diy solar

diy solar

Why not to use Daly BMS with MPPT controllers

4) Voltage gets to 3.65V and BMS disconnects the battery. PVs suddenly get in open circuit mode, and the voltage spikes to max PV voltage (i.e. 200V).
5) The BMS gets toasted right after point 4. The sudden voltage spike was over the FETs max voltage capabilities.

Won't be instantaneous so long as there is a capacitor involved, e.g. on SCC output. Inverter would have even more, but no guarantee it is connected.

10A into 1000 uF would be 10,000V/second or 100 microseconds per volt.
I wouldn't count on an interpreted language, but compiled dedicated code could be fast enough. I prefer analog hardware.
A transistor circuit could be fast enough, so long as drive of FET gate is strong. Relay or breaker would be questionable. In 1 millisecond voltage would have risen 10V, and we may have 20V headroom in BMS FETs.

I think tripping a breaker works if done before BMS disconnects (or for my AGM).
Where a BMS could trip, either turn off a series MOSFET to open circuit, or turn on a crowbar MOSFET to close.

One issue with the crowbar is that assuming failed SCC, it is sinking current through SCC and from battery. That besides capacitors of the SCC.
 
I'd think about a low resistance diode on the SCC output to avoid sinking from battery, but unfortunately they rely on having power from the batteries to operate and know what to do. However… if you wanted to do that, what about combining it with a 5-10ohm power resistor on a small wire in parallel with the main power wire with a diode? On a system with battery at 55V that would limit the battery current seen by the crowbar to about 11A, and 5ohms should still supply 5W to the SCC with only a 5V drop. SCCs from Magnum, Outback, and Schneider range from <1W to <4W idle draw.
 
Do you think a small RC snubber across the fet might be good?
I'm not 100% sure on this, but I don't think the FET would need snubber. If the switch is a relay - then yes. A proper snubber circuit can protect relay contacts from arching.

Won't be instantaneous so long as there is a capacitor involved, e.g. on SCC output. Inverter would have even more, but no guarantee it is connected.
Agree. The SCC capacitors would provide some damping.
 
Ultimately, any circuit can fail, and this is just hoping to be one more layer of protection that will be in a functional state in the event other safeguards fail.
This is my way of thinking hence why the single board computer as the base. Plus with how cheap these are just use two or three and make them check each other if you are that paranoid. But honestly at some point the law of diminishing returns comes to play.
With the Metro M4 (or the new Grand Central M4) microcontroller I can have enough horsepower to not only check the voltages and disconnect the panels when needed but also check a smoke detector or flood detector and then send a shut down signal to the Victron inverters or cut the AC power to the Victron chargers then sound an alarm or flash some lights etc. And it would all be so low power that I could but it on it's own power supply.
Seems like the best bang for the buck to be honest.
 
One issue with the crowbar is that assuming failed SCC, it is sinking current through SCC and from battery. That besides capacitors of the SCC.
If the SCC isn't damged when the crowbar closes I expect it will be soon after. The crowbar will try to discharge the battery array through the intrinsic diode in the buck transistor. No?
 
I don't think an SCC normally pulls PV input up to battery voltage.


If source is PV side, drain is battery/inductor side, and Vgs is zero, Would that hold off battery voltage? Or does it need another FET connected the other way (like a BMS or inrush limiter) for that?

I didn't want to find out in the last design I did, put a clamping diode across the (bipolar) transistor. Previous generation product would reportedly get MOSFET burned out if the (sensor) load it was driving was unplugged while active. It also served as a clamp for reverse polarity power supplies.


But the failure mode of concern in this thread seems to be when SCC has already failed and pulled battery toward PV voltage. A cooked MOSFET is just a resistor, might get pretty hot.
 
I don't think an SCC normally pulls PV input up to battery voltage.


If source is PV side, drain is battery/inductor side, and Vgs is zero, Would that hold off battery voltage? Or does it need another FET connected the other way (like a BMS or inrush limiter) for that?

I didn't want to find out in the last design I did, put a clamping diode across the (bipolar) transistor. Previous generation product would reportedly get MOSFET burned out if the (sensor) load it was driving was unplugged while active. It also served as a clamp for reverse polarity power supplies.


But the failure mode of concern in this thread seems to be when SCC has already failed and pulled battery toward PV voltage. A cooked MOSFET is just a resistor, might get pretty hot.
A typical buck converter will use a N channel fet where your document shows switch. Drain to PV panel and source to inductor. While a N channel fet is more difficult to drive it is more efficient and not hard to get an isolated supply above the rail.

Assume crowbar in on the PV panel. When the crowbar engages the battery will try to discharge through the intrinsic diode in the N channel fet. The intrinsic diode comes along with the fet. Suppose the crowbar scr triggers by accident. The intrinsic diode will not be able to survive that kind of a fault. SCRs are prone to trigger by noise and it takes good circuit design and pcb layout to minimize, but all it takes is one event to smoke a charge controller. So an awkward circuit to prevent an expensive failure may cause an expensive failure. An scr that trips a contactor sounds safe. I will probably put a contactor between panels and charge controller and open if things go haywire.
 
Not a lot of relays/contactors that can interrupt 600Vdc, more doable if a lower voltage PV array.

Since the application is PV --> MPPT --> battery voltage, less likely to be a high voltage array (which is most efficient driving higher voltage load, like an inverter.) So relay might be feasible. Of course, snubbers can help by preventing arcing. All a design and reliability exercise.

Remote-trip DC circuit breaker, or ganged to a low-current breaker as remote trip, seemed like the way to go. That's my plan. But the more fail-safe solution is "keep alive" signal rather than a "shutdown" signal.

I don't think most (good) MPPT backfeed PV to battery voltage. They may have another blocking element besides one N-channel FET. A diode (with voltage drop) or "ideal diode" (just some resistance) could do it.
 
Not a lot of relays/contactors that can interrupt 600Vdc, more doable if a lower voltage PV array.

Since the application is PV --> MPPT --> battery voltage, less likely to be a high voltage array (which is most efficient driving higher voltage load, like an inverter.) So relay might be feasible. Of course, snubbers can help by preventing arcing. All a design and reliability exercise.

Remote-trip DC circuit breaker, or ganged to a low-current breaker as remote trip, seemed like the way to go. That's my plan. But the more fail-safe solution is "keep alive" signal rather than a "shutdown" signal.

I don't think most (good) MPPT backfeed PV to battery voltage. They may have another blocking element besides one N-channel FET. A diode (with voltage drop) or "ideal diode" (just some resistance) could do it.
Yes. I expect something in a charge controller to prevent charging panels from battery array at night.
 
It's questionable if (mechanical) relays are fast enough to prevent a BMS from overvoltage events.

I'd still vote for some kind of crowbar - i.e. with a thyristor and a simple zener diode. Of course, this crowbar must then trigger some kind of interruptor (circuit breaker/blowing fuse, relay....). Otherwise, continuous current will dissipate a lot of power over the crowbar...

Concerning maximum voltage, the thyristor must handle (as long as we don't have arcs - which imho would require completely different protection mechanisms), is some "worst-case PV array OC" which should be always less than (let's say) ~500V in diy environments.

In case of a BMS with separate charge/discharge connections, the thyristor only has to withstand PV array shorting current - also here, we never have to deal with more than something like 10...20A. But if we have a BMS with combined charge/discharge port, battery maximum current has to be taken into account which could even exceed 1000A (ouch). This "huge reverse current" could be prevented by putting some kind of diode between PV Array, Crowbar and BMS.

But even with 500V and 1000A, Thyristors are available for less than 100 bucks....
 
Remote-trip DC circuit breaker, or ganged to a low-current breaker as remote trip, seemed like the way to go. That's my plan. But the more fail-safe solution is "keep alive" signal rather than a "shutdown" signal.
Agreed keep-alive would be ideal, but at this point I'm thinking for most systems as a practical matter it isn't so easy. The low power relay I linked above is DPDT, which could easily be wired in NC, however that risks burning out the trip coil of the breaker if something goes wrong. 16.7mA at 3V isn't going to tax much of any PV system, at least.

I don't think most (good) MPPT backfeed PV to battery voltage. They may have another blocking element besides one N-channel FET. A diode (with voltage drop) or "ideal diode" (just some resistance) could do it.
According to @HighTechLab in this thread, Outback Flexmax, Victron, and Midnite SCCs have magnetic isolation between PV+ and Batt+, but aren't advertised as isolated because PV- is bonded to Batt-. I know the latter is directly stated in the Outback FM80 manual. For our purposes here, however, I think he's right these controllers will always fail safe.
 
The failure mode under consideration, SCC spontaneously shorts PV input to battery, isn't protected against by a fuse.
 
Hi All, I have just experienced this exact failure in my caravan 12v system. The solar controller was a Victron Smart solar 100/50 (Apparently they fail too) and the BMS was a generic hardware 200a BMS in which I have since learnt the MOSFETs where only rated to 36v, the PV was 45-50v and my 400ah LifePo4 cells are now trashed. Expensive lesson, and one that I am not particularly interested in reliving.

I think I have a solution that may work for my particular circumstance that I would like to run by you guys to see if there's any holes in the plan. I will replace the BMS with a JBD unit rated at 80v, however due to my recent MOSFET related trauma, I would like to add addition protection.

I am looking at adding this Bluesea Remote battery switch,

https://www.bluesea.com/products/7700/ML-RBS_Remote_Battery_Switch_with_Manual_Control_-_12V_DC_500A

Its a high current latching relay, rated at 500a with remote switch that I intend trigger with the relay output of a BMV712, on high voltage,
BEFORE the BMS goes into high voltage protection.

I will be using a charge voltages of 13.8-14v, the JBD BMS protects at 3.75 (15v) so I figure I would set BMV relay to trigger at say 14.4v.
I intend to add an additional bus bar for the charging devices and install the battery switch between it and the main positive bus to Isolate any charge voltage from the rest of the circuits.

In the reoccurrence of such failure, the relay will open and all hell will be avoided. Or is that not the case?
Let me know what I have overlooked.

Cheers
 
I dont trust BMS. Instead I monitor cell voltage and audible alarm goes off when a cell-overvoltage occurs. I am following this thread for ideas/solutions on automatic cutoff solutions between charge controllers and battery.
 
The solar controller was a Victron Smart solar 100/50 (Apparently they fail too) and the BMS was a generic hardware 200a BMS in which I have since learnt the MOSFETs where only rated to 36v... the PV was 45-50v and my 400ah LifePo4 cells are now trashed.

Plenty of voltage headroom, 50Voc vs. 100V spec, something else caused the failure.
It has been noted some SCC have transformer isolation in the switcher, so failed transistors wouldn't provide a DC path. That could be a robust solution.

I am looking at adding this Bluesea Remote battery switch,

https://www.bluesea.com/products/7700/ML-RBS_Remote_Battery_Switch_with_Manual_Control_-_12V_DC_500A

Its a high current latching relay, rated at 500a with remote switch that I intend trigger with the relay output of a BMV712, on high voltage,
BEFORE the BMS goes into high voltage protection.

Rated for 12V. 50V PV and 14V battery leaves 36V to be interrupted. Probably not so high it would arc, especially at lower PV current vs. inverter or short-circuit current, but it is above spec.

We came up with the idea of PV breaker with remote trip. You might do that instead.

I think this model is remote trip. But not certain if that is 24V trip coil, not finding the details:


Mention of 24V or 48V trip controller, but still not clear what breaker trip specs are:


 
Thanks Hedges, but the root cause of the fault was absolutely the victron mppt.

It somehow lost its ability to sense it's own output and was sending the full Pv voltage through the system. It was the only charging device connected at the time.

It will show 0w output on its own display and the smartshunt reads 20-30a coming in. I have sent the vsc file to the victron distributor and they have replaced the mppt under warranty.

I did look at the midnight breakers but as far as I can see they require 24v control circuit. I'm not aware of a 12v equivalent. Also, as far as I can see the configuration I have suggested would protect from all charging device faults not just the mppt.

Is there any reason what I have suggested would work to protect the BMS and cells ?
 
Hedges has raised the correct design issue regarding the battery disconnect relay: the relay contacts are NOT rated for you full PV voltage minus battery voltage. So when it starts to open, the contacts may may arc and then fuse-weld on the first bounce.

I still prefer the opto-120 to 400VAC SS relay (SCR) crowbar the PV method; it's self latching and the opto input assures no false triggering, and they are very cheap. You should verify whether you need diode block for your PV before the charge controller or whether that is redundant. Most SSR's will trigger at 3V, so you can either use a comparator or zener diode (unsafe voltage minus 3V) from the battery to trigger it. If going simple with the zener method, you must test that circuit carefully to make sure the specific AC SSR handles the voltages below 3V OK. For the easiest solution (works with any SSR) using comparator method, you can use this or similar comparator modules:


I don't like mechanical relays for this situation as the voltage will climb very fast once the BMS has disconnected, before it is fried (and closed) by overvoltage. The relay might be fast enough save the cells, but the BMS will be fried before the relay can open. The same is true for remote triggered breakers- they may be too slow to save the BMS from overvoltage damage, once it disconnects the battery but before it is fried from voltage above it's mosfet rated voltage.

Another solution might be to look at replacing the BMS disconnect mosfet(s) with a much higher voltage one. That requires careful consideration of heat sinking, as Rds will increase with higher voltage rating. Which is why they made the dangerous mistake of not allowing for a full high voltage PV input from charge controller failure. With a low voltage mosfet, the Rds (resistance when on) is so low that little heat-sinking is needed.

Best Wishes,
Bruce
 
Last edited:
Back
Top