Circuit for detecting open/closed relay/contactor in 400-800Vdc applications

[AlfBaz]

Maybe there is still some misunderstanding about the specific fault condition to be detected and avoided:
it results from being previously switched on (both rails:complete circuit) and occurs on switching OFF.
The aim is to detect this condition in order to avoid any subsequent switching ON, because that would be unsafe.

So, when the sought condition appears, no current will flow and no fuse will blow.

Not sure if you are referring to the cct I gave. If you are and we assume there is some logic involved energizing the contactor then...
If coil is turned on and output from both resistive networks is present then all is good.
If coil is turned off and no output from restive networks is present then all is good.
If coil is turned off and output from only one restive network is present then flag fault and don't turn on contactor

[DrGeoff]

Here's an idea I had. Isolation can be as simple as an opto connected to the outputs.
You can use multiple resistors in series to get the appropriate voltage ratings and the whole thing can be potted in fire retardant high voltage epoxy.

You end up with a 6 lead device. 2wires connected to the load side, 2 to the feed side the other 4 to your detection cct

Under a fault condition there may be no voltage present.

Under a fault condition one would think/hope the appropriate fault clearing mechanism would do ITS job (aka fuses)

Which is the point.
The contacts may be welded/stuck after a fault condition. There may be no voltage/current present on the line and closed contacts need to be detected.

[AlfBaz]

Here's an idea I had. Isolation can be as simple as an opto connected to the outputs.
You can use multiple resistors in series to get the appropriate voltage ratings and the whole thing can be potted in fire retardant high voltage epoxy.

You end up with a 6 lead device. 2wires connected to the load side, 2 to the feed side the other 4 to your detection cct

Under a fault condition there may be no voltage present.

Under a fault condition one would think/hope the appropriate fault clearing mechanism would do ITS job (aka fuses)

Which is the point.
The contacts may be welded/stuck after a fault condition. There may be no voltage/current present on the line and closed contacts need to be detected.

If that's a concern then place another resistive detection string accross the feed side. If no voltage is present then fault.
At some point somebody has to open the the device to see why it's faulted

[oldway]

I'm not sure why you can't use a contactor with the circuit built-in for this but maybe you could mount a proximity detector to the outside.

If you could fire the solenoid, you could try and detect the back EMF when the contact bounces.  Some fuel systems work like this to detect the injector response time. Tthere may be reasons for not doing that (like the other contactor has failed). 

If we assume when the contacts are welded, it changes the location of the moving contact.  I wonder what the inductance of the solenoid is with the contactor closed vs open.  I just tried it with a contactor I had laying around and indeed, pick the right frequency, the inductance changed more than 50%.   You would just switch out your normal driver when making the measurement.  Maybe a train mode to work with different contactors.

This is not correct.
Pressure on the contacts is applied by a spring.
If the contacts are welded, the magnetic circuit is NOT fully closed.....there is an air gap of some millimeters.
There is is a little change of inductance of the coil, but far less than 50%
So this way to detect welded contacts is not safe at all.

[joeqsmith]

I'm not sure why you can't use a contactor with the circuit built-in for this but maybe you could mount a proximity detector to the outside.

If you could fire the solenoid, you could try and detect the back EMF when the contact bounces.  Some fuel systems work like this to detect the injector response time. Tthere may be reasons for not doing that (like the other contactor has failed). 

If we assume when the contacts are welded, it changes the location of the moving contact.  I wonder what the inductance of the solenoid is with the contactor closed vs open.  I just tried it with a contactor I had laying around and indeed, pick the right frequency, the inductance changed more than 50%.   You would just switch out your normal driver when making the measurement.  Maybe a train mode to work with different contactors.

This is not correct.
Pressure on the contacts is applied by a spring.
If the contacts are welded, the magnetic circuit is NOT fully closed.....there is an air gap of some millimeters.
There is is a little change of inductance of the coil, but far less than 50%
So this way to detect welded contacts is not safe at all.

I guess I am not sure what part is not correct.  All of it?

I did mention I had tried it and I really have no reason to provide false information, so maybe I missed something.  I had already considered that the return spring may cause some gap.   Obviously I don't have their contactor.  The one I had laying around to try uses a 24V 60Hz/ 22V 50Hz coil.   

Fully open I measured 11.59mH
Fully closed I measured 66.25mH
Relaxing the bar 7mm, I measured 20mH.

From a safety side, I am not familiar with the regulations but I assume based on the OP's posts that they are.  They were just looking for ideas.  Regardless,  I assume that your last sentence suggests that the safety was not the part you suggest was incorrect, it was measuring the inductance.   It's possible their contactor will not see this sort of change but it's easy enough for the OP to check.

[oldway]

You spoke about measuring "the inductance of the solenoid with the contactor closed vs open" and it is not really closed when the contacts are welded.

But if you already "relaxed the bar 7mm" when you measured the inductance, that's alright.

I made a very simple circuit with resistors and reed relays measuring voltage on the contacts.

No information has been given on the load, but it seems that both polarities of the load are short circuited.

Then the very simple circuit I proposed will work fine.

Measuring the inductance is not a simple circuit.
It must be done at low level and when the coil is not energized....It can't be safe because it is not simple.

[David Hess]

Measuring the inductance is not a simple circuit.
It must be done at low level and when the coil is not energized....It can't be safe because it is not simple.

It is no more complex than measuring the contact resistance through a transformer.  And if the inductance measurement is made using a self oscillating circuit, lack of oscillation can be used to indicate a failure of the sensing circuit.

Some fail-safe drive circuits use an AC signal from the controller so that a stuck high or stuck low output results in an off condition.  High reliability sensor circuits may use AC excitation for the same reason so they do not have to rely on stuck on low or high detection which is easier to fool.  For instance LVDTs which use AC excitation have an advantage over hall effect sensors in high reliability applications.  AC excitation also allows synchronous demodulation for increased noise immunity.

[joeqsmith]

....It can't be safe because it is not simple.

    

[oldway]

Measuring the inductance is not a simple circuit.
It must be done at low level and when the coil is not energized....It can't be safe because it is not simple.

It is no more complex than measuring the contact resistance through a transformer.  And if the inductance measurement is made using a self oscillating circuit, lack of oscillation can be used to indicate a failure of the sensing circuit.

Some fail-safe drive circuits use an AC signal from the controller so that a stuck high or stuck low output results in an off condition.  High reliability sensor circuits may use AC excitation for the same reason so they do not have to rely on stuck on low or high detection which is easier to fool.  For instance LVDTs which use AC excitation have an advantage over hall effect sensors in high reliability applications.  AC excitation also allows synchronous demodulation for increased noise immunity.

Not that easy at all....I wonder how you will make a self oscillating circuit with an inductor which has a free wheeling diode to suppress dV=dI/dt when current is interrupted in the coil.

And inductance varies, then you have to measure the frequency of oscillation....
A lower frequency than expected should indicate welded contacts.

How much components to make such a circuit ?

In principle, safety should not depend of a µC, nor of a PLC.

[oldway]

The explanations given by elima are very nebulous.

As far as I guess what it does not clearly explain, the battery would consist of low-capacity elements connected in series / parallel with multiple fuses protecting each group of battery cells.

The battery would be inspired by the batteries manufactured by Tesla.

The choice of a voltage of 800V is abnormally high ... It is probably imposed by the operation of the circuit that it supplies.

The battery is not grounded, at least elima does not say it.

According to the (very) simplified diagram, only the negative pole of the battery is protected by a fuse.
It seems very strange to me, because for a not grounded battery , it would be necessary to protect the 2 poles of the battery by a fuse.
The battery could then be safely isolated by removing the two fuses.

This is certainly not a common application, such as electric vehicles.
For this type of application, a much lower voltage is chosen.

The load has a capacitor of high capacity since there is precharge and descharging circuits.

The voltage is too high to be used in a defibrillator ... I would rather think of a generator of high voltage and high nergy pulses but for what purpose? I do not know.

For the problem of welded contacts, it is logical because it seems that elima would like to use a compact contactor type that does not have auxiliary contacts.

On the other hand, the welding of the contacts is a possible event in case of non-operation of the precharging circuit.

But before looking for a solution for detecting such an event, we need detailed informations about the project, inclusive informations about the load, about the contactors, and so on....

For example, it is not certain that the operation of the special contactor causes a change in inductance of the solenoid.
One of the contactors cited by elima looks like a power reed relay, in this type of contactor, the welding of the contacts does not lead to variation of inductance of the coil.

[floobydust]

It's very difficult to beat the simplicity and low cost of a mechanical (auxilliary) switch.
I would probably look to save dollars elsewhere. Tesla Motors developed their own electronic fuse, for example.
Why does the nuclear submarine project have such a low budget lol.

Back to OT:
You could measure contactor inductance. This is done with automotive fuel injectors and peak/hold drive, and automotive ignition system's misfire detection looking at primary voltage.

OP did not say if PWM is used on the contactor's coils, as they are piggies for power probably 1-2A coil current.

I would add A/D channel sampling coil voltage and look at the slope during PWM to infer inductance.
Depending on the magnetic circuit, most contactors have 2-3:1 change at least. But the solenoid-types do not have much inductance change. You would have to measure things. There is probably an arc blowout magnet too. Should be no back-EMF diode across the contactor's coil; this would slow the decay time unless a zener used too.


It might be possible to inject a HF carrier (say 100kHz with current-transformer excitation) on the battery's leads and look for the carrier after the contactor.


The contactor will get stuck during release, so I'm not sure why all the start-up procedure and testing. What about shutdown? You'd already know when de-energizing the contactor that it failed.

I wonder about the fault-chain where a DC bus event/fault happens, the contactor is opened and interrupting high-current. Not this nicey nice pre-charge current switching.
That's when the contacts will actually weld.

[2N3055]

As usual, simpler is better.. You don't need to monitor contacts, measure impedance... You need to make sure there are no dangerous voltages/currents behind contacts after relays are off...

So all you need are two voltage measuring circuits.. I drew simplified, no fuses or other protection, just relay monitoring concept.
You can make it low or high impedance whatever you need. It will monitor if there is any voltage after relays...
If all is allright neither meter will show voltage, if either or both do, danger....

By the way, safe way to do switching is to have two pairs of contactors in series. And you switch them with a slight delay, so one of them is newer carrying current. So if working one gets welded, chance is that the other one will disconnect. And you put monitoring on both so you can detect life threatening malfunctions. In which case you have to permanently disable device until repaired...

Also you want to have manual master switch that also disconnects both sides. That is for maintenance work, if battery cannot be safely disconnected.

My 2c.

[Neomys Sapiens]

Not to the circuit, but to the statement about the fuse.

[oldway]

As usual, simpler is better.. You don't need to monitor contacts, measure impedance... You need to make sure there are no dangerous voltages/currents behind contacts after relays are off...

So all you need are two voltage measuring circuits.. I drew simplified, no fuses or other protection, just relay monitoring concept.
You can make it low or high impedance whatever you need. It will monitor if there is any voltage after relays...
If all is allright neither meter will show voltage, if either or both do, danger....

By the way, safe way to do switching is to have two pairs of contactors in series. And you switch them with a slight delay, so one of them is newer carrying current. So if working one gets welded, chance is that the other one will disconnect. And you put monitoring on both so you can detect life threatening malfunctions. In which case you have to permanently disable device until repaired...

Also you want to have manual master switch that also disconnects both sides. That is for maintenance work, if battery cannot be safely disconnected.

My 2c.

When both contacts are open, the load is short circuited or has a low impedance (descharged capacitor).
Then, with 800V battery voltage, both voltage sensors P1 and P2 became a voltage divisor and will measure 400V each, what should be considered as a faulty condition, when it is not.

[2N3055]

When both contacts are open, the load is short circuited or has a low impedance (descharged capacitor).
Then, with 800V battery voltage, both voltage sensors P1 and P2 became a voltage divisor and will measure 400V each, what should be considered as a faulty condition, when it is not.

Two diodes take care of that...

[oldway]

I do not agree.

[2N3055]

And you are right..
But nevertheless, in that case voltage would be half per meter so state would be recognizable..

[oldway]

The first schematic I posted was the same as your's, but I changed my mind and I edited my post because it is not a safe design.
I prefered to measure voltage across the contacts, because voltage across welded contacts is 0V, and that's 100% sure.
Fault detection is easy: there are welded contacts if both reed relays are not in the same position. (both on or both off = no fault, one on and the other off = fault)

It is also a safe design because it will detect welded contacts if one of the coils of the reed relays is defective. (interrupted)
For safety, it must detect a fault if there is something wrong in the detection circuit.

NB: values of the resistors must be calculate again when the 24V reed relays model is chosen and their technical caracteristics are known....

[elima]

Sorry for the late response. There are plenty of replies already and most of those totally ignored something that I mentioned.
The test current should be provided from the LV system.
I can't elaborate more on the systems I am working on (hopefully you can guess why..).
It was my mistake to mention anything about high voltage dc systems because it seems that it made people over-complicate things.
The same question could be asked if you had a basic mechanical switch, connected at one side to 24V supply and the other side continues to some circuit. You don't know anything else about the rest of the circuitry/dont have access to it but would still like to detect the state of the switch. Simple continuity test. How would you do that?

I am already working on a solution and it is actually simple, very simple. Will update when I get it tested.

 

[Loboscope]

I would imagine, that the switch of a contactor which is able to trip 500 Ampere is not a tiny little part. So I would have a look, if it is possible to install a simple light barrier sensor at the contactor, so that the switch of the contactor will interrupt the light barrier when in open position.

The circuit needed to trigger a simple LED flashing up when the contactor is closed (light barrier is open) is in fact ultra simple and therefore very reliable - and durable too.
Using this construction, the sensor circuit is feed only from the LV-side and the isolation to the HV-side will be 100%.
Also this LED-indicator circuit will be safe to be interpreted:
LED on = contactor closed
LED off = contactor opend

And it will be also safe for failure interpretation: If the LED will not flash up when the switch for the corresponding contractor will be closed to "on", there must be a failure and you will call for service.

[oldway]

Sorry for the late response. There are plenty of replies already and most of those totally ignored something that I mentioned.
The test current should be provided from the LV system.
I can't elaborate more on the systems I am working on (hopefully you can guess why..).
It was my mistake to mention anything about high voltage dc systems because it seems that it made people over-complicate things.
The same question could be asked if you had a basic mechanical switch, connected at one side to 24V supply and the other side continues to some circuit. You don't know anything else about the rest of the circuitry/dont have access to it but would still like to detect the state of the switch. Simple continuity test. How would you do that?

I am already working on a solution and it is actually simple, very simple. Will update when I get it tested.

If you have 2 separate LV power supplies both isolated at 5KV,  there is a very simple solution: see schematic.
I supposed that the load has no EMF higher than the battery voltage !

As long you refuse to give us informations about the load, we can only guess ....

[capt bullshot]

The whole things looks like kind of automotive / traction stuff, and several safety standards (e.g. 26262) apply.
I'm familiar with industrial functional safety, and IMO the simplest solution would be to go for safe monitor contacts within the contactors (that's the standard method in my field of experience). Including no lengthy discussions with the certification authority.

One other way I can imagine would be a high impedance voltage monitoring at four points in your circuit: the battery side terminals and the load side terminals of the contactors. Use any suitable reference point, doesn't matter as long as there's no large AC common mode signal present.. Depending on the load (some kind of inverter, I guess), you may need a high impedance discharge / bleeder path to ideally "middle of battery", other potentials may also work.
Now measure all four voltages, and from the differences between them you can derive the states of the contactors. Yes, some MCU power may be required, but often enough some kind of MCU is already there ...

[capt bullshot]

Maybe another way:

use two small pulse transformers decoupled by diodes to inject a test current across the contacts.
Apply pulses to the LV side, reflected current gives information if the contact is open or closed.
Risk: you put some components across the contacts, that may rise safety concerns (diodes fail short, ...)
Using high impedance voltage sensing, these risks can be easier mitigated (just use enough suitable resistors connected in series)

[elima]

i am working on a circuit with a similar idea presented by oldway. First board tested and works nicely, now have to make it fail-proof and add all the needed protection.

Thanks.