For winter, I’ve configured my solar system to run with minimal power. I have the Inverter usually switched off to save power as well as other devices connected to the system. In this minimal power usage setup, my system has a constant 24 hr a day DC load of 3 watts (120mA), about 80wh a day (I expect to see this as 10wh every 3 hours in the bar graph). The Shunt and 2 MPPT 100/20 controllers are the main devices that are displayed in the VRM.
Anyway, I have noticed that the power consumption in the bar graph during the day seems to be calculated wrong. The bar graph is showing a bunch of power consumption during the day that I don’t think is occurring. At night, the power consumption records the steady 3 watt usage, but during the day the system doesn’t see this, I am guessing it’s because the DC load is a calculated number (with various rounding errors), I see it jiggling around, going from -3 watts up to 5 watts as the solar power varies. I would think these varying DC load numbers would average out to 3 watts during the day, but the bar graph shows a much larger consumption during the day, so it’s not averaging out like I would expect. For some reason, the VRM software thinks I’m consuming a bunch of power during the day when I’m not. It seems to be more than just rounding errors. Maybe the minus DC load numbers are getting ignored or the minus sign is being switched to positive, or it’s recording extra power that the solar charge controllers use during the daytime ? Anyway, I’m bringing this to your attention as it would be really nice to get this working better, or for me to understand why it’s showing this daytime consumption. I’ve included a screenshot of my VRM info showing all the phantom daytime power consumption.
At very low power levels, like only a few watts, the accuracy isn’t very high.
All Victron products have some “base load” consumption, this is typically very low, but when only measuring considering 1-3 watts of load it does become a factor.
The MPPTs consume power when they are operating for example.
In most cases reducing any measurement errors in 1-3 watt range measurements is not worth the extra time and cost.
You’ll find under normal load operations that this kind of low load noise becomes negligible and doesn’t affect the readings as much.
It has been considered to ”round down” more aggressively to eliminate it that way, but that also has its own issues.
I think the status quo where we try to give as accurate readings as we can within the hardware and communications constraints is acceptable, and for most users unnoticeable, but it is there if you look for it under these no-load conditions.
In some cases the load is real, just small (product self consumption), in other case it isn’t, and just noise (calculated DC readings from different sample rate and differently calibrated datapoints).
If you aren’t using calculated DC readings, you can disable those and it should resolve it. If you are using calculated DC readings then there is some level of inaccuracy that needs to be accepted.
In the summer time, my system generates so much more power, I leave my devices powered on all the time, which creates about a 15w base load. I haven’t been concerned about 10wh here or there, as you say, it’s just noise compared to the big chunk of power coming in each day. But, now I have it running in a minimal mode, I am seeing this daytime load which I don’t see where it’s getting the numbers for it from. If you look at the little DC Power graph in my screenshot, you’ll see that it’s basically flat at 3 watts all night, and then during the day, it jiggles, but it is jiggling from -3 watts to 3 watts (with a rare 5w), so on average, the numbers for the DC load during the day should actually be less than the 3 watts being recorded during the night. I’d expect 10wh to show up in the bar graph every 3 hrs like is recorded overnight, but this is not what I see, usually every daytime hour has 10wh, sometimes 20wh of consumption. I was just curious where these daytime numbers were coming from, as I’m not seeing anything being shown that should generate this much consumption to be recorded. Is there something in the system that secretly adds some consumption for MPPTs when they are active. You know the model numbers, and you know what their active consumption rates would be, so I can see the software doing this behind the scenes. It makes sense to me that the MPPTs are using more power while they are busy during the day, so there is higher consumption, so having it in the bar graph is probably closer to reality, but just going off the reported info I see being presented, I’m not seeing how this higher daytime consumption is getting generated, since my understanding of how the Victron stuff works is to go off the actual reported data.
guystewart
(Guy Stewart (Victron Community Manager))
4
If you disable the calculated DC does the issue resolve?
That’s a good experiment, I’ll try it and see. I just turned it off, so now there’s no calculated DC reported in the VRM. You can see there’s 3 watts missing from what the solar is doing and what the battery is getting.
I’m interested to see what will happen in the bar graph for the next 24 hours. I’d expect the consumption to be zero, but maybe I’ll be surprised.
I have a DC system that uses power, so I want to have the “has DC” enabled to keep track of it, even thought it’s not as accurate as I would like at these low power levels. I had thought the calculated DC load was just doing a simple current difference, but what if it’s also calculating the MPPT consumption, it has access to the power going in and out of each MPPT. That might explain the daytime higher load numbers. If that’s the case then, I should be able to see that in my summer graphs where I have a lot of power generation. Well, I pulled up a day in the summer when the power being generated was close to 1kw per hour, so it should have a few percent of conversion loss, but nope, it didn’t have this, so I don’t think the DC load is including any MPPT consumption, it would go up with increased power flow, and I am not seeing that.
From what I understand from our discussion, the DC load just has rounding errors that are more noticeable at low power levels, so I can live with it, or if I want to get more accurate numbers for my DC load, I can add a shunt to measure my DC system. I was hoping maybe there was something simple in the software that might be causing extra consumption to be logged during the day, like some code ignoring negative values or doing an ABS () on them, but it’s not critical.
Here’s a better example of what I’m trying to point out. My system has a steady 3 watt DC load (RaspPi / Tablet). The 24 history shown in the VRM DC Power box has a line graph that shows this steady 3 watt load overnight, but during the daytime it’s jiggles around due to the various rounding errors and synchronized data sampling times between devices. That all makes sense, and it creates a graph of DC Power usage for what seems to track what is getting displayed throughout the day. But, the consumption bar graph doesn’t match up with this graph, it shows the opposite in fact, more consumption during the day than night. You can look at the DC Power graph and see the area under the power usage is way less during the daytime. So, the two graphs have a conflict. How the bar graph is accumulating consumption is where there must be something wrong, it much be sampling the DC Load and taking larger numbers than it should, because the two graphs don’t agree with each other, and theoretically, they are both based off the same data.
