GBA Logo horizontal Facebook LinkedIn Email Pinterest Twitter X Instagram YouTube Icon Navigation Search Icon Main Search Icon Video Play Icon Plus Icon Minus Icon Picture icon Hamburger Icon Close Icon Sorted
Musings of an Energy Nerd

Bruce Harley’s Minisplit Tips

Expert advice on air-source heat pumps

If a minisplit system is mostly used for heating, consider installing a floor-mounted indoor unit. Research shows that floor-mounted units perform better for heating than indoor units mounted high on a wall.
Image Credit: Image #1: Fujitsu

Bruce Harley is a Vermont energy consultant and heat-pump specialist. To GBA readers, he is probably best known as the author of the Taunton book, Insulate and Weatherize — one of the best available books on residential energy retrofit work.

As the recent Better Buildings By Design conference in Burlington, Vermont, Harley shared a wealth of information on ductless and ducted minisplits. The title of his February 8 presentation was “Real-Life Air Source Heat Pump Performance Testing: Results and Reasons.”

Since Harley left his longtime job at Conservation Services Group, he’s been working as an independent consultant. Harley told the audience, “Now 90 percent of my work is related to heat pumps.”

Asian manufacturers have developed models for cold climates

Until recently, North American homeowners’ opinions of heat pumps were strongly influenced by the unimpressive performance of U.S.-made forced-air units designed for warm climates.

Harley explained, “Air-source heat pumps got a lot of bad press in the 1980s. Occupants noticed they cost a lot to run because of the electric-resistance heaters. Back in 2012, when I wanted to install a ductless minisplit at my house in Vermont, a contractor told me, ‘A heat pump won’t heat your house in Vermont. Maybe if you were in Texas in Oklahoma.’”

Japanese engineers have revolutionized the market for heat pumps, however, by improving their units’ cold-climate performance. “Ductless split-system heat pumps have been in mass production for 40 years,” Harley noted. “The technology has made steady advances. Now they are optimized for cold weather. It’s a rapidly growing market. Most of these heat pumps do not have an electric-resistance heater. Lots of people are installing one in an existing home with oil heat to offset the high energy bills.”

GBA Prime

This article is only available to GBA Prime Members

Sign up for a free trial and get instant access to this article as well as GBA’s complete library of premium articles and construction details.

Start Free Trial


  1. calum_wilde | | #1

    The manual doesn't list this
    The Fujitsu manual doesn't list this in the instructions. The unit must be turned off from the remote in order to accept the setting inputs suggested above. I also learned that my unit has an indication to remind the user to clean the filter that was left off from the factory.

    Thanks for the useful tips.

    Edit: "You have to flip the dip switch to the ‘remember the settings’ setting." I can't find anything on dip switches on the unit. Can anyone point me in the right direction?

  2. AndyKosick | | #2

    Why don't multi-zones turn down?
    I would love an explanation for this. It seems to me that multiple zones is where the most turn down would be desired. Is there a legitimate reason or is just a "we can always dump capacity to another zone so we'll put cheap pump in it and sacrifice efficiency."

    I have had a Mitsubishi MXZ 2C20 as my sole heat source for about 2 years now, no monitoring yet, but I've noticed much of what Bruce talks about. I ended up with the 2 zone because it was the only way to get a ducted head on a low temp unit at the time (with a warranty) and ducted was the only good option for the main floor of my house.

    On the positive side, the wall head that heats the basement is off much of the time, as the space is used only intermittently, and even then often without the heat on. This has been a savings over all I would say.

    On the negative side, the turn down on the system is pathetic considering other single zone H2i units have 11 to 1 and this barely has 3 to 1. This may be the largest contributor to the other issues that I would term "quirks in the algorithm." There's is probably a dozen things I would like to change about the way the system chooses to operate. Without out going on for too long here are a few:

    1. The wall head in the basement short cycles just as Bruce described. Next winter I'm getting a wall mounted thermostat for it.

    2. On the flip side the ducted head upstairs tries to run long cycles, to a fault in my opinion. The room temperature swings nearly 2 degrees F and the last 30% of the cycle it's dumping capacity to the head downstairs. This seems less efficient than just turning off, since it's going to end up turning off anyway.

    3. Because the unit can't run steady until it's 20-25 degrees F outside or less it's well into defrost temperatures. The problem here is that it will come on and run for about 1 minute then go into defrost. During this defrost time the room temp can drop to a full degree below the thermostat setting. Why doesn't it just defrost at the end of the last cycle so it's ready to go?

    I'm tolerant of temperature swings and understand why because I'm more attuned to its operation than the typical homeowner, but I can't help but be a little disappointed that such a sophisticated VRF system is so incapable of holding a reasonably steady room temperature.

  3. Expert Member

    This is a very useful article, but also a bit worrying. We have just installed two mini-splits in the bays of our firehall to replace the propane heat which as costing us a fortune. Everyone expects them to just do their thing, as all the rest of the equipment does. The commissioning, and fine-tuning, and monitoring that seem necessary to get them optimized, makes it appear they are still being engineered with an enthusiastic GBA reader as a buyer. Other blogs seem to support this.

  4. Robert Opaluch | | #4

    Usability nightmare
    Useful information in this article.

    If it takes a guru to make obscure adjustments to optimize minisplit operation, this technology is quite immature for the consumer market. Automating or improving the user interface is needed for operating minisplits, not to mention selecting appropriate units and options, and maybe even installing them (given all the complaints about noise).

    This reminds me of the old "command line" typed commands interface of the early days of personal computers. Geeks saw no need to go to an easy to use graphical interface. Apple and Microsoft improved computer usability, and took over the PC marketplace. Likely the same will happen with minisplits if they are to become more widely used in residences.

  5. davidsmartin | | #5

    Fujitsu does not make it easy and installers don't know much
    My installer put the unit 5 1/2 inches from the ceiling. The result is that I have to set the temperature 6 to 8 degrees warmer than the actual temperature three feet off the floor.

    Then I ordered a wall thermostat from Fujitsu. But before installing it I saw that using it would limit the lowest setting on the thermostat to 60 degrees. When I leave the house empty for a week or more I want to be able to turn the temperature down to 50, but with the Fujitsu wall thermostat I would lose that option. So I sent it back.

    Now I have ordered a remote thermostat from Flair for $99. It uses IR signals to communicate with the Fujitsu. I don't know how it overrides the Fujitsu temperature sensor. I will let you all know how it works.

  6. calum_wilde | | #6

    They're setup for a


    They're setup for a compromise between ultimate efficiency, user comfort, user expectations, and typical building design. For your setup in a fire hall I'm guessing the concrete floors are uninsulated, and the large bay doors aren't very efficient either? The factory setup is likely to be exactly what you need. The "heroics" to get back to set point temp after a disturbance is likely going to be lacking even given how much air will exchange everytime the doors are opened.

    What's being described in the article is changing some of those compromises to better suit a very efficient home, or users willing to put up with some discomfort in the name of efficiency. The average user likely wouldn't be happy with these changes.


    That sounds like a pretty bad install. There is a setting in the installation manual of at least some Fujitsu mini splits to put a offset on the set point. You might want to give it a try. But also, please do post your findings on the Flair.

  7. GBA Editor
    Martin Holladay | | #7

    Response to Calum Wilde (Comment #1)
    Q. "I can't find anything on dip switches on the unit. Can anyone point me in the right direction?"

    A. As far as I can tell, Bruce Harley is out of the office (perhaps on vacation) this week. When he returns, I hope he'll be able to help answer your question.

  8. GBA Editor
    Martin Holladay | | #8

    Response to Andy Kosick (Comment #2)
    Q. "Why don't multi-zones turn down?"

    A. I'm not sure, but I imagine that the answer is something like "Because, though better turndown ratios are technically possible, the required algorithms and equipment to perform the task are expensive and complicated." But I'm just guessing.

  9. GBA Editor
    Martin Holladay | | #9

    Response to Malcolm Taylor (Comment #3)
    You wrote, "This is a very useful article, but also a bit worrying." I agree.

    If a reading of this article is accompanied by a reading of two other recent GBA articles -- namely "Marc Rosenbaum’s Monitoring Results" and "A Follow-up From Northern Minnesota" -- a pattern emerges.

    GBA readers who are monitoring their minisplits are discovering, too often, that many of the units are cycling on and off when they should be ramping down. This problem is cropping up too often to ignore.

    I'm on record as one who points the finger at the manufacturers. These appear to be design issues. We need (a) better defaults -- defaults that correspond with normal performance expectations, (b) clearer installation and commissioning instructions, and (c) better thermostats and hand-held remotes.

  10. GBA Editor
    Martin Holladay | | #10

    Response to Robert Opaluch (Comment #4)
    You wrote, "If it takes a guru to make obscure adjustments to optimize minisplit operation, this technology is quite immature for the consumer market."

    I agree. Here's my summary: Gurus are discovering that many of these units aren't performing as expected or as described in the literature. Fortunately, even when these units perform poorly, most homeowners have developed work-arounds (usually, adjusting the thermostat setting) that keep them fairly comfortable.

  11. GBA Editor
    Martin Holladay | | #11

    Response to David Martin (Comment #5)
    You wrote, "When I leave the house empty for a week or more I want to be able to turn the temperature down to 50."

    Your expectation is entirely reasonable. The fact that some minisplit manufacturers don't allow low thermostat settings is, indeed, frustrating.

  12. calum_wilde | | #12

    Can you also ask


    Can you also ask Bruce if he is able to get information on the difference in kWh usage for before and after the cycling was fixed? I've been seeing these same concerns over cycling come up here on GBA, and I'm wondering if we could get an idea of how much extra power the cycling is using.

  13. GBA Editor
    Martin Holladay | | #13

    The residential HVAC industry is in crisis
    As evidence that the U.S. residential HVAC industry is in a state of crisis, I'd like to share a trade show display by Mitsubishi from the recent BuildingEnergy conference sponsored by NESEA. Mitsubishi should understand its products and customers, and should presumably understand the type of contractors who choose to attend a NESEA conference. Mitsubishi does not.

    The display shows a house with seven rooms. Each of the seven rooms has its own minisplit head. Yes, Mitsubishi is promoting the "put a separate head in each bedroom" approach. When I spoke to the Mitsubishi rep at the booth, he challenged my assertion that the average residential bedroom had a load that was far, far lower than the smallest Mitsubishi minisplit head.


  14. calum_wilde | | #14

    Wow, that's absurd. :(
    Wow, that's absurd. :(

  15. GBA Editor
    Martin Holladay | | #15

    A better version of the image
    Mitsubishi has published a copy of the "minisplit head in each room" graphic on its web site. The image below is clearer than the version I photographed.


  16. AndyKosick | | #16

    I'll second the HVAC crisis
    What's frustrating is that performance minded builders are turning to mini-splits because traditional equipment manufactures are failing to provide what they need, just to find that the Mini's don't quite get it either.

    It would seem that not only are contractors failing to do good load calculations, the manufactures are as well. The fact that Fujitsu seems to think this "high insulation" setting is some specialty thing and not simply the way the product should operate says a lot to me.

    That said, I may get a chance to work with one of those little Dettson furnaces in the near future and I'm looking forward to it.

  17. GBA Editor
    Martin Holladay | | #17

    Response to Andy Kosick
    You wrote, "It would seem that not only are contractors failing to do good load calculations, the manufacturers are as well."

    You're right. By the way he responded to my question, it was clear that the Mitsubishi rep at the NESEA conference booth thought that the average residential bedroom had a design heat load of 8,000 or 9,000 Btu/h. It was really depressing. It seemed as if he had never once stopped to consider that a minisplit head in each bedroom was overkill.

  18. Expert Member
    MALCOLM TAYLOR | | #18

    Seven Heads!
    If Mitsubishi get their wish, new builds may have to get a conventional mortgage and then arrange mini-split financing.

  19. Expert Member
    ARMANDO COBO | | #19

    More convinced than ever...
    If this article doesn't convince anyone that mini-splits are to be used in special situations, like a single room building, a pool house, a guest house or detached office, I don't know what will. If the "experts" can't agree on correct specs, installations, efficiencies and design, what can be said for the customer who ends up getting shafted.
    First, they are ugly, and "Form follows function" never met the mini-split.
    Second, they don't provide ventilation, so one must introduce other sources, which can be expensive.
    Third, there is no privacy unless you have a head in each and every room.
    Fourth, can be more expensive than a right-sized, right-design forced air system.
    This article just sealed it for me, and my clients. Thanks.

  20. GBA Editor
    Martin Holladay | | #20

    Response to Armando Cobo
    You wrote, "If the 'experts' can't agree on correct specs, installations, efficiencies and design..."

    Actually, the experts agree. Bruce Harley's advice is remarkably consistent with advice from Marc Rosenbaum, from the Building Science Corporation, and (for what it's worth) from articles on the topic published by GBA.

  21. bencarsan | | #21

    mitsu marketing
    Ha, yes, I've seen that graphic before. In manufacturer trainings the suggestion is that homeowners can be helped to see the need for a minisplit in every room if we just point out that heating/cooling is like lighting--and you have a light switch in every room, so....

    Needless to say that's not helpful to the poor installer wondering how many heads, and what kind, and where.

  22. GBA Editor
    Martin Holladay | | #22

    Response to Bennett Sandler
    I've said this before in other comments, but HVAC contractors behave differently from electricians and plumbers. In general, electricians and plumbers develop good plans and specs, and execute them well.

    I sometimes imagine that much of the fault lies with HVAC contractors who ignore manufacturer's recommendations. And then I learn (as your story confirms) that manufacturers are also ignorant. That makes the entire mess harder to fix.

  23. greenenvy | | #23

    Similar issues with a Pioneer unit
    A few months ago I installed a 1 ton Pioneer brand ceiling concealed mini split in a closet located between my master bedroom and master bath with the included wired thermostat mounted on the master bedroom wall. This is in central Florida. The return pulls from the master bath (with a 24X24 filter) with the supply in the master bedroom. This way I am able to create a circulation loop between these two rooms and climate control both of them well with a single unit. I know that some would advise against using a bathroom as a return source of air but it honestly works fine for us. I would consider the quality of the hardware to be very good, even though the system is much less expensive than the Mitsubishi's and Fujitsu's of the world.

    Like others stated or imply here, I expected an inverter based system to maintain temperature to a fraction of a degree but this unit seems to regularly overshoot temperature. I also find that I have to change the set temperature by two or three degrees before the unit (the compressor, specifically) will kick on. I mention the compressor specifically because the air handler runs constantly in cooling mode. I recently found a dip switch setting that I thought would turn off the air handler when the compressor is off but it seems to have no effect, at least in cooling mode (haven't tried heating mode with this setting). Supposedly the continuous runtime on the air handler is so that an internal air temperature sensor is getting accurate room temperature data but I would think that the wired external thermostat room temperature sensor would render the inline room air temperature sensor as unneeded. The fan draws very little current (.2 amps or less) but I don't like running all that air through the filter unnecessarily and the motor on the air handler will probably give out before other components since it runs continuously unless I turn off the system.

    We've not had hot enough days yet (It's 83 at the moment) to see how long it runs before shutting down when there is a significant load on the system but I can say that it presently spends a lot of time with the compressor off. Its running at a fairly low level as I type this, pulling 1.75 Amps at 244 Volts (but only 225 watts so it apparently has a very poor power factor). I maybe should have gone with a 3/4 ton unit but I am planning to leave the door open to the bedroom during the day to put as much load on the system as possible. This part of the house has east, south, and west exposure. All that said, I should probably have still gone with a 3/4 ton unit.

    So, it seems like we can summarize that much of the current mini split product available on the market today appears to have software that significantly lags the hardware. I'm no electrical engineer but this has me wondering if it is practical/possible to interface, say, a Raspberry Pi with a current model mini split. The idea would be to write custom code to operate one of the more popular units as efficiently as possible, using the full range of capabilities of the system. I would think that would primarily involve controlling the compressor speed, fan speeds and TXV settings. Different profiles or maps could be built to optimize the systems seasonally or geographically. I think that this would be a fascinating project to be involved in. I would put some money into such a project and I suspect that others would too if there was potential. Is it possible and practical? Does anyone know someone that could make something like this happen?


    System purchased:
    Simple kWh meter used for monitoring system:

  24. steve196 | | #24

    mimisplit heat pumps
    Is there any information comparing brands? Hard to sort through all the literature. I'm thinking about installing one but as things continue to change, I can't keep up. Thanks.

  25. GBA Editor
    Martin Holladay | | #25

    Response to Steve Mindel
    For builders or homeowners in a cold climate, the two brands to choose from are Fujitsu and Mitsubishi. The usual criterion for choosing between these two brands is this one: "Is there an experienced installer near my house who carries one of these two brands? Then that's the one I'm going with."

    To compare specifications, see the cold climate heat pump listings on the NEEP web site.

  26. Expert Member
    Dana Dorsett | | #26

    ...a right-sized, right-design forced air system. @ Armando Cobo
    Mini-ducted mini-splits are easier to right-size for most homes than traditional split-systems, especially for small or high performance homes, and most (but not all) have much bigger modulation ranges than modulating split systems such as Carrier Greenspeed, Bryant Evolution etc. The 1.5 ton Fujitsu 18RLFCD has about a 6.5: turn down in cooling mode, 8:1 in heating mode, compared to ~2.5:1 for most modulating split systems with bigger air handlers.

    Fixing some of the user/human interface quirks and lapses in documenation is worthwhile, but these are still better systems for high performance homes than almost all old-school force air system (heating or cooling.)

    Multi-splits are a different story, different problem. Micro-zoning always has issues as loads run up against minimum modulated output, and it's not just with ductless heat pumps. With hydronic heating & cooling the limitations can often be overcome with buffering thermal mass. Air source hydronic heat pumps & buffer tanks can be useful in micro zoned systems, but not with low-mass air coils and compressors always have some limitations in modulated output.

  27. Jon_R | | #27

    18RLFCD modulation?
    Apparently the 18RLFCD evaporator fan has only a 1.25:1 airflow modulation ratio. AFAIK, this makes it impossible to provide proper dehumidification (SHR) over a 6.5:1 cooling ratio. Thoughts?

  28. user-626934 | | #28

    Response to Jon R.
    Air handler fan modulation compared to outdoor unit modulation in cooling mode is a humid-climate issue for all mini-split systems that I know of. In humid climates, these systems would ideally modulate such that the latent cooling fraction (moisture removal) increases as the outdoor unit ramps down. This would have the effect of keeping moisture removal relatively steady, while air temperature cooling is reduced. Instead, the systems do the opposite: sensible cooling fraction increases as the outdoor unit ramps down. This is great for the SEER rating and great for dry climates, but not good for humid climates. This issue tends to be worse for the slim-duct systems, with the indoor fan modulation on the Fujitsu units particularly poor (one of my few complaints, though we use these units nearly exclusively for now).

    On the other hand, Fujitsu's "Dry" cooling mode seems to work well...boosting compressor speed to keep the coil as cold as possible, while running the indoor fan at low speed. In my own house, it makes the difference between ~60% RH vs. ~50% RH in the summer. We recommend to our clients to use the "Dry" mode in the summer if their humidity levels are tracking higher than desired.

  29. Jon_R | | #29

    lack of latent modulation
    Thanks John. The result of Dry mode is that the unit doesn't modulate down very far. Ie, the high turn down ratio becomes fictional. The result is likely to be poorer efficiency under low load conditions.

    If someone wants maximum comfort, then I suggest that each room have dehumidification (ie a method of 100% latent removal). Perhaps via open doors or via a DOAS like system.

  30. Jeremy_G | | #30

    Another energy monitor option
    I ended up with an emonTx + WiFi module from Open Energy Monitor to monitor the big loads in my house. It's more DIY-oriented than the other listed at the end of the article, though modestly priced.

    The OpenEVSE charging station for my car backs into the same platform, so I decided to extend from there instead of having two energy monitors.

    Note that they are beginning to charge for the web software platform starting later this year. It's currently free. It's all open source, so you can run it on your own hardware/cloud provider if you wish.

  31. steve196 | | #31

    Thanks Martin.

  32. dugg | | #32

    wall mount mini located near the floor?
    Any particular reason/s why a wall unit might, or might not be suitable to be mounted about 7" off the floor, and directly below a window? I'm primarily only interested in using it during the "heating" season, as I'm located in Northern New Mexico where the summers are quite mild [no a/c required]....heavily forested, and at about 8,000 ft ASL

  33. GBA Editor
    Martin Holladay | | #33

    Response to Douglas Harrison
    It's been done (see the photo below). The main objection is aesthetic -- why not purchase a floor-mount unit that has been designed for that purpose? But if you want to take a unit designed to be installed near the ceiling and install it down low, you can.


  34. dugg | | #34

    Response to Martin Holladay
    Thankx, Martin..
    i assumed as much, but I felt it best to run it by the experts.
    As for your aforementioned "aesthetics", I actually find the swoopy lines of the wall units much more pleasing..... to the stodgy boxy design that's the current offering for floor locations. The only [slight] downside to the wall units, is that they're a few inches deeper.

  35. an123 | | #35

    Re placement of a unit close to the floor, can someone tell me what the author meant by: "Or put a small air handler in the basement."? Thanks.

    1. charlie_sullivan | | #36

      I think that is suggesting putting a ducted mini split head in the basement and running short ducts up to where you would have put the low mount ductless mini-split head.

Log in or become a member to post a comment.



Recent Questions and Replies

  • |
  • |
  • |
  • |