Adding storage to my Enphase system

[400bird]

I was to build a server rack stack, I would need 7 of the EG4 100 amp hour boxes at about $1,500 each. That is over $10,000 in battery modules, and it falls a little short of the capacity I have with the Bolt cells. And nearly double what I paid for them. My hope is that in another 5 to 10 years, there will be many more ex EV batteries coming on the market, including ones with much better fire safety records.

Every once in a while, I think maybe I should have gone a different route. Math like that reminds me why I'm happy I didn't!

 

[GXMnow]

Yesterday turned out to be a really good test for my export control system. Here is the battery trace for the day.

The battery did cut out just before 5:30 am, so I had buy SCE power for almost 3 hours in the morning. We had really good sun, so it charged at a good rate all day. I am still trying to decide if I want to change how I handle the battery getting full. Dropping to 10% charge rate when the voltage is 0.05 below absorb limit seems fine for now, but I just didn't expect it to take an hour and 20 minutes to get that last 0.05 volt. But it still hit limit and stopped charging by 3:15 pm. You can mentally fill in the curve and see that the solar was still putting out everything the house needed to almost 4:30 pm. Then it started to use some battery power to zero the loads in the house. At 6 pm I microwaved something for a snack. My girlfriend and son both worked late. So we had a late dinner, cooking after 8 pm. Then ran the dishwasher, washing machine, and dryer (which are in the main panel) all the way to 10:35 pm. Each load shows up as load on the batteries. How well did that mirror the load?

Here is the So Cal Edison hourly data for the same day

I think that is pretty definitive. The cheap PZEM power meters are pretty good. The appliances that are in the main panel were all zeroed out nearly perfect. The big test is still to come. Firing up the A/C compressor. It almost got warm enough. It did hit 80 degrees out, but the inside of the house was still only in the low 70's from the cold overnights. But at this rate, it won't be long. I am going to do my annual clean out of the coils this week and make sure there are no ness in there as it has not run in 5 months. Probably time to change the filter in the furnace as well. I vacuumed the metal mesh pre filter back in December, and the paper element was just a little dirty then.

I do think I want to flip the phases between the dishwasher and the garbage disposal. The dishwasher in a bit of a current hog, pulling over 9 amps when the high pressure pump is running, and it is on the same phase as the microwave. When those were both on, I saw the XW-Pro pushing 23 amps on one phase, and just 1 amp on the other phase. It can handle it, but I would rather not push that much neutral current. The big pain in the ass will be swapping which CT the wires go through if I swap it in the main panel. If the conduit from the panel makes it to the switch box in the kitchen, I may be able to swap the wires in there instead. Those two circuits leave the main panel in one conduit all by themselves, and it only powers a combined outlet pair, the switches for the garbage disposal and the light above the sink, and then goes to the outlets inside the cabinet for the garbage disposal unit and the dishwasher.

Off to work, this will be a job for another time.

 

[GXMnow]

So no one was home most of yesterday as all 3 of us worked. I think So Cal Edison has noticed something is up.

When I log into the SCE web site, it usually comes up with the energy use bill tracker. It gives an estimate of what the next bill will be. Well, mine came up with

It started at $50 at the beginning of the month, and each day it was dropping. It showed like $20 yesterday. But I think my average daily usage for this month just went negative. I did just add a 150 milliamp offset towards export. That made it so no bar went to import while there was charge in the battery. Yes, milliamps of grid power. 0.15 amps is 36 watts. 36 watts for 13 hours is 468 watt hours exported from 5 pm to 5 am. The other 2 KWHs was all exported while the sun was on the solar panels. This morning the battery kept exporting that 36 watts all the way until the sun came up. So there won't even be that 380 watt import at 7 am.

I have to enjoy this now before my air conditioner sucks up all the power and I am back paying $120 electric bills again. My hope is the battery will at least fully zero the 4 pm to 9 pm area with that beast running.

But there is a little good news. I might place the order for some new solar panels from Real Good Solar. If I order 9 panels, they will do free shipping again. They have a Q-Cells 320 watt panel for $204.12 each. Not the best price, but with the free shipping, it brings it down to match used panels from San Tan Solar. Since these are the smaller 60 cell form factor, I think I can fit 9 in a 3 x 3 array on the garage roof. Then do 1 or 2 on a cheap DC charge controller, and get more Enphase inverters for the rest. If I put all 9 on Enphase, it might max out the amount of AC coupled solar I can feed into the XW-Pro. I found the iQ7+ inverters on Amazon for just $137 each. The iQ7+ can output up to 290 watts instead of the 240 of my current iQ7's. So I need to take that into account. (16 x 240) + (9 x 290) = 6,450 watts. That is about as far as I want to go. The XW-Pro could take 6,800 watts. 320 / 290 = 110% or 10% over panel. My current setup is 25% over panel, and even at that, it rarely maxes the inverters. So I doubt these would ever clip out. I should get every watt those 320 watt panels can make. I just have to figure out if it fits in the budget. Work is picking back up, but we are still well shy of normal.

 

[Hedges]

So no one was home most of yesterday as all 3 of us worked. I think So Cal Edison has noticed something is up.

When I log into the SCE web site, it usually comes up with the energy use bill tracker. It gives an estimate of what the next bill will be. Well, mine came up with

Paranoia - Wikipedia

en.wikipedia.org

 

[GXMnow]

Thanks Hedges

 

[GXMnow]

The previous 3 days, I ended up exporting 1.59, 2.43, and 1.68 KWH and today is looking good as well. The monthly bill tracker is back up, with the estimate down to $27 after yesterday. I am going to look over my code for the end of the solar production day. Dropping it to 10% max charge rate is looking a little too drastic. I am going to think about how I can allow it to ramp down, but not ramp back up. This is not picking of course, it is working way better than I was expecting for the quickie code I wrote. I am also thinking about the code for when the grid goes down. I am going to do the logic so the normally closed terminal puts it into grid fail mode. That way, I can just yank out the relay to put it back into normal mode if anything goes wrong.

 

[400bird]

Why are you ramping down current at the end of charge? The charge profile on the XW is handling that for me, I keep sending it the max current and the XW uses less than that depending on where it is in the charge cycle.

 

[Shimmy]

Every once in a while, I think maybe I should have gone a different route. Math like that reminds me why I'm happy I didn't!

As long as the Bolt batteries don’t burn down the house. That is where things start getting complicated.

 

[400bird]

Mine are outside 6' from the house with a smoke detector all inside a steel box lined with cement board. It's not going to keep the fire in, but should make for enough warning and distance from the house.

 

[GXMnow]

Why are you ramping down current at the end of charge? The charge profile on the XW is handling that for me, I keep sending it the max current and the XW uses less than that depending on where it is in the charge cycle.

It was running into an odd situation where the PLC was seeing the export power increase, so it kept cranking up the charge rate, and it also was sending the start bulk charge over and over after the XW-Pro wanted to go to absorb mode. So I have it change to a different routine when the battery voltage exceeds 56.650 volts. I figured it would only be minutes more to hit absorb, BUT.... These batteries are so stiff, it is taking an hour to get that 0.05 volts at 14 amps of charge current. It is basically a very nice absorb charge. I think I am going to adjust it a little and allow it to go up to 15% charge rate, 21 amps. But if it ever does drop to "no float" it won't start bulk charge again after 56.650 volts.

 

[GXMnow]

Solar production so far today is well, Crap!
We have heavy clouds, and threat of rain. It is only 8:15 am as I am typing this, so I am hoping the weather improves some. Right now my panels are putting out half what they were at the same time yesterday. But that is not enough to trigger my system into charging the battery bank. In fact my battery bank is still pushing 800 watts into the house to run the furnace. "Sell to grid" is now blocked, so it is pulling another 200 watts from the grid as well. I am actually surprised the battery is still going from the charge yesterday. It is down to 51.610 volts according to the XW inverter. It will shut down when it hits 51.500 volts. If I am not getting any charge by 10 am, I will manually hit the force bulk charge. But I may need to disable the PLC as it might just put it back into "No Float" if it sees current coming in from the grid going to the house. On the iPhone weather app, it shows the sun peeking out from the clouds in the 10 am and 11 am hours. And it did just peek out now, and it went to 700 watts of charging. Now 11% almost 800 watts. I have to stop watching it.

 

[GXMnow]

The clouds cleared for a bit and it did start to charge for a short time, but now, they have gotten even heavier. My routine saw the power start coming in from grid to run the house, so it commanded the XW to stop charging as expected.

I still want to get some juice into the battery bank, so I went and triggered the bulk charge manually, and they way I wrote the code, it is letting is continue charge. If the solar picks up, it will reset it to bulk charge, and then if it drops out, it would stop the charging. But it is forcing it to minimum 5% charge rate.

 

[GXMnow]

All in all, the sun came out a few times today. Total solar production was not great, but not terrible either. It ended up making 12.2 KWH's before sunset. With running the house, it managed to push 4.7 KWH's into my battery bank. That was enough to run everything in the house through the 4 pm to 9 pm peak rate time block. At this rate, it might even make it to 10 pm, but I will be buying grid power until the sun comes out tomorrow. If I had totally left it alone, it would have put a bit less into the battery. It probably would not have made it to 9 pm. So I am thinking about how best to make the code handle a poor production day like this. I know I am not supposed to charge from grid power, and even with my manual starts it still didn't. My house used some grid power as the solar was charging the battery bank.

 

[400bird]

It was running into an odd situation where the PLC was seeing the export power increase, so it kept cranking up the charge rate, and it also was sending the start bulk charge over and over after the XW-Pro wanted to go to absorb mode. So I have it change to a different routine when the battery voltage exceeds 56.650 volts. I figured it would only be minutes more to hit absorb, BUT.... These batteries are so stiff, it is taking an hour to get that 0.05 volts at 14 amps of charge current. It is basically a very nice absorb charge. I think I am going to adjust it a little and allow it to go up to 15% charge rate, 21 amps. But if it ever does drop to "no float" it won't start bulk charge again after 56.650 volts.

Could you have it watch the charge status and change behavior once it changed from bulk to absorb? I have mine set so at all times the max charge never exceeds solar production as a max limit.

What's the logic you're using, that it re-sends the bulk charge command even though it's already charging, just not at full requested current?

My pack is still torn down for a slow rebuild, painfully slow. I'm going to rewrite some of my logic taking in your charge command logic.

 

[400bird]

So I am thinking about how best to make the code handle a poor production day like this.

Let the pack be low on SOC and ride through to the next clear day? There's a reason you're still grid tied, right?

Also, there's a case where adding more of those nee panels DC coupled would have helped your ability to make it through SCE's peak rates in the evening.

 

[GXMnow]

Could you have it watch the charge status and change behavior once it changed from bulk to absorb? I have mine set so at all times the max charge never exceeds solar production as a max limit.

What's the logic you're using, that it re-sends the bulk charge command even though it's already charging, just not at full requested current?

My pack is still torn down for a slow rebuild, painfully slow. I'm going to rewrite some of my logic taking in your charge command logic.

When I was coding this, I had a day where clouds kept passing by. So it would go from charging from solar and exporting, to all of a sudden having to charge from grid power. This was happening over and over. So what I did was add a check for grid consumption. If it starts pulling power from the grid, it puts the XW into "No Float" to stop it from charging from grid power. I think the threshold is now at about 200 watts of grid power.

So whenever it puts the XW into No Float, it then starts watching the output side power again. I don't have an input for just the solar production at this time, so it only sees the difference of solar and the backup loads. So when the solar starts making more power than my backup loads panel, (I think that threshold is at 300 watts now) it will command it back to Bulk Charge.

I am not sure of an easy test I can do to know if the XW has switched to Absorb or No Float on it's own. So what I ended up doing was blocking it from sending the "Bulk Charge" command when the battery voltage exceeds 56.65 volts. My bulk charge ends at 56.7 volts, so just 0.05 volts short. At this point, the XW is certainly still in Bulk charge. If it drops out of bulk charge then, I let it stay out, it's full enough. But if it is below this level, and the sun comes out from behind the clouds, go ahead and put it back in bulk charge.

My charge rate percentage is initially set by ramping up or down off of the amount of grid export power. And then I have a window compare. If it wants to go below 5%, it is set to 5%, and if it is in the Bulk Charge voltage range, it can go up to 80%, any higher, clamps to 80%. After battery voltage hits 56.65, I have it set the charge rate limit down to 10%. I think I will up this to 15 or even 20% now.

Of course, I have another day of heavy clouds moving by. Right now, my 4,800 watt array is making just 519 watts. And the house is using 530 watts. The battery went to idle mode, and I am using 12 watts from the grid. The "Insight Local" Dashboard has no moving dots right now. Flow has to exceed about 60 watts for it to show energy flow. And the clouds parted a little, solar hit 840 watts, so it went back into bulk charge.

 

[GXMnow]

Let the pack be low on SOC and ride through to the next clear day? There's a reason you're still grid tied, right?

Also, there's a case where adding more of those nee panels DC coupled would have helped your ability to make it through SCE's peak rates in the evening.

Yes, last night, the pack did drop to about 50% charged (51.5 volts on 14S) at about 9:30 pm, so it made it through the expensive time. Then it just sat in standby while I paid for grid power. If there was a grid outage, I had plenty of battery left to make it to morning, but then this morning is also bad production. I can force i to grid charge a bit if I was worried of a grid outage, like if there was another close by wild fire.

I am really torn on the whole AC vs DC coupling now for the new panels.
WARNING, I over think things all the time. This is my thought process put into words.

DC coupling solves dark start and it's a little more efficient than converting to AC and back. But then there is what happens when the battery is full. With any DC coupled panels, the charge controller will basically just stop taking power from the panel to keep from overcharging the battery. It's not a big deal, but when comparing to the efficiency of charging from Enphase microinverters, I think it more than covers the efficiency difference.

Charging from solar panel to battery with DC coupling 96% efficient. The battery tops out with 1 sun hour left in the day. A 320 watt panel, 1 sun hour is 320 watt hours lost. At 96% efficient, it would lose that same 320 watts in 8,000 watt hours of solar input. 8,000 / 320 watts = 25 hours of charging. So more than 3 days of charging losses are lost in just one sun hour when the charge controller shuts down.

Charging from the 97% efficient Enphase microinverter, and the 95% efficient XW charger section. Total is only 92% efficient. So the loss while charging is double, I will lose that 320 watt hours to efficiency losses in just 4,000 watt hours of sunlight. Only 12.5 sun hours, about two good days here. But what happens when the batteries are full? Let's say the system does take another 0.5 sun hour to complete the charge, so now we still have 0.5 sun hour left in the day. Instead of throwing the power out though, it goes to the house, and out to the grid for credit, if the house is not needing that power. That is 160 watt hours, not being lost.

I know this is convoluted math, but by these calculations, the extra energy to help run the house after the battery is full is more than making up for the efficiency difference. The charging power also will not need an extra 30 minutes from the AC coupling, so the energy recouped is even more.

That leaves the dark start issue. I still think I can make my PLC bring the system back up when there is enough sun to charge the system from the Enphase inverters. I just have to decide if I want to just guess by time, or use a solar cell to measure the sunlight available. And the idea of 1 or 2 DC panels is still floating in my head.

When I first thought about adding the DC charging, I was convinced it was the way to go for the best of both worlds. And I still think there is some merit to that. But now that I have the PLC doing such nice control of the system, many of the DC charging benefits just went away. If I was truly off grid, it would be a different story, 50% DC 50% AC would be a great off grid setup. But with being able to grid export for some credit, my small import/export amounts are nothing, and the grid is my backup when production falls short.

Then the whole Arc Fault thing threw a huge wrench into it. I have now found the Enphase iQ7+ inverters for just $134 each with free shipping. So if I get all 9 of them, that is $1,206 which is now getting down to the same cost as the Schneider MPPT 60/150 charge controller and their RSD/Arc Fault box. Then I still would need 9 of the Tigo RSD units at $30 per panel as well. The cost is a wash. Q cable vs MC4 cables inline fuses, combiner and breakers.

The equipment is certainly cheaper if I went with 6 x 415 watt panels instead of the 9 x 320's, but the free shipping is only on 9 or more panels. The price of the shipping is 2 or 3 microinverters, and the cost per watt on the panels is basically the same. The 3x3 array of 60 cell panels fits very nice on my garage roof, and it is 2,880 watts of panels. The 6 x 415 is a little less at 2,490 watts. The array width ends up almost the same, but the 6 panels is a little shorter in length, hence the slightly less power. Both are 19% efficient panels. Just for my own curiosity, 9 x 60 = 540 solar cells total. And 6 x 72 = 432 cells total. So each cell is making a tick more on the larger panels. I am including Iron Ridge racking rails as well. The 6 panel setup does save a little there as I would only need 4 rails instead of 6, but the rails are actually pretty cheap. The UFO's, End stoppers, all that little stuff adds up.

In the end, I am looking at an all up cost of about $3,600 to add the 2,880 watts of panels. Shipping the cheaper 400 watt panels from San Tan would have been over $400, and those only had a 1 year San Tan warranty. AltE had some nice 425 wat panels, but again, shipping was $600. I am only at $1.25 per watt. That is still very low. By the time I buy the little things locally, it will likely hit $1.50 a watt. One big part of the savings for me is the fact I already have an Enphase iQ system. So I have the combiner and Envoy and outdoor legal disconnect. All things that basically would need to be duplicated if I went DC coupled.

Hope that didn't hurt too much to follow. Now I have all the stuff in two internet shopping carts. That is a $3,600 mouse click. Do I really want to do this? I think I can also put it all on my tax return for the 26% credit next year.

 

[GVSolar]

That's what I call a "high quality problem".

 

[Hedges]

But then there is what happens when the battery is full. With any DC coupled panels, the charge controller will basically just stop taking power from the panel to keep from overcharging the battery.

What Sunny Island does in this regard is, if DC coupled PV (SCC) has its own voltage setpoint and drives battery higher than Sunny Island wants, Sunny Island will draw power off battery by inverting and exporting to grid to achieve the voltage it wants.
Can your inverter drive grid from battery?

If SCC has communication with Sunny Island, it is told what voltage setpoint to use, so this wouldn't happen.
I just realized that having a relay to interrupt communications (and substitute terminating resistor so other attached devices can still communicate) would give control over this. Break communication when on-grid, reconnect if off-grid?

 

[GXMnow]

What Sunny Island does in this regard is, if DC coupled PV (SCC) has its own voltage setpoint and drives battery higher than Sunny Island wants, Sunny Island will draw power off battery by inverting and exporting to grid to achieve the voltage it wants.
Can your inverter drive grid from battery?

If SCC has communication with Sunny Island, it is told what voltage setpoint to use, so this wouldn't happen.
I just realized that having a relay to interrupt communications (and substitute terminating resistor so other attached devices can still communicate) would give control over this. Break communication when on-grid, reconnect if off-grid?

Yes, the XW-Pro can certainly take battery power and export it to the grid. And I can picture a way to make the logic work. Like you said, have the solar charge controller set to chare the battery bank to maybe 0.1 volt higher than the XW will charge it. When the battery voltage goes up there, have it ramp up the grid export power until the battery current falls to zero. That way, any DC charge current coming in is just being inverted and exported to the grid and the battery stops charging. It's very similar to the export control I am doing now. Just have it add in an offset if the battery is still charging above target voltage.

The one part that makes it tricky is that the XW-Pro is either a charger or an inverter. It can't do both at the same time. And since it is a single box with just one set of battery terminals, the battery current is measured internally, unless I add the Battery Monitor for $600. I would need to add a shunt that the PLC can measure on the battery side, after the Charge controller and XW-Pro inverter are combined. Too bad my BMS RS-485 does not have a ModBus map, at least that I can find. But the current measurement in that is pretty sketchy. The ripple from the XW causes the current reading to bounce over 25 amps.

I have not clicked "Place Order" just yet, but I am not sure how long they will have the free shipping on the solar panels and racking.