Help to understand automotive load dump protection for 24V

[dmg]

Hi,

A while ago I got involved in an automotive design that operated at nominal 24V. I opened a thread here because after a lot of reading my head was a mess and I thought you could help me to pick a good strategy to protect my system against 24V.

Here's that thread:
https://www.eevblog.com/forum/projects/automotive-dc-protection-(load-dumps-and-such)/msg851267/#msg851267

My requirements were:
-Power input protections against ISO16750-2 pulse A and all other pulses from ISO16737.
-ABSOLUTE MAXIMUM operating voltage of 40V (any protection should clamp below that)
-Average nominal current consumption of the device around 1A, max 5A.
-Customer doesn't know about automotive standards, final user's vehicles might or might not have celtralized load dump suppression.

So, considering my requirements, let's imagine a load dump. According to standard, as no conditions for testing have been agreed, I should withstand either max voltage with highest Ri or min voltage with lowest Ri.  Worst case would be min voltage with lowest Ri so my pulse will have these parameters:

Un = 24V (nominal voltage)
Us =151V (peak voltage)
Ri = 1ohm (alternator resistance)
tr = 10 ms (rise time)
td = 350 ms (worst case)

Under these conditions, and considering that my device should withstand 10 in a row speced 1 minute each, the protection circuitery must clamp below 40V. Also, due to the thing being for a 24V system, it the protection system can't trip below 32V.

I opened the thread and several people suggested some protection devices such as TVS. I did calculations and all TVS models suggested by people would fry, my calculations told I would need somewhere between 4 to 8 TVS in parallell to absorb a full load dump while keeping all my other requirements. Yet some people still told me that midrange TVS's would do the trick.

Back then I utlimately chose to implement a disconnect mechanism that would isolate the load during the load dump, as the numbers I got from calculations showed that absorbing a load dump (let alone 10 in a row) would be almost impossible. Also, among all the literature, papers and appnotes I've read I have yet to find a design example for a 24V system. Most papers mention the 24V requirements but only give an example for a 12V system where numbers tell that a beefy nail-sized TVS is enough for the worst case, but energy absorption requirements for a 24V system are about an order of magnitude higher.

My design ultimately worked but now the customer want's a derivate design, same restrictions but in this case I have to keep operating during most transients, so load disconnect approach won't work anymore. Do you have any design example with calculations or any advice regarding 24V load dump protection with my requirements, so I can check if I'm doing things well or if I'm missing something?

Please take into account that, as my device will take about 1A nominally, and could potentiallyand temporarily need up to 5A, that I have quite some restrictions with the amount of filtering I can provide, as not too massive inductors that can handle 5A DC are scarce and low value (up to 10 uH or so).

Thank you very much for your help.

PS. Before anyone points it out, yes, this is for a paid job.  And no, I don't have any coworker that knows better than I do about this stuff, we are specialized in radio for things that aren't vehicles... this is an outlier job that started as something different and turned into this.

[Siwastaja]

How about not trying to shunt all the energy, but instead design a pass-transistor type of regulator that can withstand the high voltage and operate normally? This way, the dissipation requirement is orders of magnitude smaller, as you only need to dissipate V difference times the current consumption of your load, not the full current the alternator can provide.

[Jeroen3]

Ask your local automotive compliance testing firm. They own a tester, they have seen designs. They know what you will need to pass certification.
Once you have a signed document stating your product passed ISO .... test, you can tell te customer to fit supression devices in their verhicles.

[Andreas]

Hello,

you might re-think your requirements.
The gap between 32V and 40V is too small.
That makes a high effort for the protection cirquit.
Think of a pre-regulator which can withstand around 60V
E.g. LT1074HV or similar  + a large varistor.

with best regards

Andreas

[dmg]

Hello,

you might re-think your requirements.
The gap between 32V and 40V is too small.
That makes a high effort for the protection cirquit.
Think of a pre-regulator which can withstand around 60V
E.g. LT1074HV or similar  + a large varistor.

with best regards

Andreas

Yes, I'm aware that this is the main part of the problem. In fact, 24V systems should as far as I know be designed with working range up to 60-65V in mind. The thing is, there's an IC we use to drive some actuators that packs a ton of functionality in it. It can work up to 38V nominal and 40V absolute maximum. We talked to the manufacturer about suitability of this IC for a 24V system and their answer was something like "It's specified for 12V systems, 24V is probably doable but you're on your own". So yeah... My pourpose with this post is to see if someone can give me some insight on how to tackle this problem (how to formulate, simulate and analyze the circuit to check if the protection is suitable) because I may be missing something important. When a lot of people tells me it's doable with midrange TVS's but my calculations say it's not then I'm probably wrong and I wanna know why.

How about not trying to shunt all the energy, but instead design a pass-transistor type of regulator that can withstand the high voltage and operate normally? This way, the dissipation requirement is orders of magnitude smaller, as you only need to dissipate V difference times the current consumption of your load, not the full current the alternator can provide.

Yes, I considered that too. For this, I did some calculations according to a Linear Technology datasheet that told you how much Safe Operating Area does your pass transistor need to have in order to withstand a surge. The conclusion was that for 12V systems a DPAK mosfet can withstand a whole worst case load dump while regulating to 40V. For a 24V load dump the required SOA jumps to impractical figures. I have to study this more in depth, because I can shutdown during a load dump which is infrequent but this method could get me past some lower energy, shorter transients.

[Jeroen3]

A current limiting resistor and a zener diode. Or is the full 5A going into the Vdd of the chip?
Like they do here.

[filssavi]

If cost is not too much of a concerne you could use a lt4366, it isa controller that uses a pass transistor nMOS to drop the voltage in eccesso from a user settable threshold, ithas a timeout (9 seconda i think) so that if the overvoltage is too long it shuts down to protect the MOSFET and it is available in manual and self rearm variants