Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Powerwall 2 + UPS Connundrum - and solution

This site may earn commission on affiliate links.
The Tesla Powerwall 2 (PW2) is an excellent protection against grid power outages, yet its design does require an additional Uninterruptible Power Supply (UPS) for any devices that are sensitive to even extremely brief power breaks.

While your fridge, microwave, most lights and other devices will likely not notice the 300-2000 millisecond relay switch-over (I'll admit I really don't know the typical Gateway relay cycle-over time as I've never been able to find Tesla specs on it) in case of a power outage, your sensitive electronics certainly will. In fact, only a few Hertz cycles loss will cause anything CPU related to fail.

Thus the need for at least a -short term- UPS to get sensitive electronic devices 'over the hump' of the power outage to final relay 'snap' into the internal house power grid and PW2 power feed. So far, so good. Almost any UPS can fill that bill and, in fact, since it is only needed for a second or two, the battery power for most are severe overkill for the task they perform while supporting the PW2.

However, the PW2 has a specific issue with most UPSs in a specific setting. I'll explain. If the PW2 is at 100% charge while the sun is shining and the user ALSO has solar panels, the PW2 will attempt to 'turn off' the inverters for those solar panels once inside the internal house power grid (which the relay broke from the grid operator no longer providing power). In order to do this, the PW2 will raise the Hertz cycle (in the US) from 60 to 65.

Now, I believe most electronic devices can handle a range of Hertz cycles (I actually have a call into Tesla PW2 support to confirm electronics working at 65 Hertz), including computers, routers, modems and other devices. Thus, they should work just fine while the PW2 is inhibiting the solar inverters from producing power by going to 65 Hz.

The problem is that MOST UPSs will only see power coming from the outlet as 'good' when it is 60 Hertz plus or minus THREE Hertz (thus 57-63 Hertz). Thus, when the PW2 is 100% charged and in a Backup situation with no grid input and solar panels that need to be inhibited, the PW2 will raise the frequency to 65 Hertz to inhibit the solar inverters - this power will then be considered 'bad' power to those UPSs. The UPSs will then simply run on their own battery power until they run out and the devices behind them will die at that time. All because they see power at 65 Hertz as unacceptable.

In an outage event that occurred with my PW2 earlier this week, while my PW2 was 100% charged and it was actively inhibiting the solar inverters by raising the frequency to 65 Hertz, my UPSs only lasted for 5-15 minutes (depending on the load) at which time they failed. Howevver, the PW2 inhibited the solar inverters for 30 minutes while discharging into the house. Since all of the computers / routers / switches / NASs / etc. (which consumed most of the power at that time) were behind UPS's, they didn't use the offered PW2 power (now at 65 Hertz and incompatible for pass-thru by the UPSs) and simply used up the batteries on the UPSs.

Thus the PW2 discharge was just for fridges and other phantom loads that slowly got it to 97% in a half hour. At that time, the PW2 decided it was OK to allow the solar inverters to create power and lowered the frequency back to 60 Hertz. The UPSs reacted by immediately charging their now-dead batteries, but the devices behind them were long dead.

BUT a solution was found that is as inexpensive as the APC and Trip-Lite UPSs I was using. It turns out that Minuteman UPSs can handle an input of 60 Hertz plus / minus 6 Hertz. Thus, it will see the 65 Hertz that the PW2 creates during those specific timeframes as acceptable pass-thru power. Thus the Minuteman should provide the conditioning to get over the short relay switch-over loss of power AND allow the PW2 65 Hertz power to then pass-thru the unit on to the devices, which will then enjoy the huge (comparative to that UPS at least) battery backup capability of the PW2.

The only testing I must do for all my electronics is if they will function with the 65 Hertz power and also function across frequency changes from 60 Hertz to 65 Hertz and back again. If the power supplies for my devices can handle those small changes without causing a crash, I have solved that problem.

Here is the Minuteman UPS I am going to be testing out next week:

https://www.amazon.com/gp/product/B015QOEP0K/ref=oh_aui_detailpage_o00_s00?ie=UTF8&psc=1

Here are the specs for the UPS that show it will see input at 60 Hertz - plus/minus 6 Hertz as acceptable power. Look at the Input Frequency Range:

Minuteman UPS - Entrust LCD Series

Although a grid loss at 100% PW2 power and during the day when solar inverters need to be inhibited by raising frequency to 65 Hertz is not a normal experience, this solution should provide the conditioning to get past the relay switch-over timeframe AND allow devices to use 65 Hertz power from the PW2.

I'll update this post when I know more about electronic devices working in a 65 Hertz environment. Testing will be occurring this weekend and next....
 
Last edited:
That is very interesting. I'll have to check this out with my current UPS (CyberPower 100VA) at a planned power outage coming next month. If it proves true, I'll change power supplies. I have found an even more robust one than the Minuteman. It is made by Eaton. A comparable model to the Minuteman you'll be looking at is here.

https://www.amazon.com/Eaton-Electr..._rd_t=40701&psc=1&refRID=3W4HK060PFFR3RBDQ1BK

It has an input frequency range of 46-70 Hz. It might be worth a look.
 
The larger new APC UPS's go from 47-65 Hz, so that's good, at least for me. Though all these gotchas (generator interfacing, only 200A gateways, no availability and uncertain timeframe for installs, etc...) are starting to make think I should just give up on trying to use powerwalls and just settle for generators for backup, and not bother with trying to make solar work in an outage.

There are plenty of hybrid solutions out there, but the non-PW solutions are all super expensive from a storage POV. If PW's could just be cheap DC batteries that could plug into these other systems it would be so much easier.

mike
 
That is very interesting. I'll have to check this out with my current UPS (CyberPower 100VA) at a planned power outage coming next month. If it proves true, I'll change power supplies. I have found an even more robust one than the Minuteman. It is made by Eaton. A comparable model to the Minuteman you'll be looking at is here.

https://www.amazon.com/Eaton-Electr..._rd_t=40701&psc=1&refRID=3W4HK060PFFR3RBDQ1BK

It has an input frequency range of 46-70 Hz. It might be worth a look.

Jeep, I was contacting many UPS manufacturers yesterday as to the 65 Hertz issue. CyberPower, unfortunately, told me that every UPS they sell only serves as a pass-thru at 60 Hertz plus or minus THREE Hertz. You should test it, but I believe you will find that the Cyberpower UPS you have will NOT see the PW2 power stream as 'good' power while it has incresed the Hertz frequency to prevent the solar inverters from functioning.

In looking at the specs for the Eaton UPS you pointed to, it DOES look like it will work. The acceptable Hertz frequencies are 46-70 Hertz so may be OK. I'm going to try the Minuteman solution but the Eaton looks like it has more range and is somewhat equivalent in price:

Eaton 5S UPS | EatonGuard.com
 
The larger new APC UPS's go from 47-65 Hz, so that's good, at least for me. Though all these gotchas (generator interfacing, only 200A gateways, no availability and uncertain timeframe for installs, etc...) are starting to make think I should just give up on trying to use powerwalls and just settle for generators for backup, and not bother with trying to make solar work in an outage.

There are plenty of hybrid solutions out there, but the non-PW solutions are all super expensive from a storage POV. If PW's could just be cheap DC batteries that could plug into these other systems it would be so much easier.

mike

The REAL issue is that the PW2 Gateway relay does not do 'instantaneous' switch-overs. There is a very short loss of cycles as the relay reacts to the loss of power from the grid and 'snaps' the relay to the other side to get the PW2 power that is at that time requested. That can run from a few milliseconds to 1000 milliseconds (as far as I know).

For MOST appliances, this can be ignored. Your washing machine, fridge, most lights, etc. might stutter for a moment but then continue on like nothing happened. The ONE application that this matters for is regarding CPU related devices. Just a VERY short loss of cycles is enough to fault them. So an interim 'conditioner' is required.

It would have been better if Tesla would have simply added some super-capacitors to its Gateway that would provide full 10-20 kW power at 120 / 240 volts for one to two seconds so that the conditioning could be done internal to the Powerwall 2, but it wasn't created that way. If it had been created that way, the relay 'snap' speed wouldn't matter as the capacitors would simply 'fill in the gap' during the switch-over of a grid outage.

But since it didn't, a UPS is required for CPU based devices. And, allowing the PW2 power to pass through at 65 Hertz is also required for it to function properly during all grid outage situations.

But I feel your pain...
 
If you are home, you can just go shut off the breakers to the solar when your UPS alarms start going off. The real problem is having the system stay up unattended when you're not home. Another option is to try to get your solar inverters configured to a tighter frequency tolerance. If they would shut down at 62Hz, then you wouldn't have any problem with any UPS units. Obviously the best case would be for them to proportionally curtail above 61Hz so that the Powerwall would never push the frequency so high to begin with.
 
If you are home, you can just go shut off the breakers to the solar when your UPS alarms start going off. The real problem is having the system stay up unattended when you're not home. Another option is to try to get your solar inverters configured to a tighter frequency tolerance. If they would shut down at 62Hz, then you wouldn't have any problem with any UPS units. Obviously the best case would be for them to proportionally curtail above 61Hz so that the Powerwall would never push the frequency so high to begin with.

To my understanding, the 65 Hertz frequency is a standard used by all solar inverters to tell them to attenuate power from the panels. I know of no user specifiable setting for the inverters to stop power production at a lower Hertz rating and there is no user specifiable setting in the PW2 to change the frequency value it creates to stop those inverters. I've found that 65 Hertz simply is the standard used across all solar inverters.

Additionally, the idea is NOT to have to flip breaker switches to make the system work at all times. Autonomous functionality is a considered plus here so UPSs that work as a pass-thru for PW2 power even at 65 Hertz are really what is required for any house / office that has computer equipment (or anything with a CPU).

This has been a huge learning curve for me but if you know of ANY solar inverters that are user configurable as to the Hertz setting to attenuate power, please let me know. However, that still leaves the PW2 having to have a setting that changes what it will output as the frequency during those periods. Tesla has let on to no such settings in the extended time I've spoken to them.
 
Some inverters have configurable parameters, but they are only available at the installer level and not intended for end-user access. I see you are also using Enphase. I may contact my installer to see if they can find out if it is possible to make this kind of adjustment. It may not be possible on the M-series but it might be on the IQ-series.
 
If you have a UL1741 compliant inverter, it will shut off at 60.5Hz, and you should never get to 65Hz from the PW. I did a 100% charge test with full solar, and the frequency slowly went up to 61Hz, by which time the inverters had shutdown. I never saw the frequency get above 62Hz and none of my APC or CyberPower UPS sounded an alarm or shutdown. I don't think the PW raises the frequency any higher than necessary to curtail solar production.
 
UL1741 inverters have only been required in California since September 2017. My solar system is already 5 years old.

However, you are correct - the Powerwall will only continue to increase the frequency if the solar continues to output surplus power.
 
If you have a UL1741 compliant inverter, it will shut off at 60.5Hz, and you should never get to 65Hz from the PW. I did a 100% charge test with full solar, and the frequency slowly went up to 61Hz, by which time the inverters had shutdown. I never saw the frequency get above 62Hz and none of my APC or CyberPower UPS sounded an alarm or shutdown. I don't think the PW raises the frequency any higher than necessary to curtail solar production.

I'm beginning to find that the rise to 65 Hertz the PW2 did to inhibit my Enphase M215 inverters either showed an error in the PW2 I own OR that my M215 inverters have a defect. The Enphase M215 inverters are only supposed to see a narrow range of 57-61 Hz - or up to 62.5 Hz if extended:

https://www.enphase.com/sites/default/files/M215_DS_EN_60Hz.pdf

The fact that my PW2 had to raise the frequency to 65 Hertz is likely identifying an error either at the inverter or the PW2 level. What a long winding road I've been on regarding this.
 
Some inverters have configurable parameters, but they are only available at the installer level and not intended for end-user access. I see you are also using Enphase. I may contact my installer to see if they can find out if it is possible to make this kind of adjustment. It may not be possible on the M-series but it might be on the IQ-series.

This may be an error on either the PW2 or the M215 inverters I have on my panels. The specs for the inverter only require a 61 to 62.5 Hertz level to disconnect:

https://www.enphase.com/sites/default/files/M215_DS_EN_60Hz.pdf

Tesla support is looking into this and my installer will likely be returning to see why the PW2 had to raise the frequency to 65 Hz to disable them...
 
I have a 10.5 kWh system with 2 Powerwall 2s. I also have an APC 1500 with the additional battery pack for my computer equipment (cable modem, wireless router, iMac, 2 canon printers). I also have had several power outages and in all cases the voltage from the solar panels and/or the Powerwalls was passed through the UPS without any problems.
 
  • Helpful
Reactions: NickFie
UPDATE on the frequency rise to 65 Hertz: After the September 23rd grid power outage event when Tesla confirmed my PW2 raised the frequency to 65.4 Hertz, I simulated the exact same event again last Sunday, October 14th with a pull to the Service Disconnect on the Main Breaker Unit.

All parameters were the same: PW2 100% charged, sun shining and solar input at 3.5-4 kW and my house was using about 1.5-2 kW at the time. All appliances remained the same. I had both the new Eaton UPS (that can pass-thru up to 70 Hertz power) and the Minuteman UPS (that can pass-thru up to 66 Hertz power) at two of my locations where critical CPU based hardware was running while the rest continued to use the previous APC and Trip-Lite UPSs.

When the Service Disconnect was pulled, all the UPSs briefly came to life and went to their internal batteries to facilitate the very short loss of 120 volt power when the Gateway relay snaps shut. But within a second or so, they all went back to using the house supplied current (now PW2 based). I was able to document via video in real-time the rise of the Hertz level from 60.0 to the maximum employed.

But THIS time, the frequency only rose to 63 Hertz. The rise took about 2-3 seconds but held steady at 63 Hertz. Thus ALL of my UPSs work with this new Hertz limit as the APC and Trip-Lites function at 63 Hertz (actually, Tesla confirmed 62.9 Hertz).

The ONLY difference between the September event and the October event was the firmware release. At the initial event I had 1.22.3 and I currently have 1.25.0. After speaking with Tesla customer tech support yesterday about this test, it was confirmed that the frequency rise is not some built-in hardware circuit on the PW2 but is actually software controlled. Version 1.25.0 differs from 1.22.3 in many more ways than the support page states - the limit on the frequency rise is one of the updates - at least against my Enphase M215 inverters as I have been able to test.
 
That sounds great for your system. However, I had a different experience on October 16th. We had a power outage for 1 hour and 23 minutes that started just before I left the house in the morning. There was a momentary dip in the lights and the Powerwalls were apparently powering the house just fine. The SOC was about 48%. I checked the Tesla app when I got to the office and found that the solar was not generating any more. It stopped briefly at the cutover and then generated for about 20 minutes before my Enphase inverters gradually all went offline and recorded errors for AC Frequency Out of Range. They did not come back online until the grid power returned.

This is on 1.25.0. I submitted a trouble report to the Powerwall Support e-mail but I haven't heard back yet.
 
After reading this, I decided to test my system. It is also on version 1.25.0

At about 9:35am, I had 3kW of solar production, 0.8kW house load, and the PW was at 92% and charging. I opened the main breaker, and the PW took over the house load. Solar was disrupted for 5 minutes due to the brief outage created by opening the main breaker. After solar production resumed, the frequency stayed around 60.04Hz until the PW reached 98% at 10:20am, where the frequency then rapidly increased to 60.54Hz at which point my inverters shut down.

I continued to let it run, and the frequency dropped to about 60.03Hz in 2 minutes. After 5 minutes, the inverters came on and more charging for about 5 minutes, until the PW increased the frequency again to 60.55Hz which shut the inverters off. At this point, I closed the main breaker to restore utility power.

I never saw the frequency go above 60.55Hz, and that is enough to shutdown my SMA string inverters. I also received a notification on my phone when the utility power was off and again when it was restored.
 
I have a precision frequency meter. It has 5 digits, but I only posted the most significant 4 digits because the last one kept changing. So 00.001 Hz resolution. It measures the frequency every second.

The frequency was close but not exactly the same as that reported through the /api/meters/aggregates data set ("battery" "frequency").
 
I have a Kill-A-Watt meter that reports the line frequency with 3 significant digits. It likely cost less than a precision frequency meter and is useful in general in tracking power usage. My SMA SB7000US also reports the frequency when it is trying to sync.