IC which Trips on Voltage and has a Reset

[laingalion]

Sorry in advance, this is probably super basic but no matter what I search I can't find exactly what I'm looking for.

What is an IC or a circuit which trips at a certain voltage and needs to be reset?
More specifically, I want the output to always be high (like VCC of the IC) until the input voltage reaches around 10V at which point the output would go low and stay low.
The reference voltage is AC at 30 kHz so it will only exceed 10V for a very short time. But as soon as the voltage goes past 10V I want the IC to trigger.


Background:
I'm trying to make a very simple over current protection for my amplifier (50A, 20V, 15-30 kHz).
At 50A I want the IC/circuit to stop the drivers.
I'm going to get an current sensing IC (ACS780 Hall-Effect sensor or something similar) to measure the load current.
Then I need another IC to trip and stay low when the current sensing IC output reaches a threshold voltage (10Vish).

After the circuit trips I want to reset it using a microcontroller. (The microcontroller will periodically check if the circuit has tripped)
I don't know if any of this is practical.


Thanks in advance

[Nusa]

If there's already a microprocessor involved, then you might as well have it monitor the voltage and decide when and how to act. Many micros already have adc inputs, so most of what you need is already there.

[laingalion]

I don't think the microcontroller would be fast enough, especially when the single frequency is up to 30 kHz (60 kHz for bi-directional).
The microcontroller will need to sample at exactly the peak, which means the sampling needs to be many times faster than the single frequency
I'm looking for something to trip within a cycle (30 us) ideally

Can a microcontroller output a PWM while also sampling from the ADC? (I'm using an Atmel  ATtiny461)
Does the interrupters used in the PWM mess up any ADC interrupters?
I figured I needed to pause the amplifier to read from the ADC. Thats why I wanted a separate circuit which trips on its own and waits until the microcontroller to check on it
Am I completely mistaken? I've never tried before

[laingalion]

Can I somehow use a 555 timer? (I've never used one before, just heard about it)
Like having the current sensing amplifier output (scaled with a voltage divider) connected to the 555's trigger or threshold pins?


Would a 555 timer be overkill for what I'm looking for? Is there something simpler?

[Zero999]

The usual way to do this is to use a crowbar circuit, which short circuits the supply and blows the fuse, when the voltage exceeds a predetermined level. The circuit is reset by disconnecting and reconnecting the power supply.


https://www.electronics-notes.com/articles/analogue_circuits/power-supply-electronics/over-voltage-protection.php

It's possible to use the TL431 as a comparator to turn off a MOSFET, connected to the load, if the supply voltage exceeds a certain level. The circuit below could do the job, with some modifications: R1 = 10k and R2 = 3k3  It also needs some hysteresis, to avoid oscillation. Add a 1M resistor between the point where R1 & R2 meed and Q2's collector. This circuit will also auto-reset when the power supply voltage returns to normal.


There was a thread about this awhile ago.
https://www.eevblog.com/forum/beginners/2n7000-strange-behaviour/

[laingalion]

Thanks for the reply

I apologize if my first post was not clear, I've revised it
The voltage reference is the output of a current sensor and not the supply voltage, would the circuits you kindly provided still work?
The voltage reference is also AC at 30 kHz, would the second circuit just reset as the reference cycles?

Some more background:
The load on my amplifier is highly highly variable, sometimes dropping to near zero. When the load's impedance drops and the current spikes, I want to stop the MOSFET drivers. The current sensor will translate the current to a voltage and I want a circuit to trip based on that voltage.

[laingalion]

So a little update
I think what I'm looking for is called a "latch"... maybe

[T3sl4co1l]

Why isn't the amplifier current-limited?  Seems like it was made to self-destruct. 

Tim

[laingalion]

Why isn't the amplifier current-limited?  Seems like it was made to self-destruct. 

Tim

Ya the ones I've made so far have self destructed lol
Sorry I've pretty new to this. When you say current limited, do you mean like a fuse?
I need something that can be reset on the fly

[Zero999]

Thanks for the reply

I apologize if my first post was not clear, I've revised it
The voltage reference is the output of a current sensor and not the supply voltage, would the circuits you kindly provided still work?
The voltage reference is also AC at 30 kHz, would the second circuit just reset as the reference cycles?

Some more background:
The load on my amplifier is highly highly variable, sometimes dropping to near zero. When the load's impedance drops and the current spikes, I want to stop the MOSFET drivers. The current sensor will translate the current to a voltage and I want a circuit to trip based on that voltage.

The circuit I posted is for use in a DC application.

I'm still confused. If the voltage is AC, then it will have both positive and negative peaks. Do you want the circuit to trip when the voltage exceeds the peak value?

Please post a schematic.

Why isn't the amplifier current-limited?  Seems like it was made to self-destruct. 

Tim

Ya the ones I've made so far have self destructed lol
Sorry I've pretty new to this. When you say current limited, do you mean like a fuse?
I need something that can be reset on the fly

A fuse is usually too slow to protect an amplifier's sensitive output transistors from overheating.

Over-current protection is normally achieved by sensing the output current and cutting off the drive to the transistors, when it exceeds a certain level. Protection against over-current is integrated into the design of the amplfier, rather than added as an after thought.

[laingalion]

Why isn't the amplifier current-limited?  Seems like it was made to self-destruct. 

Tim

Ya the ones I've made so far have self destructed lol
Sorry I've pretty new to this. When you say current limited, do you mean like a fuse?
I need something that can be reset on the fly

A fuse is usually too slow to protect an amplifier's sensitive output transistors from overheating.

Over-current protection is normally achieved by sensing the output current and cutting off the drive to the transistors, when it exceeds a certain level. Protection against over-current is integrated into the design of the amplfier, rather than added as an after thought.

Ya thats what I'm trying to add now. It's was my first attempt at an amplifier. My next version I want to add the over current protection
I'll try post a basic schematic in a sec

[T3sl4co1l]

Ya the ones I've made so far have self destructed lol
Sorry I've pretty new to this. When you say current limited, do you mean like a fuse?
I need something that can be reset on the fly

No, well, you could make it latching, but that would probably be undesirable.  Current can be limited instantaneously, preventing transistor overheating in the first place.

Is this relevant? https://www.eevblog.com/forum/projects/unstable-linear-voltage-regulator/msg359283/#msg359283

Something about H bridge, perhaps a class D amplifier?

If it's a switching amplifier, it's the perfect application for a current mode control, which is therefore current limited, at all times, by design.

Most class D amp designs are voltage mode, which gives lower output impedance, but leaves the switching current completely unconstrained.  If your speakers never short out, sure, it might run for years, decades... If you're careless, or you just don't like the idea of a shorted speaker releasing the magic smoke, current mode is the only possible choice!

If it's a linear amplifier, the traditional method is to add current sensing transistors to the output stage.  Sometimes these are foldback type, i.e., the current drops as the voltage rises, limiting power dissipation even more.

Tim

[laingalion]

Thanks so much for your inputs

So heres a schematic of a baby's first steps at an amplifier (one side of it, its a full bridge amplifier)
I used a driver which has a highside built-in. Later I will try make my own using optocouplers like everyone recommends

The load changes impedance A LOT because the transducers are designed to be in the water. Sometimes I accidentally run the amplifiers while the transducers are out of the water, thats when my amplifier self destructs

I'll take a look at the current mode control. I hope its not too complicated.
I thought latching would be fine because it would only trip when the amplifier is not being used properly
If the current mode control is a better practice I'll try learn it

[laingalion]

So heres my plan
Remember, I just made this up so please be nice if its just super ridiculous

[T3sl4co1l]

Vout is an analog level, and you're connecting it to a digital logic gate -- seems like you need some sort of analog to digital converter!  A one-bit ADC is also known as... a comparator.

D latch isn't what you want: Q is only updated when CLK is high, and remains in its previous state when CLK is low.  But a latch is the right idea.  If the input is always pulsing on and off, then you can use a D-f/f where the MCU drives CLK, D is wired to +1, and RST is wired to the comparator.  Thus, it turns on (CLKs in a fresh '1') until the comparator says off.

Of course, you need the RST OR'd with /CLK, so it turns off when CLK is low!

This can also be done with a handful of gates, in a feedback configuration.  It's sort of like a clocked AND gate.

If you run the amplifier much faster than 30kHz, you can do proper class D PWM, like an audio amp would.  Even if you keep it at 30kHz, you can control the pulse width (if the MCU has the right hardware, otherwise you can use an analog IC like a TL494), and use that to control average current.

Since the output is AC, why not consider a current transformer (CT)?  More accurate, faster, quieter and cheaper (sometimes) than a Hall effect sensor.  This has the advantage that you only need the one sensor for both switch polarities: just connect the CT to a FWB (use 1N4148 or BAV99 or BAT54S or..), to the burden resistor.

Tim

[laingalion]

Ah a comparator, that makes so much more sense than just a schmitt trigger lol
And the D flip flop setup makes a ton of sense too. Your simple explanation helped me to understand the logic better than the 30 minutes of youtube videos I just watched.
However, I think in my particular case I may want something that latches. I'll try build both

Aren't there bi-directional hall-effects? I was looking at some spec sheets, maybe I'm misreading it

[T3sl4co1l]

Hall-effect sensors are either zero offset or fixed offset, so they read AC just fine (assuming you put the right voltages on them!).  Downside: because it's common ground, you can't simply tack on a bridge rectifier -- instead, you need to use a window comparator.  (That might not be a bad thing, in and of itself: if you have a latch for each side, positive and negative, then it's not actually a window comparator (two thresholds with a single output), but two separate comparators with complementary (mirror image) thresholds, one for each latch.)

It's also a neat observation about current transformers: because the transformer's output is high impedance (ideally, a constant current: a fixed fraction of the current being measured), it doesn't matter if you connect diodes in series with it.  The diode drops are absorbed by the CT, so the output waveform (from the CT + FWB) does not have a 1.2V deadband.  This might not be all that useful to you after all (for the above reasons), but it's good info to keep in mind when it is.

Tim

[Zero999]

How about using plain old current sense resistors?

When the current is high enough to turn the transistor on, current flows through the LED and activates the opto-coupler. The transistors in the opto-coupler can be connected in parallel and used to trigger an RS flip-flop which disables the buffer. Note that you'll want pull-down resistors on the output of the buffer, since when it's disabled, it makes its outputs open circuit and therefore floating, leading to undetermined logic signals at the gate drive IC.

[T3sl4co1l]

NO --

This is not latching, so will oscillate at the frequency determined by the total propagation delay (which is mostly the opto, the slowest thing here).

It's very easy to look at something and say, "this goes up, so this needs to go down, and, there we go!"  But you must always then ask, "what happens next?"

Note also that, disabling a tristate buffer leaves an undefined state ('Z'), so weak pull-downs are required to keep a defined off state.

It might be better to use 74HC124 (quad tristate) to control which transistors are being turned off at a time, or 74HC08 (quad AND) to give defined output states, but best is to use a feedback gate that latches CLK.

Tim

[laingalion]

Hall-effect sensors are either zero offset or fixed offset, so they read AC just fine (assuming you put the right voltages on them!).  Downside: because it's common ground, you can't simply tack on a bridge rectifier -- instead, you need to use a window comparator.  (That might not be a bad thing, in and of itself: if you have a latch for each side, positive and negative, then it's not actually a window comparator (two thresholds with a single output), but two separate comparators with complementary (mirror image) thresholds, one for each latch.)

It's also a neat observation about current transformers: because the transformer's output is high impedance (ideally, a constant current: a fixed fraction of the current being measured), it doesn't matter if you connect diodes in series with it.  The diode drops are absorbed by the CT, so the output waveform (from the CT + FWB) does not have a 1.2V deadband.  This might not be all that useful to you after all (for the above reasons), but it's good info to keep in mind when it is.

Tim

I was naively hoping there would be a hall-effect which would give me the "absolute value"
I think I will go with your CT recommendation. Can the comparator  just compare the voltage across the burden resistor and the reference voltage?
Or would it be best practice to have another device measure the burden resistor's voltage?

[T3sl4co1l]

Yup, that's all.

I mean, in the sense that the comparator is a 1-bit ADC, it is measuring the voltage, so that's not wrong, either.

Tim

[Zero999]

NO --

This is not latching, so will oscillate at the frequency determined by the total propagation delay (which is mostly the opto, the slowest thing here).

I know. The schematic doesn't give the full story. Notice how the phototransistors don't connect to anything? That should have been a clue.  If you read the text below, you'll find I mentioned using the opt-isolators to trigger a filp-flop which disables the buffer.

Hall-effect sensors are either zero offset or fixed offset, so they read AC just fine (assuming you put the right voltages on them!).  Downside: because it's common ground, you can't simply tack on a bridge rectifier -- instead, you need to use a window comparator.  (That might not be a bad thing, in and of itself: if you have a latch for each side, positive and negative, then it's not actually a window comparator (two thresholds with a single output), but two separate comparators with complementary (mirror image) thresholds, one for each latch.)

It's also a neat observation about current transformers: because the transformer's output is high impedance (ideally, a constant current: a fixed fraction of the current being measured), it doesn't matter if you connect diodes in series with it.  The diode drops are absorbed by the CT, so the output waveform (from the CT + FWB) does not have a 1.2V deadband.  This might not be all that useful to you after all (for the above reasons), but it's good info to keep in mind when it is.

Tim

I was naively hoping there would be a hall-effect which would give me the "absolute value"
I think I will go with your CT recommendation. Can the comparator  just compare the voltage across the burden resistor and the reference voltage?
Or would it be best practice to have another device measure the burden resistor's voltage?

The Hall effect sensor or current transformer could go to a precision rectifier, before the comparator. Then the comparator only needs to worry about tripping on positive voltages.

Irrespective of whether a current transformer, Hall effect sensor or my idea using sense resistors and opto-isolators is used, you want the circuit to latch, so you need a flip-flop. You can make your own with transistors, a comparator or use an IC, such as the 74HC74 or CD4013.