diy solar

diy solar

Sunny Boy and Sunny Island

If you're using Sunny Island, one option for the relays is based on surplus power available.

I am, but my goal is to set up an independent system, possibly Victron-based. And I’d really prefer to have a completely standalone solution based on a simple smart plug with a relay, no long wires, dedicated circuit, or communication needed. I could use it to power a water distillation unit, space heater, little window AC, or maybe my kid’s lava lamp. The appliance in question would just have to be dumb (or smart) enough to resume operation upon application of power.
 
I'm most interested in variable loads, that could PID control to keep frequency at a target.
I though of VFD pump, although my non-PF VDF upsets Sunny Boy.
Dimmer circuits could make resistive elements variable, but also generate harmonics.
Motorized variac would be smooth.

Any microcontroller could sample to determine frequency, or an analog circuit might accomplish it.

Wish we could just buy something.

SMA use to sell a smart load, just a Sunny Boy operating in reverse.

I think there would be a larger market for zero export at CT by varying load.
 
I'm most interested in variable loads, that could PID control to keep frequency at a target.
That gives me an idea… Would be very cool if I could get my inverter based AC/heat pumps to ramp up and down with frequency. I only have the ability to influence them with a thermostat setting, and I could probably come up with a way to use Home Assistant or Node Red to modify the thermostat setting based on excess production. Tuning that PID would be tricky, since there’s a lot of latency in the frequency measurements, and also lag as the compressor ramps up and down. Passing clouds would be confounding.

Any microcontroller could sample to determine frequency, or an analog circuit might accomplish it.

Wish we could just buy something.
Same. I suppose there just isn’t a big enough market to justify developing such things. With a bit of effort, I’m pretty sure I could make something cheaply that does what I want, but bringing it to such a niche market (including UL or ETL cert) probably impossible to do profitably. I’ll keep looking, maybe there’s something out there I just haven’t found yet.

I think there would be a larger market for zero export at CT by varying load.
Which is a totally different paradigm, but the exact same problem, right? Aren’t all the zero export solutions proprietary, though?
 
I've figure tweaking thermostat setting would work. Have to determine response rate of the A/C. Frequency measurement could be made in a second or in as little as 16 milliseconds (but better to average multiple readings.) Frequency shift takes several seconds. I think a very slow PID can work. That, you can get in a digital chip. I've done one faster and one very slow with op-amps. Much easer to tune if P, I, D are separate circuits later summed, vs. single op-amp with network of R and C to program it (all interacting with each other.)

Ramp thermostat up and down very slowly, aiming for 60.05 Hz (rule 21) or 61.1 Hz (island) depending on your Sunny Boy model.
Clouds can knock it for a loop, but it would gradually bring system back to the point of just starting to curtail production.

For zero export, given a system allowed to export but with paltry backfeed credit (California NEM 3.0, except during certain peak hours), you could vary loads trying to keep CT near zero. If system is already zero export (proprietary), vary loads to keep it at 100W import.
 
In the UK we have CT controlled diverters that send any surplus electricity to resistive loads ie a water heater.

This is a DIY build example, its no longer available in UK but is now sold from France. As its designed for 230V 50htz it would not be directly usable in the US but the design could no doubt be adapted.


Water heaters in the UK are max 3kw but the Triac can handle 6kw and a second output to another load can be added.

It samples the main incoming line 50 times a sec ie to match the 50htz and then calcs how much to divert, this can and does induce flickering in fluorescent lights.

A second option is to use the smart energy function within the Solar-log made in Germany. A more expensive and less efficient solution as you need a compatible household meter, some obsolete Wemo wifi plugs and the Solar-log. But once setup you can set rules so that at certain export thresholds the solar-log can turn on Wemo plugs and then either keep that load on a for a set time or turn it off once the excess drops.
 

Attachments

  • Screenshot (1).png
    Screenshot (1).png
    261.8 KB · Views: 2
Triac dimmer circuit for the load, could flicker as you say.

Here's an interesting product (I think this is the one, can't find English documentation)



It performs PV MPPT into a resistive load by means of PWM, no inductor required (presumably has capacitor bank.)
That would be MOSFET to switch off while current flowing.

Something like that, switched at kHz or high kHz frequency, should reduce flicker. Constant pulse with PWM should perform PFC, because current draw will be proportional to voltage. Probably operate off rectified AC to keep it simple. Inductor and diode to smooth current between pulses.

CT at meter + voltage is what lets MK2PV target zero export.
For off-grid use, the high wattage 50/60 Hz dimmer switch would make even worse flicker, so filtered PFC would be preferred.

What I would like is one that implements frequency-watts, grabbing surplus power before frequency-shift of AC coupling curtails PV production. SMA used to sell Smart Load, which did that.

 
There is an Anti Flicker version of the MK2 software but it needs calibration of the hardware to match the Utility meter, the plain version of the software works out the box. I see the flicker in the workshop but then that shows there is diverting going on of 1kw or more so I stick to the plain version. Its on grid only so would not work off grid unless the program was converted to monitor frequency to control the diversion, it was looked at early on but not followed up.
 
Found this on Github, a Node Red flow for on and off grid diversion.


For off Grid its looking at SOC from an MQTT database but if that database includes frequency then you can change the Node Red flow to suit.

There are implemented two possible ways of automating the diverter. The upper one uses the battery state (bulk/absorption/float) to dinamically set the diverting setpoint, but you could replace it with a SoC percentage logic. If you set a very high setpoint, will behave the same as it was turn off. The last one turns on or off the system based on SoC.
So use a Raspberry Pi running Node Red and https://github.com/pkwagner/yasdi2mqtt to get the Freq from the SI and send excess load to a water heater based on SI frequency.
 
Back
Top