combination with a MOSfet. I'm not sure of the exact name, but now you know they exist, you should be able to find them easily yourself.

They are commonly known as photovoltaic optocouplers and that is what I ended up using.  I used a pair wired in series on the input and in parallel on the output to double the output current, but based on the switching times that was not necessary.