question

Any Ideas?

I'm using the Eastron SDM630 via 3rd party code for VenusOS. Works, but I have the same issue: Controlling the consumption to 0W is fairly tricky... I'm quite sure, if measurement frequency would be higher, the system could shrink the amount of feed-in power.

If consumption goes up&down, I easy have a few kWh per day feed-in....

Is there a solution to this?

In my opinion (and others) this is not a "meter" data-speed issue. From other threads on this subject it would appear that these overshoot power swings are caused by the speed of response of the inverter to power demand changes. The limit is said to be ~400W/sec, maybe Victron could interject here...? So, the scenario when a large AC load increases demand the inverter ramps up to meet the load requirement, the load is then switched off but the inverter at that moment is still producing power, the easiest route for the excess is back to the grid, despite grid set-point, while the inverter ramps down again.

I would think that there has to be latency in the response to ensure system stability and that the quicker reporting intervals only result in a marginal time saving overall.

Wow! Thank you for your responses!

I had the beforementioned MQTT setup with an ABB Meter with 5Hz measurement frequency running for 2 months and it hat around 40% less feed-in- & net-power "losses" than the "original" EM540 solution running at 1Hz.

I will adapt my MQTT solution to the EM540 which has double the measurement frequency of the ABB meter and see what the "losses" of this solution are.

As some of you have already pointed out, the benefit will be limited to what the inverters are (allowed?) to do. I'd be happy if victron could clarify the limitations and reasoning behind these :)

Here is some data on the raw performance of different smartmeters:

CG ET340:

ABB B23

CG EM540
All graphs show a 10 second recording of measured data. Unfortunately the first graph (ET340) did not have the filtering of equal data-points enabled, but the "steps" are clearly visible.

The other 2 measurements do have the filtering of equal data-points enabled. The "noise" of the last measurement (EM540) is due to charging my EV while recording the measurement.

When feeding the MQTT data into ve.bus the whole system is more responsive and works more accurate.

Why does victron not utilize the improved new smartmeters?

This is my grid chart for the last 12 hours using a Multi 48/3000 GX, ET112 and ESS with a grid set-point of 30W. Most people are concerned about the overshoots sending valuable production to the grid for a pittance. You can see on my chart the odd swing to the grid caused by kettle/microwave and maybe sandwich toaster thrown in there. In my case I am reasonably happy with the sloppy performance in the interest of stability but some may require razor-sharp performance so, I am certainly interested in your results, but others may require it. (I should also say that at the moment I am limiting the inverter output to 1kW, so sort of using the system more as "power-assist").
Could you please post a similar chart of your grid performance with respect to grid set-point ?

In Venus OS v3.00~4 soll das ESS schnellere Regeln haben. Besonders wohl auch beim EM540.

Das kannst du ausprobieren und sehen, ob es besser wird.

https://community.victronenergy.com/questions/164781/venus-os-v3004-available-for-testing.html