Serially linked DC-DC charge-controllers - can it be done?

[readiescards]

As I understand it per PV panel inverters (commonly called microinverters) offer:

* longer life

* extract more from each panel

* overcome the limitation of one panel in a string of panels being in the shade impacting the output of the entire string

 

Enphase make a solar PV DC-240AC inverter which seems to get good reviews but it is not compatible with offgrid systems - I think because it needs an existing reliable 240ac feed to synchonize with.

 

Thinking on this it occurred to me I could use a set of the small cheap as chips DC-DC charge controller in series to build a 48V battery bank, and have mulitple of these banks in parallel to build up a decent storage.  

 

 

This might give me:

 

* high redundancy

* overcome the issue of shading in solar PV strings

* individual battery monitoring system for zero additional cost

* total charge controller cost of 30 * £15 = £450

* ability to use a cheap basic 48V-240vac inverter that be replaced in the lifetime of the system with minimal cost - circa £450/5kW model 

* ability to replace individual 12v cells with minimal impact

 

 

So why not? Googling I can't find anyone doing this so there must be something I have over looked. I welcome some peer review comments please.

 

 

Links:

* DC-DC charge controller I've been using for 18months with out issue:

http://www.ebay.co.uk/itm/10A-LCD-Solar-Panel-Charge-Controller-12V-24V-Battery-Auto-Regulator-New-J1Z3-/311585702654?hash=item488bf482fe:g:vooAAOSw2ENW624S

* cheap 48v/240acv 5kW inverters

http://www.ebay.co.uk/itm/Pure-Sine-Wave-Inverter-Charger-Sinus-Pro-5000W-48V-240V-15A-AVR-UPS-/222409289155?hash=item33c8a06dc3:g:hNoAAOSw34FVAzYQ

 

 

[Barney12]

1 hour ago, readiescards said:

Enphase make a solar PV DC-240AC inverter which seems to get good reviews but it is not compatible with offgrid systems - I think because it needs an existing reliable 240ac feed to synchonize with.

 

Thats what we have on the two smaller outbuildings. Yes you are right; it relies on powerline technology to synchronise. 

Edited February 16, 2017 by Barney12
typo

[Jeremy Harris]

13 minutes ago, readiescards said:

As I understand it per PV panel inverters (commonly called microinverters) offer:

* longer life

* extract more from each panel

* overcome the limitation of one panel in a string of panels being in the shade impacting the output of the entire string

 

 

I'm not at all sure there's any evidence to support the first two statements, I'm afraid.  The limiting component in pretty much any inverter, of any size, is the life of the capacitors.  They have a tough time, handling high ripple currents and are seriously degraded by heat.  The capacitors in a microinverter mounted under a panel are going to run massively hotter than those in an inverter mounted in a cool location, so are, in my view, more likely to fail early.  They have a limited life even when run cool; perhaps 8 to 12 years would be typical.   Replacing a single inverter mounted away from the panels is relatively quick and easy, how easy will it be to replace under-panel mounted microinverters?

 

The second point is only true if there is shading affecting the array, where microinverters offer a good solution to allowing the whole array to continue to deliver a good output even when some panels are shaded.  If there's no shading then microinverters are not better (in fact may be slightly worse because of slightly increased fixed losses)  then a single inverter in terms of efficiency.

 

The last point is very valid, and is the only real reason for considering microinverters.

 

Edited to add:

 

Just realised that in your drawing there are more points of failure than in a single 48V charge system, so it would be less reliable.  If any one MPPT DC-DC controller goes down then the whole battery array fails.  If any one battery doesn't get charged, then during discharge the battery array will go out of balance and you may well reverse charge the flat battery if you run things down low enough.

Edited February 16, 2017 by JSHarris

[ProDave]

I would add to Jeremy's point above.

 

If you know one panel is subject to daily shading and the rest are not, then that one battery that it charges will consistently get less charge than the others, effectively limiting the capacity of the whole system.
 

[readiescards]

1 hour ago, JSHarris said:

 

If any one battery doesn't get charged, then during discharge the battery array will go out of balance and you may well reverse charge the flat battery if you run things down low enough.

 

'Reverse charge' being simply a partially discharged battery being topped up by the rest of the system or something much worse?

[jack]

2 hours ago, JSHarris said:

I'm not at all sure there's any evidence to support the first two statements, I'm afraid.  The limiting component in pretty much any inverter, of any size, is the life of the capacitors.  They have a tough time, handling high ripple currents and are seriously degraded by heat.  The capacitors in a microinverter mounted under a panel are going to run massively hotter than those in an inverter mounted in a cool location, so are, in my view, more likely to fail early.  

 

Enphase offers a 25 year warranty on their microinverters,  They've published various white papers on capacitor life, which I suppose you can take or leave given they have an interest in a particular story.  I seem to recall reading somewhere that the design of the units is such that even if the capacitors fail, they still work (presumably at lower efficiency and with significantly more ripple).  A cynic might conclude that the capacitors are expected to fail at some point, but the user won't notice the drop in efficiency, so they won't claim on the warranty.

 

2 hours ago, JSHarris said:

Replacing a single inverter mounted away from the panels is relatively quick and easy, how easy will it be to replace under-panel mounted microinverters?

 

Very easy (for us!), but that's because our panels are mounted on frames on our flat roof so can be accessed easily from underneath.  Not so easy for more typical roof mounted systems, for sure!

 

It's been a long time since I discussed this with the installer, but I faintly recall that there's another slight advantage in lower light conditions.  It's to do with the minimum light that allows string inverters to start operating versus microinverters.  I don't recall whether shading is required too, but basically on a cloudy day, there may be periods where the light falling on the panels in a conventional system will not generate enough string voltage for the inverter to start operating.  With microinverters, the light level required for operation is slightly lower.

 

We have a fair bit of shading, so they made sense for us anyway.

[Jeremy Harris]

I've seen the Enphase warranty, but I believe it's limited to replacing the faulty inverter, and doesn't include additional costs, like scaffolding etc that would be needed to get at a pitched roof-mounted array.  It could easily cost a few hundred pounds just to get a single microinverter replaced under warranty I suspect, which may skew the economics.

 

The very best high ripple current, low ESR, 105 deg C rated, electrolytic capacitors aren't going to have any form of warranty beyond a few years, especially when operated at the sort of ripple current that an inverter chucks through them, so I'm inclined to the view that Enphase are being a bit optimistic and taking a risk.

 

The minimum start-up voltage is installation dependent on a non-microinverter set up, but not installation dependent on a micro-inverter system.  Taking our case as an example, we have two strings feeding a single inverter, that has two MPPT front ends.  One string is 12 panels, the other is 13 panels.  The inverter startup threshold voltage is way off the bottom of the useful power curve from the panels, so microinverters would just be a slightly increased loss on our system (there is a fixed loss element that means that 25 microinverters would have a greater loss than our single inverter).  Then again, we have no shading at all, so all 25 panels deliver pretty much the same DC voltage at any time.

[SteamyTea]

No reason why the microinverters cannot be installed away from the modules.  Just more DC cabling.

They can then be put somewhere cold and with the option of fan assisted cooling.

My experience of inverters that cut in at a lower voltage is that they lose out at the higher end.  As inverters are generally undersized in the UK this may, or may not, cause a long term reliability issue.

 

[Jeremy Harris]

Worst case for an inverter is probably similar to that for a switched mode power supply (similar topology internally) and that's part load, rather than full load.  However, temperature plays an absolutely massive part in determining capacitor life, so if running at full load heats up the enclosure and the capacitors, then it may well over-ride the impact of the higher ripple current at part load.

 

Keeping an inverter as cool as possible is a pretty much guaranteed way to maximise its life, because the most common failure more is capacitor degradation followed quickly by switching semiconductor failure (usually IGBTs, but some use arrays of HV MOSFETS still, I think).  The semiconductor failure is due to the commutating capacitors not shuttling enough current, because their capacity has reduced and ESR has increased from ageing.  FWIW, it's the same failure mode and cause as electric car drive electronics.  It's interesting to see the lengths that EV manufacturers go to in order to cool the inverters - mine has it's own liquid cooling system, complete with a separate radiator and circulating pump, just to cool the inverters, nothing more.

Edited February 16, 2017 by JSHarris

[readiescards]

My panel connections are easily accessible - being in the roof space rather than under the panel thanks to the GCE integration panels allowing the cables to punched through the breathable membrane without impacting the rain water protection - recently tested and no leaks :-)

 

My understanding was that an invertor/charge controller per panel had an easier life as the current is much reduced.

 

If I understand the discussion points (and thanks all for the input) what I propose is NOT functionally impossible, it just might not be the most efficient/reliable solution?

[SteamyTea]

3 minutes ago, readiescards said:

My understanding was that an invertor/charge controller per panel had an easier life as the current is much reduced.

That just depends on correct sizing really.

[Jeremy Harris]

 

Just now, SteamyTea said:

That just depends on correct sizing really.

 

Absolutely.  A smaller inverter needs smaller capacitors but in terms of ripple current, the percentage relative to the size stays the same, so the stress is the same.

 

 

[MikeSharp01]

Perhaps I am over thinking this but what would be difficult about creating a 50Hz signal for the inverters to use as sync. Presumably it would not need any power behind it so could be quite simple. Depending on how far off grid you are you might even be able to extract it from the eather. ? (and / or tie it back to a GPS derived clock.)

[Jeremy Harris]

7 minutes ago, MikeSharp01 said:

Perhaps I am over thinking this but what would be difficult about creating a 50Hz signal for the inverters to use as sync. Presumably it would not need any power behind it so could be quite simple. Depending on how far off grid you are you might evenue be able to extract it from the eather. ? (and / or tie it back to a GPS derived clock.)

 

It's sadly not that easy.  Normal (rather than non-grid tie) inverters measure the impedance of the supply, so know when they are connected to the mains (very low impedance) and when they are only connected to something like another inverter or generator.

 

There are ways around this, and companies like SMA make special inverters that can be run in "island" mode, without a grid tie.  The Sunny Boy is one of those often used to do just this, but from what I can gather it isn't the easiest thing to set up.  There are details here: http://www.sma-uk.com/residential-systems/solar-system-off-grid.html

 

Edited February 16, 2017 by JSHarris

[jack]

36 minutes ago, JSHarris said:

I've seen the Enphase warranty, but I believe it's limited to replacing the faulty inverter, and doesn't include additional costs, like scaffolding etc that would be needed to get at a pitched roof-mounted array.  It could easily cost a few hundred pounds just to get a single microinverter replaced under warranty I suspect, which may skew the economics.

 

You can look at the warranty in those terms, yes, but you can also view it as a measure of confidence in the product.  If they fail fast, then they'll need to be replaced one way or the other.  Even if the disgruntled home owner has to stump up the extra costs for installation, Enphase will still bear the replacement costs of the unit(s).  It's not as though someone will just leave a dead panel up there for 20 years (will they?)  

 

Bear in mind that the software that comes with the units lets you look at the per-panel microinverter performance over its lifetime, so it isn't as though a dead microinverter can hide anywhere.

 

44 minutes ago, JSHarris said:

The very best high ripple current, low ESR, 105 deg C rated, electrolytic capacitors...

 

Exactly what they use, apparently.

 

A bit of reading, again on the understanding that these are the manufacturer's claims (TL;DR - since the capacitors are run at significantly lower temps than their rating, their lifetime is massively increased): 

 

https://enphase.com/sites/default/files/EnphaseElectrolyticCapacitorLife.pdf

https://enphase.com/sites/default/files/Electrolytic_Capacitor_Expert_Report.pdf

 

Take it with a grain of salt, but they apparently claim a MTBF of 330 years based on accelerated testing.  

 

I did a lot of research at the time and I couldn't find any credible evidence of significant failures of these products.  I'll admit it's early days though - these things have only been on the market for 8 years or so, so it's only about now we might expect to start seeing significant numbers of failures..

[Jeremy Harris]

All I can say is that I've designed and built a few three phase brushless motor controllers over the years (hobby use, and a BLDC controller/inverter uses the same power component topology as a PV inverter, except it's three phase) and it's always the commutation capacitors that are the toughest parts to source and the cause of pretty much all switching device failures.  The very best low ESR, high ripple current, 105 deg rated electrolytic capacitors I can buy have a 10 year nominal life, with no warranty over a year. 

 

The manufacturers do publish data on life versus temperature and it's highly non-linear, with relatively small decreases in temperature having a disproportionately large increase in expected life, but none give any data out beyond 15 years, which makes me think that Enphase are winging it a bit with their 25 year warranty.  It may well be that they've over-specced the ratings of the switching devices so that they can handle the increased stress as the commutation capacitors degrade and fail, but efficiency will drop a fair bit and the failure point then moves from the capacitors to the switching devices.  I know that this is the approach some Chinese brushless motor controller designs use, they just massively over-rate the switching devices and fit cheap and nasty capacitors that will undoubtedly degrade pretty quickly.  It's not an approach that is practical with big inverters, though, as they are already using banks of switching devices to get the required power handling, and increasing the number of paralleled devices brings with it another set of reliability issues.

 

 

Edited February 16, 2017 by JSHarris

[jack]

I don't doubt that you understand the main issues better than a fraction of a percent of the people reading this, but there may be other things they do that take them beyond basic brushless motor controllers.  Presumably their MTBF estimates aren't entirely grabbed out of thin air.

 

Frankly, I'm far more concerned about whether they'll still be here in 25 years to honour the warranty.  There's also the not-insignificant issue of whether what they can supply in, say, 10 years' time is compatible with the modules I currently have installed.

[Jeremy Harris]

4 hours ago, jack said:

I don't doubt that you understand the main issues better than a fraction of a percent of the people reading this, but there may be other things they do that take them beyond basic brushless motor controllers.  Presumably their MTBF estimates aren't entirely grabbed out of thin air.

 

Frankly, I'm far more concerned about whether they'll still be here in 25 years to honour the warranty.  There's also the not-insignificant issue of whether what they can supply in, say, 10 years' time is compatible with the modules I currently have installed.

 

The problem is that, in this case, the quoted MTBF seems to defy the known characteristics of a pretty well-modelled component, that's my concern.   Few suppliers of consumer-grade components like this will offer more than around 100,000 to 150,000 hours MTBF, and that's over a restricted temperature range and takes no account of the cumulative effects of having multiple components in a bit of equipment.  MTBFs can be misleading, too, as it depends how they've been calculated.  I have a suspicion that, in this case, the well-known failures of the Enecsys microinverters may possibly have coloured the way Enphase have chosen to market their product!

 

Reliability is quoted in terms of a statistical probability of failure, usually cumulative, and is always given under a specified set of operating conditions (not a range of them).  When you have components that have a non-linear reliability, or life, with a variation in one, or more, operating conditions, then you can quote an MTBF for the "best" operating conditions, yet be sure that it will be a fraction of that for the "worst" operating conditions.  Service life will generally be quoted as a number of years, but that does not mean that components won't fail within that period of time.  If you keep a bit of equipment at a constant cool temperature it's service life will pretty much always be extended, often dramatically.  It's one reason that places like data centres often feel a bit chilly if you're walking around inside one, as they provide masses of cooling both to remove all the heat the things generate and to improve equipment reliability.

 

PV inverters are basically just switched mode converters, that use commutation and filter capacitors under conditions that are pretty near identical to things like data centre power supplies, fixed telecoms power supplies, brushless motor power supplies etc, in terms of switching frequency, ripple current etc.  They are also essentially a consumer-grade device, they have to be in order to be cost-effective in a competitive market.  A consequence of this is that they use consumer-grade components, rather than high-reliability "aerospace" or "military" grade components.  There are some "ultra long life" electrolytic capacitors around, some with a predicted life of > 20 years, but they are expensive, and that long life is at the expense of a reduced ripple current rating and a relatively high ESR (the equivalent series resistance of the capacitor) both parameters that are critical to performance in a switched mode inverter, converter, power supply, or whatever. 

 

The effect of heat and ripple current on the life of electrolytic capacitors is well-modelled, and so the operators of equipment that needs to be very reliable (something like mobile phone mast power supplies, or data centre power supplies) will have routine replacement as a part of the maintenance regime to maintain the required reliability.   One of the biggest drivers in that part of their maintenance regime will be the capacitors in the supplies, as they are the single most failure-prone component in this type of system (I have the figures somewhere, but off the top of my head I seem to remember that capacitors are around 5 or 6 times more likely to fail than switching semiconductors).  Emerson have written a paper that gives a reasonably clear explanation as to why it can be challenging to determine how long capacitors will last in switched mode supplies: http://www.repeater-builder.com/tech-info/pdfs/replacing-capacitors-from-emerson-corp.pdf  which is worth a read.  It deals with high-reliability components, rather than consumer-grade components found in domestic equipment, but the principles are pretty much the same.

 

EDITED TO ADD:

 

I've just pulled off the Enphase (USA) research papers they use to justify the expected life and read them right through.  All I can say is that there is one hell of a lot of optimism in there!  They are using capacitors with a service life of 10,000 hours at the rated temperature, then arguing that because they have many smaller capacitors in parallel the service life, and reliability, increases.  I'm no expert on the statistics of failure, but I'm not convinced by the logic of their argument.  Happy to be put right by one of our resident statisticians, though!

 

The other point in the research from 2008 is that if the capacitors degrade and lose capacitance, as they will, the inverters will continue to work at a lower efficiency, but don't, technically, "fail".  It seems they have achieved this by the method I suggested they may have used, they've over-rated the switching devices so they can cope with the reduced capacity and increased peak switching current.   It seems similar to the warranty on the Nissan Leaf battery, which says that (I think, need to check) that the battery capacity is allowed to drop by 20% and the battery is still considered to be serviceable and not elegible for replacement under warranty.

 

Finally, the capacitors they are using are exactly the same as the ones I'm using in my BLDC controllers, Nichicon low ESR ones, and the research papers were written to help obtain venture capital.

 

For those interested, the two papers I can find on this are here: https://enphase.com/sites/default/files/EnphaseElectrolyticCapacitorLife.pdf and here: https://enphase.com/wp-uploads/enphase.com/2011/03/Electrolytic_Capacitor_Expert_Report.pdf

Neither seems to have been peer reviewed as far as I can find out.

 

 

The question of whether any supplier will be around in 25 years time is a very, very good one.  How many hardware technology companies stay around in their original form for more than 10 years?  It's one reason I steered clear of the mobile 'phone app controlled home automation stuff, as I suspect some of it has a life of less than 10 years, if only because the hardware and software on mobile phones is likely to change dramatically over that time and render older applications unworkable, much as has happened with PCs and tablets.  I have a perfectly good scanner that became unusable because Microsoft chose to change the way device drivers work from with the shift from XP to Win 7; when my wife upgraded iOS on her iPad the data link hardware (that was only around 4 months old) stopped working.

 

As a final point, I'm far from anti-microinverters, and if you're going for a microinverter system there is no doubt in my mind that the Enphase ones are the most reliable on the market, as borne out by their low failure rate over the past 8 years.

 

 

Edited February 16, 2017 by JSHarris
finished reading the Enphase research papers, and again to post links to the papers