PIP-4048MS and PIP-5048MS inverters

non-EV Solar, Wind and other renewable power sources
non-EV batteries and other energy storage stuff
Forum rules
Important!
This forum is for discussion of Non-EV matters.
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

PIP-4048MS inverter

Post by weber »

offgridQLD wrote:Are you thinking now of just having some kind external control over LVD and just leaving the battery V readings/setpoints inside the inverter as is?
Yes. But I'm still undecided. The problem is that the only possibly-useful "external control over LVD" I can find consists in sending a command to the inverter to raise the voltage at which it switches from battery to AC input. But there are two problems with this:
1. The maximum value we can set it to is 51 volts and it's possible that a single low-capacity cell could be in trouble at a battery voltage higher than this.
2. It's not clear from the manual whether it will ever switch the loads to the AC input when there is no AC voltage there, even when its source priority is set to "SBU" (Solar, then Battery, then Utility).

Mine is still in pieces, and I don't really want to put it back together until I've made a decision about voltage scaling. Is it easy for you to give yours a battery voltage less than 51 volts and no AC input, set parameter 01 to "SbU", run some load, then set parameter 12 to "51" and tell me if the load goes off?
When you first mentioned making changes in the inverter. My initial reaction was. Once you started fooling one reading in the inverter it could be a snow ball/ knock on effect where all other voltage dependent set points would need compensating one way or another. Perhaps adding work/complexity to achieve the initial simple task LVD.
Those are reasonable fears, that I share.
Or am I lost where you going with this or the initial reason for the tweak?
In referring to this as "LVD" (low voltage disconnect?) you may be missing what I said earlier, that I don't want to disconnect the battery at this point (which would be easy enough to do with a contactor). I only want to disconnect (or turn off) the inverter-proper which is a load, not the MPPT or AC input which are charge sources.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
offgridQLD
Senior Member
Posts: 1827
Joined: Tue, 23 Jul 2013, 16:05
Real Name: Kurt
Location: Fleurieu Peninsula, SA

PIP-4048MS inverter

Post by offgridQLD »

No problem at all testing the inverters AC transfer voltage set point and how it reacts (without a AC feed).

The only issue is my pip4048 is up at Maleny Image My wife had enough of me carting it back and forth each week in the Imiev though my daughter thought the box was a handy pillow to fall asleep on in the back seat Image.

Can test it for you this Friday night if that's any help.


Yes, LVD, low voltage disconnect. I was on track just missing the point that with a all in one design like the PIP4048 you just want the AC supply from the inverter and therefor it's DC load on the battery to stop at a set Voltage but keep the unit running charge controller and I guess the brains of the unit its self so it could seamlessly recover without a hard reset?

Kurt

Last edited by offgridQLD on Mon, 03 Nov 2014, 07:22, edited 1 time in total.
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

PIP-4048MS inverter

Post by weber »

offgridQLD wrote:Can test it for you this Friday night if that's any help.
Thanks. I'll let you know if the need passes.
I was on track just missing the point that with a all in one design like the PIP4048 you just want the AC supply from the inverter and therefor it's DC load on the battery to stop at a set Voltage but keep the unit running charge controller and I guess the brains of the unit its self so it could seamlessly recover without a hard reset?

Yes.

Here's a thought. Why don't I just ask the manufacturer if they can change the firmware to allow both parameter 12 ("back to utility" voltage) and parameter 29 (low DC cutoff voltage) to be set as high as 53 volts, at least when set via the serial interface (PBCV and PSDV commands), as this will make their inverter more generally useful with LiFePO4 batteries.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
T1 Terry
Senior Member
Posts: 1952
Joined: Thu, 30 Sep 2010, 20:11
Real Name: Terry Covill
Location: Mannum SA

PIP-4048MS inverter

Post by T1 Terry »

weber wrote:
T1 Terry wrote:Once the system is designed to produce enough input to equal the output plus a bit to get the system back to full after non optimum days, then any gain due to reduced operating temp and use of a complex controller will achieve what? The battery fully charged 20% earlier in the day? I can't see where this is of any advantage.
There are so many misconceptions here about fairly basic stuff, and this is somewhat off-topic, but here goes:

There is no such thing as "plus a bit to get the system back to full after non-optimum days". It's not unusual to have a whole rainy week where the output of your PV array is only 10% of what it is on the best days. Since we can't afford to have a week's worth of battery storage, your "plus a bit" would be plus about 900%. So what we do is design the system to balance its initial cost against the cost (including the environmental and social costs) of backup sources -- typically burning fossil fuels, or the inconvenience of voluntary austerity during those times.

So there will always be times when the extra power available from an MPPT will be useful. And on good days, having your (lead-acid) batteries reach full charge earlier does increase their life. And as Kurt says, the extra cost of an MPPT is no longer significant.

The only reasons I can come up with not to use an MPPT are if you want to have more chance of fixing it yourself if it breaks down, or if you need to locate it somewhere where the extra heat would be a problem, and so you find it preferable to waste even more power as heat in your panels instead.
The system must be designed around the fact the cell efficiency will be less during the hottest part of the yr, summer, yet the system load is likely to be the highest due to air cond use for extended periods. An MPPT controller can not improve this hot panel output, it only optimise output when the panels produce more due to colder operating conditions, but this is in excess to the systems designed requirements and as it has no other value such as resale to the grid, it becomes surplus to requirement, so of no value at all.
None of my customers run aircon off-grid. They design or modify their homes not to need it. But yes, refrigeration loads are higher in summer. But of course in winter the days are shorter so the available solar resource is lower and lighting loads are higher. In Brisbane, an unshaded array tilted north at 27.5 degrees receives on average 30% less energy per day in May, June, July, compared to Dec, Jan, Feb. And the saving on refrigeration in winter does not typically amount to anything like 30%. So an MPPT can help make up that winter shortfall.

Even on a clear midsummer day your panels are only at their hottest during the middle couple of hours. Either side of those hours the panels are cooler so there is extra power that can be harvested by an MPPT, which may allow a slightly smaller array or extend your battery life.
As far as reduced copper costs due to reduced cable size, not sure what the cost balance is here as the cabling must now handle a much higher voltage alone with additional fusing/breaker costs to accommodate this higher voltage. As the MPPT controller will not handle a full series string voltage, there has not been a great reduction in the required cabling volume, so the reduced costs would be minimal I would think.
All UV stabilised solar cable that I am aware of is good for 600 volts DC. And thanks to the explosion in the grid-connect market, DC fuses and circuit breakers are no longer very expensive. And the higher the array voltage, the fewer strings you need and therefore the fewer string protection devices you need. In fact if you have two strings or less you don't need any string protection, although you still need array cable protection from the battery.

For a given array power and cable route length, the amount of copper required to maintain the same percentage volt drop and hence the same percentage power loss, goes down as the square of the voltage. i.e. if you can double the array voltage you only need 1/4 of the copper.

Yes, the specific MPPT in the PIP-4048MS is very poor in this regard since it barely allows a 1.5 times increase in nominal array voltage over battery voltage, but most MPPTs allow at least 2 times. Phew. Back on topic at last. Image
Ouch   Image I see the MPPT topic is even dangerous on this forum Image The reason I was going down the 2 panel in series path V the 3 panels in series and trying to show the gains were not that critical was a work around for the over voltage potential with 3 panels in series.
Now I'll pull my helmet back over my ears and head back to the bunker Image

T1 Terry
Green but want to learn
User avatar
offgridQLD
Senior Member
Posts: 1827
Joined: Tue, 23 Jul 2013, 16:05
Real Name: Kurt
Location: Fleurieu Peninsula, SA

PIP-4048MS inverter

Post by offgridQLD »

"Here's a thought. Why don't I just ask the manufacturer if they can change the firmware to allow both parameter 12 ("back to utility" voltage) and parameter 29 (low DC cutoff voltage) to be set as high as 53 volts, at least when set via the serial interface (PBCV and PSDV commands), as this will make their inverter more generally useful with LiFePO4 batteries. "


You know I think you would be the guy with the best chance of convincing them.

I know the (reverse engineering) is a signature strong point and kind of fun but......

I think a relationship with the manufacture and improving the units general capability (out of the box ) has to be positive outcome for everyone.

I think if you present your case well I have the general feeling they are willing to listen to there customers and offer improvements.I have already notice they have made small tweaks to the units for distributors reselling the product under there own branding.

I would suggest a short list of shortcomings the units have that might be able to be Quick fixes from there end. As often having them fix a few issues on one go is easier as they can release the update and kill a few birds with one stone.

I know there are few people kicking around that from my observations have little understanding of the real inner workings of the units (me included) but have offered to Betta test them and have been well supported - given free samples. I'm sure they would get good value for money if they did the same for you guys.

Depending on how much involvement with the units you want to have.

Wont hurt trying.

Kurt





Last edited by offgridQLD on Mon, 03 Nov 2014, 09:55, edited 1 time in total.
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

PIP-4048MS inverter

Post by weber »

T1 Terry wrote:Ouch   Image I see the MPPT topic is even dangerous on this forum Image The reason I was going down the 2 panel in series path V the 3 panels in series and trying to show the gains were not that critical was a work around for the over voltage potential with 3 panels in series.
Now I'll pull my helmet back over my ears and head back to the bunker Image
Sorry Terry. Your suggestion that I use 9 strings of 2 instead of 6 strings of 3 with the PIP-4048MS is a valid one, and one I'm still considering.

And your suggestion of cross-connecting between strings to reduce shading losses (or equivalently making chains of panels (in parallel) then connecting these chains in series) is also valid, and a new one to me. Thanks.

But I'm still trying to come to grips with the additional cost of the large amount of extra cable and large number of extra fuses or circuit breakers this would require, versus how much extra power it is likely to give under the shading conditions it is likely to experience.

This is extremely difficult to model, given that a solar panel is nothing like a battery. A battery can be modelled very simply as a voltage source with a series resistance. A solar panel has 72 cells in series in 3 groups of 24 with a bypass diode around each group. Then each cell needs to be modelled as a current source shunted by a diode (pointing the opposite way to the bypass diodes) and both shunt and series resistances, and the diodes change their VI curves with temperature.

But it is still useful to consider the crude model where we imagine that a panel is either unshaded or it is 100% shaded and treated as being nothing but its bypass diodes.

Here's how I was planning to connect the 18 panels [_]. The lowercase x's are the poles of double-pole circuit breakers. The minus and plus signs are wires and connections, and do not indicate polarity.
        1     2     3
     +-[_]---[_]---[_]-+        a
 +-x-|                 |-x-+
 |   +-[_]---[_]---[_]-+   |    b
 |                         |
 |   +-[_]---[_]---[_]-+   |    c
-+-x-|                 |-x-+-
 |   +-[_]---[_]---[_]-+   |    d
 |                         |
 |   +-[_]---[_]---[_]-+   |    e
 +-x-|                 |-x-+
     +-[_]---[_]---[_]-+        f
But they will be physically laid out, using all available roof space, as (not to scale):
Palm tree                                                        Brush box tree
               a             b             c             d
Inverter +[_]-[_]-[_]+ +[_]-[_]-[_]+ +[_]-[_]-[_]+ +[_]-[_]-[_]+
                                         N
               e             f         W   E
         +[_]-[_]-[_]+ +[_]-[_]-[_]+     S
What I could do relatively easily would be to cross connect only adjacent pairs of strings as follows. That would require no additional circuit breakers, but an extra 6 x 2m cables, 6 x MMF connectors and 6 x FFM connectors, costing around $160.
        1     2     3
     +-[_]-+-[_]-+-[_]-+        a
 +-x-|     |     |     |-x-+
 |   +-[_]-+-[_]-+-[_]-+   |    b
 |                         |
 |   +-[_]-+-[_]-+-[_]-+   |    c
-+-x-|     |     |     |-x-+-
 |   +-[_]-+-[_]-+-[_]-+   |    d
 |                         |
 |   +-[_]-+-[_]-+-[_]-+   |    e
 +-x-|     |     |     |-x-+
     +-[_]-+-[_]-+-[_]-+        f
But considering the way that shade patches move from west to east across the array during the day, the big brush box in the morning and the palm in the afternoon, I can't see that it would give much benefit in terms of alternate current paths.
Last edited by weber on Mon, 03 Nov 2014, 11:35, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

PIP-4048MS inverter

Post by weber »

Wow! That was quick.

I wrote:
Subject: PIP-4048MS
I am working with John XXXXXX, who you kindly sent the serial protocol to. Is it possible you could update the firmware to allow parameters 12 and 29 ("back to utility" and "low dc cutoff") to be set as high as 53 volts? Even if this was only allowed via the serial protocol (PBCV and PSDV commands) it would make your inverter much more useful with LiFePO4 batteries having battery management systems. Otherwise we cannot protect one cell that goes low before all the others.
MPP Solar wrote:Dear Dave

Thanks for your mail.

I'm sorry, but we're not able to consider adjusting these parameters to as high as 53v as doing so may trigger other faults e.g. er04 (battery too low) as this narrows the the gap between "normal battery voltage" and "low cut-off" significantly and as as a result inverter cannot correctly interpret/display the battery capacity.

best
Eric
Last edited by weber on Mon, 03 Nov 2014, 11:41, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
offgridQLD
Senior Member
Posts: 1827
Joined: Tue, 23 Jul 2013, 16:05
Real Name: Kurt
Location: Fleurieu Peninsula, SA

PIP-4048MS inverter

Post by offgridQLD »

I have to say the battery capacity (SOC) monitor in the software is completely useless anyhow. it's pure voltage based. No shunt reading or compensation setting for charge efficiency or temperature compensation, like some of the better shunt based battery SOC meter's use. and even they are just a guide.

It would be no loss at all if they scrapped the SOC feature from the software all together.

Perhaps they are not use to the dealing with the stiff voltage that lithium batterys have under load.

kurt



Last edited by offgridQLD on Mon, 03 Nov 2014, 11:55, edited 1 time in total.
User avatar
coulomb
Site Admin
Posts: 6357
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

PIP-4048MS inverter

Post by coulomb »

weber wrote: This is extremely difficult to model, given that a solar panel is nothing like a battery. A battery can be modelled very simply as a voltage source with a series resistance. A solar panel has 72 cells in series in 3 groups of 24 with a bypass diode around each group. Then each cell needs to be modelled as a current source shunted by a diode (pointing the opposite way to the bypass diodes) and both shunt and series resistances, and the diodes change their VI curves with temperature.

So why not model it in LTspice or something similar that is free? It has current sources, diodes, and so on. Sounds like it would be well within its capabilities.
MG ZS EV 2021 April 2021. Nissan Leaf 2012 with new battery May 2019.
5650 W solar, 2xPIP-4048MS inverters, 16 kWh battery.
Patching PIP-4048/5048 inverter-chargers.
If you appreciate my work, you can buy me a coffee.
T1 Terry
Senior Member
Posts: 1952
Joined: Thu, 30 Sep 2010, 20:11
Real Name: Terry Covill
Location: Mannum SA

PIP-4048MS inverter

Post by T1 Terry »

weber wrote:
T1 Terry wrote:Ouch   Image I see the MPPT topic is even dangerous on this forum Image The reason I was going down the 2 panel in series path V the 3 panels in series and trying to show the gains were not that critical was a work around for the over voltage potential with 3 panels in series.
Now I'll pull my helmet back over my ears and head back to the bunker Image
Sorry Terry. Your suggestion that I use 9 strings of 2 instead of 6 strings of 3 with the PIP-4048MS is a valid one, and one I'm still considering.

And your suggestion of cross-connecting between strings to reduce shading losses (or equivalently making chains of panels (in parallel) then connecting these chains in series) is also valid, and a new one to me. Thanks.

But I'm still trying to come to grips with the additional cost of the large amount of extra cable and large number of extra fuses or circuit breakers this would require, versus how much extra power it is likely to give under the shading conditions it is likely to experience.

This is extremely difficult to model, given that a solar panel is nothing like a battery. A battery can be modelled very simply as a voltage source with a series resistance. A solar panel has 72 cells in series in 3 groups of 24 with a bypass diode around each group. Then each cell needs to be modelled as a current source shunted by a diode (pointing the opposite way to the bypass diodes) and both shunt and series resistances, and the diodes change their VI curves with temperature.

But it is still useful to consider the crude model where we imagine that a panel is either unshaded or it is 100% shaded and treated as being nothing but its bypass diodes.

Here's how I was planning to connect the 18 panels [_]. The lowercase x's are the poles of double-pole circuit breakers. The minus and plus signs are wires and connections, and do not indicate polarity.
        1     2     3
     +-[_]---[_]---[_]-+        a
 +-x-|                 |-x-+
 |   +-[_]---[_]---[_]-+   |    b
 |                         |
 |   +-[_]---[_]---[_]-+   |    c
-+-x-|                 |-x-+-
 |   +-[_]---[_]---[_]-+   |    d
 |                         |
 |   +-[_]---[_]---[_]-+   |    e
 +-x-|                 |-x-+
     +-[_]---[_]---[_]-+        f
But they will be physically laid out, using all available roof space, as (not to scale):
Palm tree                                                        Brush box tree
               a             b             c             d
Inverter +[_]-[_]-[_]+ +[_]-[_]-[_]+ +[_]-[_]-[_]+ +[_]-[_]-[_]+
                                         N
               e             f         W   E
         +[_]-[_]-[_]+ +[_]-[_]-[_]+     S
What I could do relatively easily would be to cross connect only adjacent pairs of strings as follows. That would require no additional circuit breakers, but an extra 6 x 2m cables, 6 x MMF connectors and 6 x FFM connectors, costing around $160.
        1     2     3
     +-[_]-+-[_]-+-[_]-+        a
 +-x-|     |     |     |-x-+
 |   +-[_]-+-[_]-+-[_]-+   |    b
 |                         |
 |   +-[_]-+-[_]-+-[_]-+   |    c
-+-x-|     |     |     |-x-+-
 |   +-[_]-+-[_]-+-[_]-+   |    d
 |                         |
 |   +-[_]-+-[_]-+-[_]-+   |    e
 +-x-|     |     |     |-x-+
     +-[_]-+-[_]-+-[_]-+        f
But considering the way that shade patches move from west to east across the array during the day, the big brush box in the morning and the palm in the afternoon, I can't see that it would give much benefit in terms of alternate current paths.

Is there any chance of moving one of the top row to either between the 2 rows or to the bottom row? Group D would not see sun till late morning at best, would this be correct/ Group A would be shaded from early afternoon onwards, would this be correct?

T1 Terry
Green but want to learn
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

PIP-4048MS inverter

Post by weber »

coulomb wrote:So why not model it in LTspice or something similar that is free? It has current sources, diodes, and so on. Sounds like it would be well within its capabilities.

Because I fear it would take me a long time to learn how to use it, and then even longer to debug the model. And even then I wouldn't know what data to use as input to the model, i.e. how to model the current-source current and diode-temperature for every cell over a "typical" day.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
coulomb
Site Admin
Posts: 6357
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

PIP-4048MS inverter

Post by coulomb »

weber wrote: Because I fear it would take me a long time to learn how to use it,
Well, I find it a pig to learn. But I can do the basics now.
... And even then I wouldn't know what data to use as input to the model, i.e. how to model the current-source current and diode-temperature for every cell over a "typical" day.

Ah, my mistake. I assumed you knew what you were doing Image

Edit: but seriously, would you really need to be all that accurate with temperatures etc? I suppose so, since the two options would be pretty close, otherwise it would be obvious by inspection which option would be better.
Last edited by coulomb on Mon, 03 Nov 2014, 15:36, edited 1 time in total.
MG ZS EV 2021 April 2021. Nissan Leaf 2012 with new battery May 2019.
5650 W solar, 2xPIP-4048MS inverters, 16 kWh battery.
Patching PIP-4048/5048 inverter-chargers.
If you appreciate my work, you can buy me a coffee.
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

PIP-4048MS inverter

Post by weber »

T1 Terry wrote:Is there any chance of moving one of the top row to either between the 2 rows or to the bottom row?
It's a flat roof and the panels will be on tilt stands at 25 degrees in portrait orientation, one high. So the gap between the rows is necessary to avoid self shading on midwinter mornings and afternoons.

But it would be just barely possible to have a row of 11 and a row of 7 (instead of 12 and 6). Is that what you mean? Although the 7th would be getting rather close to the chimney.

I've just remembered that what I've shown as North above, is really NNE or about 10 degrees. In fact, here's a Google Earth image with north up the page.

Image

This photo happens to have been taken on Armistice day 2011 (11-11-11) and by the look of the shadows it might well have been taken during the two minutes silence (11:00 to 11:02 am).

The existing old array is in the position that new strings A and B would occupy. You can see the chimney as a circular blob to their south-west. You can also see the motley shade from the Brush Box. The shade from the Palm tree is a larger blob on the ground, in a corner, about the same distance north-east of the existing array (that the chimney is to the south-west). A verandah roof surround the main roof on all sides, but it is not structurally sound enough for mounting PVs.
Group D would not see sun till late morning at best, would this be correct/ Group A would be shaded from early afternoon onwards, would this be correct?

That's about right. The morning shade is widespread but it is far from solid, while the afternoon shade is small but solid, and it will depend on what time of year as to what it actually hits when.
Last edited by weber on Mon, 03 Nov 2014, 18:05, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

PIP-4048MS inverter

Post by weber »

coulomb wrote:but seriously, would you really need to be all that accurate with temperatures etc? I suppose so, since the two options would be pretty close, otherwise it would be obvious by inspection which option would be better.

My thinking was, if you're going to simulate the panels to the level of individual cells then you'd want to model the patchy shade to that kind of resolution as well. But yes, it would probably be safe enough to assume all the cells in one panel are at the same temperature.

But as I said, much cruder panel-level models are worth considering. And I was about to do so, despite its off-topic-ness, when you smacked me one in the eye, presumably because I beat up on poor Terry. Image

Consider these two strings in parallel, without cross-ties. An ">" means a shaded panel, which we can treat as an ideal diode (the 3 bypass diodes in series) since it has no current source and the cell diodes can be ignored as they will be cold and therefore have a higher voltage (much lower current at the same voltage) than their hot counterparts.

An unshaded panel can be treated as a single current source in parallel with a string of 72 hot silicon diodes in series, wasting some of the current from the current source, depending on their temperature and the voltage the MPPT operates them at. Their bypass diodes can be ignored as they will be reverse biased.
        1     2     3
   A +-[>]---[_]---[_]-+
 ----|                 |----
   B +-[_]---[_]---[_]-+
     :                 :
I find it very hard to wrap my head around this stuff, so don't hesitate to question if you think I've got it wrong.

If the MPPT operates at a voltage that suits 3 series panels (3S) it will collect power only from 3 of these panels (B1,2,3). If instead it operates at a 2S voltage it will collect 4 panels worth (A2,3 and two thirds from each of B1,2,3). But with more unshaded strings in parallel there will be no benefit in 2S operation and so we will operate at 3S and we have effectively just lost the whole string that has a shaded panel.

So it's best if the shade progressively takes A1,2,3 before starting on B.

If the strings are cross-tied
        1     2     3
   A +-[>]-+-[_]-+-[_]-+
 ----|     |     |     |----
   B +-[_]-+-[_]-+-[_]-+
     :                 :
It makes very little difference. With 3S MPP we will get slightly more than 3 panels worth. We'll get B1 and a little over half each from A2,3 and B2,3. B1 will not pass more than its usual current and this will be split between the others. Because they are only being called on to provide half their possible current they will be able to operate at a slightly higher voltage than otherwise.


If instead shade takes A1 and B1, we've lost two strings.
        1     2     3
   A +-[>]---[_]---[_]-+
 ----|                 |----
   B +-[>]---[_]---[_]-+
     :                 :
And it makes no difference at all if they are cross-tied.
        1     2     3
   A +-[>]-+-[_]-+-[_]-+
 ----|     |     |     |----
   B +-[>]-+-[_]-+-[_]-+
     :                 :
Where cross ties really make a difference is when you have diagonal or checker-board shading like this. Without cross ties you lose both strings.
        1     2     3
   A +-[>]---[_]---[>]-+
 ----|                 |----
   B +-[_]---[>]---[_]-+
     :                 :
With cross ties you get one whole string made up from parts of the two strings, by zig-zagging back and forth. But how likely is it that we can arrange for shading to happen like that?
        1     2     3
   A +-[>]-+-[_]-+-[>]-+
 ----|     |     |     |----
   B +-[_]-+-[>]-+-[_]-+
     :                 :
The idea of wiring an array of 18 panels so that what is electrically a checkerboard, gets progressively shaded, does my head in.
Last edited by weber on Mon, 03 Nov 2014, 18:03, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

PIP-4048MS inverter

Post by weber »

I think this system will be fine with standard wiring. On clear days I have an array oversize factor (relative to daily load) of 3 to 4 depending on the season. So it only needs for two strings of 3 panels to be unshaded at any given time. And when it is overcast, there are no discrete moving shadows. The light is coming pretty much equally from all parts of the sky, and all that matters is how much sky the array can "see".

But that does put paid to the idea of running a heat-pump storage hot water system using the "excess" power on clear days. Probably better to install a standard solar-thermal hot water system.

Back to the topic of the PIP-4048MS inverter. I emailed the following to Eric at MPP Solar.
Hi Eric,

Thanks for your rapid response. I am a designer and installer of off-grid solar power systems in Australia, and I would like to help you make the PIP-4048MS suitable for use with batteries consisting of 16 LiFePO4 cells. The document linked below shows the relationship between voltage and state-of-charge for these cells.
http://www.cse.anl.gov/us-china-workshop-2011/pdfs/batteries/LiFePO4%20battery%20performances%20testing%20for%20BMS.pdf

In particular see the graph on page 15. You will see that the voltage is very constant at 3.30 +- 0.01 Vpc (52.80 +- 0.16 V) from 40% SoC to 70% SoC. So it is not useful to try to estimate SoC from voltage with these batteries, except near 0%, 75% and 100%.

You can also see that 51 V (3.19 Vpc) is too low for a "back to utility" setting, since the battery has less than 10% SoC at that stage, and it is possible that the lowest capacity cell has gone below 2.5 V and violated its warranty. Even 52 V (3.25 Vpc) could be only 15% SoC.

And so 48 V is of no use at all as a low voltage cutoff for a LiFePO4 battery.

But we do not necessarily need these limits to be raised. We are using a BMS which monitors cells individually. The BMS can decide when the inverter needs to be turned off, based on the voltage of the lowest cell. The BMS will communicate with the PIP-4048MS using its serial communications protocol. We just need a way to tell the inverter to turn off, but still allow the battery to be charged from solar or utility. We did not intend to leave these settings at 53 volts normally, but only to set them to 53 volts as a way of making the inverter turn off.

Is there some other way we could turn off the inverter, but leave the MPPT and utility charger running?

I have another question: If parameter 1 (output source priority) is set to "SbU", and there is no utility voltage, and the battery voltage goes below the "back to utility" setting, will it still switch the loads to utility and stop the inverter? i.e. will the loads go off?

Thanks for your help.

-- Dave Keenan
http://dkeenan.com
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
celectric
Groupie
Posts: 56
Joined: Tue, 29 May 2012, 05:14
Real Name: Tim Bates
Location: Adelaide SA

PIP-4048MS inverter

Post by celectric »

weber wrote:The idea of wiring an array of 18 panels so that what is electrically a checkerboard, gets progressively shaded, does my head in.
With the cross-ties, A2 and B2 are electrically equivalent so there's no difference between a checkerboard and a straight-across shading sequence. I think that particular cross-tie arrangement only makes sense if your shade pattern naturally falls into the checkerboard arrangement (which seems unlikely), there's no benefit to be had by making it so artificially.
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

PIP-4048MS inverter

Post by weber »

celectric wrote:With the cross-ties, A2 and B2 are electrically equivalent so there's no difference between a checkerboard and a straight-across shading sequence. I think that particular cross-tie arrangement only makes sense if your shade pattern naturally falls into the checkerboard arrangement (which seems unlikely), there's no benefit to be had by making it so artificially.

Right. Obvious when you put it like that. Thanks.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

PIP-4048MS inverter

Post by weber »

Eric wrote:Hi Dave

I'm sorry but we're unable to accommodate customized revision on our products unless it's for OEM projects with MOQ of 500 units or more.

Regarding your other questions:
Is there some other way we could turn off the inverter, but leave the MPPT and utility charger running?
no. it's ok to turn off inverter and solar can still charge, but cannot do this on utility charger as the circuitry is common with the inverter, so if inverter off, utility charging is off too.

I have another question: If parameter 1 (output source priority) is set to "SbU", and there is no utility voltage, and the battery voltage goes below the "back to utility" setting, will it still switch the loads to utility and stop the inverter? i.e. will the loads go off? as there's no utility to switch to it'll stay under battery mode until batt volt hits low cut off (prog 29) then inverter shuts down.

Eric
Last edited by weber on Tue, 04 Nov 2014, 13:34, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
offgridQLD
Senior Member
Posts: 1827
Joined: Tue, 23 Jul 2013, 16:05
Real Name: Kurt
Location: Fleurieu Peninsula, SA

PIP-4048MS inverter

Post by offgridQLD »

At least you have a answer now 500 units before they will tweak one for you Image

So just thinking from memory if you switch the inverter off via the rocker switch (bottom right of the unit)and you have PV feeding into the inverter The Mppt charger keeps charging if not the unit shuts down after 5 seconds or so.

I can confirm that on the weekend.

Wouldn't that be a simple way to kill your AC load?

Kurt
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

PIP-4048MS inverter

Post by weber »

Hi Kurt,

Yes, we could put some relay contacts in series with the rocker switch, but, as Eric said, this has the problem that it prevents genset charging, although it allows PV charging.

We also now know that we can't use the max 51 V "back to utility" setting to shut off the inverter. Only the max 48 V low voltage cutoff will do that when there is no "utility" (genset) running at the time.

I'm still trying to avoid having to do the voltage scaling because I still haven't figured out what's going on with those 1/4, 1/7 and 1/13 diff amps in the MPPT. But also because I wonder what the MPPT might do when it thinks the outgoing power is so much less than the incoming power. Or it might make some use of the supposed voltage ratio between input and output.

Another thought that occurs to me is "bottom balancing". But although 48 V (3.0 Vpc) would be a _safe_ cutoff if the cells were bottom balanced, it would be 0% SoC and not the 20% SoC we'd like to maintain.

Here's another idea. When a cell goes low, have the BMS disconnect the AC loads from the inverter output. With load-sensing mode enabled, at least the load on the battery will drop to about 15 watts. If there is 20% left in the 9 kWh battery at the time, that's 1800 Wh. It will take 120 hours (5 days) to completely flatten. Plenty of time to get some sunshine in, even if there are no humans around to start the genset.

And if it doesn't get any charge, the BMS could eventually drop out the battery isolation contactor, say when it gets down to 10% SoC. Then it would require human intervention to re-close the battery contactor once a charge source was organised.

There must already be a relay inside the inverter, (or two, if it switches the neutral as well), that switches the loads between inverter and utility/genset. I guess we just need to override the inverter's control of that.

How does that sound?

Any other ideas?
Last edited by weber on Tue, 04 Nov 2014, 15:51, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
User avatar
offgridQLD
Senior Member
Posts: 1827
Joined: Tue, 23 Jul 2013, 16:05
Real Name: Kurt
Location: Fleurieu Peninsula, SA

PIP-4048MS inverter

Post by offgridQLD »

I think at some point you have to trust (or better do the calculations to put your mind at rest) that a over sized pv array will be doing it's best each day to keep the battery alive and with just a 15w load to cover. It's going to have be some real serious (tropical cyclone washout) low output days in a row for a large array not to be able to recover 360whrs a day.

I haven't started my generator in over two years.(other than to splash oil around the motor) Ok it's nice to have it for some kind of emergency encase say (inverter or both charge controllers needed repairs)and the AC/DC charger is a nice feature of the PIP4048 for emergency back up.

But I am not relying on it to top up my batterys to avoid LVD situations I have the sun each day that should be doing that. I would think Lithium batterys that don't need long periods in absorb with limited current really can make the most of short bursts of sun that we often get in QLD even on the bad days.

I know you need to design you system to be fool proof and as automated as you can. Though I don't think it's to impractical to have the system shut down and require a manual reset (and then perhaps a manual start of a generator if needed)As it would only happen in a moment that should rarely happen if at all.

The ultimate aim is not to end up with dead batterys and having the AC load cut and then finally the system shut down is achieving that as a last resort(all be it slightly impractical if it ever happens)

So switching the AC load has my vote, bottom balancing could be a option. Though how do you then actively balance the cells after the initial (bottom balance) or don't you?

Kurt


Last edited by offgridQLD on Tue, 04 Nov 2014, 17:24, edited 1 time in total.
antiscab
Senior Member
Posts: 3183
Joined: Mon, 26 Nov 2007, 05:39
Real Name: Matthew Lacey
Location: Perth, WA

PIP-4048MS inverter

Post by antiscab »

weber wrote: Another thought that occurs to me is "bottom balancing". But although 48 V (3.0 Vpc) would be a _safe_ cutoff if the cells were bottom balanced, it would be 0% SoC and not the 20% SoC we'd like to maintain.


how big is the load in relation to the size of the battery?

how well capacity matched are your cells? (if using CALB they tend to give you a list of capacities for all your cells)

Do you have a way to get the built in charger to stop charging if any one cell gets full first (if you are bottom balancing)?

Matt
2023 BYD Atto 3 - 21k km
2017 Renault zoe - 147'000km
2012 Leaf - 101'000km - soon to be trialing a booster battery
2007 Vectrix - 197'000km (retired)
2007 Vectrix - 50k km
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

PIP-4048MS inverter

Post by weber »

antiscab wrote:how big is the load in relation to the size of the battery?
Inverter max is about 0.55C continuous, 1.1C for 5 seconds. But average load is about 0.015C. 16 LiFePO4 cells at 180 Ah, so about 9 kWh.
how well capacity matched are your cells? (if using CALB they tend to give you a list of capacities for all your cells)
Don't know because I haven't picked them up yet. Yes, CALB. But I'd prefer the system to be able to cope with mismatched cells anyway.
Do you have a way to get the built in charger to stop charging if any one cell gets full first (if you are bottom balancing)?
Yes. The BMS can send serial commands to the charger to raise both bulk and float voltage setpoints above whatever the total battery voltage is at the time.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
T1 Terry
Senior Member
Posts: 1952
Joined: Thu, 30 Sep 2010, 20:11
Real Name: Terry Covill
Location: Mannum SA

PIP-4048MS inverter

Post by T1 Terry »

weber wrote:
coulomb wrote:but seriously, would you really need to be all that accurate with temperatures etc? I suppose so, since the two options would be pretty close, otherwise it would be obvious by inspection which option would be better.

My thinking was, if you're going to simulate the panels to the level of individual cells then you'd want to model the patchy shade to that kind of resolution as well. But yes, it would probably be safe enough to assume all the cells in one panel are at the same temperature.

But as I said, much cruder panel-level models are worth considering. And I was about to do so, despite its off-topic-ness, when you smacked me one in the eye, presumably because I beat up on poor Terry. Image

Consider these two strings in parallel, without cross-ties. An ">" means a shaded panel, which we can treat as an ideal diode (the 3 bypass diodes in series) since it has no current source and the cell diodes can be ignored as they will be cold and therefore have a higher voltage (much lower current at the same voltage) than their hot counterparts.

An unshaded panel can be treated as a single current source in parallel with a string of 72 hot silicon diodes in series, wasting some of the current from the current source, depending on their temperature and the voltage the MPPT operates them at. Their bypass diodes can be ignored as they will be reverse biased.
        1     2     3
   A +-[>]---[_]---[_]-+
 ----|                 |----
   B +-[_]---[_]---[_]-+
     :                 :
I find it very hard to wrap my head around this stuff, so don't hesitate to question if you think I've got it wrong.

If the MPPT operates at a voltage that suits 3 series panels (3S) it will collect power only from 3 of these panels (B1,2,3). If instead it operates at a 2S voltage it will collect 4 panels worth (A2,3 and two thirds from each of B1,2,3). But with more unshaded strings in parallel there will be no benefit in 2S operation and so we will operate at 3S and we have effectively just lost the whole string that has a shaded panel.

So it's best if the shade progressively takes A1,2,3 before starting on B.

If the strings are cross-tied
        1     2     3
   A +-[>]-+-[_]-+-[_]-+
 ----|     |     |     |----
   B +-[_]-+-[_]-+-[_]-+
     :                 :
It makes very little difference. With 3S MPP we will get slightly more than 3 panels worth. We'll get B1 and a little over half each from A2,3 and B2,3. B1 will not pass more than its usual current and this will be split between the others. Because they are only being called on to provide half their possible current they will be able to operate at a slightly higher voltage than otherwise.


If instead shade takes A1 and B1, we've lost two strings.
        1     2     3
   A +-[>]---[_]---[_]-+
 ----|                 |----
   B +-[>]---[_]---[_]-+
     :                 :
And it makes no difference at all if they are cross-tied.
        1     2     3
   A +-[>]-+-[_]-+-[_]-+
 ----|     |     |     |----
   B +-[>]-+-[_]-+-[_]-+
     :                 :
Where cross ties really make a difference is when you have diagonal or checker-board shading like this. Without cross ties you lose both strings.
        1     2     3
   A +-[>]---[_]---[>]-+
 ----|                 |----
   B +-[_]---[>]---[_]-+
     :                 :
With cross ties you get one whole string made up from parts of the two strings, by zig-zagging back and forth. But how likely is it that we can arrange for shading to happen like that?
        1     2     3
   A +-[>]-+-[_]-+-[>]-+
 ----|     |     |     |----
   B +-[_]-+-[>]-+-[_]-+
     :                 :
The idea of wiring an array of 18 panels so that what is electrically a checkerboard, gets progressively shaded, does my head in.

My thoughts were to arrange the panels in 2 rows of 9 panels in parallel, series linked at one end or both ends with blocking diodes on the positive output from each panel. This way all the unshaded panels in the A parallel row will feed into the common conductor but the blocking diodes will prevent the shaded panels from draining any of the current back into themselves causing cell damage and wasting the energy generated. This then feeds into the common conductor connected to the negative of the B parallel row. The positives from the B row also have blocking diodes and feed into the common positive conductor linked to the solar controller.
Looking at the image in the photo, the shade does not move squarely across the roof so a series string of 2 cells bottom to top would be effected by partial shade on the lower panel but full shade on the upper panel rendering zero output from that series string.

The controller method you have chosen to use senses the average voltage and current from all the series strings and makes a calculation based on this averaged reading, not individual reading from each series string. By using 3 panels in series the strings in full sun will call the shots so to speak, the series strings with partial shade will not have the voltage or current output to compete and their output will be effectively blocked.

T1 Terry
Green but want to learn
User avatar
weber
Site Admin
Posts: 2936
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

PIP-4048MS inverter

Post by weber »

T1 Terry wrote:My thoughts were to arrange the panels in 2 rows of 9 panels in parallel, series linked at one end or both ends with blocking diodes on the positive output from each panel. This way all the unshaded panels in the A parallel row will feed into the common conductor but the blocking diodes will prevent the shaded panels from draining any of the current back into themselves causing cell damage and wasting the energy generated. This then feeds into the common conductor connected to the negative of the B parallel row. The positives from the B row also have blocking diodes and feed into the common positive conductor linked to the solar controller.
There is no need for blocking diodes with parallel panels or strings subject to partial shading. As I mentioned above, a shaded panel is cooler, so the voltage drop of its equivalent diodes is higher. It turns out that the power lost due to the small current that it does shunt from parallel unshaded panels is about the same as what would be lost due to the voltage drop of blocking diodes. So it's better to omit the blocking diodes and eliminate another source of failure.
Looking at the image in the photo, the shade does not move squarely across the roof so a series string of 2 cells bottom to top would be effected by partial shade on the lower panel but full shade on the upper panel rendering zero output from that series string.
Right. Which is why I figure it is better to place all the panels of a string as close together as possible (i.e. side-by-side in the same row), so that advancing shade takes out the whole string before it starts on the next one, so at least one, and often two, strings are unshaded at any given time.
The controller method you have chosen to use senses the average voltage and current from all the series strings and makes a calculation based on this averaged reading, not individual reading from each series string. By using 3 panels in series the strings in full sun will call the shots so to speak, the series strings with partial shade will not have the voltage or current output to compete and their output will be effectively blocked.

It's true that the unshaded strings will call the shots voltage-wise and therefore any string with even one fully-shaded cell (and I do mean "cell", not panel) will contribute zero. But this goes just as much for strings of 2 as it does for strings of 3, except that in the former case a single shaded cell means only losing the output of 2 panels instead of 3. So that is one benefit of your suggestion.

With strings of 3, if only one panel was shaded in three out of the six strings then the MPPT could reduce the array voltage and get 2/3 of the array power instead of only half. However I admit that such a shading pattern is fantastically unlikely. So strings of 3 do not help with the shading issue as I mistakenly stated earlier. They just reduce the amount of cable required and the number of circuit breakers required (from 5 to 3).

Unfortunately I cannot physically lay the panels out in two rows of 9. But even if I could, and if I were to connect them as 9 strings of 2 with cross ties (or two parallel-blocks-of-9 in series) as you suggest, I do not see any advantage in putting one parallel block in one row and the other parallel block in the other row as the north row will be far more prone to shading than the south row, since the trees are to the northeast and northwest.

I worry that, although you know better, you may still be thinking of solar panels as voltage sources like batteries, instead of current sources with a string of (non-ideal) diodes shunting them. I know it was a hard habit for me to break.

From an earlier post of yours:
This way any part of any panel in the "A" paralleled group can have shade across it but the output from the unshaded panels still feeds into the parallel cabling so any of the "B" unshaded panels can utilise the input from the paralleled 24v nom. supply to produce the required 48v nom. to charge the batteries.
Sure, the voltage will still be there, but in the example below (where ">" means shaded), current will still be limited to that of one panel, and hence power will be limited to that of one string, whether the cross-ties are there or not. Do you agree?
        A     B
   1 +-[>]-+-[_]-+
 ----|     |     |----
   2 +-[>]-+-[_]-+
     |     |     |
   3 +-[_]-+-[_]-+
BTW It probably isn't necessary for you to quote my very long posts in their entirety for me to know what you are responding to. I worry that others may find it annoying to have to scroll through my ravings all over again.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).
Post Reply