## Cheap Current Trip Circuit Needed.

19 posts / 0 new
Author
Message

One for hardware types. I need a simple and cheap isolated current trip. The easiest way to describe it is with a graph...

The simplest way would be a PTC or similar BUT I need almost instantaneous recovery. Let's say less than 100ms.

#1 Hardware Problem? https://www.avrfreaks.net/forum/...

#3 All grounds are not created equal

#4 Have you proved your chip is running at xxMHz?

#5 "If you think you need floating point to solve the problem then you don't understand the problem. If you really do need floating point then you have a problem you do not understand."

Can you affort to loose some voltage along the way?

You could use a shunt resistor which triggers a thyristor, and that thyristor could close a MOSfet gate or a BJT base current.

Your graph does not make much sens to me.

If Input current does not equal output current, then what do you do with the current difference?

It does look a bit like a Thyristor that shorts the input when the current limit is reached.

Are you comfortable with laboratory power supplies, and how the current limitnig in those work?

They typically reduce the output VOLTAGE if the output current is too high.

Plenty of schematics on the 'net.

For more detailed ideas we need to know more about what you need exactly.

There always is the problem that if there is a voltage difference over a device and it passes currents, it gets warm. ( U = I * R ).

Doing magic with a USD 7 Logic Analyser: https://www.avrfreaks.net/comment/2421756#comment-2421756

Bunch of old projects with AVR's: http://www.hoevendesign.com

Paulvdh wrote:

Can you affort to loose some voltage along the way?

Yes, happily lose up to 5V

Paulvdh wrote:

Your graph does not make much sens to me. If Input current does not equal output current, then what do you do with the current difference?

Agree. The line at the bottom was merely placed that way to make it clear which way the action flows. The current flowing, once it has tripped, can be zero, ie open circuit.

Paulvdh wrote:

Are you comfortable with laboratory power supplies, and how the current limitnig in those work? They typically reduce the output VOLTAGE if the output current is too high. There always is the problem that if there is a voltage difference over a device and it passes currents, it gets warm. ( U = I * R ).

I have no control over the source. If I try to implement any local current limiting then the power dissipation is my problem. The voltage could be up to 36V, the potential current up to 10A or more.

An ideal device would be something like a circuit breaker as fitted to your house but one which resets itself automatically after 100ms.

#1 Hardware Problem? https://www.avrfreaks.net/forum/...

#3 All grounds are not created equal

#4 Have you proved your chip is running at xxMHz?

#5 "If you think you need floating point to solve the problem then you don't understand the problem. If you really do need floating point then you have a problem you do not understand."

Brian,

- how fast does it have to trip?

- Is the trip current to be adjustable?

edit. A few ideas collection. https://www.analog.com/media/en/...

Ross McKenzie ValuSoft Melbourne Australia

Last Edited: Mon. Feb 18, 2019 - 02:30 PM

If your device can continously dissipate 360W it can just keep on pumping 10A into the target device.

A "circuit breaker" does not have a constant current characteristic.

I think you may be interested into "foldback current limiting"

https://duckduckgo.com/?q=foldback+current+limiting&ia=web

And plenty of pictures:

https://duckduckgo.com/?q=foldback+current+limiting&iax=images&ia=images

The usual idea is to limit the current to some small value, and as soon as the load current is limited by the exernal device instead of the current limiter the current increases again along the load line of the limiter.

Foldback current limiting also does not have to be lineair.

You can have 2 separate circuits. A ciruit that detects the 10A trip current and sets a flipflop (which turns of the load), combined with a sense circuit that resets the flipflop when another condition is reached.

You want your device to reset automatically after 100ms. But 100ms after what?

Doing magic with a USD 7 Logic Analyser: https://www.avrfreaks.net/comment/2421756#comment-2421756

Bunch of old projects with AVR's: http://www.hoevendesign.com

valusoft wrote:

- how fast does it have to trip?

Fast enough that the trip circuit won't get fried!

valusoft wrote:

- Is the trip current to be adjustable?

No, it can be fixed.

This is one of those odd things I get asked to quote/design, the function is to detect when the current taken by a load exceeds a certain fixed value. The loads in this case will take care of their own protection at a level well above that of the trip. I've described this as a trip because every circuit I have come up with will self-destruct if it doesn't doesn't remove itself from the system. For example...

if I use a series resistor to sense the current and bridge an AC optocoupler across it I can use a 5R resistor so that a current of 500mA generates 2,5V which is just right to turn on the opto. The resistor dissipates (0.5 x 0.5 x 5) 1.25W. However, if the 'fault' current goes up to 10A then we have (10 x 10 x 5) 500W. Now, whilst I have some resistors which will take that kind of power, they aren't exactly small.

Technically therefore what I need is a current 'detector' but one that protects itself against overload. Hence 'trip'.

It's looking like the easiest way to do this is to use something like an ACS712, or rather a few hundred of them, which although they aren't cheap they are simple.

#1 Hardware Problem? https://www.avrfreaks.net/forum/...

#3 All grounds are not created equal

#4 Have you proved your chip is running at xxMHz?

#5 "If you think you need floating point to solve the problem then you don't understand the problem. If you really do need floating point then you have a problem you do not understand."

The circuit below is among the simplest current foldback circuits.

A quick description:

When there is no load Q2 is closed, and Q1 has a steady base current through R3 and is therefore open (Mwa, 600mV drop).

When the current is so high that the voltage drop over R1 is high enough to open Q2, then the base current for Q1 is cut off.

A sense current through R4 will keep Q2 open, and the circuit resets itself within micro seconds after the current through Q2 is interrupted, or is low enough.

(R2 influences the minimum current where the circuit gets reset).

I selected the ciruit above through the image search in my previous post for it's simplicity. The picture itself is from:

https://electronics.stackexchange.com/questions/299389/how-can-i-calculate-knee-current-for-this-foldback-current-limiter

Because the Base-Emitter voltage of Q2 changes significantly (2mv/Celcius) I have not bothered to look deep into the current trip point calculations on stackexchange.

Circuits like these are usefull for gross overload protection, not for delicate precise instrumentation.

Doing magic with a USD 7 Logic Analyser: https://www.avrfreaks.net/comment/2421756#comment-2421756

Bunch of old projects with AVR's: http://www.hoevendesign.com

Last Edited: Mon. Feb 18, 2019 - 03:59 PM

Paulvdh wrote:

The circuit below is among the simplest current foldback circuits.

Neat. Thank you. Now to understand how the component values were arrived at.

#1 Hardware Problem? https://www.avrfreaks.net/forum/...

#3 All grounds are not created equal

#4 Have you proved your chip is running at xxMHz?

#5 "If you think you need floating point to solve the problem then you don't understand the problem. If you really do need floating point then you have a problem you do not understand."

It's looking like the easiest way to do this is to use something like an ACS712

Cool chip.

But you still need a pass transistor to enable you to turn off the current when the sensor, be it an ACS chip or a custom HW design, signals that the threshold current level was exceeded.

If you were to use an ACS type chip, one might consider having the input current flow through two parallel paths.

One path has a low value resistor, say 0.1 ohm, high power rating, and this path is designed to carry the majority of the load's current.

In parallel you have a higher resistance and therefore much less current flowing through this path.

This path includes your ACS chip or other current sensor circuitry.

Then, for example, for the sake of discussion, the senor path can now tripped at 1 A instead of 10 A's.

The output of both paths goes to the current control device, (PNP pass transistor, or better yet a PFet perhaps).

Note, also, (obvious in retrospect, but perhaps not obvious the first time one starts thinking about the circuit), that the current sensor doesn't have to be linear from 0 A to the trip point, (1 A in the above example).

It only has to be linear and precisely know in the region of the trip point.

As a practical matter one might have 9.9 A's, (in the 10 A trip current condition), going through the by-pass, low resistance path, and only 100 mA or even less going through the current sense path.

The ACS chip, or its equivalent, is actually a pretty smart chip.

One typically has to take the high side sense voltage and drop it through a precision voltage divider so that the top end is within the operating range of the op-amp, for the low current conditions, as even a rail-to-rail op-amp isn't actually linear all the way to the rail.

If one elected to put a small micro in the circuit, instead of an op-amp, then again the sense voltage, under low current conditions, needs to be within the linear range of the micro's ADC.

Hence the comments about the system being linear and precise at the trip point.

Some of those design concerns, ("problems"), disappear when one realizes that one only is interested in measuring / tripping on the high current end of the operational range, and one doesn't care about the non-linearity of the low current draw end, when the sense voltage (or delta voltage...) is very small.

JC

Paulvdh wrote:
Your graph does not make much sens to me. If Input current does not equal output current, then what do you do with the current difference?

Well his device switches so very fast that due to the momentum of the current already flowing in the input wire, it can't stop instantaneously but carries on flowing for a short time.

Sometimes I get irritated for no reasons (and I'm afraid it shows in some posts) but other times I'm in a better mood and  in a text editor I typed:

ACS712 is a neat chip indeed, but overkill in this situation and therefore too expensive.
Brian mentioned in #3 "happily lose up to 5V"

Brian Fairchild wrote:
Neat. Thank you. Now to understand how the component values were arrived at.

First off: because of wide spread of transistor parameters (and change with temperature, etc) circuits like these are never very accurate, and therefore no need to calculate them to 5 decimal places.
I'll give some guidelines for dimensioning the resistor values here and the order in which I would calculate them by following a few simple rules of thumb for transistors.
Because of these (simplified) rules of thumb there can be 50% deviations, and I make use of that by simply ignoring most errors that would result in deviations of a few percent.

BJT's start conducting current when the base - Emitter voltage is around 600mV, but this can be as low as 300mV when the transisor is very hot...

First step in calculating guidelines for component values.
Only look at Q1 and R3. Remove everythig else (short R1).
Transistors have a current amplification factor, called Hfe.
Start with a peek at the 2N2955 datasheet: (Also attached below this post).

http://www.unisonic.com.tw/datasheet/2N2955.pdf

Halfway page 2 you can see that the Hfe varies between 20 and 70 for acollector current of4A, and for a collector current of 10A it even drops to a factor of 4.
So, in this worst case scenario the base current must be 10/4 = 2.5A to guarantee that the transistor can pass the 10A you want.
If you have a 36V input voltage then R3 would be 36V/2.5A = 14 Ohm or less.
This is clearly a ludricous value, but it is also fairly typical for power transistors.
It is the reason that in high current situations (almost ?) always darlington or Szizalky pairs are used.
But in this case it is probably easier and more efficient to use a P-channel MOS-fet.
MOS-fet's have their own limitations, for example they almost always have a gate voltage limit of 20V, but a simple zener diode will fix that.
With a MOSfet for Q1 then the value of R3 is mostly irrelevant (Between 1k and 1Meg) and mostly defined by leakage currents and switching speed.

R1 is the resistor that determines the trip current.
If you consider only the part of the schematic aroundR1, R2 and Q2 then you'll see that there will be a base current through Q2 as soon as there is > 600mV over R1. For a 10A "Fuse" current R1 should be around 600mV / 10A = 60mOhm.
When the current rises very fast (Short circuit on the output) then during a short duration there may be much more voltage over R1. In such a (very short moment) R2 limits the base current through Q2.

Those were the most important considerations for the circuit in the normal situaton ("Fuse" is whole).
When the Fuse is tripped, Q1 is closed, you may remove it from the schematic (Use tipexx on your monitor).
The circuit will stay in the "tripped" condition as long as Q2 keeps conducting current.
I have not looked at it's datasheet but I assume Q2 has a Hfe of 200.
To be able to raise the voltage over R3 to the input voltage, Q2 must be able to conduct 36V / 560Ohm = 65mA.
This is well within the limit of most small signal transistors.
With an Hfe of 200, the minimum base current for Q2 will be 65mA / 200 = 300uA.
300uA is a pretty low current, and we're getting in the realm of leakage currents, so be a bit carefull.

We also know that to open Q2 there must be a 600mV Base-Emitter voltage.
Q1 is erased by tipex, and therefore R1 and R2 have the same current.
R1 is much smaller than R2 and the voltage drop can be safely ignored.
For a 600mV voltage drop (For Q2) over R2 there must be a current of 600mV / 560 = 1mA through R2.
Note that this 1mA is already 3x greater than the 300uA needed for the base current of Q2.
This is done on purpose to minimize problems with leakage currents mentioned above.
There must be decent current through R2 to open Q2 and R2 will keep Q2 closed otherwise, it eats the leakage path.

When the output is shorted, the base of Q2 is opened hard, the transistor is in saturation.
R4 will conduct a current of (roughly) 36 V / 9k1 = 40mA.
Of those 40mA only one is through R2.
The base of Q2 must be able to handle this 39mA. Or the other way around:
The maximum base current is specified in the datasheet of the transistor, and this gives a minimum value for R4.
Too much current through the base of Q2 and it goes up in smoke.

Previously we've calculated that Q2 will stay fully open at roughly 300uA base current.
Together with the 1mA through R2 that is a total of 1.3mA.
That 1.3mA has nowhere else to go but through R4, and the voltage drop over R4 is in that case: 1.3mA * 9k1 = 1.3*9.1 = 12V.
This means that as long as the voltage on the output is below 36V - 12V = 24V there will be plenty of current through R4 to keep Q2 open, and therefore also keep the "fuse" in the tripped position.

When the voltage on Vout start rising above 24V then the current through R4 keeps on decreasing.
At some point the current trough R4 will not be sufficient anymore to sustain both the 1mA current through R2 and the base current through Q2. When the voltage on Vout keeps rising, Q2 gets less and less base current and it goes out of saturation.
As soon as Q2 is out of saturation R3 will start pulling the base of Q1 (wipe away the tipex) lower.
When Q2 is pinched enough then R3 will start pulling current through the base of Q1 again and the fuse is reset.
The exact values where this happens are not so easy to calculate by rules of thumb and factors such as the Hfe of Q2 (with it's widely dispersed production tolerances) become significant.

-----8<--------8<--------8<--------8<--------8<--------8<---
One of the most obvious assumptions here is a "fixed" 600mV threshold Base Emitter voltage for transistors (diodes).
PN junctions have a temperature cooficient of roughly -2mV/celcius, and if Q2 is mounted close to power resistor R1 it can become quite hot. This can easily lower the trip current by 40%.
If you don't want this, simply make sure that Q2 is not near R1.
If you want to go fancy, then you can also make good use of it.
Normal fuses have an i2t value, and can be overloaded for a short time without tripping.
By making a thermal bridge between R1 and Q2 you can simulate this i2t behaviour.
This will have the added benefit that the circuit trips at a lower (steady state) voltage drop over R1 and therefore will also have lower losses of it's own.

## Attachment(s):

Doing magic with a USD 7 Logic Analyser: https://www.avrfreaks.net/comment/2421756#comment-2421756

Bunch of old projects with AVR's: http://www.hoevendesign.com

Last Edited: Mon. Feb 18, 2019 - 08:09 PM

DocJC wrote:
(PNP pass transistor, or better yet a PFet perhaps).
or better is a low saturation voltage PNP pass.

Is the common-mode voltage less than 27V?

Is the continuous current less than 6A?

If yes and yes then

copied from TND6093 - Low VCE(sat) BJT's in Automotive Applications (page 5)

Replace the over-voltage detector with an over-current detector (CSA, 431 as a comparator, feedback from Q2's collector for hysteresis)

Don't know if that circuit is low price and low cost.

An IPS may be simpler (an Integrated Power Switch has a CSA, over-current limiting or switching, over-temperature to protect the power FET, some reach to 24V automotive so 30V max)

NCS213R: Current Sense Amplifier, 26V, Low-/High-Side Voltage Out, Bidirectional Current Shunt Monitor

NCP431: Voltage Reference, Low Cathode Current, Programmable, Shunt Regulator

"Dare to be naïve." - Buckminster Fuller

@doc

I wanted to write that it is not needed to shunt current around the ADS712 because some versions of the ACS712 can handle high enough currents.

The "higher current" version is the ACS758 which can measure upto 50A and looks like:

Just before that however, I bumped into an application note from Allegro where they use PCB traces to shunt current around the sensor:

https://www.allegromicro.com/en/Design-Center/Technical-Documents/Hall-Effect-Sensor-IC-Publications/Current-Sensor-ICs-In-Current-Divider-Configurations.aspx

Doing magic with a USD 7 Logic Analyser: https://www.avrfreaks.net/comment/2421756#comment-2421756

Bunch of old projects with AVR's: http://www.hoevendesign.com

Depending on your needs, some high & low side drivers offer built-in current limiting & thermal protection.  So they aren't exactly breakers, they are in between current limiters & breakers.   The thermal recovery can be very "fast", though I don't have a specific number (prob depends on pcb cooling it back down).   I've seen a few drivers (very few) that allowed you to set the trip point.

I've used these ZXMS6005 low-side drivers (& their entire ZXMS family) on numerous occasions...they enjoy being shorted & come back to life in the blink of an eye

https://www.mouser.com/datasheet/2/115/ds32249-356086.pdf

Compact High Power Dissipation Package

Low Input Current

Logic Level Input (3.3V and 5V)

Short Circuit Protection with Auto Restart

Over Voltage Protection (Active Clamp)

Thermal Shutdown with Auto Restart

Over-Current Protection

Input Protection (ESD)

High Continuous Current Rating

Here is a high-side example

https://www.onsemi.com/pub/Collateral/NCV8460-D.PDF

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

DocJC wrote:

It's looking like the easiest way to do this is to use something like an ACS712

Cool chip.

Digikey lists both the ACS712 and ACS758 as "Not For New Designs".

Is the design for one unit now or many units into the future?

Is it for AC or DC current?

Is the trip point 1 amp?

What is the max voltage?

Brian Fairchild wrote:

...

Yes, happily lose up to 5V

...

I have no control over the source. If I try to implement any local current limiting then the power dissipation is my problem. The voltage could be up to 36V, the potential current up to 10A or more.

An ideal device would be something like a circuit breaker as fitted to your house but one which resets itself automatically after 100ms.

' 10A or more' needs to be defined - does that mean actually 15A, 20A, 50A, 100A ??

'up to 36V' gives a top end, but what is the minimum voltage this needs to operate at ?

If you are ok dropping 5V, that's 50 watts at your 10A ..

There are Electronic Circuit Breakers and Hot Swap Controllers that use external MOSFETS - search on those ?

Do you need to break +ve side, or does that not matter ?  Does it need reverse wiring protection ?

There seem to be some -48V Hot swap controllers targeting telecom applications.

For simple, low threshold voltage sense, there are some Linear LED drivers that have low voltage sense pins  - eg  AL5815/AL5816 have 200mV ref's.

One of those and a simple monostable + PowerFET, could give you a hiccup type limiter.  That defaults to on, and fires an OFF for 100ms on overcurrent.

If that overcurrent persists, it fires again....

I've also used these driver years ago for current protection..they worked pretty good

7V to 32V operation

• Less than 1μA standby current in the “OFF” state

•Available in small outline SOIC packages

•Internal charge pump to drive the gate of an N-channel

power FET above supply

•Internal zener clamp for gate protection

•60μs typical turn-on time to 50% gate overdrive

•Programmable over-current sensing

•Dynamic current threshold for high in-rush loads

•Fault output pin indicates current faults

•Implements high- or low-side switches

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

All those ready made chips take the fun out of tinkering a bit with analog electronics

Doing magic with a USD 7 Logic Analyser: https://www.avrfreaks.net/comment/2421756#comment-2421756

Bunch of old projects with AVR's: http://www.hoevendesign.com

Paulvdh wrote:
All those ready made chips take the fun out of tinkering a bit with analog electronics

Way cool thread!  So much to try out!

Jim