Automotive Load Dump: What % of Vehicles Have a Centralized Clamp Diode?

[TimNJ]

Probably better not to put 24V on the nameplate.  If it's made for 12V-nominal systems, put 12V on the plate -- it's a type rating not an exact voltage.  You don't see 400VDC or 2.5kV peak on commercial equipment (heh, well sometimes you might), the 120 / 240 / 85-265 / etc. rating is a type rating that assumes all the other things that ride on the mains.

It's not really specified that way, it's specified in multiple disparate standards, with a callout for them all from the customer (or whatever the exact circumstances are); but it might be useful to think of it that way.

Tim

Thing is, the original specified input voltage range *implied* use with 12V + 24V automotive systems and 14V + 28V aerospace systems. Interestingly enough(?), the 28V aerospace transient requirements (per DO-160G) are not nearly as brutal as for 24V automotive, so as is, the DUT can handle 12V (auto), 14 (aero) and 28V (aero). It is unclear as to whether the customer actually needs 24V automotive or they just threw it in there because it seemed logical that it should work (since 28V aero is okay). But, that's for me to figure out.

I need to check ISO nameplate conventions but I think you can probably do two separate lines, one for automotive and one for aerospace, similar to how you might handle separate input voltage ranges for 50/60Hz and 400Hz on the same unit.

Thanks!

[TimNJ]

I have implemented the LT4356 form of overvoltage protection using discrete parts.  Having it shutoff can be considered a form of foldback current limiting where the current limit decreases as the voltage across the pass element increases.  This prevents excessive power dissipation in the pass transistor.

Thought about a discrete approach too. But, probably like a lot of other people strapped for time, decided for the IC solution. The discrete approach has the advantage of using easier to get jellybean parts, which is totally valid reason these days. Cost-wise, I assume discrete is cheaper, although depends on how many parts you wind up needing, and how closely really want to match the LT4356 functionality. I imagine a few op-amps, comparators, timers, and flip-flops should do it. Even if you needed 4-5 ICs, if they are all jellybean, probably not more than $1-2 in quantity, compared to $4-5 for LT. Of course, if you want to sleep at night, probably will need to spend a lot more time evaluating edge-cases than you might care to do with an ASIC solution.

[floobydust]

I find a discrete solution has an issue during cranking, the standard for a 12V vehicle are only 4V so a mosfet partially on can be an issue. Otherwise it works fine or just use a BJT. The LT4356-3 has a charge pump for extra gate drive.
If you're switching off the pass element during over voltage events, works well- if the load has the hold time of say 350msec and can keep running during the blip.

edit: did not yet look at Maxim MAX16128/9
A bit more here http://www.flexautomotive.net/EMCFLEXBLOG/post/2016/09/15/automotive-centralized-load-dump-test-requirments

[16bitanalogue]

Apologies if this has been answered, but directly to the OP's question about a "central clamp", not that I have seen from any of the OEMs that manufacture modules; i.e., Bosch, Aptiv, etc. Each module will have EMI and suppression circuits. The instrument cluster PCB will have its own protection circuitry separate from the infotainment module. 

[Hydron]

I find a discrete solution has an issue during cranking, the standard for a 12V vehicle are only 4V so a mosfet partially on can be an issue. Otherwise it works fine or just use a BJT. The LT4356-3 has a charge pump for extra gate drive.
If you're switching off the pass element during over voltage events, works well- if the load has the hold time of say 350msec and can keep running during the blip.

Would be interested in where the 4V minimum comes from, that sounds rather pessimistic. Or is it assuming cranking on an iffy battery in the extreme frozen north somewhere?

[floobydust]

The starting voltage spec is complicated. ECU manufacturers are using SEPIC or a boost-converter to comply. It's funny old Lucas used to advise you'd burn out your headlights if they were on during cranking due to the surges... Anyhow, this is what I have for ISO-16750 values for cranking and again it's a loose standard with four severity levels and car makers have their own standard levels as well. If your product is off and can reboot during cranking, not so much of problem, the low voltages.
Below -20°C starting ICE is not fun, you have seconds to stop if the engine has flooded and shut off the fuel, before your battery craps out. Canadian cars usually come with a bigger battery (except Honda, theirs is the smallest car battery Group 51R lol). Gasoline does not evaporate much at those low temperatures so you are literally pouring in fuel for cold start enrichment.


Just a note for OP's question- "centralized" means in the alternator: "alternator with/without centralized load dump suppression" is the exact term.

[16bitanalogue]

Just a note for OP's question- "centralized" means in the alternator: "alternator with/without centralized load dump suppression" is the exact term.

Interesting. Then the internal clamps of the alternator must still be a higher voltage than what is allowed for downstream modules. That would explain why there would continue to be separate suppression circuits downstream.

[floobydust]

More modern, newer alternators using avalanche rectifiers (for at least 10 years now) still have load dump- it's just much lower in voltage.
Compare Load Dump amplitude Test B is 35V (clamped) verses 79-101V (unclamped) which is a huge difference energy-wise.

[Hydron]

@floobydust, thanks for the info - I've recently designed something for 12V vehicles which has a UVLO (falling) of 6V, and wondered if I was going to have an issue with it.
I've got a housekeeping micro on it (before the UVLO) which should be able to try and avoid the device (which takes a few seconds to turn off/on) being on during the majority of cranking events, but it's not guaranteed. Shouldn't be going anywhere near -20C thankfully, and is realistically a prototype (though I'm seriously worried if I need to change anything that we haven't already stocked up on given the current semiconductor supply nightmare - one part I'm using is already showing 2023 if I want more!).

[floobydust]

The big dip during cranking is ~70msec so if you've got the capacitance and a vehicle that is not too cold or an old battery, it might not be a problem for you. A 5V MCU and LDO and reverse-diode is cutting it close.
New is the Reset Test where you drop the voltage in 5% increments "Reset is permitted upon agreement" yet no real mention of test limits.

[Hydron]

Can't hold up for 70ms, but I can handle a 6V minimum (just) - got a FET doing reverse protection duty (be dropping more in the wiring harness than in that) and the 5V DC-DC has under 0.5V of dropout at a larger load than I'll be using. UVLO is likely actually the ultimate limit, which could be pushed a little lower if needed. Housekeeping MCU is on a 3V3 linear which is OK down below 5Vin.

Thanks for the reply; looks like I should be reading up on ISO-16750 - I was aware this was one of the auto standards to check out, but it was one of those projects where doing it quick was far more important than figuring out which standards to use and meeting them exactly, and designing for automotive is pretty new to me. Still designed the device to handle at least 40V sustained (though if it stays too long over 35V it'll shut down), 70V (probably more) transient and reverse polarity protection because it just seemed sensible to do so, so I think I'll be good given we have some latitude to dictate the conditions they'll be run in.

I gotta say though, it'd be nicer if these things were clearer, and "12V" auto supplies turn out to be a massive pain if you actually want 12V as a rail voltage without excessive pissing about. Other industries have their own PITA stuff to deal with though - CCTV's 24V AC psuedo-standard, USB PD, or some of the PoE stuff come to mind (and god forbid you need to read the standards for the latter two - CCTV basically not having any may be a blessing!).

[EPAIII]

I have not read every response and I do know that there are members here from many different countries, encompassing many different languages so perhaps there is room for different viewpoints on this. But it strikes me that the term "load dump" seems to be a very poor choice for what is being discussed. "Load dump" seems to imply the dumping or disconnecting of the load.

What is being discussed is over-Voltage protection. Perhaps a search for that would yield a better return on the subject.

[T3sl4co1l]

Well, it's due to the sudden disconnection (dumping) of a heavy load -- the battery most of all given its charging capacity, and then also exacerbated by other changes in load, say turning off the headlights or whatever.  One of those very context-dependent terms; it says what it is, if you know everything about the situation (automotive, conventional regulated alternator, battery disconnect), but if you're missing those pieces and take it at face value, yeah, it's not too helpful really.  Anyway, for better or worse, this is the accepted terminology in the English speaking world.  English being what it is, well, this is just par for the course honestly...

Also, it's... not strictly overvoltage protection, as one might argue that -- well, if it's overvoltage, it's simply, whatever the applied voltage is, there you have it; you'll just footgun yourself trying to shunt it, voltage is voltage.  So you still need to see the definition.  And as it happens, this is defined to have some source resistance, so shunting isn't actually futile, just difficult.  Example: jump starting 2x voltage is another common automotive standard, and that happens with very low source resistance; you have no choice but to ride through or shut off (including by fuse) under such a condition.

And "surge" maybe isn't a great term, it's a general word but the most common kind of surge (without qualifiers) is understood as a lightning-induced type, of the 1.5/50 to 8/20 to 10/1000 µs waveforms; these do define source impedances, and the general waveform isn't very different (sharp rise, slow fall), so, to be fair it's not terrible, there's just the particularity about how the DC is wired in (it's diode-OR'd), and it would just be a little odd to say a 80V 1,000/100,000 surge.  Though honestly, I'm kind of liking the generality, just make sure you don't leave off the units so it's very obvious it is in fact 1/100 ms or whatever.

Tim

[16bitanalogue]

I'm still struggling with the end design from a compliance perspective.

Given that central load dump suppression is a function of the alternator, I presume it still cannot cover all use cases since a vehicles wiring harness branches off from the main charging system. If a portion of the harness breaks (corrosion, loose) then at least that path's downstream modules will still suffer from a load dump pulse. Which would explain why the designs I have seen still have module level load dump suppression circuits. Therefore one would still compliance test to the 5a waveform.

[floobydust]

No, if the wiring harness breaks at the alternator's output, there will be no charging and no load dump. Downstream will just be running off the battery. The alternator is the exclusive source of load dump transients, and since the avalanche rectifier diodes are integral to it, no broken wire in the vehicle will allow full strength load dumps to start happening.

Module-level transient protection is for the other transients. Recall the tests from ISO 7637-2:
Test Pulse 1   - disconnection of inductive loads
Test Pulse 2a - disconnection of loads; wiring inductance
Test Pulse 2b - DC motors acting as generators when ignition is switched off
Test Pulse 3a - switching transients -ve; incl. wiring harness inductance and capacitance
Test Pulse 3b - switching transients +ve; incl. wiring harness inductance and capacitance
Test Pulse 4   - cranking
Test Pulse 5a - load dump (no centralized suppression)
Test Pulse 5b - load dump (centralized suppression)

[16bitanalogue]

No, if the wiring harness breaks at the alternator's output, there will be no charging and no load dump. Downstream will just be running off the battery. The alternator is the exclusive source of load dump transients, and since the avalanche rectifier diodes are integral to it, no broken wire in the vehicle will allow full strength load dumps to start happening.

I'm presuming the vehicle is already running and either a break occurs at the battery (happened on an old car of mine), the battery clamp was just loose, or intentionally pulled off (old shade tree mechanic trick). If the break happens at the alternator the battery should help to absorb the transient. If the wiring harness breaks downstream then true, the interruption of current due to the inductance of the harness to that specific module would experience some transient, but not the 'load dump' as defined.

I tried to find an alternator from Bosch or Denso with integrated clamps, but I have come up empty.

[16bitanalogue]

If anyone is interested, it appears that Motorcraft makes alternators with central load dump suppression. No, schematics have yet to be found.

https://www.motorcraft.com/us/en_us/home/our-products/batteries-starting-and-charging/alternators-gl.html

Late edit:
According to Denso North America, they do not have alternators with central load dump suppression.