jjww
Stuck in PG&E Land
No, but I do have an open grid tie request hung up with the whole PCS debacle with the 18K
(since February 2024)
WARNING! Major Issue With 18Kpv, Exporting to Grid for hours when Not Programmed To!
- Thread starter fnnwizard
- Start date
(since February 2024)
jjww
Stuck in PG&E Land
Holey crap not the dreaded PCS on the 18Kpv........
EG4TechSolutionsTeam
Online Support for EG4 Electronics
As Markus mentioned in this thread, it seems to be a data miscalculation due to the CT placement in the system. We also noticed that AC charging was enabled with 0 kW, and then 0.1 kW on separate occasions. This low charging power, combined with the CT placement, could be causing the issue. Although our engineering team is investigating this, I wouldn't raise any alarms just yet. I don't foresee any issues with your setup if it's installed according to the recommended diagrams.
Quattrohead
Solar Wizard
It will disconnect from grid as per 1741 regs, but still monitors it for return and stability of the grid.
I highly doubt it's CT location. The logs show multiple data points that do not line up. The measurements or calculations taken eslwhere is what I would look at.
The 18Kpv CT (+-1%) data is the only thing I can verify and works ( within the +-100W). 3rd party CT works perfectly fine. Here's a picture to show what I calibrated with. 4 meters +3rd party CT (+-2%) reads within 5% of each other... which one would you believe in?
The CT Placement, where I have it, is really for best practice if one has a constant load in the main grid panel, and want to monitor inverter sell back.
It's in the coding. Let's forget Fast Detection and large load surges for now. The:
GRID SELL BACK - Disable should be a GLOBAL SETTING, there's a 5 second polling period (again fast detect disable). Is this 5 second polling at a constant interval?
AC Charge - Enabled, means the inverter should not draw from battery ( of course unless grid is down)
AC-Charge Power should not matter what setting is since above 2 would (should) prevent any sellback.
I just don't understand how it can continuously sell back for hours and it wasn't all small amounts. 600-700watts at times
Attachments
Last edited:
Then explain how this works:
3x Sol-Ark 15K set up on 3 phase in "limited to home" mode, doing gen peak shave (via grid input terminals) utilizing external CT sensors. Sell is disabled. The inverters keep a 45kW genset from overloading with dynamically changing loads during the work day, peaking at 65-70kW of total loads at times. Constant draw on gen through external CTs is ~36kW. And there are non-backed up loads (e.g. gen direct) connected between CT sensors and the inverters...
↱ -------------------------------↴
Genset>CT sensors>non-critical loads>inverters>critical loads panel
⇅ ↑
batteries Solar panels
Works quite well, and I would expect a similar setup with the Eg4 18Kpv would work as well. This tells me external CT sensors can give good feedback.
By the way, I don't know how the internal ones in the 18Kpv are, but in the 15K, they are actually quite similar to the external ones, with short wire pigtail going to a plug on the circuit board....
All this said, I would still agree that faulty external CTs could factor into OP's issues! And I would also tend to feel that the unit should not need the external CTs to do what OP wants to. External CTs are more of an add-on feature that adds functionalities such as what Sol-Ark calls "limited to home", where inverter sells to CTs, but doesn't push past them.
Solar Guppy
Red Cobra Guppy
Sure, its called latency. Internal measurements are done in microseconds, the CT's in seconds. The difference is what can cause unintended sell.
The CT's use a slow serial protocol called modbus, add in the internal processor and its delays and filtering and you get what it is.
I never wrote CT's don't work, they are the system ( home ) level measurements, and is relatively stable, until loads change. Nothing bad happens what loads change except for the energy movement can go places you don't want to ( grid in this case )
Markus_EG4
I am from Germany. Don't mean to come off rude
Let's wait and see what the power company has to say. There's a lot of really useful information in this thread now. The only real truth will be the data provided by the power company. I can assure you that we have never been in any kind of dispute with a power company over selling back power before. We have tested this in R&D with an RGM meter that reads with 0.1% accuracy.
Solar Guppy
Red Cobra Guppy
20x times better that standard spec external meters used, I get this as I use own Yokogawa power analyzers for my design work, they were 0.25%
Markus_EG4
I am from Germany. Don't mean to come off rude
we use a lot of Chroma here. we have 1 Yokogawa analyzer that thing is a beast. You get what you pay for with then 100%
Markus, I pray the Power Co does not contact me about this. If they do I may have much bigger issues.
Instances of dispute with power company is not a good reason to quantify no sellback issue. Can't use that to prove a non sellback issue.
My first year in 2019 with Enphase system, I was getting charged for my sellback and 3 months of disputes with Poco finally came out to replace my meter. They over billed me over 5k for all my sell back as useage.
Since then I have over sold into the grid by way too much as I over paneled when I got an EV end of 2019. they never came out since then and have never said anything about the overage. My ROI on that system is garbage... I charge my neighbors car both his S and his X at times
Look at my SCE Useage for that 1! last month!! This is with charging more than 1 tesla, running a/c and heated pool 24/7.
Edit: Poco probably loving me. This system has been in place since end of 2019.
Now if I can find a way to beam that power wirelessly....
Attachments
Last edited:
Modbus? We are talking (on Sol-Ark) 3,000:1 ratio CT sensors, with the secondary wires (twisted pair) going straight to a circuit board in the inverter. Then there is an internal set that is basically hooked in the exact same way, just at a different point in the inverter....
There is absolutely not seconds of delay using external CT sensors! Even watching live (within 1-2 seconds) data remotely via Solar Assistant, I can clearly see the 15Ks of the previously mentioned site adjusting their output within a split second of power jumping or dropping!
I don't want to derail this thread further on this subject, but if you (or anyone else) would like to see this in action, I can get a video recording for you. Feel free to DM me, or direct me to a thread that would be suitable to discuss this further on!
Solar Guppy
Red Cobra Guppy
Got that, many inverters use external meters that the CT's connect to, it eliminates the distant issues or the inverter having to be X feet from the mains. I'm a Solis HV guy now and it uses the Actel meter.
Makes sense where you were coming from on that! Modbus would definitely bring latency into affect!SolarFlares
New Member
This is great advice......for young children. Adults acting like this is a major red flag and not something I entertain. I unwatch the thread and continue with my life. I'm sure every other person trying to help will get blamed for not reading his posts correctly or not knowing this or not knowing that.
Agree, the response time on the 18Kpv is adequate to mitigate surges for minimal backfeeding, no problem. It can easily and consistently arrest sudden stop test loads of over 7kW so far. (AC + Airfryer + Microwave = low tech testing by pulling fuse and setting 1 min timer on airfryer and mic. Airfryer takes more than 1 min to heat up so heating element is pulling almost 2k power... my 120v outlet at that location is getting abused, lol. )
I'm currently testing in Off Grid - Disable mode now and will post logs to show the total export output. I need to increase my constant load in main panel though to be sure it won't reach back to PoCo. I've also ask EG4 if they can increase resolution on the logs for me to better capture the data. Will report back.
EDIT:
I've updated 1st post to show a 3500W surge and what it looks like... see pics with title start with 7-5... the line chart suggesting it was over a period of minutes is not correct though, the raw data logs are more accurate in this instance.
To cmiller: I highlight your response and wonder if 18Kpv has similiar 2nd set of CT and wonder if this could be causing conflicting data and is an issue?
Last edited:
ksmithaz1
Solar / EV Junkie
No, this is not accurate. A "CT" is just a coil connected to a sensing unit it stands for "Current Transformer". The AC induces a small current that rolls down the wire to the sensor which can be as sensitive as you like. There are different flavors of CT's which vary the number of windings inside the unit, allowing for different scaling of the signal. I believe the 18K as a setting for this in the advanced screen, I'm not looking at it right now. The greater the current traveling down the wire the larger the magnetic wave coming off the wire, thus inducing greater current at the CT which is detected by the sensor.
No, the CT does not send modbus, it is a dumb transformer. You hook a CT to a device that samples the CT output which is ANALOG. You could convert it to modbus, and send it to another device, but it is unlikely you would do that inside an inverter. I would bet dollars to donuts, that there is an ASIC inside the inverter that is handling the CT inputs making the real-time analog information available on demand from the ARM based CPU inside the ASIC. There is no necessity to convert to modbus and poll inside the asic, you would just read the register directly to get the last sampled value, likely updated every few milliseconds at worst. The sample rate can be whatever you want it to be, and the detector is watching for fluctuations in the produced wave.
Polling the inverter over modbus to get realtime information is one thing, but has nothing to do with the ability of the inverter to poll the actual analog sensor for information. I have some modbus sensors that have both CT and voltage inputs. I poll them every 5 seconds, but the first program I wrote was polling them as fast as I could at 9600 baud, and I was getting multiple returns every second. Internally it's probably updating every 10ms or less.
ksmithaz1
Solar / EV Junkie
Apparently the CT's for these inverters is directional. Not sure how that works but it would be a requirement to know which way the power is flowing to shut it off. Electricity moves / is drawn towards the load (The receiver of the power) not forced out by source (The provider of the power). If you connect two sources to one load neither source is aware of the other. If the voltage in one source is lower than the other it will act like a load instead of a source. The funky frequency tricks/signaling is one of the ways backfeed is controlled by the power company in that scenario, but if you didn't catch an agreement that puts the box in front of your inverter it isn't going to know, and will allow a lower voltage grid to act like a small load. The only way to stop this would be to know that the grid was presenting a load by measuring the current flow traveling between the grid and the shared bus powering your internal loads.
I'm not sure why this concept is so confusing, but I think people do not understand how electricity travels. This same concept is the issue with balancing your cells in your battery bank when they are in parallel. You end up with batteries feeding each other when there is a voltage imbalance.
I'm not sure why this concept is so confusing, but I think people do not understand how electricity travels. This same concept is the issue with balancing your cells in your battery bank when they are in parallel. You end up with batteries feeding each other when there is a voltage imbalance.
They are bidirectional, but agree with everything else. Something in the inverter opened the gates to allow my batteries to supply to grid.
No forced discharged. This would be too easy to catch : ). It would show up readily in all the logs; raw data, line chart, bar chart. I have to specify these separately becasue the data is calc/gathered from different sources. I get frustrated EG4 keeps telling me my CT location is a likely culprit.
I will say this, the Off-Grid Mode Disable code seems to be working well. The Off-Grid Mode - Enable code seems to be patch work/unfinished.
I'm currently running in Off Grid Mode - Disable to test.
Last edited:
Agreed we supply Yokogawa flow meters on our skids at work they are great (different industry)
ricardocello
Watching and Learning
Sorry for being slightly off-topic here.
Those of us who do not have All-In-Ones (Blue people) get standalone Energy Meters.
My Carlo Gavazzi EM530 has dumb transformer CTs connected (5A means 200A).
However, I have to access it over modbusRTU (actually over Ethernet because I have a waveshare).
It can sample grid power at almost 10 Hz.
This is the only way to see what is actually happening with the grid within a one second interval.
The Cerbo running ESS tries to keep up, but clearly there is latency, even at 115200 baud over 10-mbit ethernet.
I would hope that all-in-ones like the 18kpv have a much more directly sampled analog CT going into a real-time control loop as you've said.
But we don't have the internal block diagrams of these AIOs to know that, we can only speculate.
---
Even slightly more off topic, but educational, the EM530 communication protocol guide says:
According to the signs of the power factor, the active power P and the reactive power Q, it is possible to obtain a geometric representation of the power vector, as indicated in the drawing below, according to EN 60253-23:
Quattrohead
Solar Wizard
You said it.
To expand -
AC Charge - Enabled means it is allowed to charge from AC as well as solar.
AC-Charge Power is how much power it is allowed to draw from the grid to charge.
Last edited:
AC Charge - Enabled, means the inverter should not draw from battery ( of course unless grid is down)
AC-Charge Power should not matter what setting is since above 2 would (should) prevent any sellback.
I just don't understand how it can continuously sell back for hours and it wasn't all small amounts. 600-700watts at times
Can you add the rest of my comment... why are you trying to misquote me? man...
AC-Charge Power should not matter what setting is since above 2 would (should) prevent any sellback.
I just don't understand how it can continuously sell back for hours and it wasn't all small amounts. 600-700watts at times
Sir, I know exactly how it is Supposed to work. Post #92
Can you add the rest of my comment... why are you trying to misquote me? man...
Last edited:
Similar threads
