Rather basic, I'm afraid, but when would you use a relay, and when would you use a transistor? In a relay the contacts wear out, so why are relays used at all?
\$\begingroup\$ To your comment about the contacts wear out. Developments have been made in solid state relays. \$\endgroup\$– DeanFeb 13, 2011 at 13:34
\$\begingroup\$ @Dean - but SSRs have their own problems: they don't offer isolation when off, and have a considerable voltage drop when on. \$\endgroup\$– stevenvhJun 11, 2011 at 6:33
5\$\begingroup\$ I know that this will not find favor with the organic non-GMO gluten-free anti-oxidant probiotic vegans, but there are relays whose contacts do not wear out because they are metallic mercury. \$\endgroup\$– richard1941Nov 17, 2017 at 20:32
Relays are on-off devices. Transistors can have their voltage drop varied.
Relays are far slower than transistors; typically 50ms to switch, and probably more. Some types of transistors can switch in picoseconds (almost 10 orders of magnitude faster.)
Relays are isolated. Transistors can be (e.g. SSR), but are often not.
Relays are electromagnetic and bring problems with them - for example, try building a relay computer with many relays. You will find that relays will interfere with each other in some cases. Transistors are not very EM sensitive. They do not emit much electromagnetic interference.
Relays consume a lot of current in the "on" state, most transistors do not.
\$\begingroup\$ You can't transmit or receive electromagnetic radiation with transistors? \$\endgroup\$– Nick TFeb 13, 2011 at 18:43
\$\begingroup\$ @Nick T, they are not very sensitive. I should clarify. \$\endgroup\$– Thomas OFeb 13, 2011 at 19:56
3\$\begingroup\$ an SSR doesn't count as a transistor. A transistor -- as the three-legged component -- isn't isolated. \$\endgroup\$ Jun 28, 2011 at 7:41
1\$\begingroup\$ Pretty sure a transistor(mosfet in this case, since we are comparing to a relay) can indeed be affected by EMI, but usually not enough on their own to actually cause gremlins to emerge. Any EMI causing fluctuations on the gate would be detectable as a current change on the drain, but it is not as much of a problem as on relays. \$\endgroup\$ Feb 27, 2017 at 19:59
\$\begingroup\$ The switching time of a relay varies greatly by manufacturer and application. 50+ ms would apply to high-current contactor relays (e.g. for switching motor loads). For control power relays, the slowest ones I've used will switch in about 20ms, and some are as fast as 7ms to close and 3ms to open. Still, your point stands about transistors being orders of magnitude faster. \$\endgroup\$– Dan A.Jan 21, 2020 at 20:00
Relays offer complete isolation between the activating circuit and the load.
They can switch AC and DC, and be activated by AC or DC.
They can be very robust.
They also have the advantage that one can often see if the device is actuated, and one can even hear the actuation in many cases.
1\$\begingroup\$ thank you! i can imagine how hearing it would be useful :) \$\endgroup\$ Feb 13, 2011 at 17:53
7\$\begingroup\$ Why the downvote? \$\endgroup\$ May 3, 2011 at 17:49
1\$\begingroup\$ +1 I also think hearing it could be useful. In case of failure, you know that the relay part is working ok. \$\endgroup\$ Oct 9, 2016 at 10:38
Aside from all the correct properties that Leon mentions, relays also have a much lower internal resistance, in fact the switch of a relay looks quite a lot like a straight piece of wire.
Any other kind of solid state switch (bjt, scr, triac, igbt) will have some resistance and drop some voltage.
In many designs where switching is infrequent and the designer of the circuit doesn't know exactly what the user is going to want to switch, a relay is a nice choice as it will switch either ac or dc at a huge voltage and current range.
In a particular application you can almost always find a solid state component that will do the job cheaper than a relay, if you can do without all the robustness and versatility of a relay.
\$\begingroup\$ thank you! hadn't even considered solid state components till now... \$\endgroup\$ Feb 13, 2011 at 17:59
4\$\begingroup\$ The Rds(on) of power N-ch MOSFETs can easily be better than the contact resistance in a relay of comparable size. \$\endgroup\$– Nick TFeb 13, 2011 at 18:41
1\$\begingroup\$ Relays are often used to bypass transient suppression components after turn-on. See the circuit diagram here: ti.com/tool/pmp10215. \$\endgroup\$ Aug 30, 2017 at 18:48
Relays are an ok choice for when the load that needs to be controlled draws more than a couple of amps, and when the switching will not be that frequent.
When you need to break (shut off) a current of several amps, load inductance can cause voltage spikes that will damage a transistor, unless you add a flyback/clamping diode to protect it. Relay contacts, being basically large pieces of metal, have much greater tolerance to this application, but even so, breaking large load currents will eventually burn out relay contacts.
If you need to switch something faster than once every second, a relay would probably have a relatively short life, and it would be worthwhile to go with the transistor option. If you don't need to switch your load faster than once every 10 seconds, you might find the relay more economical. As always, it's a design trade off.
\$\begingroup\$ thank you! i've always preferred transistors, they seem.. neater :) \$\endgroup\$ Feb 13, 2011 at 17:54
Relays can be set up to activate so many different voltage levels without adjustment. This is why they are so often seen in industrial controls. For instance say I design a controller for a valve that requires switched +15V to work. Then the company switches the valve to a current controlled style where the output has nothing to do with a fixed voltage level. This is a simple change for a relay(really no change) and probably a complex one for a MosFET.