AC Current variable trip circuit

[dan3460]

I'm looking for a circuit for my bench that will monitor AC current and it will disconnect when the current exceeds certain number. I will like to use this with my variac when starting antique radios and oscilloscopes for the first time. I have been using a dim bulb and keeping an eye on the ammeter.
So this circuit should be able to handle ac volts from 0-120V and current from 0-10A and it should have a button to reset the circuit when tripped.
I have seen a few that work on DC current, but I have not been able to find one that will work on AC.

Thanks for your help.
Dan

[dom0]

Outline: Two ~600-800 V MOSFETs source to source in the usual AC switch configuration. Gate drive e.g. with an ATX-SMPS base current transformer and a pull-down (see below!). Shunt & precision rectifier & average & comparator w/ hysteresis -> gate drive oscillator enable.

Problem: Making the circuit so fast, that it can disconnect w/o damage an instant dead short at 320 Vpeak is... challenging. Unipolar gate drive might not be enough ; you'll likely need <1 µs or even faster turn-off time overall.

I never built this ; having a variac with a quick responding analog meter is good enough for me.

[TerraHertz]

Funny, I've been thinking of the same thing since recently repairing something that had a snowball failure mode. Even using a variac at low output voltage, at some point the gear would overheat two diodes in its power supply causing one or both to short out, _then_ the mains fuse would blow. After that happened for the third time, I got sick of it. Had identified the fault, and fixed it, but it was annoying.

Such a 'fuse protector' needs to do more than shut off the power. Since it's intended for use with equipment open for servicing, it needs to fully isolate the mains once tripped. While still being fast acting enough to save fuses in the equipment, and avoid tripping the mains circuit breaker. (To save walking to the distribution board and resetting it.)

That needs an electronic switch (for speed, and able to take full short circuit current), in series with a double pole contactor (for isolation.)
For sensing, a torroidal current transformer. The circuit to set trip point and drive the switches would be easy,  I have suitable contactors and CTs, but I didn't decide on the electronic switch part yet.

Since it would normally be used between a variac (and possibly an isolation transformer too) and the equipment being debugged, it would have to withstand the load dump spike from the variac/transformer. So some big mosorbs on the input as well.

For ease of use it should have an analog mains current display, but a digital trip point setting, to allow just typing in a trip current. Plus a reset button, and a very visible "ON" warning light. And it would need its own separate power input, since the 'mains' it controls may be only a few volts from a variac set low. A 12VDC wall barnacle perhaps. Though the electronic switch part would require power rail(s) referenced to the mains.

[NiHaoMike]

Maybe some large SCRs to crowbar the supply for the short time it takes for the input contactor to open? The isolation transformer is going to have some impedance that would limit the current and should be able to survive a brief short.

[dom0]

A COTS current transducer may be already to slow with maybe 0.5 µs group delay.

[T3sl4co1l]

An electronic fuse isn't generally worthwhile because it's way too fast-acting.  It's also relatively expensive for the power level required.  You'll put in, probably about $40 worth of transistors, for your desired ratings, and only get a few microseconds worth of fault.  That's not long enough to charge capacitors during a turn-on surge (or perhaps even start warming up cold heaters!), or to reliably start a toroidal transformer (which draws a surge in the ms range).

A soft start would be good.  Perhaps switching in one, or a series of, resistor(s) before closing the circuit fully, and faulting out if any of those steps don't go well.

I've prototyped a lower rated DC fuse (40V 8A), but I've seen very little interest in it, so I don't know if I'll even bother to scale it up to line sized.

For simple solutions, it's hard to beat a handful of lamps!

Tim

[Seekonk]

I have an outlet with a number of parallel lamps in series.  I screw in more lamps for higher current.   I have used variations of this for 50 years in equipment repair.  Lamps increase in resistance when they heat up.  Typically have about 7 to 1 increase.  Anything normal won't light the lamp.

[NiHaoMike]

Or for an even greater fault to normal impedance ratio, use a capless (induction start) induction motor.

[Jeroen3]

Like this: Tele Power G2IM10AL10 ?

But how fast does it need to be? uS, mS? Because shorts will be caught by a simple B class breaker and your ears. (the breaker will rattle if you slowly ramp up the voltage)
You can DIY if you'd really want to using a LEM module a relay and a comparator, but that does not meet the speed of a passive circuit breaker.

[DanielS]

Problem: Making the circuit so fast, that it can disconnect w/o damage an instant dead short at 320 Vpeak is... challenging. Unipolar gate drive might not be enough ; you'll likely need <1 µs or even faster turn-off time overall.

The turn-off is not too critical since you can add some series inductance to slow down current rise: 100µH at 320V peak becomes a very manageable 3.2A/µs short-circuit current rise rate. You can use RCD snubbers or even just some 20mm 200/400V MOVs (130/270 MCOV) to absorb the inductive spikes on break to avoid needing ridiculously high voltage MOSFETs.

[Zero999]

An electronic fuse isn't generally worthwhile because it's way too fast-acting.  It's also relatively expensive for the power level required.  You'll put in, probably about $40 worth of transistors, for your desired ratings, and only get a few microseconds worth of fault.  That's not long enough to charge capacitors during a turn-on surge (or perhaps even start warming up cold heaters!), or to reliably start a toroidal transformer (which draws a surge in the ms range).

A soft start would be good.  Perhaps switching in one, or a series of, resistor(s) before closing the circuit fully, and faulting out if any of those steps don't go well.

What about a fast current limit followed by a circuit breaker action if the fault persists for too long? For example, limit the current to 1A, if the fault persists for longer than perhaps 100ms then open the breaker. The trouble is the transistors need to be capable of dissipating the full fault power, until the breaker opens.

[joeqsmith]

They do make mechanical ones.  The ones I have seen, you set the trip by adjusting the tension on a spring.

I did something like this when I made my last transient generator.   I just used a current sensor with an ADC hooked to a small microcontroller.  The software monitors the current and will cut the power if it gets above the set point (programmed from the front panel).   The isolation transformer I used is pretty small and limits the current to 500mA.

[DanielS]

The trouble is the transistors need to be capable of dissipating the full fault power, until the breaker opens.

I can imagine two relatively simple solutions:
1- use a thyristor or triac and progressively back off the firing angle starting from 179 degrees
2- use a motorized variac and start from 0V

Some loads might not appreciate being stuck in input voltage limbo or having odd input waveforms though.

A more complex solution would be a current-limited D-class amplifier.

[dan3460]

I do understand that there are other ways to bring to power equipment under test, I actually use them. But my original question was to see if someone had a circuit or maybe someone can come up with a circuit. With DC is easier to make that kind of circuit, my problem is AC. The unit posted by Jerone3, I don't thing will be easier to control.
The circuit should do basically two things:
1- Have a knob where you can setup a tripping current, let say from 100mA to 10A.
2- Have a button that you can press an reset the circuit.
Besides the AC problem, this circuit will need to operate on variable AC as it should be mounted in front of a variac.

[Seekonk]

I would use a 5A ACS712 with an additional shunt that can be switched in.  Then feed that into a UNO or mini328 and just drive a relay.  You will need a micro because of what you have wished for.  You will need delays, averaging, whatever to prevent all the false trips.  I've managed to get by for 50 years with just light bulbs in series.

[dan3460]

Maybe I'm posting in the wrong board. I'm looking for a way to monitor AC current and then act upon a reference point. I understand, and I also use, a dim bulb, a wonderful idea. But at this point i'm more interested in learning if such thing can be done.

[dom0]

It's totally impossible, since no one posted any hints, pointers, ... in this thread on how to implement it.

If you're looking for a schematic you just build (and not "I'm more interested in learning if such a thing can be done") then: No, I won't draw you a schematic. Do it yourself ffs.

[dan3460]

Uhfaaa... Sorry to bother you. 

[Seekonk]

Perhaps you should say what your intention is.  Are you looking for an approach based on peoples experiences or a fully tested schematic?   The latter isn't likely.  A micro is your best choice.  The ACS712 outputs 512 a voltage centered on 2.5V.  Run that into the A/D and look at values as fast as you can.    Subtract 512 and look at any positive number.  Do the same thing by subtracting the number from 512.  Any numbers over X trip the relay or add a filter to trip if say three of the last32 samples were high. 

[dan3460]

My intention is what I put in the OP. I was working on my bench and I thought it will be nice to have something that it will stop the current that you can set at a specific rate.
I was hoping for, as in other forums, that someone will post a circuit. Then someone else will criticize that circuit and post a "improved version" and so forth.
I'm a not a professional, I do this as a hobby. 

[timb]

My intention is what I put in the OP. I was working on my bench and I thought it will be nice to have something that it will stop the current that you can set at a specific rate.
I was hoping for, as in other forums, that someone will post a circuit. Then someone else will criticize that circuit and post a "improved version" and so forth.
I'm a not a professional, I do this as a hobby.

Maybe *you're* that person.


Sent from my Tablet

[dan3460]

Possible, my problem is that I have no experience with AC circuits. For example, to measure current in a DC circuit you could put a shunt low resistor and use a comparator to measure the voltage difference across the resistor, use that to drive a relay that will disconnect the output if it goes over a threshold.
How would you do that on AC? Can you do that on AC?

[dom0]

Outline: Two ~600-800 V MOSFETs source to source in the usual AC switch configuration. Gate drive e.g. with an ATX-SMPS base current transformer and a pull-down (see below!). Shunt & precision rectifier & average & comparator w/ hysteresis -> gate drive oscillator enable.

Problem: Making the circuit so fast, that it can disconnect w/o damage an instant dead short at 320 Vpeak is... challenging. Unipolar gate drive might not be enough ; you'll likely need <1 µs or even faster turn-off time overall.

I never built this ; having a variac with a quick responding analog meter is good enough for me.

[nfmax]

Since panel mounting thermal circuit breakers are pretty cheap, you could always use a set of the things with different trip currents, selected by a rotary switch...

[timb]

Possible, my problem is that I have no experience with AC circuits. For example, to measure current in a DC circuit you could put a shunt low resistor and use a comparator to measure the voltage difference across the resistor, use that to drive a relay that will disconnect the output if it goes over a threshold.
How would you do that on AC? Can you do that on AC?

A couple of posts above we gave you the answer. It's actually been mentioned at least twice.

Check out the ACS712, which is a Hall Effect current sensor. It outputs an analog voltage relative to the AC or DC current passing through. Sparkfun even sells a breakout board for it.

Other options include a current transformer.


Sent from my Smartphone

[Seekonk]

I was just thinking how a three pin TL431 makes a nice precision relay driver.  It trips at about 2.5V  and the ACS712 would be at 2.5V with a 5V supply.   Add a pot and you have something.

[timb]

I was just thinking how a three pin TL431 makes a nice precision relay driver.  It trips at about 2.5V  and the ACS712 would be at 2.5V with a 5V supply.   Add a pot and you have something.

Not a bad idea. You might even go with a SSR or Triac for a quicker response. You'd still need a circuit to latch the output, otherwise, when the current limit hits and the relay comes on, the current would drop turning the relay off and around and around we go.

[fuzzoli]

I was looking at this from the CR Magnetics website.  The full document can be found here:

http://www.crmagnetics.com/Assets/ProductPDFs/Precision%20Rectifier%20Circuit%20for%20CT%20Signal%20Conditioning.pdf

My thought was to start with their circuit which converts the CT output to a DC signal.  The output of their circuit would feed a comparator to set the trip point, followed by a 555 to set the trip "delay".  There would obviously be a need for some "glue" parts, but my thought was for the core sense/trip circuitry to be micro-free to minimize the risk of failure of the most critical part (i.e. the part that stops the current from flowing).

...more to come...

-Frank

[timb]

Here's a circuit using the ACS712: http://home.cogeco.ca/~rpaisley4/AcDcBreaker10.html

The guy uses it as a breaker for model railroad stuff, but if you use a 120/240VAC relay it should work fine.

[Seekonk]

I was just thinking how a three pin TL431 makes a nice precision relay driver.  It trips at about 2.5V  and the ACS712 would be at 2.5V with a 5V supply.   Add a pot and you have something.

Not a bad idea. You might even go with a SSR or Triac for a quicker response. You'd still need a circuit to latch the output, otherwise, when the current limit hits and the relay comes on, the current would drop turning the relay off and around and around we go.

Details, details.    I had a manager who said his best education Was working at a traveling carnival one summer.  We would go into meetings with a customer.  The customer would ask if the machine could do this.  He would respond, "You could make it do that."   The customer heard what they wanted to hear.  The correct interpretation was .....YOU could make it do that, he wasn't going to do it for you.  One of only three brilliant men I have known in my life.

[sarepairman2]

is there some kind of magnetic material with a sharp trip point?

I.E. saturating a 1:1 transformer with a DC pulse, blocking the AC signal?

[T3sl4co1l]

If you like short circuits, yes...

Tim

[dom0]

Either way, you can't switch AC off by saturating a reactor. If the reactor is in series, saturating it reduces the inductance to ~nil, "switching it on". And if the reactor is in parallel you blow stuff up.

Anyways, fast switching isn't exactly a virtue of reactors anyway, at least not for 50 Hz reactors.

[T3sl4co1l]

You'd have to have an L network (one in parallel, to short it out quickly, another in series to open circuit), and then some active circuitry to do the steering (or at least a carefully magnetized core on the one).

Tim

[Ian.M]

Current transformer monitoring the Variac output + burden resistor feeding a precision rectifier, then either a peak detector if you need a very fast trip or a low pass filter to average over a few cycles.  Next you need a comparator (or two if you want to be able to trip on excessive peak and average current at the same time) with a pot with a pointer knob and a 0-10 scale providing the reference voltage.  The comparator output(s) then drive the RESET input of a D flipflop, with a debounced pushbutton to apply power connected to its clock input and '1' on its D input so that holding in the button cant prevent tripping,  The flipflop should control a two pole contacter that isolates the Variac input.

I see little point in using a true RMS chip.  If you need that level of trip current accuracy, you'd do better to go digital and replace everything between the current sensor and the contacter with a MCU.

[Seekonk]

OP wants say a 5A trip.  At last someone mentions what does that mean.   Is it peak, average, RMS, follow a fuse thermal profile, fast or slow.   Just about nothing is resistive today.  Look at the scope image for a E27 LED lamp.  Notice the spike as a capacitor charges up.  This short duty cycle is typical of most electronics today.  Despite CE regulations I find many good size wall warts are half wave rectification.  So you turn a pot till it shuts down and back off.  Nothing precision there and within the scope of a hobbyist if he has the knowledge to repair anything.

Another idea, use another TL431 to generate 2.5V and use a pot voltage divider to create a 0-2.5V reference.  Then feed that as a reference to a current transformer sense.   Then look at that with another TL431 that triggers a SCR and relay.