diy solar

diy solar

Hacking DC supply into DC inverter mini split.

Easiest access will be right across one of the three electrolytics.
Just bring out a pair of suitable wires from there.
It will need a BIG soldering iron, the heat really spreads across the associated very wide copper areas.
 
Thank you very much for your help. This thread will probably go to bed for a while now while I shop for a set of used panels, hopefully will be testing by Spring.
 
DC fed into a bridge rectifier passes right through it right?

The 12v, 5v, etc. control circuits maybe use switching power supplies that also begin with a rectifier.

Resistance heating elements don't care.

Every fan indoor and outdoor runs on an inverter.

It possible that mini splits would accept an adequate 250v+ DC supply right on their main AC input and be entirely happy with it. Of course that would probably require batteries and straight solar would probably be a wreck of low voltage or over amperage shutdown errors.
 
It possible that mini splits would accept an adequate 250v+ DC supply right on their main AC input and be entirely happy with it. Of course that would probably require batteries and straight solar would probably be a wreck of low voltage or over amperage shutdown errors.
You would first need to make absolutely sure there are no small transformers or ac powered blower fans inside that work directly off the 50/60Hz incoming ac power. If you fed just dc straight in, it might be o/k or it may smoke some parts.

Also, a single big dc supply would need to have enough surge power grunt to start up the compressor.

Much safer to power the whole thing intact off it own inverter, or the main home inverter in a total off grid situation.

The best way of all is probably to do it your way.
Power the thing direct from the grid as originally intended, and feed supplementary solar dc into the big electrolytics that supply dc power to the compressor variable frequency drive.
The grid then supplies all the motor inrush surges, and the solar contributes whatever it can once its running.
 
Also, a single big dc supply would need to have enough surge power grunt to start up the compressor.
That is not how these things work, they have a very soft start. They slowly ramp up over 30 seconds or so to a level and then sit there for a minute or two and then do some temperature measurements to decide how much more or less they need to run to maintain the set temperature.
 
O/k fine, I have had no direct experience with this latest technology.
My HVAC experience goes back a very long way, but that is with the old direct driven piston compressors.
 
Ok I managed to acquire a couple hundred pounds of hazardous waste and have weakly proved the concept. Safety concerns can be left at the door and the thread can quarantined appropriately if necessary.

The experience was less satisfying and more so frightening that I was going to fry the control board, but everything worked.

The solar array did raise the DC bus on the electrolytics to 400v, and was quickly drawn down to the AC supply's 325v under load.

Disconnecting and reconnecting the array under load demonstrated a .4 amp reduction on the AC supply. This was distinct and repeatable. I was sort of expecting a little more than 100 watts out of it, but it is old garbage in the shade of a tree below the winter sun, so maybe not surprising.

 
That sounds most encouraging.
Be very mindful of the 400 volt limit on your capacitors, if one of those spits the dummy, there will be one hell of a bang.

Even one partially shaded panel in a series string will limit the whole string, and as you need to have maybe ten panels in series it might be difficult beating the shading problem if you have something like a chimney in the way.

Try to arrange the maximum power voltage of all your panels combined, to get reasonably close to 325 volts if possible, while keeping safely below 400v max.
 
Last edited:
The supposed 400v limit poses quite a problem, as it's rare or impossible to build an array with <400 VOC and >330 VMP. This array would have a 450 VOC in full sun.

However, YMGI's product claims it inputs of 250v plus will offset AC usage, so there's something I'm not understanding there. Maybe that has to be on the rectifier side where the DC can fill in the valleys on the pulsing DC that comes straight off the rectifier. Or maybe 250v magically feed in on the 330v electrolytics? That would be a concept beyond my understanding, or require some kind of ripple that I can't see with my meter.
 
The whole rectifier ripple thing can be quite complex to analyze, and difficult to explain in a simple way.
But basically we have two rules.
Never to exceed 400v.
And introduce a solar voltage higher than the voltage that is already on the capacitors being generated from the grid.

The solution would seem to be, still to choose our solar panels to have VMP in the most useful range, around 325 volts, and accept that the unloaded voltage can get dangerously high.
That could be mitigated with a voltage regulator to pull down the full unloaded voltage into a safe region.
It might not take very much extra loading on the panels to pull the voltage down to say 380 volts in a clear blue sky.

A bit of experimentation might be in order here. The simplest most reliable method might be to just place some small permanent load directly across all the panels. That might just do the trick. I have no idea how much load might be required, but it may not be enough to seriously diminish the total available power under normal working conditions.
The voltage curve falls off very steeply at the high voltage end, and very little load should pull the voltage down by a lot.

The next step up in complexity wold be a proper electronic shunt voltage regulator that draws whatever solar power is required to pull the voltage down to say 380 volts with no load.
Under full normal running load, the panel voltage will be lower than that, and the shunt voltage regulator completely out of the picture.
So the shunt regulator sets the max voltage and applies whatever load is necessary to hold that voltage maximum.

But if it can be done with just a simple resistor, without sacrificing much power, that solution has a lot of appeal.
 
That is not how these things work, they have a very soft start. They slowly ramp up over 30 seconds or so to a level and then sit there for a minute or two and then do some temperature measurements to decide how much more or less they need to run to maintain the set temperature.
100% correct. “Very” soft start meaning 0-XA, gradually ramping up, with steps up/down. Not like MicroAir soft start on a recip or scroll PSC compressor, where it goes from 0 to 20 or 30A or so, then quickly down to a normal 6-15A running.
 
Last edited:
Suppose you put a DC-DC Buck Converter on each panel. A converter producing a few volts less than your Max Power Point on the panels should get you a decent constant voltage output at roughly your max power output. For example, a 300W panel with a MPV of ~32v at ~9 Amps might feed a buck converter set at 30 VDC output. You could stack 12 of these panels and get 360 VDC at 9 Amps (3,240 Watts) out when the sun is shining on all of the panels.

You could also use a voltage comparitor and test the input voltage, allowing you to just turn off the converters when the sun isn't strong enough (making at least ~31v per panel). Or you could just let the regulated voltage drop out as the sun goes away.

A 10 or 20 Amp buck converter is only a few "bucks," so this would not break the bank to try it out

 
Definitely worth looking at.
My first thought was that one buck converter per panel is a lot of buck converters, and more bits to go wrong.

The usual failure mode for buck converters is for the series mosfet to fail short circuit.
A single high voltage buck converter may be tempting fate a bit.

But thinking about it a bit more, if one of several lower voltage buck converters does fail shorted (input to output), its not going to have such a dramatic overall effect on total output voltage, so it would offer a bit of redundancy.

There would then need to be multiple buck converter failures before the total output voltage rose to a dangerous level.
Many ways to skin a cat...
 
I'm not sure if I will continue any further but I like the small load idea best, if anyone has an idea for what to use or a robust form factor of resistor.

Maybe 20-50 watts and 400v (450v rated). Might like to have two of them in case one fails open.
 
Thinking about it, how about a separate resistor across each of the series panels ?
That would spread the load, and offer some redundancy as well. It could be bolted to the frame that supports your panels.
I have had a lot of success with those gold metal clad resistors, available in a wide range of power and resistance values.
Use one rated at about five or ten times the actual anticipated power, and it should last forever.

The whole concept is still unproven, but a bit of experimentation with just one panel should give a pretty good idea if the whole idea is practical.
 

Attachments

  • resistor.jpg
    resistor.jpg
    43 KB · Views: 3
Last edited:
DC fed into a bridge rectifier passes right through it right?
Issue is any power factor correction circuit following rectifier. Most PF correction circuits expect a full wave rectified sinewave waveform.

Most likely issue with just dumping correct DC level into a mini-split HV DC node is the 'incidentals' you may not understand or know exist.

There are various isolation, safety, and self-destruction prevention checks associated with HV DC power supply.

Some may use 24vac control relays requiring original AC line supply to power their small step-down transformers. The entire three phase inverter driver logic may be from a low voltage DC supply from a separate derived and isolated power supply requiring original AC input source.

You can replace all these if you know about them and how they are controlled. You would at least need to have a full schematic and be able to understand their operation.
 
Yes, beware of any small transformers or ac motors connected to the incoming ac.
Much safer to feed in your high voltage dc solar directly across the electrolytics after the rectifier.
 
If I connect an 8 panel array of used Sharp 180's on this cloudy morning I get an increased AC draw of .3 amps / 70 watts. I guess that is the effect of lighting the panels up with supplied voltage into giant LEDs? And I suppose that's what a diode would prevent.
 
You'd like to operate it at MPPT for the array.
If you can get control over pump speed (maybe by adjusting temperature setting), varying that to keep PV array at Vmp for a warm day should be pretty good.

70W backfeed into 1400W of PV array isn't much.
Current into a diode is non-linear, so if you mad that 9s string, should reduce current significantly more.

Your panels have some illumination. How does their Voc compare to voltage on capacitors? If higher, should prevent all backfeed.

I tried backfeeding panels to inspect for defects. I expected it to follow I/V curve given for the panel, but current was considerably lower than that.

 
Back
Top