Tesla Powerwall

Introductions, general chit chat and off-topic banter.
User avatar
coulomb
Site Admin
Posts: 3428
Joined: Thu, 22 Jan 2009, 20:32
Real Name: Mike Van Emmerik
Location: Brisbane
Contact:

Tesla Powerwall

Post by coulomb » Sun, 14 Jun 2015, 00:43

Gabz wrote: ... ausgrid, endevour, ergon everyone who owns the poles and wires ban this sort of system. they only allow UPS style battery connections.

I don't see why it should bother them. Back in the days of the high feed-in tariff, they didn't (and still don't) want batteries anywhere near them. But with low feed-in tariffs, why would they care? From the grid's point of view, they just see lower electricity usage (both less importing, that they get paid for, and less exporting, on which they pay roughly average market rates for the energy). So far, they can't penalise us for using less electricity. (Maybe I shouldn't even suggest this... Image )

Energex also have some sort of zero export rule that was supposed to apply from now, but was postponed. Per this Solar Choice article, "Energex also notes that energy storage devices will be allowed provided none of their power is exported." This sounds like an over-reaction to me, but now that renewable generators are like the 6th largest generator (around 1 GW in Queensland), I guess they're running scared.

That sounds like something that either the Powerwall needs to enforce, or maybe the grid interactive inverter (with some sort of measurement of exported power) will have to provide. Since most existing grid interactive inverters presumably don't have zero export capability, either the Powerwall will have to do it, or there may be a market in third party gizmos to somehow make ordinary inverters force zero export. That doesn't sound trivial to me.
Learning how to patch and repair PIP-4048 inverter-chargers and Elcon chargers.

User avatar
4Springs
Senior Member
Posts: 500
Joined: Thu, 23 Dec 2010, 01:14
Real Name: Christopher Walkden
Location: Selbourne, TAS

Tesla Powerwall

Post by 4Springs » Sun, 14 Jun 2015, 01:10

Here is a non-trivial third party gizmo project: Reposit
Looks like they are making software to interface with Tesla and others which lets people monitor and control what their power is doing.

T1 Terry
Senior Member
Posts: 731
Joined: Thu, 30 Sep 2010, 20:11
Real Name: Terry Covill
Location: Mannum SA

Tesla Powerwall

Post by T1 Terry » Sun, 14 Jun 2015, 02:01

weber wrote: Hi Coulomb,

Yes, it almost certainly has a single bidirectional DC/DC converter, used in one direction to maximum-power-point track the PV array to charge the battery, and in the other direction to simulate a PV array to the MPPT input of the inverter. However I can't see it working without actually presenting two separate ports to the world, each separated from the converter's single non-battery port by a relay of some kind.

There are no panel diodes to prevent flow of power into the panels.
And its hard to see how the Powerwall's MPPT could work in parallel with the one in the inverter and win. And why bother trying when a relay will solve the problem.

At one stage you mention "the inverter in the Powerwall", but there isn't one. You also mention a "DC/DC inverter". As normally understood, this is a contradiction in terms because a power inverter is a DC to AC device. This is the same misleading-marketing-nonsense language used by Elon Musk during the product launch.

The Powerwall doesn't need to do Demand Charge Management, for the simple reason that almost no households have a demand charge. So far, demand charges have been limited to large 3-phase customers, both here and in the US. A demand charge is a charge that is based on your highest power demand (in kW, not kWh, averaged over a 15 minute period) -- the highest during the billing period.

Dealing with household Time-of-use tariffs is a piece of cake, the Powerwall only needs to know the time of day and the times when the rates change. Hybrid inverters do that too. And the Powerwall won't be controlling any discretionary loads. That can be done with simple timers.

If there were no blocking diodes on each panel in a series string, as soon as a panel was shaded the high series voltage would over come the panels voltage and the back flow would turn it into a heater, destroying the panel after only a few weeks/mths.
So each panel requires both bypass diodes and blocking diodes if they are to be connected in a series string, bypass diodes only for parallel strings.

T1 Terry
Green but want to learn

User avatar
weber
Site Admin
Posts: 2500
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Tesla Powerwall

Post by weber » Sun, 14 Jun 2015, 03:15

T1 Terry wrote:If there were no blocking diodes on each panel in a series string, as soon as a panel was shaded the high series voltage would over come the panels voltage and the back flow would turn it into a heater, destroying the panel after only a few weeks/mths.
So each panel requires both bypass diodes and blocking diodes if they are to be connected in a series string, bypass diodes only for parallel strings.

Sorry Terry, but what you've written here is completely wrong. It can be greatly improved by changing all occurrences of "blocking" to "bypass" and vice versa, but there would still remain many falsehoods.

All panels come with bypass diodes built in. Typically each diode bypasses one third of the series string of cells making up the panel. No one uses blocking diodes any more, even for multiple parallel strings, for reasons I explained here:
viewtopic.php?title=pip4048ms-inverter& ... 332#p54049
Last edited by weber on Sat, 13 Jun 2015, 17:15, 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: 2500
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Tesla Powerwall

Post by weber » Sun, 14 Jun 2015, 03:48

coulomb wrote:I still don't think that optimising power flow to/from the battery is a piece of cake, but you've probably given it more thought than I have.

When you have a demand charge it is indeed a difficult problem. But when you only have a daily charge and different cents per kWh rates in different time slots, which are known in advance (i.e a time-of-use tariff), then don't you agree that the optimum strategy is simply to charge the battery when energy is cheapest and discharge it (to power loads but not export) when energy is most expensive?

Edit:

Off topic warning

There follows a rather long discussion on Blocking diodes in PV arrays. If you want to skip it and get back to the Tesla Powerwall discussion, click the following link:
viewtopic.php?title=tesla-powerwall&p=5 ... 529#p57530
Last edited by weber on Mon, 22 Jun 2015, 05:51, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

T1 Terry
Senior Member
Posts: 731
Joined: Thu, 30 Sep 2010, 20:11
Real Name: Terry Covill
Location: Mannum SA

Tesla Powerwall

Post by T1 Terry » Sun, 14 Jun 2015, 14:54

weber wrote:
T1 Terry wrote:If there were no blocking diodes on each panel in a series string, as soon as a panel was shaded the high series voltage would over come the panels voltage and the back flow would turn it into a heater, destroying the panel after only a few weeks/mths.
So each panel requires both bypass diodes and blocking diodes if they are to be connected in a series string, bypass diodes only for parallel strings.

Sorry Terry, but what you've written here is completely wrong. It can be greatly improved by changing all occurrences of "blocking" to "bypass" and vice versa, but there would still remain many falsehoods.

All panels come with bypass diodes built in. Typically each diode bypasses one third of the series string of cells making up the panel.
No one uses blocking diodes any more, even for multiple parallel strings, for reasons I explained here:
viewtopic.php?title=pip4048ms-inverter& ... 332#p54049

Let's just say, some people don't use blocking diodes any more, but there can be a price
ImageImageImage
These lasted roughly 2 mths in a series string. Looking closely at the face of panels that have suffered reverse current flow the lines where the grid has heated and started to separate from the adhesive film is quite evident, these examples are not as common, these were cause by faults within the panels I think, but the same thing caused the failure.
Your call if you don't want to use blocking diodes on a series string... Have you actually tested you theory?

T1 Terry
Last edited by T1 Terry on Sun, 14 Jun 2015, 04:56, edited 1 time in total.
Green but want to learn

User avatar
weber
Site Admin
Posts: 2500
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Tesla Powerwall

Post by weber » Mon, 15 Jun 2015, 05:34

T1 Terry wrote:Let's just say, some people don't use blocking diodes any more, but there can be a price.
[See images above]
These lasted roughly 2 mths in a series string. Looking closely at the face of panels that have suffered reverse current flow the lines where the grid has heated and started to separate from the adhesive film is quite evident, these examples are not as common, these were cause by faults within the panels I think, but the same thing caused the failure.
Your call if you don't want to use blocking diodes on a series string... Have you actually tested you theory?

I hardly know where to start. You haven't yet acknowledged that you mixed up "blocking" and "bypass" diodes in your previous post. So I don't know if you're still using the two terms back to front or not. Here's one of many images on the net that show which is which.

Image

The absence of blocking diodes will not do that to panels, whether they are exposed to partial shading or not. But the absence of bypass diodes certainly would. However there are no panels approved for use in Australia that do not have bypass diodes already fitted by the manufacturer.

When there are more than 2 strings in parallel, you must have a suitable fuse in each string (at the locations where the blocking diodes are shown in the above image). Those fuses are to prevent damage caused by a faulty panel in one string allowing all the other strings to feed current in reverse through the faulty string, which would also cause damage like you show. You can have blocking diodes as well if you want, but they are a complete waste of time. AS5033 explicitly says that blocking diodes cannot be used instead of string fuses. This is because diodes in current-limited situations like this have a tendency to fail short-circuit when they do fail.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

T1 Terry
Senior Member
Posts: 731
Joined: Thu, 30 Sep 2010, 20:11
Real Name: Terry Covill
Location: Mannum SA

Tesla Powerwall

Post by T1 Terry » Sat, 20 Jun 2015, 17:12

Been busy getting jobs finished and having birthdays that move me into the old bloke category so my reply is a bit late, but here none the less.
Hopefully this doesn't annoy you or deemed too far off topic or even if the Lounge area can go off topic, but if I've offended any rules, please let me know and delete this reply... or move it.
As my signature says, I'm green but I want to learn, so that will only happen if I better understand where I'm going wrong.
A reproduction of the example above but some what expanded to create a more realistic series string
Image
String A would theoretically add up to a voltage of between 68vdc and 88vdc @ between 0 amps and 22.4 amps... have I got that correct so far?

The function of the bypass diode is 2 fold, it will pass current from a higher voltage to a lower voltage in one direction but block flow in the opposite direction. Looking at the example diagram the current would flow from the bottom of the page up, but not from the top of the page down ...... have I got that correct?

If any one of the panels in string B was fully shaded, what would the Voc, Vmp and Imp of that panel be?

If the Imp was 0 amps, for any current to flow through the bypass diode the voltage above the diode would need to be less than the voltage above the diode, plus the voltage drop across the diode.... have I got this part correct?

T1 Terry
Green but want to learn

User avatar
weber
Site Admin
Posts: 2500
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Tesla Powerwall

Post by weber » Sat, 20 Jun 2015, 20:00

T1 Terry wrote:String A would theoretically add up to a voltage of between 68vdc and 88vdc @ between 0 amps and 22.4 amps... have I got that correct so far?
I'm afraid not. When things are in series the voltages add but the currents do not. The maximum current in the string would be the same as the Isc of one panel, let's say 6 A. The actual voltage and current will depend on the nature of the load. I'll assume in what follows, that it is a maximum-power-point-tracking solar charge controller (MPPT SCC) which is doing its job and extracting maximum power, and all panels are in full sun. In that case the overall voltage will be around 68 V and the current in each string will be around 5.6 A.
The function of the bypass diode is 2 fold, it will pass current from a higher voltage to a lower voltage in one direction but block flow in the opposite direction. Looking at the example diagram the current would flow from the bottom of the page up, but not from the top of the page down ...... have I got that correct?
Yes.
If any one of the panels in string B was fully shaded, what would the Voc, Vmp and Imp of that panel be?

"oc" means "when open circuit", "mp" means "when delivering maximum power" and "sc" means "when short circuit". If a panel is fully shaded and open circuit, its voltage will be zero. The maximum power obtainable from a fully-shaded panel will be zero, so Vmp and Imp are undefined. All we can say is that if one of them is non-zero the other one must be zero to obtain maximum power. But I really don't think that's what you want to know. I think you want to know what the actual voltage and current will be for the shaded panel, when in that circuit.
If the Imp was 0 amps, for any current to flow through the bypass diode the voltage above the diode would need to be less than the voltage above the diode, plus the voltage drop across the diode.... have I got this part correct?

Yes. And this is true no matter what the Imp is.

With the bypass diode around the shaded panel the voltage across the panel would be about -0.6 V and the current through the panel would be close to zero. The current through the bypass diode would be about 6 A. No problem. The MPPT SCC would eventually discover that it can extract more power by lowering the array voltage to about 3 x 17 = 51 V

Without a bypass diode around the shaded panel the voltage across it could be as high (negative) as -51 V and the current about 5.6 A. This will eventually destroy the panel through overheating. The more panels in the string, the worse this would be.

I note that none of this requires blocking diodes (which are the only kind we disagree on). And all approved panels come with bypass diodes built in -- typically there are 3 of them, each bypassing one third of the string of cells making up the panel.
Last edited by weber on Sat, 20 Jun 2015, 10:05, edited 1 time in total.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

T1 Terry
Senior Member
Posts: 731
Joined: Thu, 30 Sep 2010, 20:11
Real Name: Terry Covill
Location: Mannum SA

Tesla Powerwall

Post by T1 Terry » Sat, 20 Jun 2015, 23:29

Yes, my bad, I did know the voltage adds in series, the current in parallel.... brain fade is about my only excuse Image

The MPPT unit will only look for the optimum Vmp if it is in full output mode, feed to the grid or supply the house and take in the shortfall from battery backup or from the grid, but in a stand along once the batteries are fully charged and the house demand is less than the optimum solar output, the voltage will rise as the load reduces, zero load would be 88v from string A.
If a cell in string B is shaded, the Vmp will be 75% of the total Vmp available from string A, what stops the current from flow from string A into string B? The bypass diodes will stop any reverse current flow, so as far as I can see the only current path is a reverse flow through the panels until the 2 voltages equalise... or have I missed something here?

T1 Terry
Green but want to learn

User avatar
weber
Site Admin
Posts: 2500
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Tesla Powerwall

Post by weber » Sun, 21 Jun 2015, 04:03

T1 Terry wrote:The MPPT unit will only look for the optimum Vmp if it is in full output mode, feed to the grid or supply the house and take in the shortfall from battery backup or from the grid, but in a stand along once the batteries are fully charged and the house demand is less than the optimum solar output, the voltage will rise as the load reduces, zero load would be 88v from string A.
Agreed. I was assuming it was not a standalone with a full battery.
If a cell in string B is shaded, the Vmp will be 75% of the total Vmp available from string A, what stops the current from flow from string A into string B?
I assume we're talking about the case where there are no blocking diodes.

When the MPPT has done its job and wound the voltage down to 75% of what it was, then both strings will be supplying current to the loads, so there will be no reverse current. Both strings will have forward current.

But while it is still at the full voltage, or even worse -- if it was a standalone with a full battery and was therefore operating open circuit, then the shaded panel will not have a reverse voltage across it and its bypass diode will not be conducting.

After it cools down due to being shaded, it will have around 22 volts across it, the same as all the others and all currents will be close to zero. But while it's still hot it may have around 18 V across it while the other 3 in the same string have 23 volts across them, causing a small reverse current to flow through string B from string A. Not enough to damage it, and not for very long. It is very difficult to explain this without drawing a bunch of diagrams with stacked-up IV curves for the 8 panels. But you could do the experiment in real life.
The bypass diodes will stop any reverse current flow, so as far as I can see the only current path is a reverse flow through the panels until the 2 voltages equalise... or have I missed something here?

That's pretty well right. Assuming you only mean that the bypass diodes will stop any reverse current flow through themselves. And if the 2 voltages you are referring to are the total voltages of the two strings, I note that these must _always_ be equal, by definition.

When the shaded panel has cooled down, it will have around the same open circuit voltage as the unshaded panels. This is because open circuit voltage increases linearly with decreasing temperature and decreases only logarithmically with decreasing irradiance. The two effects roughly cancelling each other out, resulting in only a very small reverse current.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

T1 Terry
Senior Member
Posts: 731
Joined: Thu, 30 Sep 2010, 20:11
Real Name: Terry Covill
Location: Mannum SA

Tesla Powerwall

Post by T1 Terry » Sun, 21 Jun 2015, 14:29

What if it was a grid connect system? Much higher voltages on each string, one or more panels shaded on one string and there is a power outage?
Without blocking diodes the voltages must balance between the strings so the shaded string must dissipate the watts required to balance the voltage.
Am I correct in assuming the shaded solar panels will be the end of the reverse current flow as the output voltage below this point would be common between the strings?
If I have that part correct, the exposed panels will be heated by the sun yet unable to stop the reverse current flow as the voltage at positive end would be greater than the combined voltage from the panels all the way down the string to the negative side of the shaded panel.... am I still on the right track?
If this is the case, then these panels will not only be heated by the sun, but they will also be heated by the reverse current flow and this will increase as the panel heat increases, till something lets go. The shaded panel will be heated internally so unless the shade is combined with a cold wind, this panel will also heat up.

So if blocking diodes are not required to protect against these possibilities then I've got the whole thing wrong in my head

T1 Terry
Green but want to learn

User avatar
weber
Site Admin
Posts: 2500
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Tesla Powerwall

Post by weber » Mon, 22 Jun 2015, 06:37

To force reverse current through a panel you must apply a voltage to it that is higher than its actual open circuit voltage, which depends on temperature and irradiance.

I'm going to use somewhat inaccurate round figures to keep this easy to follow. But they are in the right ballpark.

Let's say the panel has Voc = 22 V at the Standard Test Conditions (STC) of 25°C and 1000 W/m^2 (one sun). They test it at 25°C in the factory by briefly flashing it with a simulated sun from lamps, so it does not have time to heat up.

And lets say the ambient temperature is 25°C and that full sun causes the panel to heat up by 40°C (to 65°C).

At a given irradiance, the open circuit voltage of a 36-cell panel decreases by about 1 volt for every 10°C rise in temperature -- a negative linear relationship. So if the panel is at ambient temperature and you suddenly expose it to full sun, it will briefly have Voc = 22 V. But after it has heated up to 65°C it will have Voc = 18 V (4 volts lower).

At a given temperature, the open circuit voltage of a 36-cell panel decreases by about 2 volts every time you reduce the irradiance by a factor of 10 -- a logarithmic relationship. So if it has Voc = 18 V at 65°C in full sun, and you suddenly shade it, so it only has 1/100th of a sun, it will briefly have Voc = 14 V. But after it has cooled down to 25°C it will be back up to 18 V again.

So you see that a panel in full sun at 65°C and a panel in 1% sun at 25°C have the same open circuit voltage. Therefore it doesn't matter how many panels are in each string (provided it is the same number in both), and it doesn't matter if all the panels in one string are in full sun and all the panels in the other string are shaded, and it doesn't matter if there is no load whatsoever on the array, there will still be no significant reverse current in any string.

But lets assume there was some fault (like a shorted panel) that caused the full current of one string to flow backwards through the other string. How bad would this actually be? Lets assume the panels are 20% efficient. This means that when a panel is operating normally, it will still be turning 80% of the sunlight that is absorbed by it, to heat. If it is open circuit then 100% will go to heat. If it has current forced backwards through it that is equal in magnitude to its forward current, then it will have 120% going to heat. This will not damage any panel, although it may speed up the normal aging process.

If however you have 6 strings in parallel, and one string has a fault in one panel, such as a shorted bypass diode) so that the other 5 strings feed backwards through the faulty string, then the good panels in the faulty string may be dissipating 200% of what they normally do. This will certainly cause catastrophic failure, with arcing and a possible fire. The higher the voltage the worse it will be. The photos you posted above look like that sort of damage.

That's why, whenever we have more than two strings in parallel, the law requires us to put a fuse in every string (one at each end of every string if Voc >= 120 V), with a trip current rating as given in the panel datasheet, or the next standard size up from 1.5 times the panel's rated short circuit current if no fuse rating is given. And with a DC voltage rating greater than the open circuit array voltage at the lowest expected temperature.
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

User avatar
lopezjm2001
Groupie
Posts: 62
Joined: Mon, 02 Feb 2015, 20:37
Real Name: John Lopez
Location: Sydney

Tesla Powerwall

Post by lopezjm2001 » Mon, 22 Jun 2015, 07:31

That was a good detailed reply Weber. I expect that using two strings in parallel, one string facing east and one facing west will become more common to shift the solar power to morning and afternoon peak periods using the same MPPT input for both strings. Thus reducing power going to grid at mid day at the low infeed tariff. This allows the designer to use a cheaper inverter which only has one MPPT input instead of two. This is one example where you expect one string to be shaded (or just exposed to ambient light) while the other is in full sun on most days. Totally legal along as they are the same number of modules in each string and all modules are the same. No need for a blocking diode.
Last edited by lopezjm2001 on Sun, 21 Jun 2015, 22:24, edited 1 time in total.

T1 Terry
Senior Member
Posts: 731
Joined: Thu, 30 Sep 2010, 20:11
Real Name: Terry Covill
Location: Mannum SA

Tesla Powerwall

Post by T1 Terry » Mon, 22 Jun 2015, 14:43

Thankyou Weber, I have a watt meter that I can put in line with string B and as long as I also monitor panel temp, well at least the last one in the string as this is the panel that would suffer reverse current flow, I can do a hands on test. I have tested and found a similar effect or the short term reverse current flow with single panels in parallel, but multiple panels in series complete with bypass diodes have always dropped current greater than the expected loss of 1 panel, so now I will try to determine why if it is not the result of reverse current flow into the shaded panel. I conduct the tests as a direct connection (with circuit breaker) to the relative voltage battery pack to eliminate any effects caused by the MPPT controller

T1 Terry
Green but want to learn

User avatar
weber
Site Admin
Posts: 2500
Joined: Fri, 23 Jan 2009, 17:27
Real Name: Dave Keenan
Location: Brisbane
Contact:

Tesla Powerwall

Post by weber » Mon, 22 Jun 2015, 15:41

I have started a new thread entitled Blocking diodes in PV arrays. Please post further discussion on that topic to the new thread:
viewtopic.php?p=57529&t=4588#p57529
One of the fathers of MeXy the electric MX-5, along with Coulomb and Newton (Jeff Owen).

User avatar
reecho
Senior Member
Posts: 617
Joined: Sun, 17 Apr 2011, 02:39
Real Name: Richard Baird
Location: Perth WA
Contact:

Tesla Powerwall

Post by reecho » Tue, 19 Jan 2016, 01:36

Powerwall Manual for those interested.....

Rusdy
Groupie
Posts: 179
Joined: Mon, 09 Jun 2014, 16:45
Real Name: Rusdy
Location: Australia
Contact:

Tesla Powerwall

Post by Rusdy » Tue, 19 Jan 2016, 17:10

weber wrote:
That's why, whenever we have more than two strings in parallel, the law requires us to put a fuse in every string (one at each end of every string if Voc >= 120 V), with a trip current rating as given in the panel datasheet, or the next standard size up from 1.5 times the panel's rated short circuit current if no fuse rating is given. And with a DC voltage rating greater than the open circuit array voltage at the lowest expected temperature.


Thanks for the off topic discussion guys! It's a very valuable learning for others (like me)! My guess the law-maker impose the 120VDC requirement due to existing AS3000 limit of 120VDC ripple free for SELV, or simply being practical about it. My point is, if one trying to do DIY (without sparky ticket), as long one stay under SELV limit, it's still legal ;).

Good point on the fuse usage (instead of blocking diode). Even for DIY, it's important to consider if there is a fault condition.

Post Reply