On 1/30/23 7:10 PM, boB wrote:> On Mon, 30 Jan 2023 16:04:10 -0800, Joerg <news@analogconsultants.com> > wrote: > >> On 1/30/23 2:07 PM, boB wrote: >>> On Mon, 30 Jan 2023 12:35:43 -0800, Joerg <news@analogconsultants.com> >>> wrote: >>> >>>> On 1/30/23 12:09 PM, Fred Bloggs wrote: >>>>> On Monday, January 30, 2023 at 2:48:21 PM UTC-5, Joerg wrote: >>>>>> On 1/30/23 10:55 AM, Fred Bloggs wrote: >>>>>>> On Monday, January 30, 2023 at 1:40:50 PM UTC-5, Joerg wrote: >>>>>>>> Attention, this is one of those rare electronics-related questions :-) >>>>>>>> >>>>>>>> Will be rigging up a small power-failure backup system. Essentially a >>>>>>>> 12V LiFePO4 battery (with BMS in there), inverter, solar panels, MPPT >>>>>>>> charger. >>>>>>>> >>>>>>>> When reading up on stuff and talking to people I found something very >>>>>>>> disconcerting. It seems that MPPT charge controllers can and often do >>>>>>>> commit suicide when operated without a battery. It even says that in the >>>>>>>> manuals a lot. The trigger event can be as simple as the BMS opening for >>>>>>>> some reason and ... *PHUT* ... MPPT is gone. A friend had that happen, >>>>>>>> twice. >>>>>>>> >>>>>>>> They can also send out a substantial voltage spike or short their >>>>>>>> MOSFETs during such events, the latter letting full solar panel voltage >>>>>>>> onto the 12V power bus. Both of which can kill rather expensive gear >>>>>>>> connected to the 12V bus. >>>>>>>> >>>>>>>> Why is that? Are the engineers designing this stuff not very bright? I'd >>>>>>>> never release a design with such "traits". >>>>>>>> >>>>>>>> Does anyone know lower-end MPPT charge controllers (20-40A range) that >>>>>>>> are properly designed in this respect? Preferably ones that are also low >>>>>>>> noise so they don't mess with measurements on the lab bench. The Genasun >>>>>>>> brand is low noise but AFAICT they only come for very low solar panel >>>>>>>> voltages. >>>>>>>> >>>>>>>> Of course, one solution is to build a massive crowbar to blow a fuse and >>>>>>>> protect the connected gear. The MPPT charger might possibly still die. >>>>>>> >>>>>>> I've seen that as a warning, in the user's manual, even for those little solar battery maintainers to keep a seldom used car battery topped off. >>>>>>> >>>>>>> You have to remove power to the MPPT charge controller before you remove the battery. >>>>>>> >>>>>> Yeah I know, and that really doesn't make sense in such a design. Any >>>>>> Li-Ion battery can literally "remove itself" via BMS action. Any >>>>>> external stuff should be resilient in that case and most MPPT chargers >>>>>> seem not to be. >>>>> >>>>> There are a bunch of different switching topologies. My guess there is a bunch of circulating magnetic energy that causes failure when the load is removed abruptly. >>>> >>>> >>>> That would be a serious case of engineering blunder. I've designed stuff >>>> like this and know how it's done correctly. I just need to know which >>>> MPPT mfg also knows this, since I don't want to roll my own again. >>>> >>>> >>>> You say your BMS disconnects the battery? Most of the time it's supposed >>>> to communicate with the charger and tell it to back off the current. >>>> >>>> >>>> That is not realistic when you are using an off-the-shelf Li-Ion >>>> battery, as is customary. They generally do not have any communications >>>> interfaces. And with a properly designed MPPT it isn't necessary. >>>> >>>> Very occasionally a large Li-Ion battery will have Bluetooth >>>> connectivity but that's mainly for relaying the charge status. >>>> >>>> >>>> Check to see if your MPPT is compatible with Li having a BMS disconnect, >>>> or that the Li is compatible with an MPPT. Send an email query to the >>>> MPPT manufacturer customer support. >>>> >>>> >>>> Everytime I asked the answer was no :-( >>>> >>>> [...] >>> >>> I have dealt with this directly. The problem is that when the MPPT >>> is charging hard and that load is abruptly disconnected, it is like a >>> tug of war with a rope where if one person lets go, the person at the >>> other end may fall down. In this case, the PWM is such that if the >>> current suddenly goes to zero, the input voltage of a synchronous buck >>> converter can go sky-high and breal the FETs. >>> >> >> There is supposed to be at least some capacitance on the output that >> would prevent this. So far all the converters I designed react >> immediately and do not let the output overshoot even with a suddenly >> disconnected load. It can be done within one cycle so you don't need >> much in terms of capacitance. One of them was an MPPT design. That one >> you could even use as a stand alone power supply, sans battery, provided >> the panels delivered enough juice. >> >> It's the same in other situations. I designed an actuator circuit where >> then the client encountered the unforseen situation where the load would >> accidentally come off. It's taken care of by a FET that immediately >> takes over and bleeds off the energy in the inductor for that one cycle. >> IOW it briefly goes linear. If something shoots up sky-high that is a >> design flaw in my opinion. >> >> >>> In my old designs, I tested for this same thing except with lead acid >>> batteries and the circuit breaker to the battery from the MPPT >>> controller. >>> >> >> I'd even test with lead acid because there could always be a gradual >> corrosion or loosening at one of the contacts. I once sat there in a >> borrowed old Chevy truck. Wouldn't start after loading my stuff and >> others around me became impatient. Popped the hood, wiggled some cables >> and the positive terminal on the battery came right off. >> >> >>> Communications from the BMS to MPPT, if it is even there, is most >>> likely not fast enough to stop this but might be in some cases. >>> >> >> True. Plus it shall not be relied upon anyhow. Comms can fail quietly, >> then the alert doesn't come and ... whaddabam. >> >> >>> Was this a Chinese product ? >>> >> >> I was told it was Californian. I'll probably receive one of the failed >> units some time in February to have a look. >> >> Anyhow, I'll include a big fat crowbar on this system because I don't >> trust design engineers in that industry much anymore. There's too much >> money connected to the 12V bus. > > > Well, there are also capacitors on the battery side but they should > hold less energy than the PV input capacitors. >I'll make sure there is some cable length so the caps might survive. Or not, but then at least my electronics loads do. With MPPT chargers it seems it's not worth to spend much money and rather treat them like disposables.> Yes, the input voltage 'shouldn't' go high but the battery side > certainly does go high. I think you are right about the input voltage > not going high, but having fixed this many years ago, I can't remember > the failure mechanism exactly. But it can go at least to the PV's Voc > voltage, but much above that, (just a few volts) the PV array should > clamp that. >It will go to Voc if the series FET shorts out. So in my case around 75V, two residential panels in series. I can also connect them in parallel but that requires much beefier cables and extra diodes in case one is shaded (they already have bypass diodes).> What was the nominal PV input voltage of this controller ? Was it set > up as a 24V array or 48V array or maybe higher voltage ? That can > also make a difference. What kind was it ? I am familiar with a lot > of the commercial ones. >I don't know yet. My friend has jury duty out of town and is gone.> It might be also that the battery voltage flying up is shoving current > the other way, too fast and even with the indutor in there can over > current the FETs. The input capacitors could make a good load for > that backwards current and hurt things as well. >I don't quite follow. How can there be a problem if the capacitors on the PV side are large enough?> Is the controller designed to work backwards reliably, too ? >Very unlikely. The one I designed wasn't either but it was in this regard bullet-proof. You could not destroy it by randomly disconnecting things. The same should go for commercial ones but obviously they can't get it done.> What I find is that the PV input caps, with much higher PV voltage > than battery voltage, is a great source of blow-up energy for the > controller but that is usually from HV to LV battery direction. > > Ultimately it is the high current from the battery and shorted FETs > that can cause the PCB to catch fire if things go very wrong. >That sounds like a seriously flawed design.> One thing to do is to catch the Vbattery side going too high *FAST* > and immediately shut off the converter(s) >Well, I am going to add in a crowbar anyhow to protect my electronics. An SCR the size of a gold ball that is capable to reliably blow a 50A car fuse if needed. If I put that sans fuse on the MPPT output it would present a dead short to the MPPT, like an empty battery. If it doesn't survive that then the MPPT is real junk. -- Regards, Joerg http://www.analogconsultants.com/
MPPT chargers dying upon battery disconnect?
Started by ●January 30, 2023
Reply by ●January 31, 20232023-01-31
Reply by ●January 31, 20232023-01-31
On Mon, 30 Jan 2023 10:40:39 -0800, Joerg <news@analogconsultants.com> wrote:>Attention, this is one of those rare electronics-related questions :-) > >Will be rigging up a small power-failure backup system. Essentially a >12V LiFePO4 battery (with BMS in there), inverter, solar panels, MPPT >charger. > >When reading up on stuff and talking to people I found something very >disconcerting. It seems that MPPT charge controllers can and often do >commit suicide when operated without a battery. It even says that in the >manuals a lot. The trigger event can be as simple as the BMS opening for >some reason and ... *PHUT* ... MPPT is gone. A friend had that happen, >twice. > >They can also send out a substantial voltage spike or short their >MOSFETs during such events, the latter letting full solar panel voltage >onto the 12V power bus. Both of which can kill rather expensive gear >connected to the 12V bus. > >Why is that? Are the engineers designing this stuff not very bright? I'd >never release a design with such "traits". > >Does anyone know lower-end MPPT charge controllers (20-40A range) that >are properly designed in this respect? Preferably ones that are also low >noise so they don't mess with measurements on the lab bench. The Genasun >brand is low noise but AFAICT they only come for very low solar panel >voltages. > >Of course, one solution is to build a massive crowbar to blow a fuse and >protect the connected gear. The MPPT charger might possibly still die.The usual result is a warranty claim, which most MPPT charge controller manufacturers were honor. That's being generous because the documentation and usually a big label on the controller will warn the user that charging without a (battery) load is a really bad idea. Lots of examples and horror stories online: <https://www.google.com/search?q=solar+do+not+operate+with+battery+disconnected> I attempted to get a list of parts replaced, but failed. I'm fairly sure all the MOSFETs blew, but there might have been some other parts involved. I've seen open batteries cause failures 3 times so far. All three in campers running whatever could be thrown together from Amazon sourced parts. I've learned from these and two more installs that it pays to buy everything from a single manufacturer. My favorite this week in Renogy: <https://www.renogy.com> Unfortunately, I couldn't find a schematic of the charger. On my last order, I attempted to demand a schematic or I would order from someone else. They complied, by sending me a hand scribbled block diagram. When I asked for an autopsy on the failed unit(s), they mentioned the disconnected battery problem, but not the BMS (battery management system). The problem seems to revolve around the battery disconnect switch, breaker or fuse. It's common advised that a high current disconnect switch be installed in series with the battery to allow a guaranteed disconnect in case of fire. The switch works nicely if the owner remembers to turn off the charger or cover the solar panels BEFORE disconnecting the battery. There are some other ways to create the problem. I have various schemes for preventing the problem, but both the owner and manufacturer will void the warranty if I become too creative. My best so far is a big contactor relay that disconnects battery, solar panels, vehicle alternator and shore power if any manner of fault is detected. So far, on one home system, it seems to work except the contacts will sometimes arc weld closed. Arc welding one pair of contacts effectively locks the contacts in the closed position, makes the idea self-defeating. Your crowbar solution might work but I suspect that the charger MOSFETs will blow before the fuse opens. Good luck. -- Jeff Liebermann jeffl@cruzio.com PO Box 272 http://www.LearnByDestroying.com Ben Lomond CA 95005-0272 Skype: JeffLiebermann AE6KS 831-336-2558
Reply by ●January 31, 20232023-01-31
On 1/31/23 2:42 PM, Jeff Liebermann wrote:> On Mon, 30 Jan 2023 10:40:39 -0800, Joerg <news@analogconsultants.com> > wrote: > >> Attention, this is one of those rare electronics-related questions :-) >> >> Will be rigging up a small power-failure backup system. Essentially a >> 12V LiFePO4 battery (with BMS in there), inverter, solar panels, MPPT >> charger. >> >> When reading up on stuff and talking to people I found something very >> disconcerting. It seems that MPPT charge controllers can and often do >> commit suicide when operated without a battery. It even says that in the >> manuals a lot. The trigger event can be as simple as the BMS opening for >> some reason and ... *PHUT* ... MPPT is gone. A friend had that happen, >> twice. >> >> They can also send out a substantial voltage spike or short their >> MOSFETs during such events, the latter letting full solar panel voltage >> onto the 12V power bus. Both of which can kill rather expensive gear >> connected to the 12V bus. >> >> Why is that? Are the engineers designing this stuff not very bright? I'd >> never release a design with such "traits". >> >> Does anyone know lower-end MPPT charge controllers (20-40A range) that >> are properly designed in this respect? Preferably ones that are also low >> noise so they don't mess with measurements on the lab bench. The Genasun >> brand is low noise but AFAICT they only come for very low solar panel >> voltages. >> >> Of course, one solution is to build a massive crowbar to blow a fuse and >> protect the connected gear. The MPPT charger might possibly still die. > > The usual result is a warranty claim, which most MPPT charge > controller manufacturers were honor. That's being generous because > the documentation and usually a big label on the controller will warn > the user that charging without a (battery) load is a really bad idea.It usually happens after the rather short warranty has expired.> Lots of examples and horror stories online: > <https://www.google.com/search?q=solar+do+not+operate+with+battery+disconnected> > I attempted to get a list of parts replaced, but failed. I'm fairly > sure all the MOSFETs blew, but there might have been some other parts > involved. >Yeah, probably the driver IC is toast as well since I am sure the gates will have punctured as well.> I've seen open batteries cause failures 3 times so far. All three in > campers running whatever could be thrown together from Amazon sourced > parts. I've learned from these and two more installs that it pays to > buy everything from a single manufacturer. My favorite this week in > Renogy: > <https://www.renogy.com> >However, I saw this warning on their Rover products as well.> Unfortunately, I couldn't find a schematic of the charger. On my last > order, I attempted to demand a schematic or I would order from someone > else. They complied, by sending me a hand scribbled block diagram. > When I asked for an autopsy on the failed unit(s), they mentioned the > disconnected battery problem, but not the BMS (battery management > system). > > The problem seems to revolve around the battery disconnect switch, > breaker or fuse. It's common advised that a high current disconnect > switch be installed in series with the battery to allow a guaranteed > disconnect in case of fire. The switch works nicely if the owner > remembers to turn off the charger or cover the solar panels BEFORE > disconnecting the battery. There are some other ways to create the > problem. I have various schemes for preventing the problem, but both > the owner and manufacturer will void the warranty if I become too > creative. >There should be ways to correct the flaws in their designs. What I don't understand is why the majority of MPPT designers can't get this done properly. When I designed an MPPT charger it wouldn't even have occurred to me that such a single-fault scenario should cause an issue, and it didn't. My client tested the heck out of it because they said that installers make all kinds of mistakes, especially if they are the homeowner. Out in the wild one has to assume the worst. For example people parallel-charging from the unregulated 12V output of an old gas generator, revving it a bit via the carburetor linkage to "goose the charge top-off". Then, after a while, the BMS comes ...> My best so far is a big contactor relay that disconnects battery, > solar panels, vehicle alternator and shore power if any manner of > fault is detected. So far, on one home system, it seems to work > except the contacts will sometimes arc weld closed. Arc welding one > pair of contacts effectively locks the contacts in the closed > position, makes the idea self-defeating. > > Your crowbar solution might work but I suspect that the charger > MOSFETs will blow before the fuse opens. >How can they blow if the battery side voltage is never allowed to exceed about 16V because of the SCR? It essentially puts a dead short across the MPPT output. If it can't even take that the design engineers should go wash dishes or whatever. I wouldn't call them engineers. -- Regards, Joerg http://www.analogconsultants.com/
Reply by ●January 31, 20232023-01-31
onsdag den 1. februar 2023 kl. 00.14.49 UTC+1 skrev Joerg:> On 1/31/23 2:42 PM, Jeff Liebermann wrote: > > On Mon, 30 Jan 2023 10:40:39 -0800, Joerg <ne...@analogconsultants.com> > > wrote: > > > >> Attention, this is one of those rare electronics-related questions :-) > >> > >> Will be rigging up a small power-failure backup system. Essentially a > >> 12V LiFePO4 battery (with BMS in there), inverter, solar panels, MPPT > >> charger. > >> > >> When reading up on stuff and talking to people I found something very > >> disconcerting. It seems that MPPT charge controllers can and often do > >> commit suicide when operated without a battery. It even says that in the > >> manuals a lot. The trigger event can be as simple as the BMS opening for > >> some reason and ... *PHUT* ... MPPT is gone. A friend had that happen, > >> twice. > >> > >> They can also send out a substantial voltage spike or short their > >> MOSFETs during such events, the latter letting full solar panel voltage > >> onto the 12V power bus. Both of which can kill rather expensive gear > >> connected to the 12V bus. > >> > >> Why is that? Are the engineers designing this stuff not very bright? I'd > >> never release a design with such "traits". > >> > >> Does anyone know lower-end MPPT charge controllers (20-40A range) that > >> are properly designed in this respect? Preferably ones that are also low > >> noise so they don't mess with measurements on the lab bench. The Genasun > >> brand is low noise but AFAICT they only come for very low solar panel > >> voltages. > >> > >> Of course, one solution is to build a massive crowbar to blow a fuse and > >> protect the connected gear. The MPPT charger might possibly still die. > > > > The usual result is a warranty claim, which most MPPT charge > > controller manufacturers were honor. That's being generous because > > the documentation and usually a big label on the controller will warn > > the user that charging without a (battery) load is a really bad idea. > It usually happens after the rather short warranty has expired. > > Lots of examples and horror stories online: > > <https://www.google.com/search?q=solar+do+not+operate+with+battery+disconnected> > > I attempted to get a list of parts replaced, but failed. I'm fairly > > sure all the MOSFETs blew, but there might have been some other parts > > involved. > > > Yeah, probably the driver IC is toast as well since I am sure the gates > will have punctured as well. > > I've seen open batteries cause failures 3 times so far. All three in > > campers running whatever could be thrown together from Amazon sourced > > parts. I've learned from these and two more installs that it pays to > > buy everything from a single manufacturer. My favorite this week in > > Renogy: > > <https://www.renogy.com> > > > However, I saw this warning on their Rover products as well. > > Unfortunately, I couldn't find a schematic of the charger. On my last > > order, I attempted to demand a schematic or I would order from someone > > else. They complied, by sending me a hand scribbled block diagram. > > When I asked for an autopsy on the failed unit(s), they mentioned the > > disconnected battery problem, but not the BMS (battery management > > system). > > > > The problem seems to revolve around the battery disconnect switch, > > breaker or fuse. It's common advised that a high current disconnect > > switch be installed in series with the battery to allow a guaranteed > > disconnect in case of fire. The switch works nicely if the owner > > remembers to turn off the charger or cover the solar panels BEFORE > > disconnecting the battery. There are some other ways to create the > > problem. I have various schemes for preventing the problem, but both > > the owner and manufacturer will void the warranty if I become too > > creative. > > > There should be ways to correct the flaws in their designs. What I don't > understand is why the majority of MPPT designers can't get this done > properly. When I designed an MPPT charger it wouldn't even have occurred > to me that such a single-fault scenario should cause an issue, and it > didn't. My client tested the heck out of it because they said that > installers make all kinds of mistakes, especially if they are the homeowner. > > Out in the wild one has to assume the worst. For example people > parallel-charging from the unregulated 12V output of an old gas > generator, revving it a bit via the carburetor linkage to "goose the > charge top-off". Then, after a while, the BMS comes ... > > My best so far is a big contactor relay that disconnects battery, > > solar panels, vehicle alternator and shore power if any manner of > > fault is detected. So far, on one home system, it seems to work > > except the contacts will sometimes arc weld closed. Arc welding one > > pair of contacts effectively locks the contacts in the closed > > position, makes the idea self-defeating. > > > > Your crowbar solution might work but I suspect that the charger > > MOSFETs will blow before the fuse opens. > > > How can they blow if the battery side voltage is never allowed to exceed > about 16V because of the SCR? It essentially puts a dead short across > the MPPT output. If it can't even take that the design engineers should > go wash dishes or whatever. I wouldn't call them engineers.yeh, I'd expect something called a charger to be at least voltage and current limited to not cook the battery and able to handle a dead/shorted/missing battery
Reply by ●February 1, 20232023-02-01
On Tue, 31 Jan 2023 13:51:52 -0800, Joerg <news@analogconsultants.com> wrote:>On 1/30/23 7:10 PM, boB wrote: >> On Mon, 30 Jan 2023 16:04:10 -0800, Joerg <news@analogconsultants.com> >> wrote: >> >>> On 1/30/23 2:07 PM, boB wrote: >>>> On Mon, 30 Jan 2023 12:35:43 -0800, Joerg <news@analogconsultants.com> >>>> wrote: >>>> >>>>> On 1/30/23 12:09 PM, Fred Bloggs wrote: >>>>>> On Monday, January 30, 2023 at 2:48:21 PM UTC-5, Joerg wrote: >>>>>>> On 1/30/23 10:55 AM, Fred Bloggs wrote: >>>>>>>> On Monday, January 30, 2023 at 1:40:50 PM UTC-5, Joerg wrote: >>>>>>>>> Attention, this is one of those rare electronics-related questions :-) >>>>>>>>> >>>>>>>>> Will be rigging up a small power-failure backup system. Essentially a >>>>>>>>> 12V LiFePO4 battery (with BMS in there), inverter, solar panels, MPPT >>>>>>>>> charger. >>>>>>>>> >>>>>>>>> When reading up on stuff and talking to people I found something very >>>>>>>>> disconcerting. It seems that MPPT charge controllers can and often do >>>>>>>>> commit suicide when operated without a battery. It even says that in the >>>>>>>>> manuals a lot. The trigger event can be as simple as the BMS opening for >>>>>>>>> some reason and ... *PHUT* ... MPPT is gone. A friend had that happen, >>>>>>>>> twice. >>>>>>>>> >>>>>>>>> They can also send out a substantial voltage spike or short their >>>>>>>>> MOSFETs during such events, the latter letting full solar panel voltage >>>>>>>>> onto the 12V power bus. Both of which can kill rather expensive gear >>>>>>>>> connected to the 12V bus. >>>>>>>>> >>>>>>>>> Why is that? Are the engineers designing this stuff not very bright? I'd >>>>>>>>> never release a design with such "traits". >>>>>>>>> >>>>>>>>> Does anyone know lower-end MPPT charge controllers (20-40A range) that >>>>>>>>> are properly designed in this respect? Preferably ones that are also low >>>>>>>>> noise so they don't mess with measurements on the lab bench. The Genasun >>>>>>>>> brand is low noise but AFAICT they only come for very low solar panel >>>>>>>>> voltages. >>>>>>>>> >>>>>>>>> Of course, one solution is to build a massive crowbar to blow a fuse and >>>>>>>>> protect the connected gear. The MPPT charger might possibly still die. >>>>>>>> >>>>>>>> I've seen that as a warning, in the user's manual, even for those little solar battery maintainers to keep a seldom used car battery topped off. >>>>>>>> >>>>>>>> You have to remove power to the MPPT charge controller before you remove the battery. >>>>>>>> >>>>>>> Yeah I know, and that really doesn't make sense in such a design. Any >>>>>>> Li-Ion battery can literally "remove itself" via BMS action. Any >>>>>>> external stuff should be resilient in that case and most MPPT chargers >>>>>>> seem not to be. >>>>>> >>>>>> There are a bunch of different switching topologies. My guess there is a bunch of circulating magnetic energy that causes failure when the load is removed abruptly. >>>>> >>>>> >>>>> That would be a serious case of engineering blunder. I've designed stuff >>>>> like this and know how it's done correctly. I just need to know which >>>>> MPPT mfg also knows this, since I don't want to roll my own again. >>>>> >>>>> >>>>> You say your BMS disconnects the battery? Most of the time it's supposed >>>>> to communicate with the charger and tell it to back off the current. >>>>> >>>>> >>>>> That is not realistic when you are using an off-the-shelf Li-Ion >>>>> battery, as is customary. They generally do not have any communications >>>>> interfaces. And with a properly designed MPPT it isn't necessary. >>>>> >>>>> Very occasionally a large Li-Ion battery will have Bluetooth >>>>> connectivity but that's mainly for relaying the charge status. >>>>> >>>>> >>>>> Check to see if your MPPT is compatible with Li having a BMS disconnect, >>>>> or that the Li is compatible with an MPPT. Send an email query to the >>>>> MPPT manufacturer customer support. >>>>> >>>>> >>>>> Everytime I asked the answer was no :-( >>>>> >>>>> [...] >>>> >>>> I have dealt with this directly. The problem is that when the MPPT >>>> is charging hard and that load is abruptly disconnected, it is like a >>>> tug of war with a rope where if one person lets go, the person at the >>>> other end may fall down. In this case, the PWM is such that if the >>>> current suddenly goes to zero, the input voltage of a synchronous buck >>>> converter can go sky-high and breal the FETs. >>>> >>> >>> There is supposed to be at least some capacitance on the output that >>> would prevent this. So far all the converters I designed react >>> immediately and do not let the output overshoot even with a suddenly >>> disconnected load. It can be done within one cycle so you don't need >>> much in terms of capacitance. One of them was an MPPT design. That one >>> you could even use as a stand alone power supply, sans battery, provided >>> the panels delivered enough juice. >>> >>> It's the same in other situations. I designed an actuator circuit where >>> then the client encountered the unforseen situation where the load would >>> accidentally come off. It's taken care of by a FET that immediately >>> takes over and bleeds off the energy in the inductor for that one cycle. >>> IOW it briefly goes linear. If something shoots up sky-high that is a >>> design flaw in my opinion. >>> >>> >>>> In my old designs, I tested for this same thing except with lead acid >>>> batteries and the circuit breaker to the battery from the MPPT >>>> controller. >>>> >>> >>> I'd even test with lead acid because there could always be a gradual >>> corrosion or loosening at one of the contacts. I once sat there in a >>> borrowed old Chevy truck. Wouldn't start after loading my stuff and >>> others around me became impatient. Popped the hood, wiggled some cables >>> and the positive terminal on the battery came right off. >>> >>> >>>> Communications from the BMS to MPPT, if it is even there, is most >>>> likely not fast enough to stop this but might be in some cases. >>>> >>> >>> True. Plus it shall not be relied upon anyhow. Comms can fail quietly, >>> then the alert doesn't come and ... whaddabam. >>> >>> >>>> Was this a Chinese product ? >>>> >>> >>> I was told it was Californian. I'll probably receive one of the failed >>> units some time in February to have a look. >>> >>> Anyhow, I'll include a big fat crowbar on this system because I don't >>> trust design engineers in that industry much anymore. There's too much >>> money connected to the 12V bus. >> >> >> Well, there are also capacitors on the battery side but they should >> hold less energy than the PV input capacitors. >> > >I'll make sure there is some cable length so the caps might survive. Or >not, but then at least my electronics loads do. With MPPT chargers it >seems it's not worth to spend much money and rather treat them like >disposables. > > >> Yes, the input voltage 'shouldn't' go high but the battery side >> certainly does go high. I think you are right about the input voltage >> not going high, but having fixed this many years ago, I can't remember >> the failure mechanism exactly. But it can go at least to the PV's Voc >> voltage, but much above that, (just a few volts) the PV array should >> clamp that. >> > >It will go to Voc if the series FET shorts out. So in my case around >75V, two residential panels in series. I can also connect them in >parallel but that requires much beefier cables and extra diodes in case >one is shaded (they already have bypass diodes). >What I meant was that if it is a bi-directional (synchronous) buck, then it is possible for the PV input voltage to be raised above Voc from the battery. The PV array, being a series of diodes, will clamp the voltage at approximately Voc, or what would be Voc, even at night. This assumes something is not right of course.> >> What was the nominal PV input voltage of this controller ? Was it set >> up as a 24V array or 48V array or maybe higher voltage ? That can >> also make a difference. What kind was it ? I am familiar with a lot >> of the commercial ones. >> > >I don't know yet. My friend has jury duty out of town and is gone. > > >> It might be also that the battery voltage flying up is shoving current >> the other way, too fast and even with the indutor in there can over >> current the FETs. The input capacitors could make a good load for >> that backwards current and hurt things as well. >> > >I don't quite follow. How can there be a problem if the capacitors on >the PV side are large enough? >When the battery breaker trips, the energy in the battery side caps might not be enough to charge the PV input caps very much because of E = 1/2 CV^2 depending on the capacitance in the controlle> >> Is the controller designed to work backwards reliably, too ? >> > >Very unlikely. The one I designed wasn't either but it was in this >regard bullet-proof. You could not destroy it by randomly disconnecting >things. The same should go for commercial ones but obviously they can't >get it done.Non-synchronous is good in that respect. Just a bit less efficient because of the diode drop on the low side of the 1/2 bridge.> > >> What I find is that the PV input caps, with much higher PV voltage >> than battery voltage, is a great source of blow-up energy for the >> controller but that is usually from HV to LV battery direction. >> >> Ultimately it is the high current from the battery and shorted FETs >> that can cause the PCB to catch fire if things go very wrong. >> > >That sounds like a seriously flawed design. > >Any design can have this problem if the bottom FETs short AND the relay contacts weld shorted to the battery. Double fault so that would be a rare occurrance.>> One thing to do is to catch the Vbattery side going too high *FAST* >> and immediately shut off the converter(s) >> > >Well, I am going to add in a crowbar anyhow to protect my electronics. >An SCR the size of a gold ball that is capable to reliably blow a 50A >car fuse if needed. If I put that sans fuse on the MPPT output it would >present a dead short to the MPPT, like an empty battery. If it doesn't >survive that then the MPPT is real junk.Crowbar should work as long as nothing breaks before the breaker trips. Electronics stuff breaks sometimes. Just have to do our best so if it does, the damage is low as possible. boB
Reply by ●February 1, 20232023-02-01
On Tue, 31 Jan 2023 15:14:41 -0800, Joerg <news@analogconsultants.com> wrote:>On 1/31/23 2:42 PM, Jeff Liebermann wrote: >> On Mon, 30 Jan 2023 10:40:39 -0800, Joerg <news@analogconsultants.com> >> wrote: >> >>> Attention, this is one of those rare electronics-related questions :-) >>> >>> Will be rigging up a small power-failure backup system. Essentially a >>> 12V LiFePO4 battery (with BMS in there), inverter, solar panels, MPPT >>> charger. >>> >>> When reading up on stuff and talking to people I found something very >>> disconcerting. It seems that MPPT charge controllers can and often do >>> commit suicide when operated without a battery. It even says that in the >>> manuals a lot. The trigger event can be as simple as the BMS opening for >>> some reason and ... *PHUT* ... MPPT is gone. A friend had that happen, >>> twice. >>> >>> They can also send out a substantial voltage spike or short their >>> MOSFETs during such events, the latter letting full solar panel voltage >>> onto the 12V power bus. Both of which can kill rather expensive gear >>> connected to the 12V bus. >>> >>> Why is that? Are the engineers designing this stuff not very bright? I'd >>> never release a design with such "traits". >>> >>> Does anyone know lower-end MPPT charge controllers (20-40A range) that >>> are properly designed in this respect? Preferably ones that are also low >>> noise so they don't mess with measurements on the lab bench. The Genasun >>> brand is low noise but AFAICT they only come for very low solar panel >>> voltages. >>> >>> Of course, one solution is to build a massive crowbar to blow a fuse and >>> protect the connected gear. The MPPT charger might possibly still die. >> >> The usual result is a warranty claim, which most MPPT charge >> controller manufacturers were honor. That's being generous because >> the documentation and usually a big label on the controller will warn >> the user that charging without a (battery) load is a really bad idea. > > >It usually happens after the rather short warranty has expired. > > >> Lots of examples and horror stories online: >> <https://www.google.com/search?q=solar+do+not+operate+with+battery+disconnected> >> I attempted to get a list of parts replaced, but failed. I'm fairly >> sure all the MOSFETs blew, but there might have been some other parts >> involved. >> > >Yeah, probably the driver IC is toast as well since I am sure the gates >will have punctured as well. > > >> I've seen open batteries cause failures 3 times so far. All three in >> campers running whatever could be thrown together from Amazon sourced >> parts. I've learned from these and two more installs that it pays to >> buy everything from a single manufacturer. My favorite this week in >> Renogy: >> <https://www.renogy.com> >> > >However, I saw this warning on their Rover products as well. > > >> Unfortunately, I couldn't find a schematic of the charger. On my last >> order, I attempted to demand a schematic or I would order from someone >> else. They complied, by sending me a hand scribbled block diagram. >> When I asked for an autopsy on the failed unit(s), they mentioned the >> disconnected battery problem, but not the BMS (battery management >> system). >> >> The problem seems to revolve around the battery disconnect switch, >> breaker or fuse. It's common advised that a high current disconnect >> switch be installed in series with the battery to allow a guaranteed >> disconnect in case of fire. The switch works nicely if the owner >> remembers to turn off the charger or cover the solar panels BEFORE >> disconnecting the battery. There are some other ways to create the >> problem. I have various schemes for preventing the problem, but both >> the owner and manufacturer will void the warranty if I become too >> creative. >> > >There should be ways to correct the flaws in their designs. What I don't >understand is why the majority of MPPT designers can't get this done >properly. When I designed an MPPT charger it wouldn't even have occurred >to me that such a single-fault scenario should cause an issue, and it >didn't. My client tested the heck out of it because they said that >installers make all kinds of mistakes, especially if they are the homeowner. > >Out in the wild one has to assume the worst. For example people >parallel-charging from the unregulated 12V output of an old gas >generator, revving it a bit via the carburetor linkage to "goose the >charge top-off". Then, after a while, the BMS comes ... > > >> My best so far is a big contactor relay that disconnects battery, >> solar panels, vehicle alternator and shore power if any manner of >> fault is detected. So far, on one home system, it seems to work >> except the contacts will sometimes arc weld closed. Arc welding one >> pair of contacts effectively locks the contacts in the closed >> position, makes the idea self-defeating. >> >> Your crowbar solution might work but I suspect that the charger >> MOSFETs will blow before the fuse opens. >> > >How can they blow if the battery side voltage is never allowed to exceed >about 16V because of the SCR? It essentially puts a dead short across >the MPPT output. If it can't even take that the design engineers should >go wash dishes or whatever. I wouldn't call them engineers.Don't forget the other side of power electronics blowing up. Manufacturing itself can screw up a good design. boB
Reply by ●February 9, 20232023-02-09
On 1/31/23 9:52 PM, boB wrote:> On Tue, 31 Jan 2023 13:51:52 -0800, Joerg <news@analogconsultants.com> > wrote: > >> On 1/30/23 7:10 PM, boB wrote:[...]>>> Yes, the input voltage 'shouldn't' go high but the battery side >>> certainly does go high. I think you are right about the input voltage >>> not going high, but having fixed this many years ago, I can't remember >>> the failure mechanism exactly. But it can go at least to the PV's Voc >>> voltage, but much above that, (just a few volts) the PV array should >>> clamp that. >>> >> >> It will go to Voc if the series FET shorts out. So in my case around >> 75V, two residential panels in series. I can also connect them in >> parallel but that requires much beefier cables and extra diodes in case >> one is shaded (they already have bypass diodes). >> > > What I meant was that if it is a bi-directional (synchronous) buck, > then it is possible for the PV input voltage to be raised above Voc > from the battery. The PV array, being a series of diodes, will clamp > the voltage at approximately Voc, or what would be Voc, even at night. > This assumes something is not right of course. >MPPT charger controllers are mostly just buck converters. Very rarely boost or SEPIC, for small stuff. Sometimes bucks are synchronous but usually not intended to be used bidirectionally. What for?> >> >>> What was the nominal PV input voltage of this controller ? Was it set >>> up as a 24V array or 48V array or maybe higher voltage ? That can >>> also make a difference. What kind was it ? I am familiar with a lot >>> of the commercial ones. >>> >> >> I don't know yet. My friend has jury duty out of town and is gone. >> >> >>> It might be also that the battery voltage flying up is shoving current >>> the other way, too fast and even with the indutor in there can over >>> current the FETs. The input capacitors could make a good load for >>> that backwards current and hurt things as well. >>> >> >> I don't quite follow. How can there be a problem if the capacitors on >> the PV side are large enough? >> > > When the battery breaker trips, the energy in the battery side caps > might not be enough to charge the PV input caps very much because of E > = 1/2 CV^2 depending on the capacitance in the controlle >Those caps will never have more than battery voltage so they won't dump anything large into the front side. When the battery vanishes those caps are the only energy storage there is on the output side and preciously little of that since the manufacturers don't want to waste money.> >> >>> Is the controller designed to work backwards reliably, too ? >>> >> >> Very unlikely. The one I designed wasn't either but it was in this >> regard bullet-proof. You could not destroy it by randomly disconnecting >> things. The same should go for commercial ones but obviously they can't >> get it done. > > Non-synchronous is good in that respect. Just a bit less efficient > because of the diode drop on the low side of the 1/2 bridge. >On a good design that doesn't matter.> >> >> >>> What I find is that the PV input caps, with much higher PV voltage >>> than battery voltage, is a great source of blow-up energy for the >>> controller but that is usually from HV to LV battery direction. >>> >>> Ultimately it is the high current from the battery and shorted FETs >>> that can cause the PCB to catch fire if things go very wrong. >>> >> >> That sounds like a seriously flawed design. >> >> > Any design can have this problem if the bottom FETs short AND the > relay contacts weld shorted to the battery. Double fault so that > would be a rare occurrance. >Which relay contacts? The problem with this design flaw is that it is caused by a single fault that might not even be a fault: The BMS opens. This could simply happen because the battery isn't happy about the temperature. Or the amount of peak load. Or as in one case I know of a simple screw terminal connection had come loose. Or whatever. The result in all these cases can be and usually is that the battery "goes away". That fact alone triggering a MPPT charger meltdown is a serious design flaw in my book.> >>> One thing to do is to catch the Vbattery side going too high *FAST* >>> and immediately shut off the converter(s) >>> >> >> Well, I am going to add in a crowbar anyhow to protect my electronics. >> An SCR the size of a gold ball that is capable to reliably blow a 50A >> car fuse if needed. If I put that sans fuse on the MPPT output it would >> present a dead short to the MPPT, like an empty battery. If it doesn't >> survive that then the MPPT is real junk. > > Crowbar should work as long as nothing breaks before the breaker > trips. > > Electronics stuff breaks sometimes. Just have to do our best so if it > does, the damage is low as possible. >Yes, but first we have to ponder and anticipate normal and frequenctly occurring scenarios such as a battery vanishing in the electrical sense. Not doing so is poor engineering. Even if the engineers are younger and may be not yet competent enough, that's what we have design reviews for. Or should have ... -- Regards, Joerg http://www.analogconsultants.com/
