Heater element matched to solar panel output = diode string

[upnorthandpersonal]

If you have 8kW array, rather just a panel or two, would it not be more useful to connect them to an inverter to power the house and then use an AC based immersion diverter? Surely an 8kW array would generate more hot water than anyone could usefully use - unless you are talking of swimming pools or hot tubs?

Hot water tanks n this case are several thousand liters. They're buffers hooked up to e.g. wood burners, and they have extra electric heating elements. These buffers are both for domestic hot water as well as the floor heating.

When adding them directly to the house, you need a grid tie inverter etc. This is just a cable running from the panels to the heating element without the cost of the inverter and need to comply with the installation guidelines of a grid tie system.

 

[SeaGal]

Increasing the inverter size to cope with extra heating (which is probably the biggest load) has several disadvantages
1. You might not be allowed to connect the larger inverter to the grid
2. Having a larger inverter used with a smaller load means you're operating in a very inefficient region of operation
3. You can be looking at over 100W of parasitic load on the larger inverters, just to operate maybe a 3W router and a 20W fridge
4. it costs more

That makes sense... In my case, however, I have the inverter to power the house anyway, so it's just excess that I use to heat water using an AC-based immersion diverter.

 

[Hedges]

Good thinking...But how do you switch DC without killing the switchgear due to arcing?

That can be a problem.

If you have 2 elements in series and short one out or open switch to include it, current flows through resistor so voltage won't jump too high.

I think some SSR are MOSFET, should work for DC. (SCR don't open until a zero crossing.)
Thermostat and over-temperature switch ought to work for pilot duty, controlling a suitably rated contactor. But safety switch I do want something failsafe. Be sure to have temperature/pressure relief valve.


Contacts and transistors could be protected with RC snubber. Most important switching inductive loads.

There is an MPPT design for heating elements. It doesn't need inductors, just has a capacitor on PV side, uses PWM into resistor to vary power.

 

[Warpspeed]

There is an MPPT design for heating elements. It doesn't need inductors, just has a capacitor on PV side, uses PWM into resistor to vary power.

That is the simplest and most efficient way to do it.

Connect a relatively large capacitor across the solar voltage source (thousands to tens of thousands of uF recommended).
That will be charged continuously by current from the solar panels.
Allow the combined voltage to rise above the rated maximum power panel voltage (MPP shown on the solar panel ratings plate).
When the upper threshold voltage is reached, connect both the combined solar source and capacitor to the heating element by switching on a mosfet.

The heating element will then discharge the capacitor, pulling the solar voltage down.
Allow the voltage to fall to a lower voltage below the rated peak power voltage.
Turn the mosfet off, allowing the voltage to again rise.

The voltage across both solar panels and capacitor ramps up and down between two set voltage thresholds.
The system cycles or oscillates at a relatively low frequency. In effect very slow motion pulse width modulation.
These voltage thresholds are set to be either side of the maximum rated power voltage, thus always maintaining the solar voltage fairly close to the maximum peak power point from twilight to full sun.

The heating element resistance should be kept "reasonable" but is not absolutely critical.
It will discharge the capacitor pretty much regardless of the actual resistance.
Power transfer efficiency will remain high over a very wide range of solar, and a reasonable range heating element resistance.

In gloomy conditions, the cycling rate will become very slow, but it will still pulse power into the heating element without overloading the solar panels.

Some kind of transient voltage suppression across the mosfet will be required, quite a few different approaches to that.
An avalanche rated mosfet will work very well (at a low enough repetition rate) so will a fast diode back to the solar capacitor, or even a simple snubber.
If the controller is located fairly close to the heating element there should not be much series inductance there to cause a problem.

If the capacitor is made large, the system will cycle slow enough (in the order of seconds) that any switching losses will be negligible.
While the system is cycling on and off, the capacitor charging current from solar remains constant.
Quite a few of these have been successfully built, and they work amazingly well.

*hint* A standard garden variety 555 timer chip contains two voltage comparators connected to the set and reset inputs of a flip flop.
This can be run in normal astable mode, using the high voltage solar capacitor, and two separate adjustable voltage dividers.
Output of the 555 will need to be inverted to drive the mosfet gate.

 

[Warpspeed]

Typical commercial MPPT charge controllers recharging a battery, will seek and hold the peak power voltage during bulk recharging of the battery.

Once the controller goes into absorb or float mode, loading on the solar panels is throttled back, and panel voltage is allowed to rise up towards full open circuit voltage with a decreasing battery charging load.

Now one of these capacitor ramping hot water heaters COULD be set up to only operate at voltages a bit higher than the main MPPT battery charging controller ever uses running flat out bulk charging.
So hot water heating power is off, until the main MPPT controller starts shedding load and the solar voltage begins to rise above the normal MPPT controller voltage operating range.
It will require a bit of very careful tweaking of voltage thresholds, but should be workable.

Its going to be less efficient than tuning the hot water ramping voltage controller range for full maximum power transfer, but with a bit of careful adjustment it might be able to use any excess solar for water heating once the battery is full ?

I have not actually tried this myself yet with a commercial software MPPT controller, so its just speculative.
Although it definitely works very well with my own constant solar voltage hardware MPPT controller.

 

[Daddy Tanuki]

so I whimped out and went the AC route. i have a 1700 liter tank (purchased) on standby (insulated) and a 5kw signeer inverter I bought last year on a whim. with 70kWh of storage the plan is to use a timer to turn on the elements from 1000-1800 dependent upon the battery voltage. will take some tinkering and adjustment to find the correct balance. if i did not have the inverter already sitting on a shelf doing nothing I would have tried some of these other setups. but the signeer is large enough that it can run a 4500 watt immersion heater, or two 2500 watt units with all of the local controls that come with a water heater.. yes I am like a golden retriever... Squirrel! but it seems to me that @SeaGal is on too something here. ac controls are common place and easy. a cheap inverter like the 5kw signeer is a shoe in for water heating. its cheap, has TOU protocols and all you need after that is a relay that measures battery voltage. do not turn on if below xxx and you are set. still sussing this out, but I have a set of elements inbound (actually several different ones for testing) along with bulkhead fittings that match the elements thread in size. I have 8 months from today to figure this out.. wish me luck.

 

[SeaGal]

... it seems to me that @SeaGal is on too something here. ac controls are common place and easy. a cheap inverter like the 5kw signeer is a shoe in for water heating. its cheap, has TOU protocols and all you need after that is a relay that measures battery voltage.

I'm not doing anything that novel... I'm using a Triac triggered in burst mode to switch between 1 and 10 (full-wave) AC cycles every 10, to delivery a controllable 0 to 3000W power to the immersion heater, in 300W steps. It's a self designed ESP32-based implementation that reads the house usage / export and battery SOC every 5s or so to decide how much heating to divert to the immersion.

A bit more detail including a link to the AC-side of the hardware is in my posting here...

Nuts and bolts of PV diverters

I have a 5.5KW (3.68KVA exported limited) SE setup with a Puredrive 10.2KWh ac coupled battery. This is my first year of operation and through the spring/summer, I've just left my 3KW immersion heater turned on all the time, cycling on the thermostat coupled to the heater. So far, this has...

diysolarforum.com

 

[Daddy Tanuki]

I'm not doing anything that novel... I'm using a Triac triggered in burst mode to switch between 1 and 10 (full-wave) AC cycles every 10, to delivery a controllable 0 to 3000W power to the immersion heater, in 300W steps. It's a self designed ESP32-based implementation that reads the house usage / export and battery SOC every 5s or so to decide how much heating to divert to the immersion.

A bit more detail including a link to the AC-side of the hardware is in my posting here...

Nuts and bolts of PV diverters

I have a 5.5KW (3.68KVA exported limited) SE setup with a Puredrive 10.2KWh ac coupled battery. This is my first year of operation and through the spring/summer, I've just left my 3KW immersion heater turned on all the time, cycling on the thermostat coupled to the heater. So far, this has...

diysolarforum.com

understood.. I am just not that smart in the IOT realm.

 

[efficientPV]

Those diode videos are of the wouldn't it be nice variety and are in no way practical. I do wish they would get across the idea that going into a pure resistance in direct connect is very inefficient for solar panels. 12V panels going into a PWM charge controller isn't that bad for loss of power because the panels can never go below battery voltage. The diode heaters do much the same because the panels can't go below the forward voltage of the diode string.

Sensing battery voltage and powering a relay driven heater is not a very good way to divert power. Battery voltage just doesn't tell you much about the state of an active battery. It is better than nothing, the motto of the solar world. It makes sense to those here because they don't know any better. It is easy just like direct connect and it doesn't take any knowledge. In a battery system with inverter it takes an extra KWH of battery to operate. Nobody thinks about that. Proportional systems are much better. Here is diversion done proportionately for a half hour period catching any short burst of energy when a cloud passes or load disappears.

 

[Warpspeed]

Have to agree, the problems of diverting power from the inverter ac output, and when to do it, and how much power, are all serious issues that need to be very carefully thought through.

That is why I started looking at what happens to the solar voltage right at the panels once the battery is fully charged, and excess solar then becomes available.

Its not a popular approach, but it gets around all the new problems introduced by loading up the inverter, while still trying to keep the battery at full charge.

 

[Solarstefan]

I tried this with success some years ago using a shunt DC to DC inverter and just adjusting the PWM for 14.4V or float of 13.8V. Once the battery voltage dropped the shunt went into standby.
Shunting all excess solar power to a hotwater tank heater. I used around 100VDC out as this was better suited to a 1800Watt resistor heater , less initial current. It was a cheap useless squarewave 110VAC 1000W inverter.
Just removed the output FETS and connected the heater to the filter caps. The DC sense feedback was already in place including a pot.

 

[Daddy Tanuki]

Have to agree, the problems of diverting power from the inverter ac output, and when to do it, and how much power, are all serious issues that need to be very carefully thought through.

That is why I started looking at what happens to the solar voltage right at the panels once the battery is fully charged, and excess solar then becomes available.

Its not a popular approach, but it gets around all the new problems introduced by loading up the inverter, while still trying to keep the battery at full charge.

thata is kind of my thought process. allow water heating only when the battery banks is above 3.4 per cell.

i bought a signeer a couple of years ago from ebay, the seller claimed it was a 240 volt unit when in reality it was a European style 230 that could be adjusted to 220 or 240 but it was not split phase. by the time it arrived here in japan the period of filing a complaint was over and I had not even received it. by the time I opened the box, read the manual I was like it sucks to be me... so been looking for a use for it for a couple of years. I ran it through a large transformer for a while just to test it out and it seems to be reasonably solid piece of kit.

The good thing is its got a TOU setup in it where you can tell it no output from xx:xx to xx:xx time. so set it to run only from 1100 to 1700 and use a voltage sensing relay to control the water heating lelements themselves. as long as the batteries are at 3.4/cell or higher. might have to dink with that setting a bit to find the sweet spot, but my battery bank is large enough that even if I set it to heat only from 3.375 or so I would be fine. it is something I am going to try just for the fun of it.

 

[Warpspeed]

I rather like SeaGal's idea of cycle skipping the heating element.
Perhaps you could arrange that to just maintain your desired battery voltage ?

 

[orangezero]

Mismatch gets pretty awful when conditions get worse. If light levels are 10% of the rated STC conditions the direct resistive connection needs 10 times more panels than MPPT matched panel. 1kW vs 10kW of panels might sting your wallet even with the current prices.

I guess you could also match the resistive load to bad weather conditions and take the hit on good weather performance... could make sense in some cases if you have no use for the excess solar energy during nice weather.

Am I not wrong in thinking "when conditions get worse" almost nothing you can do will get you much heat? Doesn't that account for a very small percentage of total heat production?

It sounds to me like most of the time 1000w of panels will give 600-900w of solar. Those bad times you are talking about would have the directly connected resistive load apply 20w of solar, while the mppt system can do much better at 200w. But when you look back at the day, the 180w of difference for those times isn't going to matter nearly as much as the 600-900w of heating you are getting in sunlight from either system.

You wouldn't be adding panels to make up for the worst of the solar production. You'd be only adding panels to account for the total difference between mppt and direct, which is only 180w at certain points of the day, nowhere close to 10x as many panels. Right?

 

[Warpspeed]

Am I not wrong in thinking "when conditions get worse" almost nothing you can do will get you much heat? Doesn't that account for a very small percentage of total heat production?

Its like building a car without a gearbox !

You can gear that car any way you like, but it still has only one gear. What is the best gear ratio ?
You can build it so its like about second gear, a bit slow to start off, and it will be howling at normal road speed.
Or maybe gear it so it like third gear. Even worse to start off, and still screaming on the freeway.

But, starting off is only a very small proportion of total running time, so you don't really need anything other than top gear right ?
Perfect !

Matching a heating element to solar panels is a bit like that.
It will work well only at one selected operating point somewhere between twilight and full summer mid day sun.
At any other conditions it will be operating way below optimum.
Or put less kindly, be bloody hopeless.

And yes, an MPPT system will behave like an automatic gearbox, shifting the operating point to suit the prevailing conditions.
This is not a small detail !!!!!!!!!!
Its the difference between success and a failed project.

 

[MattiFin]

You wouldn't be adding panels to make up for the worst of the solar production. You'd be only adding panels to account for the total difference between mppt and direct, which is only 180w at certain points of the day, nowhere close to 10x as many panels. Right?

Probably depends a lot on location. I can imagine in southern california or baja california you would get pretty good peak production and cloudy days are lot less common.
UK or Finland then on the other hand has 6 crappy days and one good day. Additionally in summer you get daylight 24/7 but the sun is at optimal angle only for couple of hours, assuming it is not cloudy at that time.

 

[efficientPV]

"the 180w of difference for those times isn't going to matter"

That is what most think. That 180W of extra power is what heats almost all of my water.

 

[Hedges]

With two heating elements, you could manually switch between single, two in parallel, two in series.
With two different values you could have 4 levels of power.
That should help optimize for different times of year.

 

[efficientPV]

With two heating elements, you could manually switch between single, two in parallel, two in series.
With two different values you could have 4 levels of power.
That should help optimize for different times of year.

That sounds like a good idea, but it isn't. Switching in an element usually drops the power in half. It isn't seasonal, but many times a day it is needed. Manual switching just isn't practical. Having a timer for the two peak hours at noon will see a slight improvement. Measuring current and switching is a much better way to do it. Nothing is going to work unless it is attached to a Buy It Now button. One current option is

AC7391 video

 

[Warpspeed]

If you do the math, you will find that if usable solar output power for a single string of panels varies from say 50 watts to 1,000 watts, the load resistance also needs to seamlessly adjust over a similar 20:1 range to maintain a constant optimum panel voltage.

You can do your element (or panel) switching steps to cover maybe a 2:1 or 4:1 switched load range, but it would be very difficult for a load to reach even half of the theoretical available power over most of that 20:1 range.
The loss in power transfer efficiency is going to be very significant.