Scratch built solar PV divertor

[Radian]

4 hours ago, Radian said:

I've got a theory about that... 🙊

 

[MikeSharp01]

On 20/05/2022 at 22:00, SteamyTea said:

'friends all over the world, all over the world.  None in this country, but all over the world'

Did the VHF test and got my call sign - which I still have, but not the Morse test did a load of moon bounce and had a satellite tracking 4 Yagi setup on my mums house - still have all the rigs and now have an HF set as you now get all bands without the Morse. Don't turn it on much - used be a regular on the crystal palace repeater - not a squeeky and used to be pretty derisive of CB types (ooooops probably offended millions of Buildhubers.)  Also still have a box of reel to reel recordings of Hancocks half hour I made in the 70s - I have checked I don;t have any that were lost!

[SteamyTea]

2 minutes ago, MikeSharp01 said:

but not the Morse test

My Father was an observer/navigator/wireless operator in Mosquitoes during WW2.  He never lost the ability to use Morse.  Guess after 6 years of it being very important it kind of sinks in.  He was in the Reserves till the early 1970s.

[Radian]

A little update. The Power monitoring section is now complete and sampling away nicely. Actual 240VAC and current is being captured at around 150 samples per cycle and looks like this (as downloaded from the webserver running on the sampling ESP32):

 

 

The tips of my AC really do look blunted like that as confirmed by x100 probing direct with a scope. When I was looking at the output of a transformer I took it the distortion was due to the saturation of the primary but no, the voltage waveform is really like that. Who knew?

 

Anyway, multiplying all the VI sample pairs over each cycle and scaling their sum with a suitable calibration value is giving me a power reading that agrees very well with both my utility supplied meter and a PZEM-004T V3 Energy Monitor I installed next to the ESP32 as a second opinion on the power measurments I'm making. Only unlike those meters, mine gives me the 'holy grail' of -ve power when pushing current into the grid (the utility meter just clips at zero and the PZEM always shows a positive value even when actually negative))

 

So the next part of the process is to multiply the power for a complete cycle by the cycle period (20ms) to get the quantity of Joules going in or out of the meter. The trick to dumping only the energy that would go back to the grid is to let out up to a maximum of 3600J then keep pulling it back in (into the dump load) if necessary. 3600J is supposed to be the threshold at which the meter records a transaction so can be used as a buffer to give an opportunity for real  loads to consume the excess but if there are no such loads then the signal to send power to the dump load is sent.

 

Normally the signal would be hard-wired to a Triac or SSR to switch on an immersion heater but I have a different plan. What I want is to be able to turn on multiple loads anywhere on the house wiring when the surplus is available. Then individual appliances can decide for themselves if they could use some power. A wireless solution is ideal so I started experimenting with MQTT. Unfortunately there seems to be too much latency between signalling that excess is available and acting on the message. Ideally the response should happen in the next cycle (20ms max. later). I also tried websockets but the complexity of using the LAN was beginning to seem like a bad idea.

 

As I mentioned earlier, I had thought about using mains signalling like that used in X10 devices. So now I'm building the hardware to 'tag' each mains cycle for consumption when required. X10 protocol is too slow to use house codes and other niceties so I'm thinking of just putting a burst at the zero crossing and qualifying it with a non-standard burst width in addition to a belt & braces check that we actually want energy diversion using MQTT messaging as a 'gate'. I've already built the 120kHz receiver and it's all quiet in this band on the mains. I guess nobody uses X10 these days anyway!

[Onoff]

What happens to these systems if you get hit by a bus? Are you going to write a manual for the muggles left behind that they can understand?

[SteamyTea]

1 minute ago, Onoff said:

Are you going to write a manual for the muggles left behind that they can understand

Still trying to work out what is going on.

[Onoff]

8 minutes ago, SteamyTea said:

Still trying to work out what is going on.

 

This thread or life in general? 

[SteamyTea]

Just now, Onoff said:

 

This thread or life in general? 

Yes, or No, not sure.

[MikeSharp01]

On 20/05/2022 at 15:08, Radian said:

Does this advance humanity or not I wonder?

Yes because it encourages development, by focusing our natural curiosity, in other directions leaving these roads un-travelled but among the known knowns so that we when we all have an automatic heart re-starter we can start playing again on this, now dusty, road.

[SteamyTea]

8 hours ago, Radian said:

What I want is to be able to turn on multiple loads anywhere on the house wiring when the surplus is available. Then individual appliances can decide for themselves if they could use some power.

Isn't that what CAN bus does.

Send out a signal that power is available, then devices take and use it, if they need it.

Edited May 27, 2022 by SteamyTea

[ProDave]

Looking good so far.

 

What I concluded (and is general knowledge any way) is most electricity meters work on an energy bucket of 1Wh.  So the meter will only increment it's counter when 1Wh of energy has passed.  A typical 3kW immersion heater will take just a little under 1 second to consume 1Wh.

 

So on that basis I did all my timing on a half second period.  The immersion heater is burst fired within that half second anything from 0% to 100% of the period to match the surplus power. And thus within that half second the same amount of energy passes both ways and the energy bucket never reaches 1Wh so never counts up.

 

Where it all gets interesting and will test your measuring and calculating algorithms is when you start firing the immersion heater, as the power that consumes will show up on your measurement.

 

The bit I never perfected is accounting for that.  In theory you should be able to measure the surplus power and calculate exactly how much goes to the immersion heater and apply it.  That never worked for me.  so instead mine works on incrementing / decrementing the immersion heater power level each cycle.  that results in a slower response to changing loads or changing generation.  On the whole it works well but no doubt causes a small amount of over / under immersion heater power when a rapid change happens.

 

If you are sampling every cycle individually, you probably then need to average 25 cycles consecutively to give you 1/2 second average and work on that, but this will introduce the time lag I have.  Perhaps a way to improve that would be rather than average over 25 seconds then do a calculation, do a rolling average so as one new reading for a cycle comes up add that to the average and the oldest one drops off, that would still give you an average over half a second but updated every cycle.

 

[Radian]

2 hours ago, Onoff said:

What happens to these systems if you get hit by a bus? Are you going to write a manual for the muggles left behind that they can understand?

This is a good reason to base the thing on a microcontroller like the ESP's rather than an embedded OS running on an SBC like Raspberry Pi. Once the code has been flashed in there's no ongoing maintenance required. A box with a flying lead is plugged into a 13A socket near the meter and another interrupts the cable going to the immersion. Simple as.

 

4 minutes ago, ProDave said:

So on that basis I did all my timing on a half second period. 

I think that's what make things a bit more difficult. If the energy content of the meter buffer (bucket) is tracked on a cycle by cycle basis then at the end of any given cycle a decision can be made whether to start emptying it by applying the dump load. This is why I'm striving for low latency.

 

Having energised the dump load, the subsequent cycle measurement(s) will reflect the effect that both the dump load and natural house loads are having and the buffer can be allowed to drain to some minimum threshold before letting it refill by removing the dump load. I can visualise this and code for it but as soon as I decouple it from the cycle by cycle basis and respond in other time bases I find it gets very confusing!

[MikeSharp01]

10 hours ago, Radian said:

A little update. The Power monitoring section is now complete and sampling away nicely. Actual 240VAC and current is being captured at around 150 samples per cycle and looks like this (as downloaded from the webserver running on the sampling ESP32):

 

I think I am still not clear why your current trace is the shape it is in terms of nuances (get the switched mode thing I think) EG the tiny burble just right of the current peak on the down slope. I can see that you are looking to handle things cycle by cycle and I am wondering given that the next cycle can be either import or export, perhaps randomly - might be fun to find the likelihood of each type of cycle following on under differing circumstances (just a few), just how quickly a load needs to respond, at both ends of the cycle, to be sure you do not end up in: a) a fight between appliances to use power they thought was there (they were told so) but isn't because another load got there first - too many pile in and b) a ringing loop that actually causes loads to switch on/off exactly out of synch with when you want them to. The single dump load probably solves the first, assuming you can overcome the issues @ProDave highlights, but how much of a potential export cycle will you get assuming you can send the message once you are sure the cycle would otherwise be an export cycle and then not leave it on for the next cycle until you are sure this is going to be an export cycle ad nauseam. All that while keeping track of the 3600 J  / 1 Wh point. Fascinating exercise.

 

PS I think this crossed with your post - I think I get it now - perhaps.

 

Edited May 27, 2022 by MikeSharp01
Post crossed

[Radian]

4 minutes ago, MikeSharp01 said:

I think I am still not clear why your current trace is the shape it is in terms of nuances (get the switched mode thing I think) EG the tiny burble just right of the current peak on the down slope.

 

It's quite a messy picture and changes a lot throughout the day. The central peak is clearly the recharge of numerous capacitors via bridge rectifiers around the house. I think the current to the right of the peak will be from the more inductive loads like the fridge freezers and fans. When I put on a big resistive load like oven and kettle combined, its a bit more like you would expect. I just captured this now and only one freezer is on:

 

 

Doing this, I have just spotted the small phase shift between current transformer and voltage conversions. Expecting the current to lag I thought I'd got the sampling sequence the other way around to compensate but I need to check that.

[joe90]

2 hours ago, Onoff said:

What happens to these systems if you get hit by a bus? Are you going to write a manual for the muggles left behind that they can understand?

This is exactly why our Jeremy scrapped his weather compensating heating control system that he designed, no one else would understand it if he was not around (or got too old to remember) he adopted my Luddite approach to most things, a single room stat 🤷‍♂️.

[ProDave]

21 minutes ago, joe90 said:

This is exactly why our Jeremy scrapped his weather compensating heating control system that he designed, no one else would understand it if he was not around (or got too old to remember) he adopted my Luddite approach to most things, a single room stat 🤷‍♂️.

I have started writing a "home manual" documenting everything in the house so that when I am gone, it should at least help someone else understand some things.  That has gone as far as even listing the code for my PV diverter, but I suspect if that stopped working it would be binned or replaced.  Given my daughters complete lack of interest or knowledge in energy efficiency etc (she shrugs her shoulders and says so what when I point out she has been out all day and her tv was left turned on)  if she is the one that has to deal with it, it would not get replaced.

 

Really surprising that teenagers growing up now have no comprehension of using energy efficiently.  Perhaps you only "develop" that interest when YOU are presented with the bills to pay from your income?

[joe90]

47 minutes ago, ProDave said:

Really surprising that teenagers growing up now have no comprehension of using energy efficiently.  Perhaps you only "develop" that interest when YOU are presented with the bills to pay from your income?

+1, and not only teenagers 😱

[Ajn]

28 minutes ago, ProDave said:

Perhaps you only "develop" that interest when YOU are presented with the bills to pay from your income?

"Perhaps" is a good choice. Part of our deal with remaining son and his wife was they use our assests in the combined build but they pay all bills and services into the future. However I have an ingrained interest in the "where does it go" but theirs is to the extent of not wasting stuff  but not the detail. He has no interest in my gadget crap so I will not be writing any manuals so I have just done an overlay on the existing and turned it down so my system has priority. So if anything happens to me and my plan B system fails then they switch it off and revert to plan A. All the instructions are already written for those and it's not my fault if they have lost them I did put all that stuff in a nice folder. Not that I would care by that time anyway.

 

That said their nearly seven year old is showing an interest in the science stuff so there maybe hope yet.

[Onoff]

1 hour ago, ProDave said:

Perhaps you only "develop" that interest when YOU are presented with the bills to pay from your income?

 

I get a WhatsApp message from upstairs to "advance the hot water".

 

First thing I do when I walk in at night is turn any advance settings off and turn the room stat down to something near normal.

[SteamyTea]

7 minutes ago, Onoff said:

turn the room stat

What's that then?

Put everything on a timer. Midnight to 7AM.

They will learn to live with it, 7,000,000 people have.

[ProDave]

8 minutes ago, Onoff said:

 

I get a WhatsApp message from upstairs to "advance the hot water".

 

First thing I do when I walk in at night is turn any advance settings off and turn the room stat down to something near normal.

We found that when running the B&B, you would go into a room in the morning after the guest had left, to find the room thermostat at 30 degrees,  the boiler burning away to keep that running, and the windows wide open.  To say it "annoyed" me is an understatement.

 

I am convinced daughter showers however long is necessary to empty the tank of hot water before turning the water off.  I simply cannot imagine just what one does in a shower for half an hour, even if you have got the excuse of having long hair and needing 3 lots of shampoo and conditioner with instructions to leave each in for several minutes before rinssing. 

[SteamyTea]

2 minutes ago, ProDave said:

I simply cannot imagine just what one does in a shower for half an hour

I can, but I went to Public School.

[Onoff]

14 minutes ago, ProDave said:

even if you have got the excuse of having long hair

 

Mine did have hair that reached down to the backs of her knees. It a fit of teenage rebellion she cut it off one night, cue tears from SWMBO! 😂

 

Secretly well please as it's reduced the frequency I have to unclog and clean the Geberit wall drain. Both of mine are shortsighted but you would have thought they'd have at least felt the water lapping at their ankles!

[MikeSharp01]

4 hours ago, Ajn said:

remaining

Should I be sorry to here that - if so I am?

[Ajn]

1 hour ago, MikeSharp01 said:

if so I am?

thank you.