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

Community and Q&A

Determining Minimum Capacity for Indoor Minisplit Units

ianrking | Posted in Mechanicals on

Hi, there. I am having much trouble understanding what the minimum capacity for an indoor unit connected to an Mitsubishi MXZ-C outdoor unit should be. Specifically, I have an MXZ-2C20NA2 outdoor unit with an MSZ-EF09 and MSZ-EF15 connected to it.

Mitsubishi’s submittals says that the rated capacity of the MXZ-2C20NA2 is 5,700BTU/hr minimum to 20,000BTU/hr maximum. For the MSZ-EF09 and 15 indoor units, a capacity range is not given – only a nominal rated capacity (9,000BTH/hr and 14,000BTU/hr, respectively).

If you look at the engineering information for the MXZ-2c20NA2 (\M-Series_Technical%20Data%20Book_01-2021.pdf), Mitsubishi supplies an indoor unit capacity vs. compressor frequency curve for each class of indoor unit connected (06, 09, 12, 15 kBTU/hr). If you take the nominal capacity of any class indoor unit and multiply it by the applicable capacity factor when the compressor is operating at it’s lowest frequency (~18-20Hz per graphs), you get approximately the minimum capacity of the MXZ-2C20NA2 given in the specs (5,700BTU/hr).

For example:
For a 15-class unit operating at ~20Hz, the capacity factor is approximately 0.43. 14,000BTU/hr * 0.43 = 6,020BTU/hr ~5,700BTU/hr.

Does this mean that a single MSZ-EF15 – or any class indoor unit for that matter – should be able to modulate to a capacity as low as ~5,700BTU/hr What happens if two indoor units are operating at the same time? Can they split the minimum capacity? And could that capacity be split according to the demand or must it be split using the ratio of the two units’ nominal capacities?

Thank you!

GBA Prime

Join the leading community of building science experts

Become a GBA Prime member and get instant access to the latest developments in green building, research, and reports from the field.


  1. Jon_R | | #1

    It's been asked before and as I recall, was never answered. Do note that this isn't a "VRF" system (eg a City Multi-S) - the indoor units might just be on or off at full rated capacity.

  2. ianrking | | #2

    Hi, Jon. Thanks for your reply. I sincerely hope that’s not the case. That would also mean that all of the engineering data provided is moot.

    Even Mitsubishi’s consumer level material says that the units modulate! From their own website:

    “ Inverters, on the other hand, are always running in the background, adjusting the compressor speed in real time. By ramping up power to the compressor when needed, an inverter provides a more accurate, on-demand approach to temperature control. If your room temperature’s already a comfortable 72 degrees, for example, an inverter may slow the compressor’s engine to a crawl. If your room’s cooking at 104 degrees, the inverter will push the compressor into high gear, keeping you out of the hot seat.”


    “INVERTER-driven units, on the other hand, are able to dynamically adjust their compressor speed based on a room’s temperature. Because the unit only draws enough power to maintain a steady room temperature, via the compressor, it “changes” its capacity to suit the room it’s in.”

  3. Expert Member
    Akos | | #3

    Some units work better than others. I have an older LG multi split that works pretty much as Jon R says, which is terrible for comfort and efficiency.

    It seems the Mitsubishi unit runs better, I did a bit of very rough testing on a similar system (2C20 with 12k wall mount and 12k ducted). The indoor heads do modulate, generally you can get down to pretty close to or even a bit bellow the documented minimum operating point of the outdoor unit. The range is still not that grade, I could only get about a 2:1 turndown the 12k units.

    Unfortunately didn't get around to sweeping the operating range with two heads running. Documentation on lot of these is really unclear. Instead of multi split, the safest bet is to go with one-to-one setups, these work and work well.

  4. ianrking | | #4

    Hi, Akos. Thanks for your reply. I’m also thrilled to hear that you’ve done testing that seems to confirm that the individual heads do in fact modulate. I monitor my outdoor unit and definitely hear/see it spinning up and down as the day goes by to either compensate for variable heat gain or outdoor temperature/humidity conditions.

    Unfortunately, single zone systems were not possible for me. On our NYC apartment terrace we have minimal space to dedicate to HVAC equipment.

    The 09-class MSZ-EF unit in my 240-square-foot bedroom works great. The 15-class unit MSz-EF in my 415-square-foot living room/kitchen/bathroom space also works great except overnight. We live in a 1-bed apt on the 16th floor of a 30-story Depression-era skyscraper with brick barrier walls, concrete floors, and no insulation. The public halls and elevators are also not air conditioned. We also face south and west and have no shading whatsoever (above every other building). With no AC on the indoor temps easily reach 80+ degrees. And with various apartments around us unoccupied with no AC on at various points in time we get a lot of infiltration from neighboring units.

    During the day both our units work great together to reduce to and maintain the desired indoor temperatures and humidity. At night, I like to keep both units on but the bedroom is set way down (we like it cold cold cold) while the living room is set to 73-74. With the low overnight loads the 15-class unit will cycle on and off rather than modulate which results in drastic humidity swings. And for some reason as the one unit cycles on and off the other unit is affected and the temp and humidity also swing slightly tho not as much. I’d think that even overnight the heat gain for the entire apartment is above the minimum capacity of the outdoor unit (5,700BTU/hr).

    I was wondering if it were possible that each class of unit itself had some sort of minimum operating capacity - which is obviously undocumented - but the engineering manual published by Mitsubishi does not support this possible understanding. Clearly the higher capacity heads can modulate very low on a single system. For some reason there’s no single system that supports the EF-series indoor unit.

    Just can’t figure this out….

  5. ianrking | | #5

    FWIW a poster on this thread states that the heads do in fact modulate….and that lower limit is controlled by the outdoor unit.

  6. Jon_R | | #6

    I wonder what one gains when they upgrade to VRF such as the City Multi-S? Perhaps just simultaneous heating and cooling?

    Keep in mind that latent performance is about CFM/ton and has little to with turn-down ratio and compressor cycling.

  7. Expert Member
    Akos | | #7

    Interesting data sheet from ianrking's link:

    It actually shows operating range depending on which heads are installed. When you run two heads at the same time, the min capacity goes up.

    I don't know what is the difference is between a 2B20 and 2C20 but I would guess they are similar.

    As for night time cooling on the oversized head, there is no simple solution. You can help a bit by configuring your unit to fan-off when the cooling is satisfied, this will keep it from re-evaporating moisture from the coil between runs.

    1. Jon_R | | #8

      Interesting that the numbers don't add up. For example, two 6K heads lists 6000 as the only cooling output on each but 7800 as the minimum total. So not at all clear that the heads will ever do 7800/2 = 3900. Even if they do, the turn-down ratio is poor - sometimes 1:1 and sometimes 1:1.5, depending on timing.

    2. ianrking | | #9

      It’s a very interesting submittal, yea. What’s most interesting to me is that if it is to believed, you literally cannot oversize a head because you always have the same minimum capacity of the system. (In this case, 5,400 BTU/hr for a single head running and 7,800 BTU/hr for a pair of heads.) A single 15,000 BTU/hr head has the same floor as a single 6,000 BTU/hr head. With this understanding I put the 15,000 BTU/hr in my living room/kitchen/bathroom area because when I am cooking I would like the extra capacity there.

      And I also think that the columns in the submittal are somewhat mistitled. The left two columns I believe are actually the maximum rated capacity of each connected head. This is based on the total rated capacity, or ratio thereof based on the fraction of the total connected rated capacity at that head. The third column I believe is the modulating range of the unit a certain combination of heads are on at any given time. At least that’s my understanding.

      I tested out my system today with just the 15,000 BTU/hr head running and set the setpoint to the room temperature and it did in fact run very low. The MXZ fan was barely turning. That seems to confirm my understanding of the submittal described in the above paragraph.

      I have thought about setting the thermal fan to off - yes. It’s a good suggestion. I just wish it didn’t require a non-reversible alteration to the unit. Why it is not just a switch is beyond me.

      1. Jon_R | | #11

        IMO, poor latent performance is driven by making the specs look better. So SHR is measured at rated capacity, but SHR is lousy at the common partial load capacity. Ie, metrics are having a very negative effect on usability.

  8. Deleted | | #10


  9. ianrking | | #12

    So I think this most recent understanding makes sense. The capacity range of the entire MXZ system (i.e. the outdoor unit and all attached indoor units) is controlled by the capacity range of the outdoor unit. However, the capacity range of the outdoor unit varies depending upon what indoor units are on at any specific time. If one indoor unit is on the upper bound is the maximum capacity of that particular indoor unit but the lower bound is that of the MXZ with one unit on. (In the case of the MXZ-2C20, that is 5,700BTU/hr.) If multiple units are on, the lower bound of the entire system is somewhat higher. (For the same MXZ-2C20, this is not currently documented but for the B generation unit it is 7,800BTU/hr.)

    How the refrigerant is divided when operating at or below the system’s lower bound still not well understood but I believe whatever the behavior is is the source of my observed issues. The lower bound when both units are on is actually higher than the documented 5,700BTU/hr. (Maybe about 7,800 BTU/hr.) All of that refrigerant needs to go somewhere but with both units on there’s no where for the system to put it without contributing to actual cooling. Thus the cycling of at least one of the units on and off…

  10. ianrking | | #13

    FYI. Mitsubishi publishes the same data of the MXZ-A generation! I do not understand why they do not publish this information for the current MXZ-C generation.\MXZ_2A20-3A30-4A36_Service_OB444H_6-11_(2-8_43-51).pdf

  11. gamix | | #14

    Newer SM, or smart multi Mitsubishi condensers can operate at lower capacities than older models. I am no expert. NEEP publishes info on MXZ C series.!/product_list/

    For instance, I'm discovering this (outdated) unit, MXZ 5C42NAHZ has good efficiency & COP numbers, but to everyone's point I think bad turn-down shows up as "min" "24,000" out its max 48k btu heating ability. The newer MXZ SM42NAMHZ multi head, at same rating, shows much better turn-down of 48k, down to 9.8k btu. I take this to suggest that suitable multi-heads can take advantage of modulation all the way down to this cumulative level. No?

    I'm faced with a cheaper-older vs. newer-pricier set of options and believe the NEEP data may tell some secrets. Another interesting point is the *older* unit has about a half a point *better* COP, which is probably because it is afforded a more limited operating range (means a lot when going from 2.3 to 2.89). This is where frustration makes sense, if you can't call any less than 24k btu to obtain that stated efficiency (without effectively becoming a single-stage condenser). I think our whole discussion is about realizing more "Variable" benefits. Am I thinking about this right?

Log in or create an account to post an answer.


Recent Questions and Replies

  • |
  • |
  • |
  • |