Relay vers Mosfet

[TimB]

In a board I'm working on I have to control 2 valves they draw around 400ma at 12v each so using a 5amp mosfet works fine.

However I'm looking at changing the control valves and they have bigger coils and take a lot more power. More like 800ma.

Since the 12v supply line to the is on the board I'm thinking best not to overload the track. So was thinking use a relay and just wire the power outside and switch a line via the relay. Using this method I think will be advantages and much more flexible but wanted to get any input I can on pros/cons etc

BTW the duty cycle is 100% for one valve and 1% for the other. No fine timing is needed. Just keep one valve open 100% of the time and the other a few seconds every 5 mins.

[RGV250]

Hi Tim,
I cannot comment much on relays as we always specified solid state switching for solenoids, one thing to look out for here is that some solid sate relays only switch AC and do not turn off if the load is DC. Not saying how I know
We always used 24VDC for the control. The only thing I would say if you do go for mechanical relays is check if the solenoids have built in flyback diodes and if not consider adding them.

Bob

[TimB]

Thanks Bob

The use of 24v is important for me to consider as this is all on a wall mounted electrical box. There is water in pipes inside  so my aim was to have a separate power supply and as I need 5v for the pic kept it to 12v to save on regs eg more if I used 24v

However a small relay will handle what every they throw at it. BTW I always add a flyback diode on the board

One question. Transistor or Mosfet to drive the relay?

[trastikata]

If it is a wall box and there's a possibility for lightnings to hit the building or around it, BJT's might be a more reliable option than MOSFET's.

[david]

If it is a wall box and there's a possibility for lightnings to hit the building or around it, BJT's might be a more reliable option than MOSFET's.

Better still a Thyratron!   Actually I would skip the relays and directly switch the valves with dual Nch FETs in an SO-8 package.

David

[See_Mos]

I have a lot of applications that use 1 amp 60 volt mosfet solid state relays in DIP 6 package but I had a shock just now when I checked R/S and Farnell, the prices are many times higher than they used to be. Depending on how many you need I may be able to let you have a few.

If you need isolation then relays are a cheap solution, SSR's are a neat solution, Otherwise I would stick with direct MOSFET drive.

 Is widening the power tracks an option? With the usual copper thickness a 1mm wide trace can handle about 2.5 amps and as the second solenoid is only on for short periods that would be more than enough.

[John Lawton]

There will likely be some current surges so it is important to route the power tracks so that current to the loads flows only to and from the power supply (connections) and not via the low level PIC circuitry. This can especially be an issue for the 0V line.

[See_Mos]

There will likely be some current surges so it is important to route the power tracks so that current to the loads flows only to and from the power supply (connections) and not via the low level PIC circuitry. This can especially be an issue for the 0V line.

correct!
This is a topic we discussed a long time ago on the old forum

[ken_k]

Hi Tim
I had a look and DC solid state relays are not as cheap as they used to be; one can make one up using a optocoupler with a photovoltaic (DC) output,these are not cheap.
If you are using a FET and back EMF diode across the solenoid coil the you could use a cheap SMD optocoupler with suitable voltage collector emitter rating rating, a suitable SMD optocoupler should cost about 20 cents.
I'm assuming the FET gate drive is OK at 12V.

Note: change *.txt  to *.asc if running LTspice file.
opto relay.txt   

[dr-zin]

If I understand your needs correctly, I would go with plain vanilla mechanical relays.  They are designed with coils that operate at 5, 9, 12, or 24VDC routinely (other values in between on occasion), or 120 or 240VAC.  They can easily pass or block many amps of current with almost no losses.  They come in multi-pole versions that SSRs can only dream about, although that is not an issue here.  I would control them using a low-side NPN or high-side PNP bipolar transistor, with their bases directly connectable to a PIC output pin (Okay, maybe use a small current limiting resistor [33-470 ohm] in between depending on voltage switched and transistor gain).

Note, your circuit that is ON 100% of the time may not even have to be powered at all; merely route your power through the NC (normally closed) and Common contacts of the relay and just take advantage of the relays fat clapper contacts to handle multiple amps of current (do get an appropriately sized one however) without ever energizing the coil.  You can't do that with an SSR!

While not glamorous, a mech. relay is very efficient... Consider this:  an SSR routinely has a metal plate on its underside so it can be bolted to a metal chassis to absorb the heat it produces, but a mechanical relay NEVER needs heat-sinking.  The only reason I would ever use a more expensive SSR was if the switching speed was greater than 3-5 hertz.

[John Drew]

While agreeing with dr-zin in part, I do find relays less reliable.
They need at least a little current through resting or working contacts to keep them 'wet'.
I have often had to clean contacts in certain situations, especially RF applications.
I've also had a VERY small number of hardware failures and open circuit coils.
Reliability wise I'd suggest solid state, for cost I'd prefer relays.
Cheers
John

[tumbleweed]

There's also the issue of the relay contacts arcing and welding that you don't have with an SSR.

 

[david]

I would only expect relay contact welding if I had an incandescent or capacitive load where inrush currents may be many times greater than the steady state.  Tim has indicated a figure of 800mA which will be a DC or steady state value and is defined by the resistive component of the valve's coil.  Any inductive component will surely only slow up the rate of rise in the turn on current.
SSRs for AC applications are usually based on (4 layer) junction devices and hence will have a forward drop in either direction of about 1V regardless of current which means about 1W per amp of conduction which would require a bit of heatsinking for some loads.  SSRs for DC applications will be FET based (channel resistance - not junction drops) and amount to what Ken has already suggested (but with a killer FET). 
The requirements are quite modest and many SMT FETs are rated 20A or more thanks to very low channel resistances.
To me, the key decision is if you want galvanic isolation or not and I don't really see a need for it provide the layout precautions mentioned earlier are observed. i.e. valve load current does not flow via logic grounds.

Cheers,
David

[okmn]

In a board I'm working on I have to control 2 valves they draw around 400ma at 12v each so using a 5amp mosfet works fine.

However I'm looking at changing the control valves and they have bigger coils and take a lot more power. More like 800ma.

Since the 12v supply line to the is on the board I'm thinking best not to overload the track. So was thinking use a relay and just wire the power outside and switch a line via the relay. Using this method I think will be advantages and much more flexible but wanted to get any input I can on pros/cons etc

BTW the duty cycle is 100% for one valve and 1% for the other. No fine timing is needed. Just keep one valve open 100% of the time and the other a few seconds every 5 mins.

what about  infineon smart switch or called other name  power distribute switch  for automotive
BTS series