Flaky SolarEdge Ethernet?

[jasonvr]

I've got a SolarEdge SE6000A and have had it up and running for about a year at this point. It was originally installed with the GSM kit until I realized the limitations of that, so I installed a WiFi bridge in the garage and ran an Ethernet cable to the inverter.

Every once in a while I would get a notification from SE stating that it hadn't received data in more than 2 hours and it would clear itself. I think I got that maybe 3 times in a year.

However, over the last 2 weeks, it has happened a ton. I just tested connecting my laptop to the WiFi bridge instead and it is just fine (i.e. it's not the bridge). Sometimes the inverter will report not even being able to get a DHCP address. Currently it is in a state where it has an IP address, but all the pings fail. I can ping the inverter from my PC, so the network connectivity is fine. I did a packet capture on my router and there are no packets coming from the inverter at all, even when it is supposedly running its tests.

Earlier this week I restarted the inverter after sundown and it immediately cleared up the issue. However, now it is back. I don't want to restart the inverter in the middle of production, so there's not a lot I can do. Even if I don't reboot it, I have no doubt it will suddenly start working this evening

I do wonder if it is heat related at all as it is quite hot outside.

Has anyone else seen this kind of behavior?

[JSchnee21]

The SolarEdge servers seem to be really lagging over the past couple weeks. I've have several days in which the data displayed on the SE portal was several hours late. It generally catches up by late in the afternoon / early evening. But, I haven't received any errors.

My 11.4k inverter is outside, but it's pretty well shaded (on East side of house) and it doesn't get all that hot here in NJ. My inverter temps usually peak ~145-155F in the Summer.

Is your inverter inside the garage or outside? Do you log it's temp on PVOutput.org?

Best,
Jonathan

[jasonvr]

Inverter is outside on the west facing wall, so it really gets the brunt of it in the afternoon. Today the reporting cut off at noon though. I do log the inverter temp thru PVOutput. Last reported reading today was 126F. During our heatwave a couple of weeks ago I saw it hit 158F. I've certainly had delays recently as well and yeah, it catches up in the evening on its own.

Be aware that the data alert might not be set up unless you did it. I think I had to enable that one, but I don't exactly remember

[jasonvr]

OK, sun went down so I could experiment
1) Pulled Ethernet cable from bridge and put it back in allowing reset on inverter - no effect
2) Turned off inverter via the rotator switch - no effect - link light still illuminated on bridge (i.e. circuit still active on inverter)
3) Threw main disconnect switch to completely shut down inverter and turned back on (rotator still off) - after initial DHCP delay everything worked and data was uploaded

So, it really does appear to be something in the inverter. I'm gonna give it a couple of days, then I'll be contacting either my installer or SolarEdge

[ButchDeal]

OK, sun went down so I could experiment
1) Pulled Ethernet cable from bridge and put it back in allowing reset on inverter - no effect
2) Turned off inverter via the rotator switch - no effect - link light still illuminated on bridge (i.e. circuit still active on inverter)
3) Threw main disconnect switch to completely shut down inverter and turned back on (rotator still off) - after initial DHCP delay everything worked and data was uploaded

So, it really does appear to be something in the inverter. I'm gonna give it a couple of days, then I'll be contacting either my installer or SolarEdge

ok you need to ready your manual.
Turning the rotary switch (the DC disconnect) is NOT turning your inverter off. It is just disconnecting the DC array and it should not be done without turning off the inverter first. There is a small toggle (on / off) switch on the inverter bottom just above the DC disconnect.
Turning off the AC disconnect is going to reset the inverter but is a hard reset.

[jasonvr]

D@mn, you're right of course. I knew there was an order of operations but I completely forgot about the toggle switch. I obviously don't have to do this that often. The good news is that since I was in night mode, I don't think it was super bad to have forgotten the little toggle (i.e. there was basically no voltage from the array at that point). I basically did the computer equivalent of pulling the power cord instead of a graceful shutdown. I'll see what kind of behavior I get today and try again in the right way if I have a comms interruption again

[ButchDeal]

We have had zero problems with the ethernet connections on many models for several years.
I would suspect your ethernet bridge though.

[NukeEngineer]

I had similar issues on my SE6000Hs units. I disabled DHCP and set the IPs of these to static, and these uploading problems disappeared.

[jasonvr]

We have had zero problems with the ethernet connections on many models for several years.
I would suspect your ethernet bridge though.

Considering that I could ping the IP of the inverter while it was failing to communicate with the servers, I'm failing to see how the bridge could be suspect. It's obviously allowing bidirectional traffic to the inverter at that point.

I had similar issues on my SE6000Hs units. I disabled DHCP and set the IPs of these to static, and these uploading problems disappeared.

Were you able to set the IP statically in the inverter? I've got a static DHCP reservation set in my router, but didn't know you could set a static IP thru the inverter itself. I assume I have to remove the cover and use the internal buttons for that.

[NukeEngineer]

Yes you can set a static IP in the inverter. For the older A models, I'm assuming you'd need to use the buttons on the comm board, so you'd need to remove the cover.

[ButchDeal]

Considering that I could ping the IP of the inverter while it was failing to communicate with the servers, I'm failing to see how the bridge could be suspect. It's obviously allowing bidirectional traffic to the inverter at that point.

A local ping is pretty low level, The router most likely dropped the reservation and started blocking the traffic.

[NukeEngineer]

A local ping is pretty low level, The router most likely dropped the reservation and started blocking the traffic.

Unlikely, most home routers do not filter traffic to just IP addresses it has leased. This is why setting a static IP (on your same subnet, of course) on a device and never informing the router about it, will *mostly* always work. Worst case here is that the inverter didn't update the lease and some other device got leased the IP it had, but then you'd see IP conflict messages on the other device if it was a PC or something reasonably capable.

Unless, of course, his router has this as a fancy feature.

I went through these similar tests as he. I could ping inverter, router showed a valid DHCP lease, etc., but the inverters basically refused to upload data every few days. Setting their IPs to static instantly removed this headache.

[ButchDeal]

Unlikely, most home routers do not filter traffic to just IP addresses it has leased. This is why setting a static IP (on your same subnet, of course) on a device and never informing the router about it, will *mostly* always work. Worst case here is that the inverter didn't update the lease and some other device got leased the IP it had, but then you'd see IP conflict messages on the other device if it was a PC or something reasonably capable.

Unless, of course, his router has this as a fancy feature.

I went through these similar tests as he. I could ping inverter, router showed a valid DHCP lease, etc., but the inverters basically refused to upload data every few days. Setting their IPs to static instantly removed this headache.

Most home routers DO filter and even without the filter they don't router to addresses they have not leased or have arp entries for. Router security is pretty common these days especially in the ones provided by the ISP. The NAT to hide the non-routable address space used on the LAN can get confused as well. Static addresses work as it does its own arp table updates and all the traffic is outbound, The arp table updates are broadcast and the router updates its NAT table based on them.

However with the wife extender, many times you have a second DHCP NAT going on with the wifi extender.

The reason the Static worked is because the router is poorly handling the DHCP renewal. This is common as not all routers implement the DHCP correctly and timed lease/renewals can be handled differently. Several cheap routers copied the same buggy code base, most have patches but many are out of support or have manual patch methods.

[NukeEngineer]

Most home routers DO filter and even without the filter they don't router to addresses they have not leased or have arp entries for. Router security is pretty common these days especially in the ones provided by the ISP. The NAT to hide the non-routable address space used on the LAN can get confused as well. Static addresses work as it does its own arp table updates and all the traffic is outbound, The arp table updates are broadcast and the router updates its NAT table based on them.

However with the wife extender, many times you have a second DHCP NAT going on with the wifi extender.

The reason the Static worked is because the router is poorly handling the DHCP renewal. This is common as not all routers implement the DHCP correctly and timed lease/renewals can be handled differently. Several cheap routers copied the same buggy code base, most have patches but many are out of support or have manual patch methods.

It sounds like you're confusing internally generated traffic with externally (incoming) generated traffic.

I've never encountered a home router that enforced DHCP'ed only clients behind it's NAT, never! Even if that were a feature, what's the point? Someone who has PHY access to your network can do "whatever they want" on it. Forcing an internal attacker to first acquire a DHCP lease seems like a screen door on a vault.

Based on your last sentence of your first paragraph, sounds like you're agreeing with my initial point anyway.

[ButchDeal]

It sounds like you're confusing internally generated traffic with externally (incoming) generated traffic.

I've never encountered a home router that enforced DHCP'ed only clients behind it's NAT, never! Even if that were a feature, what's the point? Someone who has PHY access to your network can do "whatever they want" on it. Forcing an internal attacker to first acquire a DHCP lease seems like a screen door on a vault.

Based on your last sentence of your first paragraph, sounds like you're agreeing with my initial point anyway.

Just because you havent doesnt mean that others, perhaos others with more exoerience in lan work havent. I didnt say it was a security feature, i said it was buggy code, yu lept to that assumption.

In any case the double NAT/ DHCP is most likely the issue. I have seen that repeatedly.

It sounds like you didnt inderstand that sentance to me.

I also have considerable eperience with networking equipment, way more than with solar equipment.