Over voltage protection circuit for 14v battery system

[OwenH]

I am designing a battery-operated device that acts as a passthrough between a battery and a camera. The device requires overvoltage (OV) protection to prevent damage to the camera from a user installing a 26v battery with the same mechanical interface as a 14v battery. I have created a working prototype OV protection board using the MAX16126 Load-Dump/Reverse-Voltage Protection IC. This chip is a gate driver for back-to-back MOSFETs and comparators that detect overvoltage. The version I selected has an auto-retry feature.

My concern is how to handle situations where the circuit trips and the camera may brown out before the circuit resets. The delay before retrying is 150 ms, so there is a risk of this happening. I am also aware that there may be specific use cases where nuisance tripping could occur, such as a high load at startup causing the voltage to drop below the UV trip point of the circuit or an overshoot from a switching power supply.

I chose to use an IC for this protection because it offers a low-resistance solution with MOSFETs that is very compact and integrated. I am seeking input and experience from others who have implemented this type of protection and its related pros and cons.

Thanks,
Owen

[BeBuLamar]

Perhaps a single chip linear voltage regulator? Depending on the current needed it could be hot dropping almost half of the voltage in the regulator.

[JustMeHere]

Check this out:

 Load Dump and Cranking Protection for Automotive ... https://www.ti.com/lit/pdf/snva681

[OwenH]

I guess ideally what I ideally need is to check for overvoltage when the battery is attached then stay latched on until the next power cycle.  I don't actually need TVS protection for the camera as its already built in I just need to protect from over voltage from the installing the wrong battery.

I already have this design and it appears to work exactly to the data sheet. I guess I'm just afraid someone will plug in an accessory that could cause it to think there is a fault due to inrush current.  Maybe it will be ok if I make the UV threshold very low so that when the voltage drops on inrush it won't trigger the MOSFET switch. 

Is it reasonable to think that there shouldn't be any transient voltage that would cause a nuisance trip when powered by a battery?  I can understand where a switching power supply could cause a problem.

[JustMeHere]

Check out the SCR Crowbar circuit.  It's for OV protection

https://www.electronics-notes.com/articles/analogue_circuits/thyristor-scr-triac/overvoltage-protection-crowbar-circuit.php

[Kim Christensen]

You could power it via a Boost Buck Converter and not worry which battery the user connects.

[Psi]

Have you considered a polyfuse and a TVS/zener to clamp and trip the polyfuse.

[OwenH]

Thank you for all the suggestions. I’m afraid that a buckboost won’t work considering I need to support up to 18 amps. I guess I should have mentioned that. I’m also space constrained leaving me with about 24x24mm of board space.  I think the idea of a poly fuses is problematic given the current.  A crowbar with a standard fuse might work but wouldn’t be resettable. I also wonder if a crowbar with a TVS would be problematic considering the 14 v battery has a voltage range of about 11v to 17v and the 26 volt battery can be from about 20v to 33v.

[coppercone2]

I don't think there is a good way to do this. Fuse + return to manufacturer for improper use. Why are some yahoos plugging in random crap ?

Most battery powered devices trip a fuse (good) or break if you give them the wrong voltage. its not like standard for this not to happen. Its standard to damage them, and it makes sense for a BOM standpoint. I think if you fuse it so the device does not fail catastrophically you are already at a high quality level.

seriously, even for non technical people, "too much juice = broken". I think its over rugged and you are damaging your budget implementing that.


Does anyone expect a device to function normally after a 200% voltage over load from a properly connected power source? People expect things to survive some spikes or out of spec stuff, but this is like gross overload. Like no one thinks its a good idea to put a 24V battery in a 12V car. Like having battery chemistry differences change it by 20%... ok... thats being nice, but double voltage?!@

Label the input to say "accepts 12V battery". If you can't get the label changed, something is really wrong and people want you to work miracles, and you got instructions from a moron.

And having a weird circuit that does this non standard thing might end up costing more returns then a few mega negligent users, even if it was some how made. Very unnecessary and unorthodox and rather complex for a battery input.

[Psi]

Presumably the issue only exists on first connection and if the battery voltage is too high you just don't do anything.

So if you add some delay before series pfet turn-on then you have time to detect if the voltage is too high and block the pfet from turning on before the delay is over.
If the battery voltage is too high it never connects the pfet and you never have/need any retry system.
It will just stay disconnected until you remove the battery and power it up next with a lower voltage battery which would start the timer again.

A automotive load dump chip is not really suited for this situation because load dump can happen at any point during operation. It needs a retry system to reconnect after the issue resolves. Where as using the wrong battery only happens on first power on and you must remove the battery to install the correct one.

Unless i misunderstood the issue?

[OwenH]

I thought about using a Pfet but I want the low resistance of the nfets.  In fact I'm using a pair in order to reduce the resistance even more.  But the concept of using a switch that only looks for the over voltage when the battery is installed is what I ideally would like to use.

I guess I don't have enough experience to know how common a load dump event or transient voltage would be.  These cameras have switching regulators internally to power the camera electronics and include their own protection electronics.  They also have output accessory ports where video monitors, transmitters and other devices can be plugged in. So an accessory with a large capacitive load at startup I fear might cause the Load dump IC to trip. Unfortunately I don't have regular access to these camera as they are in the 50k to 80k $ range.

Could I use a simple Crowbar with a TVS and a fuse? I was under the impression that a TVS might not start conducting fully at the voltage I need.  I would need to allow <17v and start blocking at anything >17v.  Maybe with a very large 17v Zenor diode? It would need to be able to trip a 20amp fuse.

[BeBuLamar]

If you want to devise a circuit that would work I have no idea now. But if you want to solve the problem of people installing wrong battery then perhaps you should looking into these.
1. How do people have batteries of wrong voltage? Because your company use both and they are available and easy to get?
2. If so what if you devise the connection system where you can not connect the wrong battery. Because there is still some risk although not as great by installing the 14V battery in a 26V device.
3. Of course the best approach is to educate the people. Trying to idiot proof is simply trying to outsmart your people which isn't good for moral.

[Kim Christensen]

I thought about using a Pfet but I want the low resistance of the nfets.  In fact I'm using a pair in order to reduce the resistance even more.

Since the power source is a battery, you could still use N-FETs to implement PSI's idea. Just put them in the negative lead instead and design the trigger circuitry differently.

[Psi]

I thought about using a Pfet but I want the low resistance of the nfets.  In fact I'm using a pair in order to reduce the resistance even more.

Since the power source is a battery, you could still use N-FETs to implement PSI's idea. Just put them in the negative lead instead and design the trigger circuitry differently.

Yeah, just have to be sure battery negative is fully isolated from anything else so nothing external can bypass the nfet.

Might cause issues on stackable batteries, the kind you put on a camera but you can also add modules on the other side of the battery module.

[OwenH]

In regards to batteries with different voltages using the same mechanical interface, this isn't really the case.  When I wrote the original post I simplified the explanation to avoid making the the problem too confusing.  What I have is a pass through mounting device that goes between the battery and the camera.  This device has battery plates that are interchangeable and it could be possible for the user to install a 26v battery plate with a 14v battery plate.

I have been on the fence if I have a responsibility to idiot-proof this or if very clear and permanent warnings on the product itself should be sufficient.  This is a professional product and the user should have the experience to use it properly, but when you have cameras in the $80k range, there is a lot at stake. 

Now it is also true that most of these professional cameras can actually support up to 34v, so it is only a handful of products that would be at risk. I need to do a bit of research, but these 14v only cameras may have OV protection built in. (I think if I was to design an expensive camera with a changeable battery interface, I would consider OV protection as mandatory.)

If I do continue with a OV switch I was wondering about using a high voltage PIC like the PIC16HV753 and utilizing the comparator to check the voltage before turning on the MOSFETs after a delay. It looks like maybe this would work with just the PIC, resistors, TVS diode and a MOSFET gate driver? The PIC chip is actually less expensive than the Load dump IC.   This would likely have a similar part count and cost to the design I made with the load dump protection IC.  I'm curious if the idea of a PIC would be considered over complicating the problem? 

[shapirus]

If I do continue with a OV switch I was wondering about using a high voltage PIC like the PIC16HV753 and utilizing the comparator to check the voltage before turning on the MOSFETs after a delay.

Why would you need a PIC? Unless I miss something, you could use a single comparator (of course rated for the maximum expected input voltage) to compare the input voltage against a reference and turn on the MOSFET switch, or a relay, through which the main circuit is powered only if it does not exceed the allowed value (and not before a power-on delay set by an RC circuit has elapsed). Or even use two comparators to make a window comparator which allows only a given range of input voltage.

up: I went ahead out of curiosity and drawn up a basic (no hysteresis etc. -- may need to add some resistors or schmitt triggers, if necessary) circuit that does this, using a minimal number of components. Simulates well, but I didn't try it in hardware. Is it similar to what you wanted?



(also attached is a Proteus design file so you can play with simulation.)

p.s. now that I am looking at it, the U1B's non-inverting input must rather be fed from a voltage divider between IN and GND with resistor values like 100k for the upper half and 30k for the lower half, instead of REF. Setting it to REF makes the circuit's behavior dependent on the TL431's turn-on time which isn't good, and the capacitor charge time to the REF level will depend on the IN voltage.

p.p.s. since you need to support up to 18A, a relay will probably be a better choice than a MOSFET, however there are MOSFETs with Rds(on) well below any relay's contacts resistance, so it may become a question of cost, availability and size.

[shapirus]

Another update: the circuit shown above is actually dangerous and will not work properly. For a brief moment at power-up, when the U1:C's output transistor is not yet opened, the MOSFET's gate will receive a full input voltage potential which will either fully open it, or destroy it.

R&D in action :).

Need a proper NANDNOR gate or an inverting Schmitt trigger replacing U1:C which will not set high on the output pin until powered up.

...so something along the lines of:



An IC may (or should) be used in place of the two discrete transistors, but I'm not sure how easy or cheap it can be to get one that is rated for at least 30V supply.

A fuse in series with the rail powering the protection circuitry plus a TVS diode connected between the MOSFET's gate and GND rated so as to begin to conduct and blow the fuse at just under 20V will serve as an additional protection of the load, as well as the MOSFET, should the OV protection circuitry fail to work as expected.

[OwenH]

The reason the PIC interested me is that in one IC for $1.20 you have a shunt regulator, comparator, ADC, Voltage reference, and the ability to program in a delay all in a 3x3mm package.  I would have to spend a bit more time familiarizing my self with the PIC and its shunt regulator and see that I could make it work with the voltage range of the circuit.   Of course one down side would be the need to program it.  Am I missing something else that would make this a bad option? I like the idea of minimizing the BOM.

In regards to the MOSFETs I chose to use 2 pairs of Toshiba TPH2R306NH1,LQ which give me a resistance of about 4.6mΩ.  Testing at ten amps gives a temp rise of about 5 degrees without any heat sinking.  In the actual product the PCB has a thermal connection to the aluminum enclosure and the PCB is 4 layer 2oz copper with many thermal vias and copper planes.   

[shapirus]

The reason the PIC interested me is that in one IC for $1.20 you have a shunt regulator, comparator, ADC, Voltage reference, and the ability to program in a delay all in a 3x3mm package.  I would have to spend a bit more time familiarizing my self with the PIC and its shunt regulator and see that I could make it work with the voltage range of the circuit.   Of course one down side would be the need to program it.  Am I missing something else that would make this a bad option? I like the idea of minimizing the BOM.

Then it sounds like a winner, especially considering the space constraints. Yes, it'll need extra human time spent, but only once, for development and proper testing.
Also need to take special care, both on the software and hardware side, that on the power-down->power-up transition the switching MOSFET always has a definite zero Vgs, regardless of input voltage, as well as when the user replaces the batteries of different voltages very quickly (that's a corner case, yet not impossible).

Once powered up and the program is ready to run, you have endless possibilities on setting up any delays and voltage constraints you need.

If it's more than a one-off job, it will probably save production costs, since it's a single component that won't need any special selection or matched passive components to make it work and the program will stay 100% identical from batch to batch, thus reducing the number of potential points of failure.

It will however lack the magical attraction of purely analog circuits.

[Kim Christensen]

Am I missing something else that would make this a bad option?

Well, a PIC with a shunt regulator is going to be relatively power hungry. Not going to be much of an issue when running since the camera draws 18 amps, or if the battery is not left connected. But if you were going to leave it connected to the battery, it may be an issue if stored for a longer time period.
With an MCU like a PIC, you'll need a resistor across the gate-source of the MOSFET to ensure it stays off when the PIC is still in it's reset state. (Pins will be floating inputs on reset, and before the software initializes them as outputs)