What I'm learning about DIY Enphase and Hoymiles Microinverters

[Ampster]

I do not understand why, if the Hoymiles are CA Rule 21 compliant, that they could not be AC coupled to a grid forming inverter that can AC couple? I understand the IQ8 with the System Controller 2 can form a microgrid. That is only part of what is necessary for effective AC coupling. What I am not clear about is whether it has the algorithm and can use frequency/Watt to control any CA Rule Compliant GT inverter in an AC coupling situation? Do @ncsolarelectric or @svetz know for sure?

 

[ncsolarelectric]

I do not understand why, if the Hoymiles are CA Rule 21 compliant, that they could not be AC coupled to a grid forming inverter that can AC couple? I understand the IQ8 with the System Controller 2 can form a microgrid. That is only part of what is necessary for effective AC coupling. What I am not clear about is whether it has the algorithm and can use frequency/Watt to control any CA Rule Compliant GT inverter in an AC coupling situation? Do @ncsolarelectric or @svetz know for sure?

It has UL 1741-SA compliance. Period. It can be used in conjunction with the DTU-Pro and the CSS WattNode meter to manually set the Export control to zero or some other limit. The same meter SolarEdge uses. The DTU cloud has the control for export control, power factor, and other DER functions. I did not use those things. Hoymiles has videos of these features on their YT channel.

Update: I misspoke earlier and have corrected my comment.

 

[ncsolarelectric]

Interesting and confusing result just now. I am currently running on the Enphase microgrid and the HM-350NT is working. It has a green LED saying the grid is good and it is putting out power. Why it didn't work previously, I can only assume at this time that because the system was not at full sunlight, with full inverter power, it didn't have enough "stiffness" to work, but now that the system is fully powered, it's working. Interesting.

Hmmm... Could it be that now I have the DTU connected and so it's working?

PS: Testing takes forever when there are 15 minutes between data points recorded.... stay tuned.

 

[fromport]

Hmmm.. Could it be that now I have the DTU connected and so it's working?

There is no logical explanation behind that.
Eg: if an IQ7/8 (both 1741-SA capable) is hooked up to any micro grid, it would work without any connection to an envoy.
It either detects a grid or not (either the real grid or the micro grid)
I will taper down as soon as the frequency start to shift.

Simple check: unplug the DTU and see if the LED on the HM stays green ;-)

 

[svetz]

What I am not clear about is whether it has the algorithm and can use frequency/Watt to control any CA Rule Compliant GT inverter in an AC coupling situation? Do @ncsolarelectric or @svetz know for sure?

As long as it is recently UL compliant (>= 2018?) then any inverter will have the feature to frequency shift to control power.

Interesting and confusing result just now. I am currently running on the Enphase microgrid and the HM-350NT is working.

That's more like it. Don't suppose you got the frequency of the array or the amps off the HM? I'm still thinking it's one of the scenarios in #103.

The only reason for the Hoymiles not to work is if there's some control specifically telling it not to (the only control for the HM is if the microgrid is unstable and anti-islanding kicks in or possibly frequency throttling power).

I've seen my array micros off when the battery IQ8s have a stable micro-grid, so know the Enpower or Envoy can turn off the array micros if desired, most likely that's why you didn't see it producing.

 

[Ampster]

It has UL 1741-SA compliance when used conjunction with the DTU-Pro and the CSS WattNode meter.

I presume you are talking about the Hoymiles? If I understand it then, the Hoymiles needs the DTU-Pro and the CSS WattNode meter in order to be UL 1741-SA compliant, correct? I have the same question which @fromport has raised. Are the Hoymiles natively UL 1741 SA compliant like the IQ7/8s.

 

[ncsolarelectric]

I presume you are talking about the Hoymiles? If I understand it then, the Hoymiles needs to DTU-Pro and the CSS WattNode meter in order to be UL 1741-SA compliant, correct? I have the same question which @fromport has raised. Are the Hoymiles natively UL 1741 SA compliant like the IQ7/8s.

I misspoke earlier. I only know that the DTU-Pro and the WattNode meter are required to set the manual Export control limits. The COFC does not say they are required to be UL 1741 compliant. The label, the datasheet, and the COFC all say they are UL1741-SA compliant. See attached.

Also, I did disconnect the DTU and it kept right on working, now it is overcast and it is still working. I'm stumped.

 

[Ampster]

As long as it is recently UL compliant (>= 2018?) then any inverter will have the feature to frequency shift to control power.

I am still trying to understand these two aspects of AC coupling. I understand the UL 1741SA spec defines how a GT inverter responds to frequency/Watt signals. The unknown factor for me is what is the standard governing the grid forming inverter and its ability to measure current and change frequency in response, so that the GT inverter will modulate. You used the term, "control" and that defines is for me. The simple "response" of the GT inverter is what I think is defined in 1741, but the ability to "control" a GT is as far as I can tell not defined in 1741.
My assumption is that UL 1741SA compliance does not guaranty that a grid forming inverter has the algorithm to AC couple with a GT inverter.

 

[ncsolarelectric]

Test aborted due to cloud cover and thunderstorms. I need consistent irradiance to know it's the inverter throttling back and it's not a cloud overhead.

Update: The S-Miles Cloud recorded an alarm. The inverter saw a grid underfrequency at 4:00pm, at about the time I turned on the PV breaker to reconnect to the grid. The event lasted 5 minutes and then it reconnected. I think perhaps the off-grid Enphase inverters are phase-locked to the grid via the System Controller 2, which can sense the other side of the transfer switch. No big deal.

 

[ncsolarelectric]

I am still trying to understand these two aspects of AC coupling. I understand the UL 1741SA spec defines how a GT inverter responds to frequency/Watt signals. The unknown factor for me is what is the standard governing the grid forming inverter and its ability to measure current and change frequency in response, so that the GT inverter will modulate. You used the term, "control" and that defines is for me. The simple "response" of the GT inverter is what I think is defined in 1741, but the ability to "control" a GT is as far as I can tell not defined in 1741.
My assumption is that UL 1741SA compliance does not guaranty that a grid forming inverter has the algorithm to AC couple with a GT inverter.

I've been watching videos on both SA and 1547. The standard IEEE-1547 (2018)+ lays out the requirements, the end results that they want for DERs to connect to the grid. It does not tell you "how" to do it or provide any kind of algorithm. 1741-SA covers some of those requirements and provides guidelines for safety, but again, it doesn't tell you "how" to do it, just what your design needs to comply with. The how-to is up to the inventor.
You can design it however you like, but it must pass the tests required by UL-1741-SA.

 

[fromport]

I've been watching videos on both SA and 1547. The standard IEEE-1547 (2018)+ lays out the requirements, the end results that they want for DERs to connect to the grid. It does not tell you "how" to do it or provide any kind of algorithm. 1741-SA covers some of those requirements and provides guidelines for safety, but again, it doesn't tell you "how" to do it, just what your design needs to comply with. The how-to is up to the inventor.
You can design it however you like, but it must pass the tests required by UL-1741-SA.

I found this in the SMA documentation:

So higher than 62Hz and lower than 57Hz it should shut off in 160 milli seconds.
Rule 21 (california) starts power limiting at between low 58.5Hz en high 60.5Hz



More interesting info in that document:

Source:

https://ressupply.com/documents/sma/SB_3.0-7.7-US-41_Grid_Support_Details.pdf

 

[ncsolarelectric]

I found this in the SMA documentation:

So higher than 62Hz and lower than 57Hz it should shut off in 160 milli seconds.
Rule 21 (california) starts power limiting at between low 58.5Hz en high 60.5Hz

More interesting info in that document:

Source:

https://ressupply.com/documents/sma/SB_3.0-7.7-US-41_Grid_Support_Details.pdf

Happy I retired from product development.

 

[Herb]

It looks like you missed an opportunity to AC couple your Enphase to your hybrid inverter. The Enphase was never intended to be just a GT system which by its nature will not produce power unless it is AC coupled to an hybrid inverter

Well I checked the price to get the sunlight backup upgrade but man it was out of my range, and if you add batteries is like buying the system all over again. Since the last post I have added a Growatt all in one and another battery bank, off grid of course, and it will run the house for at least 3 days 24/7, again not a huge system but I can charge a bank while using the other and just rotate them to keep going. I can disconnect an array of panels from the Enphase and bypass the micro inverters straight to the charge controller and run the 2 banks and that will do for me and at a fraction of the cost. Cheers!

 

[ncsolarelectric]

That's more like it. Don't suppose you got the frequency of the array or the amps off the HM? I'm still thinking it's one of the scenarios in #103.

Frequency varies at 15-minute intervals from 59.96 to 60.03 for the majority of the day.
The HM put out about 30W more than the IQ8+, but you can see it is limited by irradiance on the panel.
The dip in the Grid Voltage starting around 14:30 hours is where I switched to the Enphase microgrid and back to grid at 16:00. That is the longest my system has run off-grid, mostly from sunlight.

 

[Ampster]

I can disconnect an array of panels from the Enphase and bypass the micro inverters straight to the charge controller and run the 2 banks and that will do for me and at a fraction of the cost.

I presume what you are saying is the Growatt cannot AC couple. If it could, you could use the panels without having to disconnect the micros.

 

[svetz]

I found this in the SMA documentation:

Thanks for posting that! I knew they were talking about increasing the UF1 so utilities could request DER/ESS to output more power, but wasn't aware it was in the spec. Should have known since some states are already offering ESS incentives. Not Florida though, suspect we'll be waiting for the national incentive.

Well I checked the price to get the sunlight backup upgrade but man it was out of my range, and if you add batteries is like buying the system all over again....

You might want to bookmark this thread, earlier ncsolarelectric said Hoymiles is coming out with an AC battery solution to go with their microinverters. It would be sweet to get all of the benefits of microinverters and AC batteries at a low price.

Frequency varies at 15-minute intervals from 59.96 to 60.03 for the majority of the day.

Unfortunately, your cursor on the graph is pointing at the least interesting points; had it been over a non-nighttime low power output (e.g., 15:15 when it's 50% throttled and 16:00 when it's 80% throttled) and shown the frequency we'd know if we could eliminate #1 from #103.

...The dip in the Grid Voltage starting around 14:30 hours is where I switched to the Enphase microgrid and back to grid at 16:00... Looks like the bit to the left is the interesting part of the curve then. So, at 14:30 when the microgrid first formed the output throttled back to about ~150W, that could have been the battery charging or something?? So, what... batteries filled up and then the HM throttled down to ~20W output? As the "curve" isn't a discrete value (it's the average over the reporting time), you need the values at each increment to make sense of it. What would be helpful is to get the frequency where it looks throttled and the IQ8 output at the time of the various Hoymiles output. You can get that from the field lastReportWatts in this api link and use an epoch converter to convert lastReportDate to a human time (it only updates in 5 min intervals and is the average over the last 5 minutes).

 

[ncsolarelectric]

I'm gathering a ton of data, but it's inconclusive due to inconsistent cloud cover. I can't tell if it's throttling back or a cloud is going over. It ran on Microgrid from about 9:45AM to 2:30PM, with a couple of transfers back to the grid to test the phase-lock.

When running with about 600W load, the HM-350NT put out a peak of 356W. That's 57W higher than the IQ8+ puts out at the peak on my system. This was while on the microgrid, at the same time the IQ8+ were curtailing the output to keep production at 600W. So however the IQ8s are controlling their own output power, it doesn't curtail the HM inverter.

I also confirmed that when I turn on the AC breaker (not when it connects to the grid), the IQ8s phase-lock to the grid like a UPS would do. This rapid frequency shift "sometimes" knocks the HM-350NT offline with an over/under frequency fault. It happened twice, but a third time it didn't happen.

The day got progressively more cloudy after that short data point. I'm thinking maybe I should connect it to a 48Vdc power supply and let it run at constant power, to hell with the cloud cover. Then I can see if the output actually throttles back.

Also, I connected my SENSE power meter to the solar system output. Now I have 1-sec interval data of the true watts. Nightime loss is less than 1W/inverter.

 

[svetz]

... due to inconsistent cloud cover. I can't tell...

Summer storms... got to love them. No worries, better weather is coming.

When running with about 600W load, the HM-350NT put out a peak of 356W. That's 57W higher than the IQ8+ puts out at the peak on my system.

That's not good. If at 600W load the HM was supplying 356W, then the other 9 IQ8s were supplying (600-356) / 9 = 27 watts each. That sort of sounds like scenario 2 where the HM isn't throttled and the IQs are. Don't suppose you ran the query https://envoy.local/ivp/meters/readings or looked at the HM output to get the frequency at that time? Don't forget you can also run the query http://envoy.local/api/v1/production/inverters to get the individual IQ8 output.

Hmmm, how are you measuring the load? Could that be case #3 (where the load is an incorrect calculation)? A clamp meter reading on the load will confirm the accuracy.

...however the IQ8s are controlling their own output power, it doesn't curtail the HM inverter.

The thing that's weird is that from #139 the HM does seem to be throttled back at times. I suppose even current sources have an upper voltage limit...might be useful to grab the microgrid voltage when the HM is throttled back. The HM display shows it, from #139, you just need to rest the cursor over what looks to be the throttled output. The meter query will give it to you too.

UPDATE: The upper voltage from the datasheet is 264V.

...This rapid frequency shift "sometimes" knocks the HM-350NT offline with an over/under frequency fault. It happened twice, but a third time it didn't happen.

Even when it goes offline (anti-islanding), it should reset after 5 minutes. I see the same behavior with my IQ7+s, although for me it also depends on some other factors as to if they start up immediately or not (e.g., load, battery SoC); although the power always has UPS-like switch over. Sometimes, if the load is low and the batteries are > 96%, I've seen the IQ7s shutdown even if they were running before.

I'm thinking maybe I should connect it to a 48Vdc power supply and let it run at constant power, to hell with the cloud cover. Then I can see if the output actually throttles back.

Given the clouds may affect the IQ8s (and therefore what the system is thinking), I think to do a true test all the panels have to be subject to the same light and would advise patience for a true understanding.

Also, I connected my SENSE power meter to the solar system output. Now I have 1-sec interval data of the true watts. Nightime loss is less than 1W/inverter.

Nice to see some data on BiduleOhm's post #68, thanks! Is that 1W as in 1W/hr or 1W/night? The SENSE is a CT and plug that measures amps and voltage, right? So, as the video says, since there are capacitors in the microinverters we should expect current in the wire, but as that power isn't really consumed (it's reactive) there's no overall energy loss while on-grid. Off-grid it's a bit different, the inverter's battery still has to invert to provide the current, so it's subject to the losses from the inverter inefficiency (as well as copper-loss). But, my personal suspicion is that once converted, the inverter only has to continue to supply the copper-losses, that is the unused energy is still conserved and isn't inverted twice. Might just be wishful thinking.

Did you by chance run the query in #67 (https://envoy.local/ivp/meters/readings) to get the nighttime reactive power?

 

[ncsolarelectric]

Summer storms... got to love them. No worries, better weather is coming.


That's not good. If at 600W load the HM was supplying 356W, then the other 9 IQ8s were supplying (600-356) / 9 = 27 watts each. That sort of sounds like scenario 2 where the HM isn't throttled and the IQs are. Don't suppose you ran the query https://envoy.local/ivp/meters/readings or looked at the HM output to get the frequency at that time? Don't forget you can also run the query http://envoy.local/api/v1/production/inverters to get the individual IQ8 output.

Hmmm, how are you measuring the load? Could that be case #3 (where the load is an incorrect calculation)? A clamp meter reading on the load will confirm the accuracy.


The thing that's weird is that from #139 the HM does seem to be throttled back at times. I suppose even current sources have an upper voltage limit...might be useful to grab the microgrid voltage when the HM is throttled back. The HM display shows it, from #139, you just need to rest the cursor over what looks to be the throttled output. The meter query will give it to you too.

UPDATE: The upper voltage from the datasheet is 264V.


Even when it goes offline (anti-islanding), it should reset after 5 minutes. I see the same behavior with my IQ7+s, although for me it also depends on some other factors as to if they start up immediately or not (e.g., load, battery SoC); although the power always has UPS-like switch over. Sometimes, if the load is low and the batteries are > 96%, I've seen the IQ7s shutdown even if they were running before.


Given the clouds may affect the IQ8s (and therefore what the system is thinking), I think to do a true test all the panels have to be subject to the same light and would advise patience for a true understanding.


Nice to see some data on BiduleOhm's post #68, thanks! Is that 1W as in 1W/hr or 1W/night? The SENSE is a CT and plug that measures amps and voltage, right? So, as the video says, since there are capacitors in the microinverters we should expect current in the wire, but as that power isn't really consumed (it's reactive) there's no overall energy loss while on-grid. Off-grid it's a bit different, the inverter's battery still has to invert to provide the current, so it's subject to the losses from the inverter inefficiency (as well as copper-loss). But, my personal suspicion is that once converted, the inverter only has to continue to supply the copper-losses, that is the unused energy is still conserved and isn't inverted twice. Might just be wishful thinking.

Did you by chance run the query in #67 (https://envoy.local/ivp/meters/readings) to get the nighttime reactive power?

I'm not set up with the Token to view the envoy.local stuff. I didn't even know I could do that. When I clicked on your previous links, I just saw a 404 Error page. This time it actually connected and asked for a Token. Idk yet what to do there, it's all new to me. Not sure I can handle another item on my to-do list right now either. I've got a warp drive to build!

1. The load is measured through the Enphase Enlighten Live Data app.
2. As I said, at some data points it appears that the HM throttled back, but it could also be cloud cover. Can't distinguish between them yet. Today is completely overcast, with no blue sky at all.
3. At some data points it is obvious the IQ8s were throttled back and the HM was full throttle while on the microgrid.
4. I'm glad to know that the HM is performing similar to an IQ7. It appears safe and works well IMO.

One thing I would like to point out. The IQ8s do not form a microgrid and manage zero-export without their Production CT, Consumption CT, Envoy, and IQ System Controller 2. They don't work stand-alone in this mode. I'm not expecting the HM inverter to work stand-alone and do those things either. I just wanted to see if it was safe to run them in parallel with a microgrid, and it appears so because the Enphase can phase shift and kick them off if necessary.

On the other hand, I am waiting on the Chint meter I ordered, and I also found out that the WattNode meter is the same one that SolarEdge uses. And that meter is much cheaper than the standard WattNode. I'm just waiting for a good paycheck before I spend any more money on this experiment. I also need to pay my NCUC and Utility fee to expand my system to include these HM inverters permanently. I may also use those Powerfield PowerRack buckets. They are very cost-effective for a ground mount!

Update:
Here was the most interesting data point of the day yesterday. 59.80 Hz, exactly when I closed the PV breaker to put the system back on grid. The Enphase inverters were phase-locking and it was exactly when the DTU took a reading. Lucky? Skills! This was the one-time I did this and the HM inverter did not shut down for a frequency alarm.

 

[ncsolarelectric]

Regarding nighttime losses. I mean <1 Wh/h per inverter. There are 10 inverters, the SENSE meter reads Watts not VARS. It shows 9 or 10 Watts all night, with 0 decimal places. It is understandable that the output inductor core has core losses, the wire has losses, and the capacitor ESR has losses. 1W at 240V is only 4.16 mA per inverter. 10W nighttime losses for a 3kW inverter is excellent, IMO.

 

[svetz]

... the SENSE meter reads Watts not VARS...

Pretty sure the SENSE uses a CT (which measures amps) and a plug (which measures volts), it then from those measurements calculates watts.
Reactive power is calculated by looking at the instantaneous voltage and amps on each leg.

... It is understandable that the output inductor core has core losses, the wire has losses, and the capacitor ESR has losses. 1W at 240V is only 4.16 mA per inverter. 10W nighttime losses for a 3kW inverter is excellent, IMO.

Not arguing that. The hypothesis is your SENSE is measuring the voltage/current in the wire which must exist as there are capacitors on the AC side of the microinverter. But since that power isn't consumed (except for very copper losses and capacitor losses which are really low), the actual consumption is less than that and may be less than what the video says if the power is conserved on the AC side of the inverter. Can't really know that without measuring the DC current, which I don't see any easy way to accomplish.

 

[ncsolarelectric]

Pretty sure the SENSE uses a CT (which measures amps) and a plug (which measures volts), it then from those measurements calculates watts.
Reactive power is calculated by looking at the instantaneous voltage and amps on each leg.


Not arguing that. The hypothesis is your SENSE is measuring the voltage/current in the wire which must exist as there are capacitors on the AC side of the microinverter. But since that power isn't consumed (except for very copper losses and capacitor losses which are really low), the actual consumption is less than that and may be less than what the video says if the power is conserved on the AC side of the inverter. Can't really know that without measuring the DC current, which I don't see any easy way to accomplish.

The SENSE meter is a Watt meter, not a VA meter. It knows the phase angle and discards the VARS. The CTs are at the System Controller, so there is 1W resistive loss, not 1VA, per inverter when including the wires, capacitors, inductors, connectors, etc.

Update: Actually, it is easy to see the DC side of the Hoymiles. Panel current was 0.00 all night. The only losses are on the grid side.

 

[svetz]

I'm not set up with the Token to view the envoy.local stuff. I didn't even know I could do that. When I clicked on your previous links, I just saw a 404 Error page.

Oh yeah... that can be really frustrating if you've never done it before. Really really useful though. As you can login to Enlighten you are already all setup... you just need to understand how it works.

You're getting the 404 because your home router isn't resolving envoy.local to the IP address. So, just replace that with the IP address of your envoy. Say the Envoy is 192.168.0.10, then the query becomes http://192.168.0.10/api/v1/production/inverters.

When you do that, you'll get a 401 error and an authorization page that looks like this:


Click the OK button to get rid of the dialog, then click the blue bar circled below:


That will popup a familiar looking login screen (not shown). Login with your Enlighten username and password and it'll take you to a screen with a white background... don't touch anything... and it'll redirect you to a screen that looks like this:

​

Your browser is now authorized and connected to your Envoy. IMPORTANT: Now, only that one browser is authorized. Leave that tab up to keep the authentication alive. Open a new tab in the same browser and paste in the same URL in again (e.g., http://192.168.0.10/api/v1/production/inverters - don't forget to use the correct IP address). Now that you're authenticated instead of getting redirected you should see the JSON response:

[
  {
    "lastReportDate": 1659275889,
    "devType": 1,
    "lastReportWatts": 146,
    "maxReportWatts": 296
  },...

 

[svetz]

The SENSE meter is a Watt meter, not a VA meter.

Since you didn't believe ... I took the trouble of confirming it for you... From https://blog.sense.com/how-the-sense-home-energy-monitor-works/:

...Two current sensors clamp around the main power feed inside your electrical panel, and then Sense connects to a 240V breaker; .... The sensors take advantage of the relationship between electricity and magnetism to non-intrusively measure the current, while the voltage is measured through the same circuit breaker that supplies the monitor with power.

So it measures magnetism to calculate amps and it measures voltage (okay you EEs...don't bust me here, I know you can't measure voltage directly either ; -). From those measurements it reports watts.

 

[ncsolarelectric]

Since you didn't believe ... I took the trouble of confirming it for you... From https://blog.sense.com/how-the-sense-home-energy-monitor-works/:

So it measures magnetism to calculate amps and it measures voltage (okay you EEs...don't bust me here, I know you can't measure voltage directly either ; -). From those measurements it reports watts.

Yes, I know how CTs work, I've been using them for decades. Most power analyzers use CTs, then have microcontrollers to measure the phase angle and do the calculations to determine Watts and VARS. SENSE reports WATTS and kWh just like a utility calculates the bill. It doesn't display VARS. I think you may have some idea that CTs can't be trusted. Believe me, they can be trusted down to 0.02% if they're made right and I agree, these are cheap plastic, but they are certainly accurate enough for what I'm doing.