Resettable short circuit protection circuit that really works?

[e100]

Searching the web there are lots of similar looking theoretical op-amp based "should work" solutions, but no actual "proven to work without oscillating" solutions.
Does anyone have a circuit that has been proven to work in real world usage?

My particular need is to protect a 9v DC pool chlorinator electrolysis cell that draws 15A at the recommended salt concentration. Periodically you have to add salt to compensate for losses and this is a tricky operation as adding too much salt too quickly without giving it a chance to mix properly can put pockets of concentrated saline through the cell with a consequent jump in current that the rectifier and transformer aren't designed to handle.
You need to run the chlorinator during this operation as you are using the current draw as the primary indicator of salinity.

[Randy222]

Does it need to turn off, or just limit to a max?

Current Limiting is the game.

https://electronics.stackexchange.com/questions/262088/limiting-current-without-dropping-voltage

[e100]

Does it need to turn off, or just limit to a max?

At the most basic level it just needs to trip off at some adjustable threshold.

[Randy222]

Maybe Littel Fuse PTC RGEF series? PTC is ideal, you won't need powered circuitry to get the protection.

https://www.littelfuse.com/products/polyswitch-resettable-pptcs/radial-leaded/rgef.aspx

[fourfathom]

I don't have an answer for you, but do have questions and comments:

Do you want it to recover from overcurrent without requiring a power-cycle? 

If so, the PTC devices may not work for you as they will stay in the "tripped" (high resistance) state until the current is dramatically reduced.  Also, the range trip time and level is somewhat loose, so probably not best if you want a tight limit.  I've used these for short-circuit protection and been very happy with the behavior, but if you want to pass 9A and shut down at 10A you should look elsewhere.

A regular current-limit circuit could be your answer, but it obviously has to dissipate the power, or be a fancy switching regulator design.

I designed one protection circuit that rapidly shut down the power when the current limit was reached, then retried a few times per second.  For this to work, your system needs to tolerate the high-current transients.  I'm not suggesting that you build one of these, but you should consider your options.  There are bi-metallic thermal fuses that behave in this manner (with a slow cycle rate) -- my vacuum cleaner has one of these in the powered beater-brush head.  Would that work?

[e100]

Do you want it to recover from overcurrent without requiring a power-cycle? 

'Press a button' seems like the simplest solution as the person adding the salt stays until a stable salinity level has been achieved.

If so, the PTC devices may not work for you as they will stay in the "tripped" (high resistance) state until the current is dramatically reduced.  Also, the range trip time and level is somewhat loose, so probably not best if you want a tight limit.  I've used these for short-circuit protection and been very happy with the behavior, but if you want to pass 9A and shut down at 10A you should look elsewhere.

There are bi-metallic thermal fuses that behave in this manner (with a slow cycle rate) -- my vacuum cleaner has one of these in the powered beater-brush head.  Would that work?

PTC and other thermal devices are sensitive to the ambient temperature. For example, a PTC setup for winter conditions will have a 30% lower trigger threshold at the height of summer which increases the chance of a nuisance trip, and as you said, once tripped it'll stay that way for several hours until the timer turns everything off in the evening.

[PCB.Wiz]

Searching the web there are lots of similar looking theoretical op-amp based "should work" solutions, but no actual "proven to work without oscillating" solutions.
Does anyone have a circuit that has been proven to work in real world usage?

My particular need is to protect a 9v DC pool chlorinator electrolysis cell that draws 15A at the recommended salt concentration. Periodically you have to add salt to compensate for losses and this is a tricky operation as adding too much salt too quickly without giving it a chance to mix properly can put pockets of concentrated saline through the cell with a consequent jump in current that the rectifier and transformer aren't designed to handle.
You need to run the chlorinator during this operation as you are using the current draw as the primary indicator of salinity.

So there is a current meter the user can see ?

Rather than a button, you could make a hiccup circuit, that flips in a series power resistor, and continues to do that at some slow duty cycle, until the trip peak reduces.

[nali]

Sounds like a job for an e-fuse?

E.g. https://www.ti.com/lit/ds/symlink/tps25984.pdf?ts=1728273033150&ref_url=https%253A%252F%252Fwww.ti.com%252Fpower-management%252Fpower-switches%252Fefuse-hotswap-controllers%252Fproducts.html

I've used these in smart meters, which must provide a DC output to a Comms Hub, which must trip off then retry in the case of an overload.

[Randy222]

If you want active and very accurate current protection, you op-amp a sense resistor.
Inline the sense resistor (heavy metal loop type that can handle 15A, etc) with the load and if AC do it on the gnd side leg, each side of the sense resisor to the op-amp inputs. You adjust amp gain depending on the output stage logic or silicon control. As example, the output of the amp can drive gate of a pFET, when the output of the amp goes high (use a pull down resistor on the amp output, 1k-10k) , when amp goes high the pFET will turn off. The fet or igbt needs to be able to handle the 15A, so one with low 'on' resistance (R_DS) is better.

Edit: for clarity, my example needs a 2nd opamp as comparator and a voltage reference on input to comparator. This way you can fine tune the threshold at which comparator turns on or off the current. The 1st amp feeds the other input to the comparator, etc.

[e100]

Sounds like a job for an e-fuse?

Things are a bit complicated as the chlorinator power supply is simply a transformer and rectifier with no smoothing.

All the e-fuse devices I've seen derive their operating power from the supply being monitored so in this case there will be a dead zone from 0 to about 2.5v on every half cycle where the device isn't able to run.

[Jeroen3]

Wouldn't it be easiest in this case if you get a more lossy transformer that is able to have it's output shorted?

[e100]

Searching the web there are lots of similar looking theoretical op-amp based "should work" solutions, but no actual "proven to work without oscillating" solutions.
Does anyone have a circuit that has been proven to work in real world usage?

My particular need is to protect a 9v DC pool chlorinator electrolysis cell that draws 15A at the recommended salt concentration. Periodically you have to add salt to compensate for losses and this is a tricky operation as adding too much salt too quickly without giving it a chance to mix properly can put pockets of concentrated saline through the cell with a consequent jump in current that the rectifier and transformer aren't designed to handle.
You need to run the chlorinator during this operation as you are using the current draw as the primary indicator of salinity.

So there is a current meter the user can see ?

Rather than a button, you could make a hiccup circuit, that flips in a series power resistor, and continues to do that at some slow duty cycle, until the trip peak reduces.

Good idea, but that feature needs to be built on top of a reliable circuit that can do that switching without having oscillation problems, hence my search for a proven design.

[PCB.Wiz]

Good idea, but that feature needs to be built on top of a reliable circuit that can do that switching without having oscillation problems, hence my search for a proven design.

It is not trying to regulate the current, there is no linear region to oscillate.
It does a time-delay hiccup, where it switches in a (large size) load resistor.
You could even use a relay here, if you want proven and simple. It only cycles when someone is there, mixing it not quite right ?

[e100]

Good idea, but that feature needs to be built on top of a reliable circuit that can do that switching without having oscillation problems, hence my search for a proven design.

It is not trying to regulate the current, there is no linear region to oscillate.
It does a time-delay hiccup, where it switches in a (large size) load resistor.
You could even use a relay here, if you want proven and simple. It only cycles when someone is there, mixing it not quite right ?

What information are you displaying to the user?

I could imagine an on/off switching system that runs at a 1% duty cycle so you only pass 1 out of every 100 half cycle pulses and capture the peak current and display that to the user once per second.

[PCB.Wiz]

Good idea, but that feature needs to be built on top of a reliable circuit that can do that switching without having oscillation problems, hence my search for a proven design.

It is not trying to regulate the current, there is no linear region to oscillate.
It does a time-delay hiccup, where it switches in a (large size) load resistor.
You could even use a relay here, if you want proven and simple. It only cycles when someone is there, mixing it not quite right ?

What information are you displaying to the user?

I could imagine an on/off switching system that runs at a 1% duty cycle so you only pass 1 out of every 100 half cycle pulses and capture the peak current and display that to the user once per second.

I was assuming you had a meter display already ?
It could be as simple as a reed switch with a few turns of current wire around it, and a heavy relay and resistor + cap for timing
Reeds look to be available with about 10% AT tolerance so you adjust current trip via turns and maybe some magnetic steering.
Once high enough, the reed flips to charge the cap across the relay, which then flips the relay to ballast resistor, which opens the reed.
The relay stays flipped until the voltage drops below the relay holding voltage, then it releases. If current is still too high, it repeats.
Charge time is set by cap + resistor and discharge time is set by relay coil resistance and cap.
An LED + Res across the relay coil could indicate this working, although the relay clicks and current meter dips may be enough.

[tom66]

Sounds like a job for an e-fuse?

Things are a bit complicated as the chlorinator power supply is simply a transformer and rectifier with no smoothing.

All the e-fuse devices I've seen derive their operating power from the supply being monitored so in this case there will be a dead zone from 0 to about 2.5v on every half cycle where the device isn't able to run.

Then you want a device with an external FET.  You would run the IC from a local semi-regulated supply (a half-wave rectifier and a small cap would work, maybe add an LDO if you want) whilst passing the unregulated supply through the FET and current sense circuit.
https://www.ti.com/product/TPS24771

I expect this would work OK, since the device only cares about the current passing through the sense resistor, but I haven't tested it.  TI may have dev kits available to experiment with though some surgery may be needed to change what supplies Vdd.

You would set the fault timer up to be around the order of a few seconds by changing the fault timer capacitor, which would cause the device to periodically retry until the short circuit is removed.

[e100]

Good idea, but that feature needs to be built on top of a reliable circuit that can do that switching without having oscillation problems, hence my search for a proven design.

It is not trying to regulate the current, there is no linear region to oscillate.
It does a time-delay hiccup, where it switches in a (large size) load resistor.
You could even use a relay here, if you want proven and simple. It only cycles when someone is there, mixing it not quite right ?

What information are you displaying to the user?

I could imagine an on/off switching system that runs at a 1% duty cycle so you only pass 1 out of every 100 half cycle pulses and capture the peak current and display that to the user once per second.

I was assuming you had a meter display already ?
It could be as simple as a reed switch with a few turns of current wire around it, and a heavy relay and resistor + cap for timing
Reeds look to be available with about 10% AT tolerance so you adjust current trip via turns and maybe some magnetic steering.
Once high enough, the reed flips to charge the cap across the relay, which then flips the relay to ballast resistor, which opens the reed.
The relay stays flipped until the voltage drops below the relay holding voltage, then it releases. If current is still too high, it repeats.
Charge time is set by cap + resistor and discharge time is set by relay coil resistance and cap.
An LED + Res across the relay coil could indicate this working, although the relay clicks and current meter dips may be enough.

The existing current meter is a basic moving coil meter that displays the average (RMS?) current.

I'm not an engineer and this is the first time I've heard of a hiccup circuit. Is your magnetic breaker system with a time delay going to work with the unsmoothed lumpy waveform of the rectified AC chlorinator power supply?

[Zero999]

Use a magnetic circuit breaker. Don't bother trying to roll your own. It will comply with the safety standards and a home made circuit won't.

The transformer should also have a primary side breaker.

It is also a good idea to add bimetalic switch to the transformer, to cut the power if it overheats. Fix it to the core (tap holes or glue with thermal epoxt resin) and wire it in series with the primary.

[Randy222]

Sense resistor, one op-amp, one comparator, one voltage reference, one Omron G3NA SSR.

When current gets to the threshold, this setup will go into oscillation, because when the current gets cut the amp output will drop to zero, SSR will again turn on.
It's basically a current clamp, so even if you get a short ckt it will clamp to the set max amps. You can control the oscialltion using a resistor and cap.

The G3NA series has app note for DC loads.

[PCB.Wiz]

The existing current meter is a basic moving coil meter that displays the average (RMS?) current.

I'm not an engineer and this is the first time I've heard of a hiccup circuit. Is your magnetic breaker system with a time delay going to work with the unsmoothed lumpy waveform of the rectified AC chlorinator power supply?

Good question.
If this is raw rectified that will cause problem in a simple circuit.
Whilst you can with care use reed switches with pulsed fields, that use intends the pulse to remain above some holding level, so it can work on a simple ON/OFF case, but here you have a widely vasrying current that sweeps up and down with the mixture changes.

https://www.te.com/en/products/relays-and-contactors/electromechanical-relays/intersection/operating-dc-relays-ac.html

That means you need something a little smarter than can detect with a threshold, then count or filter samples above that threshold.
A response time in the ballpark of 100ms~800ms would be about right for something with soft overloads and thermal mass like your transformer and rectifier.

Addit : AliExpress have something like this, with a wire sense current transformer.  Tho most of these seem to use 10A relays, a bit wimpy, but it could control a larger automotive relay.

LG-C25X    DC24V   AC 0-5A / 0-10A / 0-20A / 0-50A / 0-100mA    ①Relay output;   ②Delay time setting 0-99.9S.   ③LED display
https://www.aliexpress.com/item/32809338166.html

If you wire these so the relay drops the sense current, they will naturally hiccup.

This is cheaper, but does current detect only
https://www.aliexpress.com/item/1005004503711502.html

These must use some sort of parasitic powering, hence the minimum current given.
Here is a smaller unit, that has a higher min current
https://www.aliexpress.com/item/1005006412086239.html

[Zero999]

Sense resistor, one op-amp, one comparator, one voltage reference, one Omron G3NA SSR.

When current gets to the threshold, this setup will go into oscillation, because when the current gets cut the amp output will drop to zero, SSR will again turn on.
It's basically a current clamp, so even if you get a short ckt it will clamp to the set max amps. You can control the oscialltion using a resistor and cap.

The G3NA series has app note for DC loads.

Add a diode to the comparator circuit to make it latching. It can be reset with a push button switch.

Here's a simulation, for educational purposes only. The tripping current is also dependant on the MOSFET's R_ON, which forms past of the sense resistor. It typically has a positive temperature coefficient meaning this circuit will trip more easily when the MOSFET is hot, which is a good thing.

Warning: an E-fuse circuit should have a real fuse or circuit breaker, as there's a risk the transistor can fail short circuit.

[T3sl4co1l]

What's wrong with a dumb current-limiting resistor? Electrolysis is a simple process, nothing fancy required here.

Tim

[e100]

Sounds like a job for an e-fuse?

Things are a bit complicated as the chlorinator power supply is simply a transformer and rectifier with no smoothing.

All the e-fuse devices I've seen derive their operating power from the supply being monitored so in this case there will be a dead zone from 0 to about 2.5v on every half cycle where the device isn't able to run.

Then you want a device with an external FET.  You would run the IC from a local semi-regulated supply (a half-wave rectifier and a small cap would work, maybe add an LDO if you want) whilst passing the unregulated supply through the FET and current sense circuit.
https://www.ti.com/product/TPS24771

I expect this would work OK, since the device only cares about the current passing through the sense resistor, but I haven't tested it.  TI may have dev kits available to experiment with though some surgery may be needed to change what supplies Vdd.

If I'm reading the datasheet correctly this will not work as the absolute ratings section says that the 4 pins (VDD, SET, SENM, FSTP)  on the supply side of the mosfet must be kept within a few millivolts of each other.
https://www.ti.com/lit/ds/symlink/tps24770.pdf
 

[e100]

What's wrong with a dumb current-limiting resistor? Electrolysis is a simple process, nothing fancy required here.

Do you mean switched in circuit as a temporary thing while the salt is being added?

[T3sl4co1l]

Just always on. What's wrong with that?  Or if you prefer lower losses -- a series capacitor or inductor in front of the transformer/rectifier.

Tim