I'm building a power system using four x 48 V 100 Ah lithium batteries in parallel (48 V 400 Ah nominal). The batteries are already purchased, there is no going back. The cells cannot be accessed, they are in a sealed case and opening it voids warranty. The packs have pre-BMS cutoff signaling capabilities which will be used (easy; those just trigger when active), but those signals are not 100% guaranteed if cutoff conditions change very rapidly.
The battery bank is being charged by a 48 V 100 A alternator with an advanced regulator that can also detect a variety of pre-BMS cutoff thresholds, so 99.9% of the time all is well. However, aside from a "keeper battery" bank of UPS AGMs in parallel that will need to be replaced routinely (undesirable), there is currently no SAE-approved industry solution for a 48 V load dump during alternator charging (12 V and 24 V do exist; Sterling).
In the unlikely but possible event that all four batteries cutoff during charging, a disastrous voltage spike will occur and ruin everything in its path (alternator, inverter/charger, etc). A solution may exist next year for absorbing the spike (Balmar) but 1 year could become 3 years.
It's possible a continuous 5% (of 400 Ah) load will prevent the worst of the spike, but with significant voltage ripple and no guarantee equipment won't be damaged.
In the meantime, with four batteries in parallel, I'd like to effectively use each one as its neighbor's "keeper battery", such that a detection circuit goes high 12V+ if one or more battery experiences a cutoff. This may not happen for months or years, so the circuit must remain in watchdog mode for extended periods and remain reliable.
So it boils down to detecting the loss of a 48 V nominal source (40-60 V source becomes open circuit), and using that loss to trigger a separate 12V+ source signal.
There exists a separate 12 V nominal source (11.9 min - 14.5 max) that will always be available for this purpose. It will feed the alternator regulator's feature-in port, which can monitor for 12V+ nominal and stop charging. I'm trying to obtain exact values, but I believe it will work from 11-15 V and sink 20 - 150 mA. Time to detect the loss of 48 V should be < 2 seconds, time to restore is fast as reasonably possible.
I've explored both signal relays, power relays, solenoids, and even some more complicated voltage detection circuits using R-C networks and transistors. None seem to meet all of the job requirements, which are:
Continuous duty; this is critical. Relays that with the coil energized for months/years tend to breakdown (e.g. NC contacts tied to 12V+; goes high when coil is de-energized). Latching relays could be used but I'd need to turn a constant 48 V nominal into a momentary pulse (bosch style relay with R-C on contact path to-ground?), and do the same when 48 V is gone.
Low power consumption, because this circuit must remain active 24/7, and while the battery bank is massive with solar and alternator charging, keeping power consumption to a minimum is ideal. Signal relays are good for this, but tend to be in the "don't energize for long periods" category, although I found one that Hasco says can remain energized if I limit the input to 48 V using a zener + resistor pair (zeners only come in 47 V type, so I'd limit to that). Some power relays are rated for continuous duty and only draw 1 W (still high, but okay), but they often have high minimum contact load requirements (1 A or more; so now you need an R-C across the contacts to generate a whetting current to draw more than the 20 - 150 mA signal current; I'm also not certain the R-C can just be added outside the relay on its leads, but I think so; anyone know?)
Compact due to space limitations; solenoids are typically bulky and expensive, so they're not ideal.
Works from 40 to 58 V (e.g. relay coil must latch at 40 V and max coil voltage is at least 60 V, ideally higher)
Simple, to avoid failure points, and work in a wide range of temperatures -20 to 65 C
Those are the main requirements. I can't help but think there is an easier way to detect the loss of a 40-58 V signal.
I've seen circuits that use an RC, a transistor, and a diode, but I can't tell whether these are simple to build in a DIY application and reliable. They may be.
I've also looked at voltage detection devices, but the readily available ones for 48 V tend to use 120 or 240 VAC for monitoring power, or they draw power from the device being monitored, which won't work if that device experiences a cutoff.
I think I can implement just about any viable solution. I once earned an EET degree, but it's faded from memory, so I'm having to relearn a lot. Can do. Will do. But don't want to waste time going down wrong paths. Any advice?
So far, this looks most promising in the signal relay department (Hasco says it can be energized continuously with zener + resistor to clamp coil to 47 V max), or this in the power relay. I had someone tell me that power relay is rated for continuous duty but I see no mention of that in the datasheet and I'm awaiting a response from TE.
Please don't troll, or leave snarky posts. If you have a real idea, I'll be extraordinarily grateful.