Two relays in series with flyback diode.

[Harald Kapp]

You measure the voltage drop across drain-source of the mosfet. From I = V / R with V = measured value and R = Rdson you get Ids = Vmeasured/Rdson. The difficulty here is that Rdson can vary considerably, so you will have to trim the circuit.
A much more precise solution puts a small sense resistor (e.g. 0.1 Ω) in the connection between source and ground. Then the same rule applies ass above, but this time with a known and much more stabel sense resistor.

In both cases of course the measured voltaeg will be very small (you don't want to drop much voltage becaus you want the voltage at the load (relays), so you'll have to amplify the measured voltage, then you can feed it to e.g. an analog input pin of the microcontroller and use the ADC to get a measurement.
The measured voltage will probably be quite noisy, so some low pass filtering may be reuqired, e.g. in the analog front end (amplifier) of in the digital domain (avaraging after A/D conversion).

 

[(*steve*)]

A single shorted turn on a dc relay isn't going to do a lot to the current passing through it or to it's operation.

 

[Harald Kapp]

sorted

shorted? Right.

But relay coils can have shorts across multiple windings...

 

[(*steve*)]

A shorted turn will probably change the turn on/turn off times. That may be important in this application.

Also, what happens if a set of contacts on one relay Wells themselves together?

 

[Tha fios agaibh]

what happens if a set of contacts on one relay Wells themselves together?

Probably a more likely scenario, of which a redundant relay to kill power is an option.

I don't know if it would suit this application, but another approach is to use logic to monitor the relays. For instance If output turns on or off, you monitor an input off the relay contacts wired in series. In other words, if Output goes low and input doesn't shut off in x time, controller sends a fault.

If critical, a watchdog timer circuit to very that the microcontroller is operational might even be necessary.

 

[Harald Kapp]

Thats right davenn. I need to switch 4 sets of poles (4PDT) but it is critical for my circuit that theese contacts are at the same state even under fault conditions.

We're doing a lot of gueswork here.
@HellasTechn : could you state in more detail what your critical application is?
Also: what is the fail safe state of the relay contacts? You state the have to be in the same state, so in case of a welded contact on one relay it will be required that the other relay contacts all stay closed, too? That really calls for a safety relay.

 

[HellasTechn]

@HellasTechn : could you state in more detail what your critical application is?

Well critical for me it is
I am building a system where 2 computers (PC1 and PC2) will talk over a single rs422 bus through the relays to an other computer (PC3). The mcu taps on the rs422 output of PC1 and PC2 and based on a code i wrote determines which PC (PC1 or PC2) should be connected to PC3. When i power down the mcu the relays go to normally open state and in turn route the signals to a switch for when i want to manually control the routing.

What i call fail safe state for the relays is to have them either both engaged in normally open position or both disengaged in normally closed position. If one is in NO position and the other is in NC position i risk a short circuit between PC1 and PC2 outputs via my hardware switch.

A shorted turn will probably change the turn on/turn off times. That may be important in this application.

No that is not really important. A few microseconds can not harm the PC's output.

Also, what happens if a set of contacts on one relay Wells themselves together?

That can not happen because the relays only handle signals. To damage relay contacts you would need a few amps passig through.

don't know if it would suit this application, but another approach is to use logic to monitor the relays. For instance If output turns on or off, you monitor an input off the relay contacts wired in series. In other words, if Output goes low and input doesn't shut off in x time, controller sends a fault.

Im not sure i understand this approach.

shorted? Right.
But relay coils can have shorts across multiple windings...

Well as long as it can engage the relay its ok. If it coms to the point that the relay can not engage then it will probably draw enough current to either blow the fuse or the relay coil.

 

[Harald Kapp]

That can not happen because the relays only handle signals.

An important information we should have had earlier.
Do you use a full-duplex (4-wire) or half-duplex (2 wire) configuration?

Why the need for 2 relays?
For a 2 wire configuration a 2 pole changeover relay (double pole double throw) is sufficient to select between 2 connections.
For a 4 wire configuration a 4 pole changeover relay (quad pole double throw?) is sufficient to select between 2 connections.

With this one relay you can select between PC1 and PC2.
You'll need a second relay to select between automatic an d manual control. But this second relay needs a drive circuit separate from the one above because the operations of these relays are independent.

No need for expensive fail safe mechanisms.

 

[HellasTechn]

An important information we should have had earlier.

Not quite like that.

o you use a full-duplex (4-wire) or half-duplex (2 wire) configuration?

I use half duplex configuration.

No need for expensive fail safe mechanisms.

Yes, though in my case because the data bus is RS422. I need 2 wires per PC.
Your diagram is correct only i need double the contacts you show. I made a drawing of my diagram below.

As for reliabilit i think i am good enough because in case something hapens to the mosfet, the relays will both be affected and in case something happens to a relay they will both go down.

 

[Harald Kapp]

Yes, though in my case because the data bus is RS422. I need 2 wires per PC.

Yes. I simplified my diagram. Each connection means of course 2 wires and 2 switches.

I'm sorry, your diagram doesn't make sense to me.
What are the connections to PC1 auto and PC2 auto (+ and - each)? as far as I see they lead nowhere.
Have a look at my diagram:
The wires from PC1 and PC2 go to

1. a relay which is controlled by the mcu to select either PC1 or PC2 as the partner for PC3.
2. a manual switch which makes the same selection manually in case the automatic selection is inactive.

This relay and switch perform this part of your description:

The mcu taps on the rs422 output of PC1 and PC2 and based on a code i wrote determines which PC (PC1 or PC2) should be connected to PC3.

Then the output from the relay and the switch go to a second relay which selects which of the two to use:

1. up position will select the automatic path via relay 1
2. down position will select the manual path via the switch.

This second relay performs this part of your description:

When i power down the mcu the relays go to normally open state and in turn route the signals to a switch for when i want to manually control the routing.

 

[HellasTechn]

I'm sorry, your diagram doesn't make sense to me.

I am sorry for confusing you. I dont have the rest of the diagram now.

Yes. I simplified my diagram. Each connection means of course 2 wires and 2 switches.

I got a little confused by your diagram. But now it makes sense. When i get the rest of the diagram ill breadboad it and see

 

[HellasTechn]

Come to think about it the missing part is that the PCx + and - tags from the 2 relays are then fed to 3rd relay. The 3rd relay acts just like the manal switch feeding PC1 and PC2 one at a time to PC3 so the output of the manual switch and the output of the relay are in parallel and that might cause the issue if relay 1 and relay 2 goes open or shorted.
Again as mentioned in an earlyer post, keeping relay 1 and relay 2 in serries is a safe choice.