Hz is all over the place when I am on battery

Hz is all over the place when I am on battery

I have two Powerwall 2 units with SolarEdge Inverter. When the grid goes down and I am running on battery power, the HZ fluctuates from 60Hz to 71Hz in the 80s and I observed and videoed it exceeding 100 Hz.

Are there any settings that I should be looking at to get the Hz stable at 60Hz?

gregb | March 13, 2018

I have very little specific knowledge of Tesla Powerwalls or SolarEdge inverters, but it sounds to me like neither the Powerwalls or the SolarEdge are the "master". These devices are designed to synchronize to the electric grid, but in the off-grid case they are essentially referencing each other and "chasing their own tail" so-to-speak. The solution would be for one or the other (Powerwall or SolarEdge) to become the "master" and generate the reference 60 Hz. signal. If this is not possible, then another device could be added to provide the reference. This could be a small box that generates a low power (sub-Watt) 120 VAC, 60Hz. reference. Such a device would at its simplest, have a quartz crystal oscillator as the time reference.

rickwise | January 16, 2020

Did you use an oscilloscope or frequency counter? Frequency counters can false trigger on noise. Since your readings are all greater than or equal to 60Hz, I guess that you used a frequency counter. If I had the equipment I would use an oscilloscope in x/y mode with a 60Hz sinusoid applied to either the x or y input and the system under test applied to the other input and look for a stable 1:1 Lissajous pattern. Phase not important.

rogerdata | January 17, 2020

I agree with @rickwise. If you were actually getting 70-80 Hz any motors such as refrigerator or washing machine etc. would be running way faster then they should. I would think some motorized appliance would have already failed. So probably noise.

jeff | January 17, 2020

You didn't say what you were using to monitor the Hz, so there's no idea of how accurate it is, but raising the Hz from the normal 60cps to around 63-64Hz is how the inverters control the panel output if the battery (assuming you have one) is fully charged and there's more solar energy than your house is using. Raising the Hz lowers the panel output. It will return to 60cps when the sun goes down or more demand is present and there's enough sun. 75-80Hz is way beyond this system so I wonder what you are reading this with? I have both a UPS that shows Hz and a Kill-O-Watt meter which shows pretty much the same, and they've never read that high.

rickwise | January 17, 2020

Inverters use switching techniques that generate noise. Small amounts but still noise. These inverters are run by microprocessor that have a crystal oscillator. I would expect 60Hz with a tight tolerance all the time. Cycles per second is an old term. Maybe you have used a vibrating reed frequency meter. They are pretty noise immune. About using the o-scope in the way that I previous suggested, just use it like most people and that would be fine. I just like Lissajous patterns even though I probably murder the pronunciation.

gregbrew | January 19, 2020

As Jeff described above, during grid failures, with full batteries and excess solar production, there's nowhere for the extra power to go, so the TEG commands the batteries to shift their output frequency from the nominal 60Hz to about 63Hz. This variance is enough to convince the PV inverter that the grid is bad, so it disconnects the PV panels. The residence now runs on the PWs at ~63Hz until some threshold of depletion is reached (I don't know what that threshold is). At that point, the TEG instructs to batteries to go back to a 60Hz output to the house. The PV inverter sees that the grid is now stable (it's really the PW output...remember that the actual grid is down), and the PV supplies the house and recharges the PWs with excess. Once the PWs are full, the process repeats until the Sun goes down or the grid comes back up.

jeff | January 20, 2020

@Gregbrew: I notice that Tesla has revised the firmware and now, the app shows the PV's charging the battery to 100%, which they had previously said was impossible, the PV's shut down via Hz increase, and then the gateway switches the house to the battery, letting it drain down to around 94-95%, at which point it turns on the PV array, runs the house and uses the extra to recharge the battery back to 98%, and then continues to use the PV's to run the house until sundown, at which point it blends the PV's and battery until there's no more sun, ultimate running on the battery alone. We have grid but we use it sparingly, and normally not at all.

I don't understand the benefit o this newf overcharging, then draining, then recharging (hopefully there's enough sun left) and they using the PV's to run the house until sundown and thence on to the PW. It didn't used to do would charge the battery to 98%, and they leave the battery alone unless it was very cloudy...little sun, or the sun set. As the sun lowered, you could see the house drawing more and more on the still does this but now it has this new overcharge/drain/recharge cycle. Any thoughts on this?

jeff | January 20, 2020

Excuse the spelling errors...I don't see an 'edit' feature.

jeff | January 20, 2020

Also, mine is now having trouble remaining connected via cellular (AT&T) to Tesla. It worked flawlessly for two years, but now t would disconnect around 11:30am - 12pm and remain disconnected until about 5:30am the following morning, at which point it would automatically re-establish the cell connection and now you could use the app to monitor, until around noon. This happened repeated and at nearly the exact same times, 4-5 days in a row, perhaps more. I reconnected the gateway via our house wifi and it is staying connected and reporting and everything appears normal. Anyone else have issues with the AT&T cell connection for the gateway to Tesla? More specifically, here in Puerto Rico, where AT&T as had problems?

Patrick | January 20, 2020

Jeff - FWIW our new system runs TEG release 1.43.3 and works as you described in your first 1/20 post above. We are assuming this is normal as the solar inverter output must be throttled back to match the home consumption once the Powerwalls are fully charged. So far we have achieved our goal to get as close to 100% self-powered as possible and eliminate/minimize the use of grid power.

Sounds like this behavior is different than what you saw before but it shouldn’t be a problem. A quick call to Tesla support should confirm this.

We’re having any cellular data issues and we also have a hardwired Internet connection to the TEG. Suggest reporting the ATT issues to Tesla as well.

jeff | January 24, 2020

@Patrick: Thanks for your input. we're one version behind in firmware at 1.41.2 but it sounds like we're both having problems with the cell connect. I've reported this to Tesla and so far I don't see anything being done. This WAS working fine for the two years since we installed the system, but lately it is totally intermittent. Tesla says they are sending a message to our installer to send out a tech to see if there's anything they can do to improve the cell connection, which I highly doubt, since that's AT&T's department. It's not that the PW won't connect...there is NO cell service for it to connect to. This is seen in the TEG GUI on the network page. There's no green dot next to the cell connection, meaning it's unavailable. Frequently. cell comes and goes, so I revised PW to use our house wifi. But that's a problem since we don't have unlimited internet (HughesNet). I've inquired of Tesla as to what the data total for a 24 hour day might be, since this system connects and reports frequently. So far, no answer. Tesla seems to be dragging their heels on this. And the really weird part is back when they installed it, the chief of the crew told me they were unable to connect reliably to my house wifi, even when we put an access point right next to the PW! So they configured it to cell, and it's been working that way for 2 years. After I reconfigured it to our wifi, it stays connected to wifi. Go figure. And we've yet to hear from the tech who's supposed to come to the house. But, all said and done, the system works perfectly in all other respects, and as you pointed out, we are close to 100% off-grid. We only turn on the grid when we're using the electric clothes dryer...we have a gas stove and water heater, so the dryer it the only juice hog in the house.

Patrick | January 25, 2020

Good luck getting things resolved. To clarify - just realized I left out the word “not” regarding the cellular data issues - ours is working fine with AT&T. We also connected hardwired Internet service from the local provider. Both connections have been active with no issues since our initial installation in Dec 19.

Jones | January 29, 2020

I had some very odd problems when I first installed Powerwalls. I already had in place 4 small UPS units for a bunch of server and electronics gear. When off grid and on Powerwall backup, every single one of them interpreted the synthetic Powerwall frequency as “bad” and went into protect mode for the attached equipment. Removed the UPS units and installed high end surge suppressors with good noise filters. Works just fine now. The UPS interpret anything outside 59.7-60.3 Hz as unstable power. I measured the output of the Powerwalls as 60.4-60.5. Talked to the design engineer about it (he was a bit surprised) and he confirmed the higher frequency of the Powerwall so that it syncs with grid within a minute and then reconnects.

bp | January 30, 2020

We have a new 15.4KW solar/4 PW system that was activated about a month ago.

Yesterday I ran a full charge/off-grid test to verify the system worked correctly - and it didn't.

The PowerWalls charged to 99%. I flipped the breaker inside the Tesla Backup Gateway to take the system off-grid. According to the Tesla app, the PowerWalls began supplying power to the house, and the solar panels (because the PWs were full) stopped supplying power.

So everything seemed to be working.

Then I went inside the house. All of the UPS systems were issuing alarms. A couple shut off completely and refused to come back online. I was able to get one of our 3 air conditioners to run. Another unit failed to start. And I discovered later that a breaker was thrown in one of our bedrooms with exercise equipment plugged in.

While I wasn't able to measure the frequency, I suspect it was way too high - and that caused all of the problems.

I then flipped the grid breaker to put the system back on the grid - and that didn't change anything - the system stayed off grid running from battery power. I tried turning on the ovens to increase the battery draw, which I hoped would bring the charge level down enough to get back to 60Hz - and the lights in the house flashed when both ovens were turned on, and they didn't seem to be operating correctly.

I tried restarting the Backup Gateway, and that didn't change anything. After a few minutes (I suspect there is a 5 minute timer on grid/off grid switches), the system finally shifted back to grid power. I was then able to get almost everything working, with the exception of 1 UPS that was now showing a fault and refused to turn on.

The system had been in BACKUP POWER mode, so the PowerWalls were not being used - keeping the power level at 98%. I tried multiple changes of reserve power % and changing between all of the modes in the Tesla app - and the Backup Gateway continued to refuse to pull power from the PowerWalls.

I then called Tesla's Energy support group - and after waiting on extended hold, in the middle of the call, the system started to pull power from the PowerWalls.

Tesla verified our system was set to use high power frequencies to shut off the solar power when off grid and the PowerWalls were full. The tech escalated the issue to their "tier 2" support, and I should hear back in a few days. Our Chilicon Power microinverters only need 60.5 Hz to shutdown - which should avoid the problems we saw during the above test.

They also pushed the 1.44.2 firmware update to our Backup Gateway - which will hopefully correct the problem we saw with the switch from backup to self powered modes.

For anyone who hasn't run this test, I strongly recommend fully charging the PowerWalls and then going off-grid to verify you don't have this problem - and then get surprised when you really need the system to work when the grid is down.

Patrick | January 30, 2020

Bp - thanks for posting your experience. Good advice to check everything carefully with the grid disconnected.

Not sure if it would’ve made any difference but I’m wondering if simulating the grid failure by turning off the main breaker in your meter base rather than the TEG breaker might be a better choice. Not sure what else is connected to the breaker inside the gateway...

We conducted grid outage load testing for about an hour (flipped off main breakers in meter base) using our three Powerwalls and didn’t notice any issues with our two HVACs. That said, we don’t have any small UPS units in the home and our home consumption was only running about 60-70% of full expected load.

So far I assume the frequency shifting is working fine in our case, at least in terms of throttling back the inverter output. Once the Powerwalls are fully charged the inverter output is throttled up/down to match the total consumption of the home - have seen this during every sunny day so far.

Will do a series of longer grid failure tests next month and will report back if we see any issues. We’re still running 1.43.3

gregbrew | January 30, 2020

I don't know that the inverters actually "throttle" up or down. On my string system, with no grid, the inverter is simply shut off at 100% PW charge, and turned back on when the PWs are at some predetermined state less than 100%.

Patrick | January 30, 2020

gregbrew - FWIW once the PWs are fully charged I can watch the inverter output vary in real-time (on both the TEG portal and mobile app) to match the actual home load. I've seen it vary from as low as 100W to as high as 6-7kW. Just sharing what we've seen so far...

As stated earlier, we have a zero-export configuration with the latest 10kW SolarEdge hardware and software. Since no energy can be sent back to the grid it make sense that our inverter would need to do this...

gregbrew | January 30, 2020

If the SolarEdge hardware contains micro-inverters, it would make sense that the panels can be shut down individually. That's a really cool and useful feature.

Mine is set up as two strings, 10 x 265W on one, and 9 x 265W on the other. My single ABB inverter box actually contains two 2.1kW MPPT circuits that report separately on the front panel, but as a single value to the Tesla app. Too bad it's not set up to shut down one when the house load can use it, and put the other one on standby.