Power input protection variants

[crypton]

Hei.
Couldn't find topics with similar keywords, maybe there exists smth similar?

But nevertheless, I've been crunching my head around a certain power input protection. Don't have much PCB space to use (as usual ).

My idea is to have a transient overvoltage protection on the main DC power line - to dissipate exess transient voltages so they wouldn't go over 50V.
Constant power @42V is expected (for higher Vin I can expect more current consumption. But I don't expect functionality above 50V constant DC).

Input range
(7-10V) 12V - 42V (Buck converter can go up to 55V, but I use 50V input caps).

System (accumulated from different chips' datasheets) could consume 2.5W max peak (assuming ALL of the chips consume max amount of current at the Same time - unlikely, but that's the top-notch max I could put together).
I assume system to run on much-much lower current at nominal ~250mW or smth like that (~20mA @12V / ~6mA @42V).

I have sketched a power input circuit and would like to discuss ideas - the good and the bad about current approaches and how would it be good to solve such a situation.
(For me input voltages go over what I can test in real life right now, so I've mostly thought about these things only in theory).

Circuit: (any way to embed images into posts from attachments?)


My trail of thought:

Imagine, someone hot-plugs in 42V battery pack or smth similar with high inrush current capability.
I imagine having some voltage peaks from such connection - how big exactly -  beats me. I've used 55V overvoltage as an example.
So I focus currently keeping everything under 50V.

1) First thing to clamp the voltage is some 48V TVS (D2).
The resettable fuse (F1) would be a current limiter for the TVS and in the end - against dead short in the system. - Although at it's values (140mA Ihold, 300mA Itrip, 60V), this PTC can provide protection from low voltages as then currents can reach that high. With higher voltages the current consumption is not so high and PTC fails to protect.

At worst (if transient ever goes that high) the voltage after this point is 77.4V (clamping voltage of TVS). I've assumed it probably won't go that high and took the breakdown+margin so ~55V after this point.

2) D1 would be protection against applied reverse voltage.
But to me, it seems I can still short the input from TVS and thus should use bidirectional TVS ?.

3) Since I still might have voltage over 50V I should provide means to supress this energy a bit further. For that I have used a Zener (D3).
Problem with the zener is that I need a certain current, to get it to regulate @47..48V and still not to blow it and also provide sufficient current to the system (due to limiting series resistor R1).

After some trial and error I've taken some intermediate consumption value that I think would be grossly over what system would need at worst case - maybe (not gonna go for the max peak) - and also protect the device. So I've thought ~1.2W would be OK to provide for the system.

So I've found a Zener that is 1.5W. When voltage is over 50V, Zener becomes the load - basically. So I've sized the series resistor R1 so that it limits the zener's current near it's maximum.
Assuming 55V overvoltage (which in this case can be constant).

1.5Wz / 47Vz = 31.9mA -> 30mA, just to be more 'safe'.

In R1 point of view, (55Vin - 47Vz) / 0.03A = 266.6 Ohm -> 270 Ohm.
And thus power through R1 -> 8Vr*0.03A = 0.240W -> 1206 / 1210 SMD package will do.

4) Ok awesome, but will the resistor limit my system ?
Assume we have expected max input voltage, how much can the resistance value allow current for the system ?
42Vin / 270 Ohm = 156mA. @42Vin, system might ask for ~30mA max. This will dance on the limit of R1 power capability, but should manage.
12Vin / 270 Ohm =  44.4mA.
BUT, assuming that at lower voltages, to keep up the demand of 1.2W from the system (that might happen, but might not, as this could also be gross over estimation), the buck converter draws more current from the 12V and thus exeeding my resistor's capability. But does live on my nominal 250mW @12V assumption..

Am I screwed?

Are there any better ways to achieve that kind of supply range and consumption demand or should I consider these things somehow differently?

Thanks, if you've reached so far, much appreciated!

[CopperCone]

You probobly want to put the reverse protection diode in front and make sure that the tvs can survive long enough to trigger your fuse. May not be pretty if your source is capable of 55amps.

A better solution would be a line regulator/circuit breaker ic thats protected upstream by beefy hv devices and a good fuse.

Also 50v caps on a 42 volt rail that can go to 55 is kinda marginal and i personally regard it as crappy. I would use higher voltage capacitors here

Also the inrush current depends on your capacitors parasitics and wire inductance. The benefit of a load switch ic is that you can do controlled current to actively regulate inrush.

As for suggestions on specific efuse, load switch, protection integrated circuits, someone would need to pay me a fat wad of cash to go through the offerings of various manufacurers to make some kind of industry review of the situation. There are tons.

[crypton]

As for suggestions on specific efuse, load switch, protection integrated circuits, someone would need to pay me a fat wad of cash to go through the offerings of various manufacurers to make some kind of industry review of the situation. There are tons.

Sure, sure I am more interested in the sequence logic of the components and parameters to look /match against for. Finding exact manufacturer's components is not the goal.

I've wondered if I'd make the protection so that it would start to clamp voltage and dissipate power already at 42V or so - so basically bring the maximum nominal voltage more down. Because I'd really keep the input below 50V. If someone does use higher voltages, well then one should not apply voltages higher than stated...

[CopperCone]

Check out the ltc4364. It does most of what you want i think. It is very feature packed as far as ic protection goes an other chips will be similar. Some of the even more advanced protection ic will function as a buck converter to ensure uptime during overvoltage events but normally act as a short. Ltc7860 does this.

[crypton]

I started to consider http://www.linear.com/product/LTC4367.

That led me to a similar potential solution - use a 100Vin max capable buck converter instead.
For me LM5008 seems to do the job
 - it has a rather small footprint, compared to other solutions.
 - cost is manageable, compared to the current converter MP2459GJ
 - with the chip's 100V capability, start up ramp and current limit features give added bonuses.

I will still probably have a fuse, a TVS and 100V caps. At least this way I could be more confident having high voltage peaks.


Nevertheless, the initial schematic could also be doable, but then I figured, I would need to bring the max level to about 33Vmax (a TVS with this working voltage, has a max clamp of about 47..48V).
So power capability is definitely crippled from expectations but I guess that's the trade off.

EDIT:
Found a topic discussing similar aspects. Pretty much people talk about same things I have now concluded and more:
https://www.eevblog.com/forum/eda/tvs-diode-layout-considerations-for-automotive-application/