SteamyTea Posted July 18, 2022 Share Posted July 18, 2022 3 hours ago, SteamyTea said: Does your meter have 2 LEDs, one for direction and one for quantity? I seem to remember that my neighbours smart meter has 2. Follow up. My old meter has two LEDs. First picture is with the CC OptiSense sensor in place, which I notice has its own red LED that flashes in synch with the meter one, so should be easy to test out my homemade logger. Second us just the two LEDs. Link to comment Share on other sites More sharing options...
TonyT Posted July 18, 2022 Share Posted July 18, 2022 Isn’t one of the Leds a Indra red receiver, I’m sure I’ve seen the DNO with a laptop program them…. Via a clip on led transmitter years ago mind… Link to comment Share on other sites More sharing options...
Radian Posted July 18, 2022 Author Share Posted July 18, 2022 3 minutes ago, TonyT said: Isn’t one of the Leds a Indra red receiver, I’m sure I’ve seen the DNO with a laptop program them…. Via a clip on led transmitter years ago mind… Yes it looks like an IEC 62056-21 optical bi-directional comms port. Link to comment Share on other sites More sharing options...
MikeSharp01 Posted July 18, 2022 Share Posted July 18, 2022 3 hours ago, Radian said: Of course these "revenue grade" meters eschew the likes of CT's and use volt-drop across known resistances to quantify current. I'd do the same were it not for the invasive introduction of a suitable resistor after the main isolating switch in the Consumer Unit. CT's are a pita due to their non-linearity and phase errors. Why not use the wire to the consumer unit as your resistor and measure the voltage drop across a length of that. 1 Link to comment Share on other sites More sharing options...
Radian Posted July 18, 2022 Author Share Posted July 18, 2022 1 minute ago, MikeSharp01 said: Why not use the wire to the consumer unit as your resistor and measure the voltage drop across a length of that. There's a thought. Actually there's a relatively benign way of doing that... the combined TN earth point could be at one end of a Kelvin connection pair and the other at the Neutral in the consumer unit. The same current is flowing in the Neutral (CT clamps work just as well on either conductor). And I'm quite happy referencing my voltages to Neutral as it is. Might have to look into that a bit more. Link to comment Share on other sites More sharing options...
ProDave Posted July 18, 2022 Share Posted July 18, 2022 12 minutes ago, Radian said: There's a thought. Actually there's a relatively benign way of doing that... the combined TN earth point could be at one end of a Kelvin connection pair and the other at the Neutral in the consumer unit. The same current is flowing in the Neutral (CT clamps work just as well on either conductor). And I'm quite happy referencing my voltages to Neutral as it is. Might have to look into that a bit more. But changes in it's low resistance which will likely be dominated by the resistance at junctions, could make calibration subject to change? Link to comment Share on other sites More sharing options...
Radian Posted July 19, 2022 Author Share Posted July 19, 2022 9 hours ago, ProDave said: But changes in it's low resistance which will likely be dominated by the resistance at junctions, could make calibration subject to change? Maybe, although the TN junction ought to be reliable enough and the other Kelvin tap would probably be on the first connection to the Neutral busbar with just the main breaker switch in between. Could always resort to the old "Steel dressmaking pin through the insulation" trick but that would be criminal (well, it would be on the other side of the meter). Link to comment Share on other sites More sharing options...
MikeSharp01 Posted July 19, 2022 Share Posted July 19, 2022 10 hours ago, ProDave said: But changes in it's low resistance which will likely be dominated by the resistance at junctions, could make calibration subject to change? Maybe you could continuously / occasionally recalibrate if there was a portion of the day / fraction of a second when your CT could measure a known load (all dimensions) and then see if the resistance measured current value had moved. Link to comment Share on other sites More sharing options...
Radian Posted July 23, 2022 Author Share Posted July 23, 2022 I've been misinterpreting the meaning of the impulse LED coming on constantly - I thought it was signaling power going back into the grid (Joule bucket overflowed). That's not what it means at all. It simply stops flashing in either on state or off state when it sees no chargeable power consumption through the meter. In fact, if my firmware ever detects it changing state, I know the Joule bucket is effectively empty. This might still be handy for synchronising my model which can only ever 'dead reckon' the level in the Joule bucket based on instantaneous power measurements (that may be a bit off). Link to comment Share on other sites More sharing options...
MikeSharp01 Posted July 23, 2022 Share Posted July 23, 2022 40 minutes ago, Radian said: either on state or off Why would it come on if no chargeable current is flowing surely each complete pulse is 1Wh so turning it on and leaving it on means more than nothing is happening. Unless it comes on at 90% (or some such) of 1Wh expecting it to fill shortly but it does not. So then the falling edge is the full 1Wh and on means almost there - seems daft though. Link to comment Share on other sites More sharing options...
SteamyTea Posted July 23, 2022 Share Posted July 23, 2022 44 minutes ago, Radian said: I've been misinterpreting the meaning of the impulse LED coming on constantly Is that because you can export? I cannot export (got no PV) so I can only have a flashing light, not one cone constantly. Though I can be off constantly when now power is being drawn (well below 3600 J). I am on Mk3 of my meter, basically on a nicer sensor, but with a temp sensor in it as well (so I know how hot the meter box gets). Link to comment Share on other sites More sharing options...
Radian Posted July 23, 2022 Author Share Posted July 23, 2022 2 minutes ago, MikeSharp01 said: Why would it come on if no chargeable current is flowing surely each complete pulse is 1Wh so turning it on and leaving it on means more than nothing is happening. Does seem daft but googling around people ask questions such as "why is my meter LED glowing red even though my main switch is off?" It seems to be more likely to be off than on but I think that's just where it is in the measurement cycle when it detects no power flow. 3 minutes ago, SteamyTea said: Is that because you can export? Yes - If it's really sunny, I can manage a couple of hundred Watts or so past my base load. Again, this is just a temporary lash-up of second-hand PV modules thrown against a S facing fence to test my diverter project. So far 5.1kWh has 'leaked' past since April. Link to comment Share on other sites More sharing options...
ProDave Posted July 23, 2022 Share Posted July 23, 2022 You may have identified the behaviour for your meter, but in respect of the red light I don't think there is any standard. I have seem them solid red when exporting and I have seen them flashing when exporting (but not counting up on the import register) My present meter shows solid red when exporting and the meter reading display now alternates between the reading and rED Meaning Reverse Energy Detected. A scheme devised before solar PV became common to alert the meter reader to the fact someone had been naughty. Link to comment Share on other sites More sharing options...
Radian Posted July 23, 2022 Author Share Posted July 23, 2022 Robin Emley put up this video showing what his meter does: Very helpful for me. I wish I'd gone through all his videos like this before I started 🙄 2 Link to comment Share on other sites More sharing options...
kommando Posted July 23, 2022 Share Posted July 23, 2022 This is Robin's video where is uses an old wheel meter to show how his box works. What it also shows that using calibration changes the wheel can drift forwards or backwards when the diverter is dumping load. 1 Link to comment Share on other sites More sharing options...
Radian Posted July 23, 2022 Author Share Posted July 23, 2022 While it's true that calibration doesn't matter all that much, there is a potential issue when the excess to be diverted is small. A few tens of Watts difference between what the meter sees and what the diverter thinks it sees may mean the Joule bucket is quietly filling without the diverter being aware. If this remains the case for several minutes the meter may overflow and start exporting the few Watts represented by the measurement discrepancy. It will certainly push the equilibrium point one way or the other so the system will continue to rock back and forth, and possibly overflow (in either direction) as a result. Apart from CT non-linearity, phase error and VI calibration, one other source of discrepancy can be the anticreep parameter which Robin includes to 'tidy up' the accumulator for the Joule bucket when testing with no power passing through the CT. This can rob a few Joules when import/export is on a knife-edge. Link to comment Share on other sites More sharing options...
MikeSharp01 Posted July 23, 2022 Share Posted July 23, 2022 So really the energy monitor using the red light is not going to be universally useful but will work in non export situations or we need to add more features / intelligence to get a more universal solution. Link to comment Share on other sites More sharing options...
Radian Posted July 23, 2022 Author Share Posted July 23, 2022 1 hour ago, MikeSharp01 said: So really the energy monitor using the red light is not going to be universally useful but will work in non export situations or we need to add more features / intelligence to get a more universal solution. Oh, I think it's very useful in import-only applications like SteamyTea has. The minimal intervention has a lot going for it and it can be incredibly accurate using interrupt triggered timing to measure the period, and perfectly acceptable accuracy when polled. I'm planning to keep it as part of my energy management (diverter) for the time being as a calibration reference. Link to comment Share on other sites More sharing options...
MikeSharp01 Posted July 23, 2022 Share Posted July 23, 2022 10 minutes ago, Radian said: Oh, I think it's very useful in import-only applications Absolutely - just need to work out how to use it in reverse so as to get accuracy both ways. Simple question with a complex answer - did we already discuss this? Why not get another meter and connect it backwards so it flashes on export! 1 Link to comment Share on other sites More sharing options...
SteamyTea Posted July 23, 2022 Share Posted July 23, 2022 4 minutes ago, MikeSharp01 said: Why not get another meter and connect it backwards so it flashes on export! Probably already got that on the PV side. So should be just a case of conditional counting. How is your ESP32 unit coming on? Link to comment Share on other sites More sharing options...
Radian Posted July 23, 2022 Author Share Posted July 23, 2022 1 minute ago, MikeSharp01 said: Why not get another meter and connect it backwards so it flashes on export! Logically sound I think, but in terms of intervention - rather extreme! Measuring power and flow direction using a CT is really quite simple and inexpensive by comparison. Nearly all the necessary resources are available from an inexpensive CT, ESP32 and a bunch of resistors. It also gives you the luxury of trialling code for different algorithms over-the-air and dynamically tweaking parameters over MQTT. In one of those videos Robin was re-compiling his entire code and re-flashing just to change a few values. Link to comment Share on other sites More sharing options...
MikeSharp01 Posted July 23, 2022 Share Posted July 23, 2022 46 minutes ago, SteamyTea said: How is your ESP32 unit coming on? Very well, just conditioning the photodiode and getting the code working, just can't decide where to split the client / server activities. Might get to share something tomorrow. Link to comment Share on other sites More sharing options...
Radian Posted July 31, 2022 Author Share Posted July 31, 2022 Well, I'm pretty cheesed-off right now. I seem to have lucked-out with my utility supplied smart meter. Unlike the vast majority of meters, mine appears to have a teensy tiny Joule buffer - something less than 450J by the looks of things. To arrive at this conclusion I added a 'live stream' of my cycle-by-cycle plot using websockets. It makes capturing elusive events much easier. Here's a long 'still' from a recent experiment... ..So here I had set things up to measure the size of the Joule bucket (what I've been assuming to be 3600J) by switching off the dump load and letting a couple of hundred PV watts leave the meter and go out into the grid for a few minutes. This is to ensure that the bucket is totally brimmed to overflowing. Then my generation is allowed to fall and power to slowly go from -ve (export) to +ve (import). A close up of the transition... The grey vertical on the left is the last meter impulse tracking my export (yes, the meter impulse LED flashes both for import and export). Then the power (green trace) climbs from -ve (export) to +ve (import). The assumed content of the Joule bucket (blue trace) then falls from the maximum clipped level (3600J assumed) as imported power is draw out from the bucket into the house. But rather than falling to zero, it only gets down 1/8th the way (around 3600/450) before the first import meter pulse arrives and I'm charged for the power. So between me and the supplier (theoretically) being billed for export and import between the two impulses, there's only a zone of around 450J where power may be pushed out and pulled in without charge. This is disappointing as it makes measurement tolerances and response times extremely tight. Link to comment Share on other sites More sharing options...
Radian Posted October 31, 2022 Author Share Posted October 31, 2022 Time for an update. I settled with working within a 450J window and it's not too much of a hassle. The diverter has been in constant operation for 3 months now and has only let 6% of what it has diverted out to the grid. However, last week I installed a new UPS for some computer kit and immediately noticed a problem. The switching supply in the UPS puts a tragic amount of wideband noise on the mains when charging the battery and this was swamping my X10 like 120kHz signalling. I was using this as a low latency signal to "burn" cycles remotely around the house. But as elegant as it was, it's no good if it can't be relied on. Back to the drawing board (I always suspected this might be an issue). I then revisited the use of MQTT but nothing I did could get latency down below a max. of 160ms. At 3kW this is about 500J so the buffer could easily over/underflow and hence possibly cost me unnecessary import. UDP was still too slow with a max. latency of 80ms. Something else was needed. This is when I started looking at RF modules like the Nordic Semis NRF devices. But when researching I found comparisons with a thing called ESP-NOW. Turns out I don't need another RF module as Expressif have implemented a low-latency peer-to-peer communication protocol that can run concurrently with the WiFi stack on the ESP8266 and ESP32. It works by allowing small packets to fly between devices point-to-point. They're not passed through access points or switches and have a latency of a couple of milliseconds as a consequence. Perfect! It also has a broadcast capability so my goal of being able to distribute dump loads around the place is nicely catered for. The packets are only 250 bytes but that's plenty to have a 'burn' flagbit and a frame count (to spot missing packets for a QOS check) as well as a cycle-by-cycle voltage and power measurement to keep the immersion unit updated with the bigger picture. Being point-to-point also means the entire network, along with MQTT etc. can go down but the measurement unit and distributed dump load switches will all stay working. The range isn't great though and I've swapped the ESP32's with ones that can have an external antenna to improve the range. However, even with 5% packet loss, with packets sent at the end of every cycle it would be unlikey for 5 out of 100 to go missing in a row and lose 100ms (300J). Also, it's easy to rebroadcast the packets with a repeater that can extend the range. Somehow ESP-NOW manages to step in and answer all my prayers for the sake of a tiny extra code overhead. 2 Link to comment Share on other sites More sharing options...
SteamyTea Posted November 1, 2022 Share Posted November 1, 2022 9 hours ago, Radian said: ESP-NOW I hope @MikeSharp01 sees this, then he can tell me how to make a better energy monitor. I wish I understood all this communication business better. Link to comment Share on other sites More sharing options...
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now