Any experience with a DC-powered mini-split air conditioner?

I would suggest, that for this whole class of variable speed inverter driven high COP units,
they all use an input power conditioner to provide DC for the inverter. That varies with 48V,
solar panels, or AC line. But I would expect the rest of the components to be pretty much
the same.

There are 4 installed here, but 240 VAC driven. I have only used them (so far) for heating,
rating them about 1/3 higher capacity than cooling. They have had to work pretty hard,
with up to an 80 degree F differential between inside and outside. They are far quieter than
older equipment, esp inside. With lighter loads about the only way to tell they are running, is
to note the fan turning.

The power consumed is entirely dependent on the BTU load. But they seem to be 2 or 3
times as efficient as an older style air to air heat pump, and even more against resistance
heating. Indication is they probably are more economical than propane for most temps,
and comparable to older geo thermal. I do not think they are close to natural gas heat
today. Running a 48VDC plant could be considered a big operational cost increase.

Less than a year is not much experience here. But they totally outstrip earlier equipment,
and for DIY rate of maybe 40% of retail, they soon are cost effective. A prime consideration
for these MITSUBISHI and FUJITSU units was ability to perform at outside temps well
below zero F.

One more thing, I would expect this class not to have the startup surge issues of non
inverter equipment. Bruce Roe

Thanks for responding, bcroe. I thought for sure there'd be more people in the DC A/C field. Maybe I have to plow the field and plant some seeds to see ;-0.

In web searches, I found very little info, though there was one person who insisted that DC inverter A/C units (plus car radios, etc.) "have an internal 48vDC bus." Which I can't substantiate, and that doesn't make any sense at all. Instead, on a quick look at Alibaba, I found block diagrams of certain 48v DC mini-split units, which indicate that at least THOSE units have a boost board taking the 48vDC to 360vDC (=basically the same as rectified and filtered 220vAC) for the compressor--that substantiates what bcroe said.
A true 48vDC mini-split may just be the "holy grail", though I did find some 1 to 1.5 ton mini splits that run on 120vAC. That's somewhat tempting, though it'll run my Chinese inverter pretty hard.

I have not (YET!) pried the cover off any of these mini splits. But expectation here is a much
refined and standardized inverter + compressor is at the heart, with input circuitry set up to
supply the standardized voltage from whatever. 360VDC is where my solar array runs, and
running them direct sometimes might be possible.

Rectifiers are easy, DC to DC has become easy, DC to AC and starting up induction motors
not so easy. Bruce Roe

I've always been a little perplexed by this fascination with direct DC powered appliances. Perhaps it's a little contradictory but I think it makes things more complicated. I'm gonna need 240vac... might as well use it to power everything.

Inverter capacity is not that expensive. I have a 4.4kW Magnum that you can get now for ~$1800. I use it to run a 2 ton mini split and it still has >50% capacity available. As was mentioned, any inverter driven HVAC has no starting current. My Magnum runs my 2 ton like a champ. My lights dip more when my fridge kicks on than when my AC does.

IMO you're gonna get higher quality at a lower cost buying an AC powered mini-split of which millions are produced vs a DC powered mini-split of which maybe thousands are produced.

nwdiver Agreed; with a good inverter (which I don't have), A/C can be run quite easily. The only thing lost out on is total system efficiency--meaning that more panels have to be purchased to run things--not to mention more batteries if nighttime use is desired. And that can quite easily offset the additional cost of a pure DC unit.
For example, to run an 1800W load off AC, my inverter is ~90% efficient...meaning that that same load requires nearly 2,000W from my solar panels. That's nearly one whole solar panel (in my case) just for conversion losses.
Be my guest .
You're right about it making things more complicated. I just happen to have a smaller system, and am doing everything I can to conserve a watt. Just keeping in mind that wall warts/switching power supplies have losses from 10-40% (older transformers being worse)--and that's on TOP of the inverter's 10% loss. "Phantom" loads are the worst--they're consuming power to do basically nothing (and then running a 15-20% loss on top of that.) Take a Yamaha piano keyboard I have: the wall wart runs 7mA @ 120vAC 24/7 with the keyboard off (+the inverter losses = 22Wh/day). The keyboard itself uses virtually zero power when off--and it runs on 12vDC to boot. Am I pinching pennies? Probably...


For comparison purposes, I found a 1.5 ton 16 SEER unit which runs an economical $980 @ HVACDirect.com (not including shipping) Problem is, the EER (BTU/W) rating is a pitiful, paltry 7.7! At full load, it consumes 2,600W. Add to that a 10% loss for the inverter, and I have 2,860W at the MPPT--which adds it's own 5% loss to the panels' power. Total panelage required? 2,990W. I won't mention that my array is only rated for 2,940W...in full sun. (Obviously, the mini-split will use less when throttled down...but I'm using the worst-case scenario so I'm not surprised later.) To make this work, I'd need to replace my panels (limited space, so I'd have to get more efficient ones), a bigger charge controller (it's the real bottleneck), a new inverter, PLUS the A/C. (Ouch.)

Sure, there's more efficient units--but they cost about the same as the 48vDC one I'm looking at. A 1.5 ton 22.5 SEER HyperTek costs around $1,500 (not including shipping), and has a much better EER of 13; full load power of 1,230W. (A Carrier mini-split with the same SEER would run you over $2,000 with shipping.) Still, if I pipeline that through the 10% loss of my inverter, that's a total of 1,353W at the solar array. (And at least for me, I would still need a split-phase quality inverter, $$$.)

Conversely, I'm seriously considering buying a 1.5 ton 48vDC mini-split pretty much direct from China (which is where most all the rest of them come from anyway). It has an EER (BTU/W) of 20, and is rated to use ~900W at full load (18000BTU). Assuming that specification is accurate, that's a really hard class act to follow--and well within reasonable ratings of my 3KW system. Sure, it may cost nearly $1,800 (INCLUDING shipping)--but that's a whole lot cheaper than a bigger system PLUS a cheaper A/C.

I'm starting contact with the seller; if this pans out, and (with me as the guinea pig) it really DOES live up to spec, would anyone else here be interested in a 48vDC mini-split with an EER of 20? (Trying to decide if it's worth pursuing a bulk order or something...)


bcroe I wouldn't recommend toying with 360vDC...unless you REALLY know what you're doing. That being said, there's a very fair chance that 220v supply coming into the "inverter compressor" board will run through a bridge rectifier, and straight into a 400vDC filter capacitor. AT YOUR OWN RISK, after disconnecting the 220v supply to the board, that's where you'd connect your panels. Probably even beat a Chinese 48v unit at efficiency, 'cause you'd cut both the MPPT and inverter out of the circuit!

Mini-splits being split between indoors and out have varied length copper tubes or line-sets...getting accurater refrigerant charge loaded takes special skills and a borrowed or rented tool/equipment if expected EER rating is to be achived. Yes, there are some pre-charged units and even some pre-charged line-sets but together accurate charge measurement and DIY can be tricky. This is an exercise of just being 'close enough'...

Have not experienced any pre charged line sets here. I am suspect of them leaking in the
long term. Typical instructions for a precharged say no additional charge up to 15 feet.
Mine are not very long. Dealing with the lines is the only tricky part of mini splits, I had to
buy all new equipment to deal with the R410A pressures. Bruce Roe

Right, this is just another option to look into if net metering has issues. Bruce Roe K9MQG since 1958

You're completely right. I have been through HVAC school, so that aspect of the system doesn't bother me at all. Have access to gauges, vacuum pumps, cutting and flaring tools, lineset benders, etc. A mini-split install isn't exactly a job I'd recommend to a DIY'er who's just purchased a left-handed crescent wrench to tighten the muffler bearings on their car.
All of the units I've seen have been precharged--but that may vary from manufacturer to manufacturer.

At least the ones I'm familiar with didn't have soldered fittings--they were flared compression fittings (could possibly leak if not tight enough). But then again, this is a solar forum, not an HVAC forum .

So I got a response from the seller about the 900W load rating on a 1.5 ton DC mini split, and I quote verbatim: "Sorry the date is wrong, 18000btu is 1400W power usgae." Oh, well...like they say, "if it seems too good to be true, it probably is."

That may be the peak you must allow for. But average over time will vary greatly
with the actual thermal loading, could easily be half or less. Bruce Roe

1400 W sounds about like a possible running power draw.

(18,000)/((3.412)*(1400)) = 3.76 ~ a reasonable est. for a running C.O.P. in cooling mode.

You're completely right. However, I was trying to establish worst-case scenario, and determine A/C efficiency.
It's China...they can be known to overstate things (like those 9,900mAH 18650 Li-ION cells--talk about an off-gridder's powerwall dream! Until you buy them and are lucky to get 900mAH out of them.) The problem in this case is that the EER (W/W) is stated as 5.58 (if it was BTU-H/W it would be 20 EER). They list 18000BTU as 5,000W of work (mathematically it equates to 5,275W). And 5,000W / 5.58 EER = 896W input power. As shown in the specs below:

EDIT: can't upload tables. Ended up with just a partial post, and HTML displayed. Grr...maybe I can copy-paste a picture (NOTE THAT THIS IS FROM A DIFFERENT LISTING, shows EER OF 5.18 on the 18000BTU unit).



Basically, if I take the updated 1,400W for 18,000BTU output, that gives me an EER (BTU-H/W) of only 12.8, or very close to a 220vAC mini-split (when NOT figuring in inverter losses). So much for the "holy grail"...

Thanks for doing the arithmetic. A C. O. P. approaching 4 is pretty good.

I expect there is rather minimal energy loss in the input section, whether it is a
rectifier, a DC-DC converter, or an MPPT panel input, so that factor would not
give one system a big advantage over another. That thanks to solid state
advances in recent decades. Massive heat sinks of earlier times are minimal
now. Bruce Roe