Testing a simple and cheap only fridge solar system

[lorenzom]

This is a simple offgrid isolated fridge system (not really meant for driving other appliances).
It is also realized with very usual components, just added a custom control board (very simple though)

As it is just a proof of concept, this test build goes a bit extreme :
- 85W solar panel for 12V systems (used but still >90% output)
- old Solsum 8.8X charge controller (used, no mppt, but ok with this solar panel)
- 12V car battery (75Ah, not new but OK): as I never discharge more than 5%, this should last a while
- 600W modified sine inverter (a cheap one of those I repaired)
- tabletop fridge, compresor marked 50W , 2 bottles of 1.5l full of water in the freezer
- DIY control system: cost of components perhaps 10$: PCB, relay, 8pin microcontroller (attiny13), low power transistor, led, 4 resistors


Here is the fridge:


Here is the power source:



And thess are the battery and controllers:



Some mods are needed:
- on this charge controller: take out the regulated 5V to supply the control board, take out charge (sun) led info, and battery status led info. With a more sofisticated controller, like my victon bluesolar you just want to read the serial port
- on the inverter: bring out the 2 on/off switch wires, this will go to the control board relay
- on the fridge: bring the thermostat wires out, they have to be separated from the motor wires as they will be read by the microprocesor.

Some comments about the fridge:
this is a very simple one: Hyundai MN11, 82l internal volume, 50W QD35Y compresor, thermostat is mecanical (no electronics).
- walls are only 4cm thick, and include condensers on the sides! (where the extracted heat is released). Compare this to the 9cm thick insulation of my DC chest fridge, with a separated and ventilated condenser.
- the internal lamp should be changed for leds able to work on battery voltage
- At the beginning the fridge didn't always want to start from this inverter (just if still warm, but not too much!). I had no more powerful inverter, but managed to reduce the inrush current putting a resistor in serie with the starter PTR (equivalent would be to change the PTR from actual 15ohm value to 33ohm). Now starts fine in every condition
Starter (provisional !) mod:




Control board (left), charge controller (right):

The logic implemented in the microcontroller is quite simple:
- always shutdown at night
- during daylight, start the inverter only when the battery reaches full load, and for a limited time
this is the most important feature to protect the battery: in this case you will always cycle from battery full to a limited discharge (no more than 5% in my case)
- of course, shutdown the inverter when the thermostat trips
- you can also extend the limited time if the battery stays at full load, shutdown immediately if the battery charge status drops too much, blink a led to show some info, enable a car socket if there is excess PV...
- as I did it with my DC fridge, the control board is fed from the charge controller, so its supply current has to be very low. Essentially driven by interrupts, this control board draws less than a milliamp.

Some measures, made with ac/dc current clamp:
inverter running with fridge attached draws 6.5Amps from battery @ 12.6V
AC current fridge running : 0.4A total, 0.01A goes to start winding
The solar panel has almost the best fixed orientation for actual season. Charge was measured at best 4.5A (but I'm not there all the day)

Yesterday (almost 12h daylight at this season), the fridge worked during 9x 35min cycles (weather sunny, but some haze) , and generated estimated 1.5l ice (room temperature 22ºC). That means ~= 430Wh energy used (supposing 6.5A @12.6V)
Today, very hazy,then cloudy, it worked no more than 5x 35min cycles (~=240Wh used). With 23ºC room temp, internal was still around 11ºC (good for beer!)

Time counter on top, min-max thermometer inside


Obviously, I would not put fish or meat in this fridge. Autonomy is too short, and cold is lost very quickly.
Repeating, this is more a test system to show how it is posible to run a fridge from PV with few resources.
With a better insulated fridge and more PV power (but limited by the charge controller), I am almost sure you would have a usable system.

We can compare this data with what my chest fridge controller(freecold RSI 150, ~=125l internal volume) recorded from the victron 75/15 attached to 340W panels:
yesterday 280Wh were harvested , today 270Wh, to run nothing more than the chest fridge (I blowed my inverter in a failed security test connecting it to mains )

The bigger chest fridge could almost have been run from the 85W panel with the old SolSum non mppt charge controller!

We can make some math to find out where the difference in efficiency comes from (next post)

 

[Leeds]

Very neat test.

I've always danced around modifying the fridge itself as I've wanted that component to be plug-n-play and end-user replaceable.

But your mods on the starter and choosing to use the built in thermistor instead of dropping in a second one really is a cool approach.

 

[efficientPV]

I have just about the same microprocessor driven system. Just modified mine to have a much longer hot start delay as my battery has become very weak and even a minor charge will raise battery voltage. Also modified my MSW inverter to have a shorter fixed on pulse. That dropped startup current to half and reduces run power. Many MSW inverters actually increase on time when HV is low due to load. This is the attempt to keep average resistance power constant. For motors long on times at startup makes things even worse.

 

[conquistador]

Can you elaborate a bit on "- of course, shutdown the inverter when the thermostat trips" This seems a clever feature, especially as most energy for a fridge would be lost by running the inverter in idle most of the time.
How do you interface the fridge thermostat with the microcontroller and how do manage wake up when the thermostat wants to turn the fridge on again? I understand that the thermostat is then without power once the inverter has been shut off.

 

[lorenzom]

Can you elaborate a bit on "- of course, shutdown the inverter when the thermostat trips" This seems a clever feature, especially as most energy for a fridge would be lost by running the inverter in idle most of the time.
How do you interface the fridge thermostat with the microcontroller and how do manage wake up when the thermostat wants to turn the fridge on again? I understand that the thermostat is then without power once the inverter has been shut off.

Sorry for my late response.
You are totally right about so much energy lost running the inverter in idle most of the time.
It was one of my priorities to tackle this problem.
The initial idea was very simple: as the thermostat is a simple switch, use simply this switch to start/shutdown the inverter. This means you have to:
- modify your fridge, disconnecting the thermostat and bring its 2 wires out (it implied also for me to quit the internal light bulb, because it shared one of the thermostat wires)
- modify your inverter connecting these 2 thermostat wires to the on/off inverter switch (actually also a simple switch)
That would be the simplest solution.

Now, using a microcontroller (Attiny13 in my case), the thermostat drives just a digital IO port (internally pulled up). To command the inverter, just connect a relay output to the inverter switch. I use a 12V DC relay, + connected to battery, and -driven through a transistor from a microcontroller out port.
I was able to find a 5V supply inside the charge controller, which is used for the microcontroller. To limit drastically the additional current draw, it is sleeping most of the time, and just wakes up when there is something to do (upon thermostat, battery status, charging changes, ...)

Hope this will answer your question.

 

[lorenzom]

Sorry for the time past since last post (busy with other things).
Good news is the system has been working flawlessly all the time: the (used) car battery is still alive, not showing any sign of wear,
AND I've been doing some measures, so
--------------------------------------------
lets do some maths
--------------------------------------------
We know that a usual AC fridge/freezer compressor can be run from a simple modified sine inverter.
But how much power/energy do we loose by:

- using an inverter to drive the compressor ?​

- using modified sine instead of pure sine ?​

Here is some data to try to answer these questions

AC fridge we are using here (a very basic model):

compressor wansheng QD35Y (3.5cm3) marked 50W 0.40A ,​

specs: QD35Y nominal power 49W Cop 1.15 -> cooling output 49W*1.15= 56W​

​

​

First, let's look at how much we loose by driving our AC fridge from an inverter, compared to main grid

COMPRESSOR AC power/energy draw:​

1- running directly on 230V 50Hz grid:​

active power measure 5 min after start, by counting time for x disk turns of an old mechanical Energy Counter (reapproved 1945! but verified with my oficial home counter): 63W. This value is quite higher than specs​

​

2- running on modified sine inverter (from battery):​

measured AC voltage 230V, frecuency 49Hz, AC draw 0.4A with current clamp​

power, measured in same conditions as before: 64W​

​

used AC energy: this takes in account the starting energy (because of the compressor inrush current, waay about the running current):​

the same energy counter gave 11.12kWh for 167.2h running time: (little synchronous time counter, time corrected with 49Hz inverter frecuency). Resulting mean power is 66.5W, with one start every 20min (this is the actual running time limit). The 2.5W overhead can be explained by the higher power needed to start (1500W during 3 secs would match).​

​

COMPRESSOR+INVERTER DC power/energy draw:​

DC draw from the inverter feeding only the fridge (measured with DC clamp): 6.5A @12.6V = 82W , increasing to 6.7A @12.5V = 84W. So we need to add 20W losses by using this inverter: its almost 1/3 of the power needed to run the compressor!.​

That's consistent with​

- 7.5W inverter no load power draw (measured with RC wattmeter, DC clamp shows more but is not very precise at such low values),​

- 84% conversion efficiency (cheap 600w rated output inverter running at just 11% of its rated power).​

No surprise at all: to be able to start such a compressor, we need a big inverter (about 10 times its rating), which not only will sip more power at no load, but also works outside its best efficiency range.​

​

​

What about using MODIFIED SINE: is this a bad idea or not?

It seems that running the compressor on modified sine doesn't use much more power. Because we use here an electromechanical energy counter, there are by construction no computing errors due to modified sine: the instantaneous active power V(t)*I(t) is what drives the turning disk.​

On the other hand, our inverter has a somewhat lower frecuency, so the compressor will run a bit slower and output less cooling power. I can't say how much this 2% frecuency change does affect the cooling output, but just assume there is no big difference.​

​

Conclusion:

Running a classic AC compressor from a modified sine inverter will cost the added power losses in the inverter (30% more in this case) due essentially to no load current and conversion efficiency. Just be aware this is running the inverter ONLY when the fridge needs it (to save most of the wasted inverter no load power).​

Now if we want it to run from a battery charged by solar panels, we should be able to overcome the intermitency of solar power: nights and posibly clouded days. That's where we need storage I will be discussing in the next post.

 

[conquistador]

Sorry for my late response.

Thank you anyway, this was interesting. I've never seen a fridge thermostat too closely, but your procedure looks good. My pure sine wave inverter 12V 1000W beeps for overload at every fridge start, but then runs fine. This inverter might be at the lower limit for running a fridge. I've measured 65W running and then 150W when the compressor did not run, that was probably a defrosting cycle, this lasted for 45min.
I'll try your system in spring, when I will have access to an old fridge at my holiday place. There is a push switch which locks down when on, no software switch so everything should be fine. Thanks again for replying.

 

[conquistador]

Just one more little thing:
So you took out the thermostat out of the 230V circuit completely, and you use it only to pull INT0 of the attiny to GND to wake up from deep sleep . Then from the schematic, you must have put L of the power directly to the compressor?

 

[lorenzom]

You are right: thermostat is totally disconnected from 230V circuitry (and has to be if you want to protect your microcontroller)
The compressor is directly connected to the inverter output. To start the fridge, just run the inverter (by its on/off switch, or from a relay output connected to this switch contacts). Really nothing complicated.

 

[lorenzom]

The Attiny is doing some more tasks than just actuating the relay:
- it has to recognize the solar charger and battery status, looking how its 2 leds behave (not so simple for this old charger)
- starts the inverter only when the battery reaches full state (and thermostat is on) and starts a running time countdown when the battery left full state. This helps to never discharge deeply your battery, (my 20 mins running time correspond to 5-10% of battery capacity)
- shuts down the inverter when the thermostat switches off, or countdown finishes
- another countdown before restart to let vanish the backpressure in the compressor
- also blinks shortly a led as live signal.

 

[conquistador]

- shuts down the inverter when the thermostat switches off, or countdown finishes
- another countdown before restart to let vanish the backpressure in the compressor
- also blinks shortly a led as live signal.

Very good ideas, I'm thinking of adding a 2-3 s wait time after turning the inverter on, and then connecting the compressor motor so that the inverter does not need to start with a near short circuit on the output.

 

[lorenzom]

Hello all, here again, to share my experience of this system after 2 years of use.

I had hooked on the inverter AC output a time counter, and an electromechanic wattmeter.



This is what has been summed in 760 days:

• 1928.8h running time (corrected for 49Hz),
• 137.66kWh AC inverter output (fridge only, + these measure instruments)

We can deduce:

• mean AC wattage: 71W (accounting for start inrush currents and modified sine losses)
• mean daily AC use: 181Wh/day (estimated 234Wh/day on DC side, supposing inverter transforming efficiency at 84%, and adding 19Wh for inverter no load)
• battery capacity use: from used time, (supposing 6.5A DC draw): 167 cycles (@ full capacity 75Ah)

Some thoughts about the components, pros and cons:

custom control system

+ Designed to never run the inverter at no load, you save plenty of energy.​

Start fridge only when battery reaches full state and for a 20min run. This limits strongly the discharge depth and is meant to maximize battery life.​

Brain is powered directly from solar regulator, and needs less than 2mA @5V.​

- With this sort of control, the fridge of course doesn't run at night.​

85Wp photovoltaic solar panel

​

85W peak power: with this solar panel, the extracted energy is equivalent to what my 180l DC chest fridge uses, with its 20liters of ice cold storage.​

Im am living in tropical conditions, so have already good irradiance, although exposition could be improved due to obstacles about horizon (losing almost 2hours of sun these days).​

​

- With this less efficient system, 85Wp is actually undersized.​

pwm solar regulator (Solsum 8.8x).

+ Only 8A max, matches battery voltage and solar panel, so PWM is not too bad. Looking at the charging current, you can see how charging current (and efficiency) increase with battery voltage (good thing because we are never far from full charge),​

- Does'nt allow for a much bigger panel. Also there is no charge at all when irradiation is low. A MPPT regulator would perform better, but would also need a higher voltage solar panel.​

car battery (maintenance free, 12V 75Ah, used).

+ Against common knowledge, a car starter battery can be used in a solar system, as long as you only allow very shallow discharge. This is the case here: 6.5A for 20min is less than 8% depth discharge. Running during sunny hours, depth discharge is even lower because solar charge current has to be substracted.​

This is a second hand battery (no information about its previous history). It has shown no problems yet.​

As car (starter) battery, it is designed to start thermal engines with strong currents during short times. This is well adapted to the high inrush current you have when starting the induction motor of a classic fridge compresor.​

HF modified sine inverter

+ No problem when the fridge is running, I don't notice more noise nor heat due to modified sine. Power draw was shown to be not very different The cheap 600W inverter, charged with continuous 75W runs cold and never needs to start the fan.​

- The only challenge was to be able to start the compresor whith its high current surge. The simple solution was a little mod on the fridge: replacing the PTC starter disk with a more resistive one has proven to be effective.​

fridge

+ A very basic model, with no electronics, simple RSIR AC compresor. Thermostat was eventually disabled to get maximum cold. On normal days (sunny in the morning, clouds building up afternoon), inside temperature at the end of the day was around 10ºC below ambient, enough for some fresh beer.​

- I never managed to get 2 bottles of water to freeze completely, so thermal inertia was low, and inside temperatures would rise during night or cloudy days. This fridge has only 3.5cm insulation thickness, it's not great, but is what most domestic fridges have. Also, it features a hot wall condenser behind the lateral skin, which does'nt seem to be an efficient design compared to the usual external tube condenser, because part of the extracted heat seeps back directly through the insulation.​

This system was still working, but I am now improving it, replacing the fridge with a more efficient freezer. This will be described in another thread.