Protection circuit for devices powered on a car

[Foaly]

Hi,

I'm making a protection circuit for my car's battery and I'm looking for some advice from EE gurus.
My goal is to be able to power a few devices from my car (most notably my laptop, DSLR camera, and camera charger). I've found a 20A line in the electrical box that is not connected (its for the electrical roof motor, which I don't have on my version of the car), so I'm making a wire harness to use it. Obviously I don't want to damage any electrical circuit in the car so I'm making a protection circuit board with the following features :
- undervoltage cut-off (to make sure I won't excessively discharge the battery and I'll always be able to start the engine)
- overcurrent protection
- reverse current (voltage?) protection (in the unlikely event that a surge happens on the device side, I don't want it to propagate to the rest of the car)
For the first two points I'm using a BTS50080 power switch with a TPS3710 for the voltage detection. For the last point, a simple Schottky diode. The schematic is attached.

My question is, is there an important protection feature that I'm missing, and/or is there a mistake in the way I designed this circuit? I believe the whole thing is pretty simple but you know, there is always something more to learn. And also, I don't want to burn my car 

Thank you in advance for any advice or idea!

[JimRemington]

The electrical system in an automobile can exhibit voltage surges of > 125 V and you definitely need to protect your circuitry against that possibility.

I've been using the recommended TVS diodes for years without problems. The 1.5KE series can handle momentary surges of 1.5 kW and cost < $1.
http://www.littelfuse.com/~/media/electronics/application_notes/littelfuse_tvs_diode_automotive_circuit_protection_using_automotive_tvs_diodes_application_note.pdf.pdf

https://www.digikey.com/products/en/circuit-protection/tvs-diodes/144?k=&pkeyword=&s=15790&imin1729=&imax1729=&imin1730=&imax1730=&FV=ffe00090&mnonly=0&ColumnSort=0&page=1&quantity=0&ptm=0&fid=0&pageSize=25

Add an in line fuse.

[Jeroen3]

Undervoltage will have problems with cranking. Better make a time limit, say max 30 minutes if input isn't >13V. That only resets until the voltage is >13V for 5 minutes again.
Or just buy one of those taxi modules for secondary battery.

[Foaly]

Hi and thank you both for your answers,

The electrical system in an automobile can exhibit voltage surges of > 125 V and you definitely need to protect your circuitry against that possibility.

I've been using the recommended TVS diodes for years without problems. The 1.5KE series can handle momentary surges of 1.5 kW and cost < $1.

I didn't know those TVS diodes so I read about them and I looked at the specs from different manufacturers. For exemple, ONSemi/Fairchild offer the 1V5KE15A with the following specs :



If I understand correctly, this means that this diode will be completely blocking at 12.8V (battery is 12.6V max), start conducting between 14.3V and 15.8V, and never allow more than 21.2V. Ideally, I would have liked to ensure that the output was always below 19V. Is it realistically possible, in the context of a car battery and electrical systems, that such a surge happen that would let this diode have a voltage of more than 19V?

Also, could you confirm this is the correct way to use the TVS?

Add an in line fuse.

I forgot to mention that I already have the 20A fuse built in the car for this line. The BTS50080 also offer built-in current limiting protection. I thought about adding an SMD polyfuse but they didn't come with the power requirements that I needed and I guessed it wasn't necessary anyway?

Undervoltage will have problems with cranking. Better make a time limit, say max 30 minutes if input isn't >13V. That only resets until the voltage is >13V for 5 minutes again.
Or just buy one of those taxi modules for secondary battery.

My first tought was that it wasn't important to loose power on this line when cranking because most devices are battery chargers (laptop, camera...), but after thinking about it I should make it cleaner and take that into account.
What do you think about this circuit then? I added D4 and calculated R2 so that when the input voltage is decreasing, the sense voltage will only decease at the rate of C1 discharging through R2 and R5 (in this case it should take about 1.6s if I haven't messed up the maths, but I could bump up R2 a bit, and it shouldn't prevent 1nA from flowing to the SENSE pin).

[Siwastaja]

That diode clamping at 14.3V might conduct slightly all the time when the car is running. Pick the next one for some margin.

Look up "load dump".

Use series resistance before the clamp. Calculate R so that you don't get too much voltage drop preventing the operation of your devices during low voltage, but at the same time, make sure it limits the current to levels the clamp TVS can withstand even if Vin is, say, 100V, for tens of milliseconds, which is possible with load dump. Also make sure the resistor can handle the power. You can use a PTC "polyfuse", you can even thermally couple it to the resistor so that it protects for both overcurrent and overtemperature.

This resistor is also important for damping any LC parasitics which might generate or amplify existing spikes.

Doing non-cost-restricted one-offs is easy, just use a beefy, expensive TVS with a beefy, expensive power resistor and a PTC.

[David Hess]

If you are going to clamp the input then you might as well have a crowbar circuit because if it is effective, it will blow the fuse.  For that reason I would use series protection instead of shunt protection.  A micropower boost converter or even switched capacitor inverter can provide the bias supply to power a source follower so any load dump fault gets removed.  The same source follower may provide current limiting and low voltage shutdown.

I would not worry about reverse current protection.  The impedance is so low that the fuse will blow before damage can be done by this.

[Foaly]

@Siwastaja : I tried to calculate such resistance but I might be doing it wrong because I can't get it to work. Limiting the current to 67A during a 150V surge requires minimum 2.5 Ohm resistance, but limiting the voltage drop to 2V at 12V and 10A requires maximum 0.2 Ohm. And anyway, those 67A would blow up the 20A fuse.

I'm wondering (and this is for David's answer as well) if the car doesn't already provide such surge clamping protection inside the electrical box or even before that? Power electronics is not my area of expertise (as you probably have gathered by now) but it surprises me that such surges could happen on the +12V rail anywhere in the car, when there are dozens of ICUs, sensors, and the accessory socket connected to it without apparent protection of this scale.

Furthermore, if I understand correctly those surges mainly happen when the battery is disconnected from the alternator so that should happen fairly often, and it shouldn't blow the fuses every time (be it because of thr TVS or the crowbar circuit).

By the way, again if I understand correctly, those load dump surges happen at the alternator sides when the load (battery) is disconnected (because of the high charging current and the large alternator inductance). But when this happens, my circuit is connected on the battery side, not the alternator side, so it shouldn't see the surges?

[David Hess]

I'm wondering (and this is for David's answer as well) if the car doesn't already provide such surge clamping protection inside the electrical box or even before that? Power electronics is not my area of expertise (as you probably have gathered by now) but it surprises me that such surges could happen on the +12V rail anywhere in the car, when there are dozens of ICUs, sensors, and the accessory socket connected to it without apparent protection of this scale.

The primary protection on the car's alternator and battery provided +12 supply is the low impedance of the battery.  But batteries may fail either by being disconnected or having a cell connection crack and this happens often enough that all of the electronics must be able to handle the power spike produced by the alternator.

Furthermore, if I understand correctly those surges mainly happen when the battery is disconnected from the alternator so that should happen fairly often, and it shouldn't blow the fuses every time (be it because of thr TVS or the crowbar circuit).

Exactly, the fuse is there for safety and not for handing a load dump.

By the way, again if I understand correctly, those load dump surges happen at the alternator sides when the load (battery) is disconnected (because of the high charging current and the large alternator inductance). But when this happens, my circuit is connected on the battery side, not the alternator side, so it shouldn't see the surges?

The alternator and battery are connected in parallel.  When the battery is removed, the car systems are still connected to the alternator.

[Foaly]

Okay I had misunderstood then, I thought there was some kind of dedicated charging circuit that cut the alternator from the battery (through a relay or something) when the battery was full to prevent overcharging it (the car systems being powered on the battery at this time), and connected it back again when the voltage fell below some threshold.

In this case, this condition should only happen when the battery breaks or is manually disconnected when the car is running, and it is normal for the fuses to blow in this situation? Why isn't there a general, large clamping protection system in parallele on the 12V bus to protect all the electronics? Otherwise, I can't see why my phone charging on the accessory socket wouldn't fry in this situation (I'm pretty sure the cheap car USB charger doesn't provide beefy TVS and resistors protection).

Thanks again for your time 

[David Hess]

Okay I had misunderstood then, I thought there was some kind of dedicated charging circuit that cut the alternator from the battery (through a relay or something) when the battery was full to prevent overcharging it (the car systems being powered on the battery at this time), and connected it back again when the voltage fell below some threshold.

The output from the alternator is directly connected to the battery.  The charging current is controlled by adjusting the current through the alternator's field.

In this case, this condition should only happen when the battery breaks or is manually disconnected when the car is running, and it is normal for the fuses to blow in this situation? Why isn't there a general, large clamping protection system in parallele on the 12V bus to protect all the electronics? Otherwise, I can't see why my phone charging on the accessory socket wouldn't fry in this situation (I'm pretty sure the cheap car USB charger doesn't provide beefy TVS and resistors protection).

The fuses do not normally blow if the battery is disconnected and there is no surge suppression at least on any of the vehicles I have worked with.  All car loads have their own internal protection and there are automotive standards for what is required.  The USB charger has some sort of series protection like I described.

[Foaly]

Okay, understood. In this case I should do like the USB charger and use only a series protection? I don't understand what you refer to by saying

A micropower boost converter or even switched capacitor inverter can provide the bias supply to power a source follower so any load dump fault gets removed.

Could you link an example circuit that would be appropriate in my case?

[Siwastaja]

@Siwastaja : I tried to calculate such resistance but I might be doing it wrong because I can't get it to work. Limiting the current to 67A during a 150V surge requires minimum 2.5 Ohm resistance, but limiting the voltage drop to 2V at 12V and 10A requires maximum 0.2 Ohm. And anyway, those 67A would blow up the 20A fuse.

Oh, then you need 0.2 ohm and need to make sure the TVS is rated to survive until the fuse blows. Self-resetting polyfuse would be a good choice because replacing fuses sucks. That fuse should provide the 0.2 ohm already so no need for a separate resistor.

Maybe the shunt type protector (TVS crowbar + resistance + fuse) is not the right choice for this job, since you have high power requirements, and maybe you should look into series protection.

I'm wondering (and this is for David's answer as well) if the car doesn't already provide such surge clamping protection inside the electrical box or even before that?

Usually not. Each module is protected separately, typically, unless you know better for some specific case.

Power electronics is not my area of expertise (as you probably have gathered by now) but it surprises me that such surges could happen on the +12V rail anywhere in the car, when there are dozens of ICUs, sensors, and the accessory socket connected to it without apparent protection of this scale.

It's just quite an arbitrary design choice that each module includes its own protection. Granted, it could be centralized, but it would require a separate protection module, whereas when each module protects itself, it can be often integrated simply on the same PCB.

Furthermore, if I understand correctly those surges mainly happen when the battery is disconnected from the alternator so that should happen fairly often, and it shouldn't blow the fuses every time (be it because of thr TVS or the crowbar circuit).

Load dump happens when the alternator disconnects from the battery, and this is not very usual, definitely not normal "everyday" operation, but it can happen. Yeah, I think it's OK if your device resets during this, but it's annoying if you need to manually replace fuses.

By the way, again if I understand correctly, those load dump surges happen at the alternator sides when the load (battery) is disconnected (because of the high charging current and the large alternator inductance). But when this happens, my circuit is connected on the battery side, not the alternator side, so it shouldn't see the surges?

Yeah, if you somehow really physically install battery posts electrically and mechanically separated from the normal battery posts, this could be the case, but I highly doubt so.

For example, if an oxidized battery terminal connection lifts during a bump, then you have the whole electrical system connected directly to the alternator - even though the current goes through the battery connectors, but they are not connected to the battery. Hence the load dump.

[Siwastaja]

Okay I had misunderstood then, I thought there was some kind of dedicated charging circuit that cut the alternator from the battery (through a relay or something) when the battery was full to prevent overcharging it (the car systems being powered on the battery at this time), and connected it back again when the voltage fell below some threshold.

Lead acid technology absolutely requires this rather continuous charging to higher-than-open-circuit-voltage for good battery life, so it's not called "overcharging". Overcharging would happen well above 14.5V. The charging regulator circuit (keeping the voltage around 14V) is integrated in the alternator itself, but it cannot react quickly enough if the battery disappears suddenly, hence the load dump. Even if there was a mechanical relay, it would have some delay too.

[Siwastaja]

Why isn't there a general, large clamping protection system in parallele on the 12V bus to protect all the electronics?

Then it should be dimensioned to handle a much larger power, since it needs to support the combination of all the possible loads peaking at the same time. Then, it would also protect things that don't need protection - mainly the starter motor. Designing a low-drop 1000A protection is not going to be cheap.

Also, there's the history. Since old cars didn't have electronics that required protection, there was no protection. Piece by piece, more electronics was added, so it was logical they protect themselves and don't require external modifications, for compatibility.

In the end, how it works now, it's a distributed system. The distributed nature of protection is not a poor choice at all. It probably isn't any more expensive to do this at the module level, it's more robust, always backwards compatible, and protects against unexpected transients that somehow get coupled in the cabling. And if one device has a fault in its protection
circuit design, only that device is affected.

Otherwise, I can't see why my phone charging on the accessory socket wouldn't fry in this situation (I'm pretty sure the cheap car USB charger doesn't provide beefy TVS and resistors protection).

It won't blow exactly because it has the protection (not necessarily using the resistor + TVS approach, probably an integrated series type solution is much cheaper), unless it's utter crap completely broken by design. Automotive protection is well specified and standardized, this is a normal thing in designing anything for a car, a lot of relatively cheap ICs available that state "load dump" compatibility as the main marketing point on the first page of the datasheet.

If it isn't protected against these well documented phenomena, then it's not a compliant car accessory. Does happen with $1 Ebay things.

[HwAoRrDk]

Each module is protected separately, typically, unless you know better for some specific case.

I once tore down a power window control module for my car, just out of curiosity, and was surprised to find that the module didn't appear to have any power input protection apart from a series polarity protection diode and a low-value series resistor. Just power straight in, through those two components, and into some kind of 5V regulator. Perhaps my model of car does indeed have some kind of centralised protection. Either that, or the manufacturer considers something relatively low-worth as a power window module to be 'expendable' in case of serious load-dump.

[Siwastaja]

Or, the "some kind" of 5V regulator is specified for automotive spikes & load dumps. Widely available and only some ten cents more expensive than your "typical" 7805.

This also explains: why do it on module level?

Well, compare a load dump compatible 5V regulator to a 7805: it only needs a transistor rated to slightly higher breakdown voltage (such as 100V, instead of, say, 40V), and some simple low-current control circuitry that keeps that transistor off and keeps it inside the SOA. Not much more complex than what 7805 already has. So, instead of 5 cents to manufacture, it costs 6 cents. Add some automotive qualifications and paper work, and the retail price for the regulator may be 50 cents instead of 20 cents. Not a big deal.

Doing a "centralized" version would require a 1000A transistor and probably cost at least $100.

[jbb]

I'm a little late to the party, but something like the LTC4364 hits many of these points.  It offers the following:

• Surge stopping to strictly limit Vout
• Reverse current blocking
• Electronic circuit breaker function
• Undervoltage cutout

It is unfortunately a little expensive (NZ$14 in singles from DigiKey), but how much is your design time worth?

[David Hess]

Could you link an example circuit that would be appropriate in my case?

I did not find the exact circuit online but figure 2 here shows the idea.

Since you would want low power operation, the bipolar transistor is replaced with a power MOSFET which requires no gate current.  The problem then becomes supplies a lightly higher voltage than the input because the MOSFET requires a gate voltage a couple volts higher than the output.  Some MOSFET gate driver ICs include a charge pump and could be used.

Or a depletion mode power MOSFET could be used to simplify the circuit but they are more expensive and not all that common.

[Foaly]

@Siwastaja : Thanks for the detailed explaination, I understand a lot better. Indeed, I didn't try to search "load dump" into a parts distributor before but it gives lots of ICs designed to withstand these kind of conditions. There doesn't seem to be a lot of general-purpose protection chips though, unfortunately. I found the MAX16128 which looks like it could do the trick but I still need to dive more into the datasheet to understand how it works.

@jbb : Thanks, the LTC4364 looks a lot like the MAX16128 that I found. It's more expensive but it's available at my usual supplier, while the Maxime is not. I'll compare the specs of both.

@David Hess : I won't pretend I understand this circuit in details but I get the general idea. The rest of the article is also interesting on its own. I think I'll stick to one of the chips above for simplicity; a few more bucks for a protection chip is nothing compared to the expensive laptop and camera that will be connected to it.

[David Hess]

@David Hess : I won't pretend I understand this circuit in details but I get the general idea. The rest of the article is also interesting on its own. I think I'll stick to one of the chips above for simplicity; a few more bucks for a protection chip is nothing compared to the expensive laptop and camera that will be connected to it.

There are a bunch of automotive specified power ICs which might be suitable.

One of the first audio amplifiers that I designed and built from scratch used a 16 volt CMOS operational amplifier that died to voltage surges from the +12 volt automobile supply within a couple days so I ended up adding a resistor and zener diode to protect it which was enough because of the low supply current involved.

[Foaly]

Ha, that means those surges happen really frequently then. I'll definitely need to test the output of the circuit with an oscilloscope before connecting anything valuable to it. 

zener diode to protect it which was enough because of the low supply current involved

I'm not sure I see the relationship between these two facts -- is it because, due to the fact the op amp needed only a small current, you could have a relatively large-value resistor on the input and dampen the spike into the zener this way?

[David Hess]

zener diode to protect it which was enough because of the low supply current involved

I'm not sure I see the relationship between these two facts -- is it because, due to the fact the op amp needed only a small current, you could have a relatively large-value resistor on the input and dampen the spike into the zener this way?

The current requirements were so low and consistent that a simple zener shunt regulator was sufficient.  In this particular design, the discrete output stage had voltage and current gain so there was no need for a large output voltage range or current from the operational amplifier itself.  If more power was required, then a zener shunt regulator would not have been suitable.

The operational amplifier was a TLC272 or something similar (1) with a nominal maximum operating voltage of 16 volts and an absolute maximum operating voltage of 18 volts.  Originally the whole audio power amplifier had a single polarity protection diode on the positive supply feeding a big aluminum electrolytic capacitor but within a couple weeks, the operational amplifier died presumably do to a voltage spike.  So I added a resistor and 1 watt 15 volt zener diode to only the supply for the operational amplifier and never had another problem.  Under normal conditions where the 14.2 to 14.5 volt charging voltage was reduced to 13.7 to 14.0 volts by the polarity protection diode, the zener diode would never even conduct.

I have designed and used input voltage protection circuits like I described with a bias supply and power MOSFET before to good effect but they are more complicated.  Done right, they can be practically foolproof and even protect themselves.  Automotive designs usually get by with something a little simpler although aircraft and military designs might not.  If you can live with it, a PTC resetable fuse and power zener diode can be a good choice.

(1) I am not recommending the TLC272 for anything in particular and it is a pretty mediocre part.  It was just what I had in my free samples collection at the time and the overall design was not picky.  An LM358 would have worked and likely survived any voltage spike do to its 36 volt absolute maximum rating.

[Foaly]

Thanks for the explanation David, I understand better.

So to conclude in case someone comes across this topic and is interested, I ended up starting from scratch and implementing jbb's suggestion, the LTC4364. It not only offers lots of protection features but also completely replaces my original design by providing an integrated UVLO and current limiter. It can even act as a regulator to clamp the voltage to a predefined value when a surge happens (as opposed to simply cutting the current path, which would reset the load). To be more precise I closely followed the schematic in Figure 1 (page 11 of the datasheet : http://www.farnell.com/datasheets/1648693.pdf) which looks like it could handle pretty big surges. The most difficult part now is waiting for Oshpark's package...

Thanks everyone for your help!

[floobydust]

Careful with the LTC surge stoppers, they have weaknesses.
Jr. engineers designed them-in on 24V truck-powered systems and products started failing.
Traced it down to -ve voltage transients destroying the IC and mosfet. I wouldn't use them again.

Instead I use 6.6kW automotive TVS found in ECM's: 8SMA27.
It's big enough to take load-dump and clear a 15A fuse if it shorts.

[NiHaoMike]

I wonder how things are going to change now that more and more cars are using switching power supplies to supply the 12V rail.
http://techno-fandom.org/~hobbit/cars/ginv/i1mech.html