Solar + Powerwall: Lights have started flickering when off-grid

[Yonki]

@Yonki, if it is any consolation, I understood right away what you meant when you said you needed to get an isolation transformer. I faced the same problem when diagnosing a Sony TV from 1980s where the internal ground was connected to the hot side of AC power! For anyone unfamiliar with using an oscilloscope, if I had attempted to connect the scope's ground clip to a ground point in the TV circuit in the usual manner without plugging the TV or the oscilloscope into an isolation transformer it would have immediately shorted out the 120V power at the full current capacity of the circuit, causing a big spark and probably burning up part of the TV circuit or the 'scope or probe.

Ha! I learned the same (hard) way. As a teen, I had a part time job fixing computer terminals and printers (this would have been around 1980). I was pretty comfortable using a scope, until I had to troubleshoot a bad power supply on a Silent 700. Turns out this was my first experience with a switching power supply, and I didn't understand that the high-voltage section wasn't isolated from the mains. Tried to connect the ground lead to what seemed like a reasonable place on the power supply and BAM - scared the crap out of me, rendered the PSU fubar, and taught me a lesson I'd never forget!

Thanks for sharing!

 

[BGbreeder]

I have blown more than a few things up over the years, too. Fewer as I have learned, or gotten older. Usually due to a failed part that was injecting line voltage in somewhere that I wasn't expecting it. My personal version of the "measure twice, cut once", is "check the voltage twice, then do whatever you were going to do..." Way to much experience with people doing "what the heck?!" type things with high power items. I just had one recently where there turned out to be a hidden (as in buried, as in not to code...) splice box that changed wire colors in a circuit.

All the best,

BG

 

[oholladay]

Yes! Both still about 118.5 Vrms, but one certainly looks less noisy than the other:

Flickering:​

View attachment 711038

Not Flickering:​

View attachment 711039

Waveform is the same, but there's a lot less...crap? Jitter? Less frequency variation, possibly less amplitude variation - not sure why it seems less "thick" - I'm pretty sure scope is set to the exact same settings.

Notably the flickering one is bouncing around between 59.2 and 59.9 Hz, while the clean one is between 59.9 and 60.2 Hz, so when running off the "good" Powerwall/inverter, maybe my clocks won't be running so slow (we've been losing 8 or 10 minutes a day).

I have occasionally noticed flickering on mine, but when synced with the grid and when on off-grid mode. One time, it started flickering on grid and the only way I was able to get it to stop was to switch to off-grid.

Regarding frequency, I have noticed that, when my charge is below ~75%, it operates at around 59.6Hz. It starts to creep up closer to 60 once it gets above 80% charge. I don't have an oscilloscope, so I rely on the API data to tell me frequency.

This post has a good graph showing frequency throughout the day. It's pretty clear when he switches to "off-grid mode." Even when the frequency is relatively constant, it is still much choppier than when tied to the grid.

 

[Yonki]

I discovered the cause of the problem, and it probably has nothing to do with Powerwalls (though it might still have something to do with Tesla's installation).

I do a lot of 3D-printing, and the temperature of the printbed is controlled by turning a 600W heater on and off many times a second (this is called PWM, or Pulse-Width Modulation). For some reason the stress of turning a 5A load on and off many times a second, causes some LED lights to blink (all incandescent lights and many LED lights are not affected, but some are really sensitive). In my old house, only LED lights on the same AC circuit (circuit breaker) were affected. And prior to getting Tesla solar installed, I never saw it in my new/current house.

However now when I turn the printer on, I get it everywhere throughout the house. I'm not sure if Powerwall inverters struggle with high frequency loads (600W being switched on and off around 30 times a second - it's a bit unusual), or if running the whole house through a 10 meter section of AWG 3 wire (instead of the AWG 0/2 that it was originally wired with) is passing enough of an IR drop to cause voltage fluctuations throughout the entire house.

Whatever the case, I'm going to find a way to condition the AC going to/from the printer so this noise gets squashed at the source. So as far as Tesla is concerned, I'm letting them off the hook (though it still sorta sucks that Tesla turned my 200A service into a 100A service with AWG 3 wiring).

Appreciate all the suggestions/support received!

Cheers,
Fred

 

[BGbreeder]

I would look into a "noise/EMF suppressor" They commonly come in 15 and 20 amp 120V versions. I am guessing that your 3D printer didn't come with an FCC class B rating... You could even get an isolation transformer between the plug and the wall, if the EMF suppressor isn't enough, and then add another suppressor on the wall side of that transformer.

I would also double check your grounds, and ground paths.

All the best,

BG

 

[Yonki]

I would look into a "noise/EMF suppressor" They commonly come in 15 and 20 amp 120V versions. ... You could even get an isolation transformer between the plug and the wall, if the EMF suppressor isn't enough, and then add another suppressor on the wall side of that transformer.

Unfortunately EMI suppressors are aimed for high-frequency transients that are fairly easy to suppress with common-mode chokes and LC filters. While this problem is technically EMI, the frequency of the interference (varying from 1Hz to 60Hz) is both very low and in the same band as the "signal" (60Hz), so a typical EMI suppressor would be impractical if not impossible. I'm looking at rectifying the AC to DC, where I can much more easily filter the PWM'ed waveform with a (large) inductor and filter capacitor.

I am guessing that your 3D printer didn't come with an FCC class B rating...

Well it's home-made, so no...but I don't think it violates any regulations. In fact when I'm hooked up to grid power there's no issue at all. The load looks like you're turning your space heater on and off several times a second - a little unusual, but not an issue for a large distributed electrical grid that expects to see thousands of heaters turning on and off in thousands of homes every second.

But switching a 5A load on and off many times a second really messes with the Powerwall's inverters. Here's what the AC looks like right at my printer's power input:

The voltage is dropping by 15% (it's totally stable when I switch to grid voltage). I think my earlier scope photos looked cleaner because I'd turned off my printer when I took them (everything's on the same workbench) - so they weren't actually capturing the waveform that was making my lights flicker.

So ultimately, the problem is caused by an interaction between my printer and my Powerwalls. Ideally Powerwalls wouldn't respond this way to step load changes, but it's pretty challenging designing a small box that can perfectly emulate a gigawatt distributed grid all by itself. I'm sure this wouldn't be a problem with a Tesla Powerpack, though - maybe I should ask for a quote.

Thanks again for the suggestions.

Fred

 

[jboy210]

I discovered the cause of the problem, and it probably has nothing to do with Powerwalls (though it might still have something to do with Tesla's installation).

I do a lot of 3D-printing, and the temperature of the printbed is controlled by turning a 600W heater on and off many times a second (this is called PWM, or Pulse-Width Modulation). For some reason the stress of turning a 5A load on and off many times a second, causes some LED lights to blink (all incandescent lights and many LED lights are not affected, but some are really sensitive). In my old house, only LED lights on the same AC circuit (circuit breaker) were affected. And prior to getting Tesla solar installed, I never saw it in my new/current house.

However now when I turn the printer on, I get it everywhere throughout the house. I'm not sure if Powerwall inverters struggle with high frequency loads (600W being switched on and off around 30 times a second - it's a bit unusual), or if running the whole house through a 10 meter section of AWG 3 wire (instead of the AWG 0/2 that it was originally wired with) is passing enough of an IR drop to cause voltage fluctuations throughout the entire house.

Whatever the case, I'm going to find a way to condition the AC going to/from the printer so this noise gets squashed at the source. So as far as Tesla is concerned, I'm letting them off the hook (though it still sorta sucks that Tesla turned my 200A service into a 100A service with AWG 3 wiring).

Appreciate all the suggestions/support received!

Cheers,
Fred

Have you tried connecting a true sine wave inverter UPS to the printer? That might buffer the device from the rest of the house and visa verse.

 

[BGbreeder]

@Yonki, makes sense to me. The oscilloscope videos were really helpful. It seems like you get a cycle and a half of lower voltage and then it resets. That flicker would drive me nuts.

I would tend to agree with @jboy210 that a pure sine wave inverter should help.

I am sure that you have thought of these things, but could you make your printer be 240 to see if that helps? Can you redo your pwm heater algorithm to be higher frequency, or put a large capacitor across it to even things out a bit? Or are you trying to heat just in time? I suspect that a 240V to 120V isolation transformer might also work to draw the load more evenly from both phases. The load is so small that it must be nudging some hysteresis in the Powerwalls circuits.

All the best,

BG

 

[oholladay]

Since you mentioned the PWM issue, I decided to do more testing and noticed today that the flickering really picked up on LEDs connected to a dimmer switch (which use PWM themselves) when my washing machine motor was at its full spin speed (so maybe not drawing 5A, but perhaps at least 3A). It is driven by an LG inverter motor, which apparently uses PWM to drive the motor. I wonder if it's the combination of these two PWM devices that really makes the flickering noticeable.

In the past, I've noticed flickering (even when on mains) when my laser printer is printing. I wonder if the fuser uses PWM.

By the way, I like seeing the oscilloscope output. Very interesting.

 

[Yonki]

Have you tried connecting a true sine wave inverter UPS to the printer? That might buffer the device from the rest of the house and visa verse.

Haven’t tried it, but that is probably the best (only?) off-the-shelf solution. Unfortunately a double-conversion (output always comes from inverter connected to DC storage) UPS looks like it’s a minimum of $700 to $1,000 new. Going to try my low-cost rectify, filter, and run bed off DC solution first.

 

[Yonki]

I am sure that you have thought of these things, but could you make your printer be 240 to see if that helps? Can you redo your pwm heater algorithm to be higher frequency, or put a large capacitor across it to even things out a bit? Or are you trying to heat just in time? I suspect that a 240V to 120V isolation transformer might also work to draw the load more evenly from both phases. The load is so small that it must be nudging some hysteresis in the Powerwalls circuits.

That’s what I’m thinking, too. Since the switching can happen as frequently as every cycle, I think the Powerwall might be reacting to the load change on the next AC cycle, only to have the load then switch in the other direction, so now the Powerwall reacts to that, etc. The frequency of the load changing is too close to the frequency of the inverter’s control loop.

Using 240V to balance the load is a good idea, but the existing heater is 24ohms, so now the currents would be twice as large. I could buy a 240W bed heater (which would be a 96 ohm load, so the current would be halved as well as balanced across 240V), but that’s not the first thing I want to try because it’d be a PITA for several reasons (there’s no easily accessible 240V nearby, I have 2 printers, and changing the bed heater is a PITA in itself).

PWM at a higher frequency would be a great solution - the Powerwall would see basically the same load every cycle. However the SSR (solid state relay) I’m using to switch the heater on and off switches at the zero crossings of the AC only. So I’d have to go with some other switching device that allows switching at any voltage, and there‘d be more serious (traditional, but still relatively low frequency so more expensive/bulky to suppress) EMI effects as well to contend with.

So yeah…a double-conversion UPS keeps looking like the best solution. And a UPS is a good idea for a 3D printer anyway - while I was doing various Powerwall testing, I lost power to the house for a second or two, ruining a print that had been running for a few hours. Some prints can take days. Since Powerwalls/Gateways don’t always switch over perfectly smoothly after a grid issue, it’s good to have a UPS.

Cheers,
Fred

 

[Yonki]

Since you mentioned the PWM issue, I decided to do more testing and noticed today that the flickering really picked up on LEDs connected to a dimmer switch (which use PWM themselves) when my washing machine motor was at its full spin speed (so maybe not drawing 5A, but perhaps at least 3A). It is driven by an LG inverter motor, which apparently uses PWM to drive the motor. I wonder if it's the combination of these two PWM devices that really makes the flickering noticeable.

Makes sense. You just reminded me of another (semi-practical if very big) solution - have the AC from the Powerwall drive an electric motor with a flywheel and another motor (functioning as an AC generator) on the shaft. The output of that generator can drive the switching load and the flywheel smooths out the load the Powerwall sees. Maybe not practical but an interesting way to visualize what needs to be done (even the load over time).

In the past, I've noticed flickering (even when on mains) when my laser printer is printing. I wonder if the fuser uses PWM.

Probably! It‘s such a great, efficient way to control voltages and power.

 

[manole]

I only use the my 2 Powerwalls for back up since I got them this Spring. I decided to test them out today for more than a few minutes and I’m getting flickering on LEDs with Jasco dimmers as well. I also noticed my bathroom exhaust fan sounds noticeably less powerful and my electric oven produces a higher pitched whine than normally. I’m sure the more I look for anomalies the more I’ll find. I’m guessing one or both the powerwall inverters are not producing as clean/steady power as Tesla is stating. Anyone have any luck debugging this further with Tesla support? My fear with submitting a ticket is that they’ll send multiple rounds of below average electricians that have to clue how to debug this (I’ve had different problems in the past but same outcome with Tesla support).

 

[Yonki]

I only use the my 2 Powerwalls for back up since I got them this Spring. I decided to test them out today for more than a few minutes and I’m getting flickering on LEDs with Jasco dimmers as well. I also noticed my bathroom exhaust fan sounds noticeably less powerful and my electric oven produces a higher pitched whine than normally. I’m sure the more I look for anomalies the more I’ll find. I’m guessing one or both the powerwall inverters are not producing as clean/steady power as Tesla is stating. Anyone have any luck debugging this further with Tesla support? My fear with submitting a ticket is that they’ll send multiple rounds of below average electricians that have to clue how to debug this (I’ve had different problems in the past but same outcome with Tesla support).

My advice would first be to do a controlled (more controlled than my first experiment, where I came to a wrong conclusion) experiment looking at the lights, fan, and oven under 3 (or maybe 4) conditions:

1. Connected to grid power
2. Disconnected from grid power with only Powerwall A on (Powerwall B switched off with the switch on the side)
3. Disconnected from grid power with only Powerwall B on (Powerwall A switched off with the switch on the side)
4. Disconnected from grid power with Powerwall A and B on (optional because I doubt this will revel anything 2 or 3 doesn’t)

If 2 and 3 give you different results, you probably have a faulty Powerwall. If they give you the same result, and it’s different from 1, then this is just what Powerwalls do. I’d still bring it up to Tesla, maybe they have a solution. But I don’t see how a good Powerwall would affect your oven and bathroom fan. LED dimmers are notoriously finicky with some LED types with grid power - do you see flicker with your dimmers set to max brightness?

The procedure I found reliable for turning off Powerwalls when disconnected from grid power is simply flipping the switch on the side - you should hear a click, then that Powerwall is off. You can then turn it back on, wait for the click(s), then turn the other one off without losing power to the house. I can do that with no problem on my two good Powerwalls (and I finally have an appt for a Tesla guy to come out and fix my bad one).

Let us know what you find!

 

[manole]

Thank you for the detailed response.

The plot thickens. I tried again just now with both powerwalls on to set a fresh baseline and no anomalies for the last 30 min since I turned off the grid!? The only difference is now solar inverters are contributing a little bit of energy (it’s very cloudy and solar is only producing around 300 watts). So I turned off the solar breakers to eliminate that variable and still no anomalies. Perhaps the Powerwalls produce dirty energy when they’re closer to 100% when I first started this experiment and after depleting a bit they are cleaner. Another theory is that they simply need time to warm up… I don’t know.

I’ll keep experimenting and monitoring and keep this thread up to date in case anyone else has similar experiences.

Also, to answer the question of dimmer level, when I was having this issue earlier today, the flickering went away when the dimmer was at 100%.

 

[KyleDay]

Thank you for the detailed response.

The plot thickens. I tried again just now with both powerwalls on to set a fresh baseline and no anomalies for the last 30 min since I turned off the grid!? The only difference is now solar inverters are contributing a little bit of energy (it’s very cloudy and solar is only producing around 300 watts). So I turned off the solar breakers to eliminate that variable and still no anomalies. Perhaps the Powerwalls produce dirty energy when they’re closer to 100% when I first started this experiment and after depleting a bit they are cleaner. Another theory is that they simply need time to warm up… I don’t know.

I’ll keep experimenting and monitoring and keep this thread up to date in case anyone else has similar experiences.

Also, to answer the question of dimmer level, when I was having this issue earlier today, the flickering went away when the dimmer was at 100%.

I’ve also had your similar experience when I experienced an unexpected grid outage. My lights were humming and my garage door motors operated at higher/faster frequencies than normal. Everything was “weird” for a while until power came back. But since then, nothing weird. I’ve measured voltage and had a few outages and nothing strange has happened since. I wonder if it was a bug that got fixed.

 

[zƬesla]

I've been having fun with LED bulbs as well. Replaced all incad. bulbs with LEDs in a chandelier recently only to find considerable flicker. Setting the dimmer to 100% helped but not entirely. Replaced the dimmer switch which appeared to fix the issue, however occasional power fluctuations (w/PWs) cause a slight flicker at onset that so far I've been able to live with. Also replaced the garage opener bulbs with LEDs and found that we could open but no longer close the door remotely from the car. Turns out the LED bulbs were causing interference when they were on, so the CFL went back in. Hard to justify the $10 for a shielded LED garage bulb for how little they're on.

 

[oholladay]

Thank you for the detailed response.

The plot thickens. I tried again just now with both powerwalls on to set a fresh baseline and no anomalies for the last 30 min since I turned off the grid!? The only difference is now solar inverters are contributing a little bit of energy (it’s very cloudy and solar is only producing around 300 watts). So I turned off the solar breakers to eliminate that variable and still no anomalies. Perhaps the Powerwalls produce dirty energy when they’re closer to 100% when I first started this experiment and after depleting a bit they are cleaner. Another theory is that they simply need time to warm up… I don’t know.

I’ll keep experimenting and monitoring and keep this thread up to date in case anyone else has similar experiences.

Also, to answer the question of dimmer level, when I was having this issue earlier today, the flickering went away when the dimmer was at 100%.

My situation below is slightly different, but there could be some relationship. The other day, I had a grid outage while I was exporting (so Powerwalls were full). Immediately, the Powerwalls starting outputting 66Hz (as determined from the API) to force the solar inverters offline. My wife noticed a very strange hum from the microwave and many other things sounded different. It eventually came down to 63Hz and then, once the Powerwalls could charge, back to 60Hz (causing the solar inverters to come back online). I’ve also noticed that when the Powerwalls are below ~80%, the frequency they output when off grid is around 59.6Hz.

It is true that the frequency variation when the Powerwalls are off grid is much greater than grid power. There is a good graph of this posted in another thread. However, the sine waves I’ve seen from oscilloscope photos measuring off grid Powerwalls look fairly clean.

 

[gpez]

My situation below is slightly different, but there could be some relationship. The other day, I had a grid outage while I was exporting (so Powerwalls were full). Immediately, the Powerwalls starting outputting 66Hz (as determined from the API) to force the solar inverters offline. My wife noticed a very strange hum from the microwave and many other things sounded different. It eventually came down to 63Hz and then, once the Powerwalls could charge, back to 60Hz (causing the solar inverters to come back online). I’ve also noticed that when the Powerwalls are below ~80%, the frequency they output when off grid is around 59.6Hz.

It is true that the frequency variation when the Powerwalls are off grid is much greater than grid power. There is a good graph of this posted in another thread. However, the sine waves I’ve seen from oscilloscope photos measuring off grid Powerwalls look fairly clean.

Call Tesla to get that max frequency dropped. 66hz is not pleasant for a lot of electronics, motors included. @oholladay @manole @KyleDay My grid outage frequency issue is resolved!

 

[RKCRLR]

Some of my LED lights flicker when on backup and the Powerwalls are close to fully charged due to the higher frequency of the Gateway commanding the solar off. The lights don't flicker once the SoC drops to around 90%.

What I don't understand is why some of my digital clocks run fast during backup. Initially I chalked it up to the frequency shift due to solar generation control. But lately I've had some outages where the solar wasn't producing much and the Powerwalls weren't fully charged. Some of my digital clocks still ran fast.