Here's a weird one

[romex jockey]

Whoa, you just reminded me: I once installed a solar system at a police station: they had a some other common-mode things (intercoms I think) and the only way I could get it to work was put a choke on the entire PV system. Worked perfectly. I don't know whether others have used this trick. These were Enphase microinverters, maybe 10 years ago, that couldn't find eachother/the Enlighten box until I added the choke. I don't know why that would come into play here, it would be interesting to know if the trips happened at the same time each day (i.e. 30m after sunset or something).

This is most curious, can you describe this for us?

~RJ~

 

[jaggedben]

This is most curious, can you describe this for us?

~RJ~

The simple method is to wrap the line conductors through one or more ferrite torroids. I've done this a few times, it's probably what the previous post was referring to. The more expensive method involves a line filter device, which I've so far avoided. The pain in the neck is when the inverters and monitor are wired with loads in between them such that you cannot isolate them together to choke them off. But it can still be helpful depending on where the noise is coming from.

 

[tortuga]

Well for all the rf engineer's and (curious geeks like me) here is their patent: 9634723
And judging from that, in particular:


Along with the chokes mentioned there is some adjustment that can be done to lower the dBmV.
Quite frankly when people try to re-invent TCP/IP networking I get a laugh, it reminds me of the early days of the internet.
I avoid using / selling any system that uses a patented media layer.
One of the local utilities here almost got sold some patented proprietary power-line data solution for all their smart meters, and my engineer mentor was working on the project, this what he avoided, vendor lock -in, shortage of techs who can diagnose problems, security holes etc (anyone on the same transformer likely is on this 'network').
(They decided instead to implement a regular GSM network in the 400Mhz range at much more cost)
It would be a much better design to use one of the various IEEE 802 'open source' media layers.
but I digress...

 

[tallgirl]

They have quite a few communications related patents, including 10,615,742, which is one of mine.

Complaining about very low level communications being patented is really misguided. There's no market for mixing and matching technologies like microinverters. The right place to complain about vendor lock-in is at a higher level, like inverters or data loggers. When I was the Chief Architect at Draker Labs, we integrated utility-scale inverters which spoke all manner of vendor-specific crap into a common data set -- that would have benefitted from what is now emerging as the SunSpec standard.

If you read 10,003,300, which is another of mine, it deals with control features related to ride-through. Different power converters are going to handle ride-through in their own unique ways until some kind of standard emerges. We were trying to solve issues in small / dispersed grids like Hawaii and Australia with distributed PV causing overloaded circuits and/or excessive voltage rise -- positive voltage gradient in medium voltage distribution lines -- with bang-bang controllers. That also sucked and would have benefitted from standardization, but many of those problems are "new".

 

[RumRunner]

They have quite a few communications related patents, including 10,615,742, which is one of mine.

Complaining about very low level communications being patented is really misguided. There's no market for mixing and matching technologies like microinverters. The right place to complain about vendor lock-in is at a higher level, like inverters or data loggers. When I was the Chief Architect at Draker Labs, we integrated utility-scale inverters which spoke all manner of vendor-specific crap into a common data set -- that would have benefitted from what is now emerging as the SunSpec standard.

If you read 10,003,300, which is another of mine, it deals with control features related to ride-through. Different power converters are going to handle ride-through in their own unique ways until some kind of standard emerges. We were trying to solve issues in small / dispersed grids like Hawaii and Australia with distributed PV causing overloaded circuits and/or excessive voltage rise -- positive voltage gradient in medium voltage distribution lines -- with bang-bang controllers. That also sucked and would have benefitted from standardization, but many of those problems are "new".

Hi TG.

That’s impressive.
Since this is as you claimed that SunSpec“ is an emerging standard. . . is it using the same architectural model as the generic seven-layer OSI stack that an old cudger like me is familiar with ?

I took Network Engineering almost twenty five years ago and never would have thought a new Architectural model be a common garden variety that is supplanted by the four-layered OSI, SunSpec standard.

I realized Solar Systems were not even on the horizon at the time . .so I would be totally lost if a radical approach in network systems were thrown under my feet. LOL

The patents you hold would be on the physical layer I imagine?

Keep those valuable techs info coming and I’m sure a lot of people can benefit from them.
All the best.

 

[tallgirl]

Hi TG.

That’s impressive.
Since this is as you claimed that SunSpec“ is an emerging standard. . . is it using the same architectural model as the generic seven-layer OSI stack that an old cudger like me is familiar with ?

I took Network Engineering almost twenty five years ago and never would have thought a new Architectural model be a common garden variety that is supplanted by the four-layered OSI, SunSpec standard.

I realized Solar Systems were not even on the horizon at the time . .so I would be totally lost if a radical approach in network systems were thrown under my feet. LOL

The patents you hold would be on the physical layer I imagine?

Keep those valuable techs info coming and I’m sure a lot of people can benefit from them.
All the best.

I don't hold the patents. They were assigned to SolarBridge when I wrote microinverter and datalogger firmware there. They went to either SunPower or Enphase as the assets of SolarBridge were sold various ways. But yes, those are primarily physical layer patents, though I believe one of them isn't even in the communications stack. One of them is about managing power distribution on medium voltage -- say, 10-20kV -- feeders where distributed PV was pushing so much power backwards that either the circuits were going to be overloaded, or the voltage was going to be outside of the FERC ranges.

When I was at Draker I learned about utility interconnection, which is a different matter. We had projects with complex switchgear which was boosting anywhere from 50kV to 500kV. Most of that was over garden-variety Modbus. The reason for all these weird protocols for consumer-scale PV is the desire not to run a bunch of extra communications wires.

Back to the OSI model ...

Recall that the "bottom" of the OSI model is very low level, so it's completely appropriate to start well above the physical layer, and to stop short of the application layer. SunSpec sits on the boundary between the media layers and the host layers. SolarEdge, for example, provides standard RS-485-based Modbus communication to their inverters (or they did). When I was writing data logger firmware my goal was to provide TCP/IP-based Modbus. Then I left SolarBridge and ran off to Draker because they had much bigger projects -- the largest was 23MW. These days I do industrial controls.

 

[RumRunner]

....Back to the OSI model ...
Recall that the "bottom" of the OSI model is very low level, so it's completely appropriate to start well above the physical layer, and to stop short of the application layer. SunSpec sits on the boundary between the media layers and the host layers. SolarEdge, for example, provides standard RS-485-based Modbus communication to their inverters (or they did). When I was writing data logger firmware my goal was to provide TCP/IP-based Modbus. Then I left SolarBridge and ran off to Draker because they had much bigger projects -- the largest was 23MW. These days I do industrial controls.

OK, I got that.
What I still have to get my head wrapped around on-- is: the initial [start] “WELL ABOVE” the physical layer that you claim.
How far up the ladder would you start before reaching the Destination Address Information Layer, which is the topmost Application Layer.

From the bottom (lowest) least intelligent ladder, you would have to negotiate five layers i.e. Data Link, Network, Transport, Session and Presentation before you can communicate with the most intelligent Application Layer.
During the programming phase, you will need to pick which layer you would have to start to make sure you are able to establish communication.
With so many protocols coming out of the wood works--it seems you would need to focus on your target to be able to communicate flawlessly.
As an example, let’s take the Modbus Plus protocol from Schneider Electric. This protocol is different from the old generic Modbus.
They created their own proprietary systems (Modbus+) that will only recognize hardware that they designed from their R & D laboratory.
They also use the serial interface.

In other words they are saying:
“you use our systems with the peripherals that we made to make things work.“

Other PLC manufacturers use the TCP/IP protocol which is considered the de facto standard.

This was based on the relay logic. (coil and NO NC contact)
Front line technicians already have the basic understanding of this system.
So, only minimal hand holding mentoring is necessary.

FWIW:

I trained with the Modicon engineers during the early introduction of Programmable Logic Controllers (PLC)
Modicon was the brand name by Gould Electronics Inc. for their PLC products before the company was acquired by Schneider Electric. Actually the company changed hands from Japanese owned to Chinese to German. Only when they got involved with military projects that they reverted back to US ownership. They were forced to sell their foreign assets because of the DOD “buy American” policy.
Although Schneider is not US wholly owned --some divisions are still under the “buy American” material acquisition requirement.
The MODBUS nomenclature stuck even after acquisition.

MODICON is where MODbus got its name.

So much for histrionics.

Have a nice day.

 

[jim dungar]

Other PLC manufacturers use the TCP/IP protocol which is considered the de facto standard.

MODICON was the original name/company which was then bought by Gould. Modbus+ was developed by Modicon (well before Schneider Electric acquired them) it was their equivalent to AB's proprietary 'blue hose'. Modicon was one of the first to major PLC lines to offer 'inter component' communications over TCP/IP. I remember when it was introduced some 20 years ago with their Momentum product offering.

 

[tortuga]

@tallgirl wow thank you for your reply!

Complaining about very low level communications being patented is really misguided. There's no market for mixing and matching technologies like microinverters.

With all due respect in my humble opinion there is more to this than mixing and matching.
cat5 ethernet cable and tcp/ip do not nuisance trip AFCI breakers.
I have tools in my truck to troubleshoot an ethernet connection.
I have no tools to troubleshoot some proprietary protocol.
Thats the amazing success of wired ethernet.
Even Apple whom love to kill off open standards cant kill it.
I am working from the perspective of solving the AFCI issue, If I was sent out to solve this AFCI problem I would immediately want to run a separate com wire.

The reason for all these weird protocols for consumer-scale PV is the desire not to run a bunch of extra communications wires.

That desire to not to run a bunch of extra communications wires is spreading like this virus going around.
Total unrelated anecdotal example:
One of my co-workers recently was troubleshooting the emergency phone in an elevator.
It used to be a POTS line, but a portion of the link got upgraded to VOIP over WIFI.
They were only requesting me to troubleshoot the WIFI, its not a big office building so it *could* have worked.
I insisted they run a cat5 all the way to the elevator room from the main patch panel, the economics were not good.
Problem solved.
Yes its more work and perhaps I am sliding fast down the silvery slope of becoming an old school curmudgeon, things attached to a building should be 'wired' as in comm cable and things that go in your pocket or bag should be 'wireless'.
And data signals should stay in there own cables.

They have quite a few communications related patents, including 10,615,742, which is one of mine.

Wow much respect. You probably write in assembly.
My critique is the manufacturer needs to offer the option of using ethernet for these edge cases.
At least they are not using batteries to power the communications like the dimmer folks.

 

[jaggedben]

@tallgirl wow thank you for your reply!


...
My critique is the manufacturer needs to offer the option of using ethernet for these edge cases.
...

How would you provide for running ethernet to dozens of NEMA6 microinverters on a roof without adding significant cost to the equipment? They don't 'need' to do this. It's an economic decision.

All of the micro-inverter PLC troubleshooting service calls I've ever done - and there have been a few doozies - wouldn't amount to 1% of the extra labor and material to run ethernet to every inverter. (I didn't think it was even worth it with Tigo, where you only had to run the ethernet to one device on the roof.)

 

[tortuga]

How would you provide for running ethernet to dozens of NEMA6 microinverters on a roof without adding significant cost to the equipment? They don't 'need' to do this. It's an economic decision.

All of the micro-inverter PLC troubleshooting service calls I've ever done - and there have been a few doozies - wouldn't amount to 1% of the extra labor and material to run ethernet to every inverter. (I didn't think it was even worth it with Tigo, where you only had to run the ethernet to one device on the roof.)

I just like options, 1% seems like a low price to pay for the existing electrical system to function.

Like I said it would be nice for the guy with tripping AFCI breakers to have the option of running cat5, not do it for every system.

I got actually got kicked out of the 'Value Engineering Team' years ago. LOL

jaggedben I enjoy reading your comments and opinions on things and agree some type of WIFI standard then would probably be better.
My old boss was a stick in the mud, and I never want to get like that, so I am here to learn.
When first looking at the data plans for the last job we had I thought "geez were never gonna get this done".
There was cat5 for the usual computer and phones then there was lighting realys, alarms, doors, roof, basement, cameras, something in the bathroom, and several to the elevator, I mean just about everything.
I was amazed to watch our guys do all the data cabling pretty fast.
They enjoy the work and it pays well,
Now its done and the office runs smoothly.

 

[PWDickerson]

I just went out on a service call to a site where we installed a residential PV system with a Solectria inverter a couple years ago. The owner built the workshop first (where the PV is installed), and he is just now finishing the house. There are no AFCI breakers in the workshop panel. The PV system in the detached shop appears to be tripping the 2-pole AFCI breakers in the home. But the single-pole AFCI breakers aren't tripping.

We had a few issues early on with these Solectria inverters (actually manufactured by Delta I believe) tripping AFCI breakers, and we started installing ferrite beads on the inverter's AC conductors routinely. The inverter in question has had the ferrites installed from the beginning. Tech support at Solectria suggested that I put a couple more wraps around the ferrites and see if that resolves the issue. Can anyone think of a reason why the 2-pole breakers are affected and the 1-pole breakers are not?

 

[jaggedben]

...Can anyone think of a reason why the 2-pole breakers are affected and the 1-pole breakers are not?

Just theorizing in the most general manner, but...
Since the inverter output (I presume) is line-to-line, any noise from the inverter would probably be more evident on a breaker that measures more than one line. For the same noise to propagate to the neutral it would have to travel all the way to the transformer, through the coil and back out on the neutral. All that wire and coils might act as a filter.

 

[synchro]

The inverter in question has had the ferrites installed from the beginning. Tech support at Solectria suggested that I put a couple more wraps around the ferrites and see if that resolves the issue.

Are you putting both phases through a ferrite core, or using a separate ferrite core on each phase? Normally putting both phases through the core is what you want since this attenuates common mode noise, but the core won't saturate because the magnetic fields from the large inverter line currents cancel out. I agree that more turns around the ferrite should help if you can fit them on it.

 

[winnie]

I agree that more turns around the ferrite should help if you can fit them on it.

2631181381 Fair-Rite Products Corp. | Filters | DigiKey

Order today, ships today. 2631181381 – Clip Together (Order Qty 2=1 Assembly) Free Hanging Ferrite Core 490Ohm @ 100MHz ID 1.465" Dia (37.21mm) OD 2.445" W x 1.190" H (62.10mm x 30.23mm) Length 2.500" (63.50mm) from Fair-Rite Products Corp.. Pricing and Availability on millions of electronic...

www.digikey.com

The above is a rather large split ferrite; you can pre-form conductors into loops, put the ferrite into the loops, and then clamp the ferrite halves together. With the above it is easy to get one turn (2 passes of each conductor) of #2 THHN for 3 phases. Could probably do two turns (3 passes) of #2 for single phase.

-Jon

 

[PWDickerson]

The inverter in question is a single phase 6.6 kW using #8 AC conductors. I put one ferrite core (split-core type) on each leg as instructed by tech support. I was able to get three loops around the core or 4 conductors passing through its center. It buzzed a lot more after I increased the number of loops.