Transformerless inverters, the differences

[david schweickert]

Hello,
I am going to be installing some power-one transformerless inverters and know of a few changes that need to be made such as using PV wire that is thicker insulation than use-2, and use of an ungrounded module I believe. I the NEC 690.35 it refers to an "ungrounded Photovoltaic Power systems".
Q. Does this section include transformerless inverters
Q. what other changes should i include in my materials and work. Thanks a bunch.

 

[jaggedben]

and use of an ungrounded module I believe.

It's not the module that's ungrounded, it's the system. That means there is no grounded conductor, meaning no white wire on the DC side. There is still an equipment ground though.

Q. Does this section [690.35] include transformerless inverters

You could say it applies since all transformerless inverters require ungrounded systems, AFAIK.

Q. what other changes should i include in my materials and work. Thanks a bunch.

As mentioned, no white wire on the DC side for an ungrounded system. Also, if you have a disconnect, both negative and positive should be disconnected. Otherwise, follow the instructions in the inverter manual and 690.35, but it should be otherwise similar.

 

[ggunn]

Hello,
I am going to be installing some power-one transformerless inverters and know of a few changes that need to be made such as using PV wire that is thicker insulation than use-2, and use of an ungrounded module I believe. I the NEC 690.35 it refers to an "ungrounded Photovoltaic Power systems".
Q. Does this section include transformerless inverters
Q. what other changes should i include in my materials and work. Thanks a bunch.

I'm pretty sure that both DC+ and DC- conductors must have OCPD (fuse or breaker) in ungrounded arrays if there are 3 or more strings, so that means a different type of combiner. In cases where home runs are routed all the way back to the inverter, it is common to combine DC- conductors at the transition box; with ungrounded arrays you cannot do that.

I didn't know about USE-2 not being acceptable for string wiring in ungrounded arrays; can you cite a reference?

Of course, modules frames and racking still must be grounded via the EGC.

 

[BillK-AZ]

I didn't know about USE-2 not being acceptable for string wiring in ungrounded arrays; can you cite a reference?

690.35(D)(3)

Note that USE-2 is not referenced by the above. Major difference is two layers of insulation and a 1000 volt rating.

 

[jaggedben]

In cases where home runs are routed all the way back to the inverter, it is common to combine DC- conductors at the transition box; with ungrounded arrays you cannot do that.

I don't see why the combining location is any different in ungrounded arrays. Can you enlighten me?

 

[ggunn]

I don't see why the combining location is any different in ungrounded arrays. Can you enlighten me?

No problem. It's the because of the difference between a transition box and a combiner box. With a combiner box, there are multiple string inputs and a single DC output conductor pair to the inverter. In a transition box, the strings are not combined; the purpose is to transition from PV wire to THWN-2 (or whatever) and the PV source circuits are home runs to the inverter.

If there are three or more strings, the home run DC+ conductors connect to separate fused inputs at the inverter. In a system where the DC- is grounded, the DC- conductors are simply bussed wherever they are combined, so it is common practice to combine them in the transition box to save on wire, conduit size, etc. for the run to the inverter. However, in an ungrounded system, both the DC+ and DC- conductors are fused where they are combined, so if you are running home runs from the PV back to the inverter, the DC- conductors must be kept separate as well.

 

[jaggedben]

In a system where the DC- is grounded, the DC- conductors are simply bussed wherever they are combined, so it is common practice to combine them in the transition box to save on wire, conduit size, etc. for the run to the inverter.

I've never encountered that practice. It seems like a bad idea when it comes to diagnosing ground faults. In my opinion, grounded conductors should be combined in the same location as ungrounded conductors, preferably in a terminal block such as you find in combiners, or something similarly easy to disconnect.

 

[ggunn]

I've never encountered that practice. It seems like a bad idea when it comes to diagnosing ground faults. In my opinion, grounded conductors should be combined in the same location as ungrounded conductors, preferably in a terminal block such as you find in combiners, or something similarly easy to disconnect.

I won't comment on the advisability of the practice, but I know for a fact that it is commonly done and it passes inspections. I don't see where it would be a problem for diagnosing ground faults, though; differences in voltage on the DC+ conductors show where the fault is.

Advisability aside and true to the subject of the thread, this can be done with conventional inverters where the DC- is grounded but not with a transformerless inverter where the DC- is not grounded. One reason for that is ground fault detection; in a transformerless inverter, ground faults are detected by a difference in current between the DC+ and DC- conductors, so when an inverter has multiple DC inputs, the DC- must remain separate as well as the DC+ for each input.

 

[jaggedben]

I won't comment on the advisability of the practice, but I know for a fact that it is commonly done and it passes inspections. I don't see where it would be a problem for diagnosing ground faults, though; differences in voltage on the DC+ conductors show where the fault is.

Differences in voltage usually don't show when the grounded conductors are connected to each other, and that's the problem. Who wants to go around opening transition j-boxes (possibly under the array) and undoing splices in the grounded conductors in order to diagnose a ground fault? Not me.

Advisability aside and true to the subject of the thread, this can be done with conventional inverters where the DC- is grounded but not with a transformerless inverter where the DC- is not grounded. One reason for that is ground fault detection; in a transformerless inverter, ground faults are detected by a difference in current between the DC+ and DC- conductors, so when an inverter has multiple DC inputs, the DC- must remain separate as well as the DC+ for each input.

I thought the reason was that overcurrent protection is required on both source circuit conductors (when using more than two strings), so you can't just combine in a j-box. Multiple inputs aside, if you are combining strings for a single input I believe this can be done at any location, regardless of whether the inverter is transformerless. I think it's advisable to combine both + and - at the same location, but not required by code or inverter design.

 

[ggunn]

Differences in voltage usually don't show when the grounded conductors are connected to each other, and that's the problem. Who wants to go around opening transition j-boxes (possibly under the array) and undoing splices in the grounded conductors in order to diagnose a ground fault? Not me.

Sure they do; I have found ground faults in arrays without disconnecting a thing, and whether the DC-'s are grouped at the transition box or at the inverter is of no consequence. And who said anything about splices? A transition box when it's used this way has terminal blocks for everything.

I thought the reason was that overcurrent protection is required on both source circuit conductors (when using more than two strings), so you can't just combine in a j-box. Multiple inputs aside, if you are combining strings for a single input I believe this can be done at any location, regardless of whether the inverter is transformerless. I think it's advisable to combine both + and - at the same location, but not required by code or inverter design.

Of course, if you are combining into a single input that is a different scenario from what I was talking about. The question posed was what the differences are between transformerless and conventional inverters. Here's one: If you are running separate DC+ connectors all the way back to multiple inputs on a transformerless inverter, then your DC- connectors must remain separated as well because of the way that transformerless inverters detect ground faults. Such is not the case with a conventional inverter. Whether or not you would choose to wire a system with combined DC- and discrete DC+ into multiple inputs on a conventional inverter is of course your call, but I work in the industry and I know that it is commonly done, and inspectors in the jurisdictions I work in have no problem with it.

The reason that three strings or more must be individually fused (which is another issue entirely from the subject of this thread) is if three or more strings are connected without fusing and a ground fault occurs within a string, the outputs of the unfaulted strings will feed the fault and cause an overcurrent condition in the faulted string. If there are only two strings connected and there is a fault in one of them, the fault current is limited to the Isc of one string.

 

[jaggedben]

Sure they do; I have found ground faults in arrays without disconnecting a thing, and whether the DC-'s are grouped at the transition box or at the inverter is of no consequence.

As someone who as diagnosed a few ground faults myself, I find it astonishing you would say something like this. The problem is the worst in two-string arrays, because the voltages from each ungrounded conductor will appear identical to ground as long as the grounded conductors are still connected. Maybe you have never had to diagnose a fault on a two-string array. On multi-string arrays it can be different, but it still usually makes things simpler if you disconnect everything. (On ungrounded conductors this can be more easily done by opening fuse holders). If you have a method that doesn't require this I'm certainly curious.

And who said anything about splices? A transition box when it's used this way has terminal blocks for everything.

Okay, good. As I said, I've never encountered this practice.

The question posed was what the differences are between transformerless and conventional inverters. Here's one: If you are running separate DC+ connectors all the way back to multiple inputs on a transformerless inverter, then your DC- connectors must remain separated as well because of the way that transformerless inverters detect ground faults. Such is not the case with a conventional inverter.

Normally I would take 'multiple inputs' to mean separate MPPT tracking inputs. (Obviously in such a case you don't want to combine conductors on either side of the circuit that don't go to the same input. I don't see this as something you would do on a conventional inverter any more than a transformerless one.)

But you seem to be using 'multiple inputs' to simply mean multiple available connection points at the inverter (with or without included overcurrent protection as appropriate). In that case I would maintain that it makes no (electrical) difference whether you combine your DC - at the inverter or elsewhere, as long as you use overcurrent protection appropriate to the system. If you wanted, you could still combine your DC - conductors remotely from the inverter and run your DC + conductors to the inverter and combine them there, so long as you included proper fusing at both combining locations.

I don't understand why the inverter's method of ground fault detection makes any difference to this question. Transformerless inverters are not incapable of accepting combined circuits. The only way I can make sense of what your saying here is that the requirement for fuses on both + and - will mean different or additional equipment that will probably discourage the practice you have mentioned.

Whether or not you would choose to wire a system with combined DC- andr discrete DC+ into multiple inputs on a conventional inverter is of course your call, but I work in the industry and I know that it is commonly done, and inspectors in the jurisdictions I work in have no problem with it.

I didn't say it was a code violation or would cause a problem with system function. Still, I like to have my + and - conductors for any given source circuit combined at the same location, wherever that is. It makes diagnosing ground faults easier, and it makes the system easier to figure out for someone who did not install it.

 

[ggunn]

As someone who as diagnosed a few ground faults myself, I find it astonishing you would say something like this. The problem is the worst in two-string arrays, because the voltages from each ungrounded conductor will appear identical to ground as long as the grounded conductors are still connected. Maybe you have never had to diagnose a fault on a two-string array. On multi-string arrays it can be different, but it still usually makes things simpler if you disconnect everything. (On ungrounded conductors this can be more easily done by opening fuse holders). If you have a method that doesn't require this I'm certainly curious.

Semantics. Of course I open switches to the strings (fuse holders or breakers) to measure the voltages on the DC+ conductors to find the fault. Disconnecting the DC- connectors from each other is not necessary and it makes no difference if they are grouped at the combiner or at the inverter. And of course, on a two string array you have to separate the DC+ conductors, just the same as any other array.

Normally I would take 'multiple inputs' to mean separate MPPT tracking inputs. (Obviously in such a case you don't want to combine conductors on either side of the circuit that don't go to the same input. I don't see this as something you would do on a conventional inverter any more than a transformerless one.)

But you seem to be using 'multiple inputs' to simply mean multiple available connection points at the inverter (with or without included overcurrent protection as appropriate). In that case I would maintain that it makes no (electrical) difference whether you combine your DC - at the inverter or elsewhere, as long as you use overcurrent protection appropriate to the system. If you wanted, you could still combine your DC - conductors remotely from the inverter and run your DC + conductors to the inverter and combine them there, so long as you included proper fusing at both combining locations.

I don't understand why the inverter's method of ground fault detection makes any difference to this question. Transformerless inverters are not incapable of accepting combined circuits. The only way I can make sense of what your saying here is that the requirement for fuses on both + and - will mean different or additional equipment that will probably discourage the practice you have mentioned.

Let me try again to be succinct. By multiple inputs I mean something like the multiple fused inputs on a Sunny Boy conventional inverter. You can connect four strings to it without an external combiner box by running home runs from the strings to the inverter. You will likely want to transition from the USE-2/PV wire to THWN-2 as close to the array as you can in a transition (junction) box. At that point you may combine the DC- conductors onto a single conductor and connect to any of the DC- terminals on the inverter, and many do. Whether the DC- connectors are grouped at the J box or the inverter makes no difference in chasing ground faults.

However, on a transformerless inverter (at least with the SMA models) with multiple fused inputs, if you run home runs to the inverter for the DC+, you must also keep the DC- conductors separate, because the inverter detects ground faults differently. On a conventional inverter, there is a fuse or breaker internally that connects DC- bus to ground, and current flowing through it is the trigger for detecting the fault. That method will not work for an ungrounded array, so the way that SMA does it (and I would expect that others do something similar) is to route the flow for each input circuit through parallel ammeters on both the DC+ and DC- lines and compare them. A ground fault will generate a difference between the ammeter readings and trigger the alarm.

This means that you cannot home run the DC+ conductors and group the DC- connectors as you can on a conventional inverter because there would be no way to balance the ammeter currents. This is a difference between conventional and transformerless inverters, which was the question posed by this thread.

As I said earlier, if you prefer to also home run the DC- if you home run the DC+, there is nothing that says that you cannot do that, and that method works for both conventional and transformerless inverters. Whether one method is desirable over the other for conventional inverters is a totally different discussion.

 

[BillK-AZ]

Here's one: If you are running separate DC+ connectors all the way back to multiple inputs on a transformerless inverter, then your DC- connectors must remain separated as well because of the way that transformerless inverters detect ground faults. Such is not the case with a conventional inverter.

The situation with the Power-One PVI-6000 transformerless inverter is not as quoted above (per discussion with Power-One). They use a toroid sensor through which all the DC from the array passes. If there is a balance (no ground fault) there is no magnetic field. If one wants to there are provisions for combining all inputs so that the inverter acts like a single inverter without the two independent sets of inputs. Normally there are 4 inputs, call them A, B, C, and D, with A-B connected in parallel just past the terminal block, same for C-D. There are no fuses in the basic model, if string fuses are needed, they must be in external combiner boxes. Best used with only 4 strings, that way no fuses are needed if they are wired with no more than two strings per MPPT. For the PVI-6000 it seems that a single pair of DC wires is not a practical option because each of the terminals that are available are limited to 18A.

 

[jaggedben]

Disconnecting the DC- connectors from each other is not necessary and it makes no difference if they are grouped at the combiner or at the inverter.

Yes, it does make a difference (disconnecting them, that is), but we should start a new thread if we want to argue about that.

Let me try again to be succinct. By multiple inputs I mean something like the multiple fused inputs on a Sunny Boy conventional inverter.

Electrically, using the SMA integrated disconnect is no different than using any other combiner designed for grounded systems. (It really should be called the SMA Integrated Disconnect and Combiner, but I guess that's too long for everyday.)

The inverter actually only has one input.

However, on a transformerless inverter (at least with the SMA models) with multiple fused inputs, if you run home runs to the inverter for the DC+, you must also keep the DC- conductors separate, because the inverter detects ground faults differently.

I believe that how the inverter detects ground faults is not relevant. AFAIK SMA doesn't make single phase inverters with multiple MPPT inputs (unlike, say, Power One), so all strings are ultimately combined whether it is transformerless or conventional. All GFDI action (at least with Sunny Boys) happens inside the inverter, after strings are combined, regardless of the type of inverter. You can combine the strings remotely or close to the inverter, and combine + and - at the same place or separately, with both types of inverters, according to your pleasure.

What needs to be said is that, even if you run all your DC + and - conductors to the inverter, you cannot use only the regular SMA integrated disconnect to combine your strings for a transformerless inverter. The reason has nothing to do with ground fault detection, rather it is that the SMA integrated disconnect lacks overcurrent protection on both sides of the circuit as required for ungrounded systems, and this is a code violation. You need to use something like the combiner that SMA offers precisely for such a purpose, which has fuseholders for both negative and positive conductors. I believe this is merely a matter of meeting code and making the system relatively safe in case of faults. That is, I believe that if you did use the regular SMA integrated disconnect on their transformerless inverter, the GFDI would still function properly, since that all actually happens in the inverter, not the disconnect, for both types of inverters.

If I were installing a system with an SMA transformerless inverter, I could, if I wanted, use a regular fused combiner at the array, one designed for a grounded system, but only combine the DC - conductors there, and pass the multiple DC + conductors through the combiner down to the SMA disconnect. It might even make economic sense to do that.

 

[BillK-AZ]

....

If I were installing a system with an SMA transformerless inverter, I could, if I wanted, use a regular fused combiner at the array, one designed for a grounded system, but only combine the DC - conductors there, and pass the multiple DC + conductors through the combiner down to the SMA disconnect. It might even make economic sense to do that.

The SMA transformerless inverters have only a two-wire DC input, their installation instructions show the use of a combiner box that fuses both + and - of each string. SMA sells a 6-input DC combiner (TLUS SBCBTL6) with 12 fuse holders for use with the SMA transformerless inverters.

 

[ggunn]

The SMA transformerless inverters have only a two-wire DC input, their installation instructions show the use of a combiner box that fuses both + and - of each string. SMA sells a 6-input DC combiner (TLUS SBCBTL6) with 12 fuse holders for use with the SMA transformerless inverters.

http://www.solar-net.nl/zonnepanelen/files/images/Manual-SMA-SB2100TL.pdf

In section 6.2 it shows two two-conductor inputs.

 

[ggunn]

Yes, it does make a difference (disconnecting them, that is), but we should start a new thread if we want to argue about that.



Electrically, using the SMA integrated disconnect is no different than using any other combiner designed for grounded systems. (It really should be called the SMA Integrated Disconnect and Combiner, but I guess that's too long for everyday.)

The inverter actually only has one input.



I believe that how the inverter detects ground faults is not relevant. AFAIK SMA doesn't make single phase inverters with multiple MPPT inputs (unlike, say, Power One), so all strings are ultimately combined whether it is transformerless or conventional. All GFDI action (at least with Sunny Boys) happens inside the inverter, after strings are combined, regardless of the type of inverter. You can combine the strings remotely or close to the inverter, and combine + and - at the same place or separately, with both types of inverters, according to your pleasure.

What needs to be said is that, even if you run all your DC + and - conductors to the inverter, you cannot use only the regular SMA integrated disconnect to combine your strings for a transformerless inverter. The reason has nothing to do with ground fault detection, rather it is that the SMA integrated disconnect lacks overcurrent protection on both sides of the circuit as required for ungrounded systems, and this is a code violation. You need to use something like the combiner that SMA offers precisely for such a purpose, which has fuseholders for both negative and positive conductors. I believe this is merely a matter of meeting code and making the system relatively safe in case of faults. That is, I believe that if you did use the regular SMA integrated disconnect on their transformerless inverter, the GFDI would still function properly, since that all actually happens in the inverter, not the disconnect, for both types of inverters.

If I were installing a system with an SMA transformerless inverter, I could, if I wanted, use a regular fused combiner at the array, one designed for a grounded system, but only combine the DC - conductors there, and pass the multiple DC + conductors through the combiner down to the SMA disconnect. It might even make economic sense to do that.

I am at a loss to see what we are arguing about. Yes, of course, an SMA conventional inverter (the inverter itself) has only one input and the four inputs are to an integrated combiner box. Separate MPPT inputs are on some inverters, but not on Sunny Boy conventional inverters, and that has no bearing on what I said. Regardless of the semantic differences, it is a common practice to home run the DC+ conductors to the separate inputs on the inverter (or on the integrated combiner at the inverter, if you like) but to combine the DC- connectors at the array in a transition (junction) box. SMA says not to do that with a transformerless inverter; I attended a workshop on SMA transformerless inverters last fall and that's how it was explained to me. If you disagree with that, take it up with SMA; I am only passing it on.

 

[jaggedben]

The SMA transformerless inverters have only a two-wire DC input, their installation instructions show the use of a combiner box that fuses both + and - of each string. SMA sells a 6-input DC combiner (TLUS SBCBTL6) with 12 fuse holders for use with the SMA transformerless inverters.

Thanks for the clarification. I'll admit I've only seen the diagram of the disconnect in the manual for US products, on page 30 at http://files.sma.de/dl/10707/SB8-10TL-IUS103811.pdf My company hasn't actually started installing these.

 

[jaggedben]

I am at a loss to see what we are arguing about. ...

Well, you were saying that the method of ground fault detection affects combining location and equipment. I was saying the ground fault detection method has nothing to do with it, it's only the overcurrent protection requirements that make a difference.

If installing a two string system with no requirements for overcurrent protection on the PV source circuits (in accordance with the exception in 690.9 A) it would be allowable to combine both DC+ and DC- at the transition box and run them both to the disconnect on an SMA transformerless inverter (US models, that is). If SMA says otherwise I would be interested to hear their reasoning.

 

[ggunn]

Well, you were saying that the method of ground fault detection affects combining location and equipment. I was saying the ground fault detection method has nothing to do with it, it's only the overcurrent protection requirements that make a difference.

If installing a two string system with no requirements for overcurrent protection on the PV source circuits (in accordance with the exception in 690.9 A) it would be allowable to combine both DC+ and DC- at the transition box and run them both to the disconnect on an SMA transformerless inverter (US models, that is). If SMA says otherwise I would be interested to hear their reasoning.

Both, yes, but not (one and not the other), parentheses added for clarity. That's what SMA told me, anyway. The SMA guy could have been mistaken.