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

Mitsubishi heat pump gives up below 20˚ F

tdechant | Posted in General Questions on

Our Mitsubishi heat pump at our “pretty good” split-level ranch has been giving up when the temps dip below 20˚ F. I’ve been racking my brain trying to figure out what’s happening, though I admit I’m no HVAC expert. Here’s our setup:

MSZ-FH09NA (living space)
SEZ-KD12NA4 (bedrooms)
SEZ-KD09NA4 (lower level)
MXZ-3C30NAHZ (outdoor unit)
MHK1 thermostats for each indoor unit

What’s puzzling is that the system performs just great above 20. We’ve had several days where the nighttime lows have dipped into the low 20s, and it kept up with the set point just fine. It could get a little cold in the living space, but cranking the fan to max brought it right back up. 

But now that we’ve had a minor cold snap (last night’s low was 10, today’s high was 20), it’s almost like the system has given up. When we woke up, it was 5˚ below set point in the bedrooms and 8˚ below set point in the living space. As the day went on, it got worse. I finally turned on our backup baseboards at 4:30 to take the edge off when the kids got home. Based on last year’s super helpful thread on cold-weather heat pump performance, that doesn’t seem right.

I’ve checked the outdoor unit both during regular operation and during a “wait” cycle, and there isn’t any ice on it. Upstairs it’s about 60-62 right now, and both the mini-split and ducted unit putting out 70-71 air at best (measured with an IR thermometer). 

Any ideas on what’s going on?

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.

Replies

  1. brettske | | #1

    First thing I would think of would be refrigerant level. I have a mitsubishi hyperheat that had a lot of trouble heating recently and would not blow warm air (below 90 degrees...not enough to heat the room). Some refrigerant had leaked out due to a loose connection. Tech refilled it and it was pumping out 120 degree warm air.

    Brett

  2. tdechant | | #2

    Thanks, Brett. I'll have them take a look. How long did it take them to track down the leak?

  3. bfw577 | | #3

    You almost certainly have a leak at a flare fitting. You have 12 on your system and they are the number one source of leaks.

    Are you saying the supply air coming out of the units is only 70-71? Thats really low. On my 12k Midea and Gree I have seen supply temps as high as 146 at max capacity. They usually cruise around 120.

  4. tdechant | | #4

    That's good to know, BFW577, thanks. .When you say 12, are you talking about the SEZ-KD12NA4?

    1. bfw577 | | #5

      You have 1 outdoor unit with 3 heads connected? Each unit has 2 flares on the inside unit and 2 outside. If so you would have 12 flares.

      When you get it fixed I suggest you record the supply temps at max capacity. Turn every unit to fan high and set the setpoint to like 86 or the highest it will go. It will then run at max capacity. Use a thermometer to measure the duct output and take note of the outside temperature and make a chart. I have monitoring equipment on mine and had a leak in a flare. I watched the performance gradually drop and it was as clear as day as duct supply temps dropped over time. It is critical for these units to have the exact refrigerant charge. A couple ounces low and these units drop off a cliff in performance.

      Here is a graph of my 12k Gree Sapphire supply duct temperatures.

  5. _jt | | #6

    Did you look at the condenser?

    I have a MrCool unit (i.e.: not the best, not really in the same class as the name brand units) - it's been running great in the low teens. Air has been coming out around 120, and power draw is typically around 1000 watts.

    That being said, a few weeks ago, it was 2000 watts and not keeping up. Looked outside and the condenser was frozen over. It turned out that it was placed under my roof in a way that rain and roof melt fell directly in front of the condenser, immediately freezing. We got a 1 inch thick slab of ice. Condenser didn't give up - but started working very inefficiently.

    Easy to fix with a hat. :) Now it's been working well.

    1. bfw577 | | #7

      Be careful running it in cold weather. I looked at those Mr Cool units but they do not have basepan heaters to defrost ice. You have a very high risk of ice bulging and cracking the condenser coil open. The unit would be ruined if that happens. Do you have the DIY model where it comes with the precharged linest that just clicks together? The Mr Cool are actuallly just rebadged Mideas along with the countless other chinese brands.

      1. _jt | | #10

        Yes - I have a Pioneer too - exactly the same. Not the DIY model. Although my installer did have to add more freon to the unit than the pre-charged lines had. So pre-charged lines aren't that perfect.

        I've been doing the daily inspection - which I recommend everyone should do.

        Once I put the hat on no more ice build up. Defrost mode has been sufficient to prevent ice build up. I've been estimating (given ballpark CFM + IR outlet temperature + loop ampmeter) - COP above 2 even when temps are in the teens. (Better than advertised as far as I can tell.)

        Rated temp is 5 F - which we don't really get where I am. (15 F is 99% temp) At that temperature I think the pan heater would make a big difference.

        It does seem to make a big difference to run the wall unit fan at high speed all the time. (Both in terms of the theoretical calculation as well as room comfort)

    2. bfw577 | | #11

      I have monitoring equipment on mine as well and found running the fan on high all the time increased efficiency by 25+ percent. It also heated my house much more evenly.

      1. _jt | | #16

        You were right. This morning power usage jumped up, went outside and ice is building up on the drain pan. Will have to either increase the slant (the other mini split is fine) or work out a better way to defrost it moving forward.

        From what I can tell power and energy monitoring is key for efficient mini split usage.

  6. walta100 | | #8

    Asking how long it takes to find a leak is like asking how long it takes to find a murderer, every case is different some are obvious and other can never be found.

    You best hope is to have the installer find the leak. If he takes pride in his work he will be ashamed to admit there is a leak but once he is convinced and does the hard work required to locate the leak he is unlikely to charge you for all the time he will spend looking for it.

    Leaks go back to your selection of contractors and the best workmen make junk equipment into a great system and poor workmanship can make the best equipment into junk.

    I think it is a good idea to keep records of your systems performance at several different outdoor temps and record the supplied air temp at high speed operation. So you will know your system is slipping.

    Walta

  7. jameshowison | | #9

    So, if a system leaks refrigerant what's the equivalent carbon emissions? I know that R410a is something like 1,700 times as bad for climate change as straight carbon (GWP).

    Anyone want to make an estimate of (for example) miles driven in a car or number of trees that would need to be planted to make up for a typically sized refrigerant leak?

  8. tdechant | | #12

    Update: A tech came out to inspect our system, and while it’s a little low on refrigerant, it’s not too far off spec, so it’s unlikely we have a leak.

    In the meantime, when it gets cold, we’re going to try setting the MHK1s significantly higher. The heads all seem to put more effort into warming the air when the set point is farther away from room temp.

    One last note: the tech said the new units run a defrost cycle on a timer. If the coils get quickly back up to temp, then it drops out of defrost and goes back to warming. If it doesn’t warm up quickly, that’s a sign of frost on the coils so it continues the defrost cycle until it’s warmed up. My hypothesis is that the timed defrost interrupts the usual slow-and-steady warming that heat pumps normally do, and for whatever reason it doesn’t make much of an effort to catch up. Though that’s all speculation.

    The tech and his Mitsubishi expert at the shop were both stumped by the behavior, and since it had warmed up a bit outside by the time they arrived, weren’t able to replicate it.

  9. bfw577 | | #13

    Did they evacuate all the refrigerant from the system and weigh in the proper charge? That is the only way they would know it was low on refrigerant. These units absolutely need the exact refrigerant charge to operate efficiently. I would find out if that was done.

    Also, I have read many service manuals and technical literature and no mini split I know uses a timer for defrost. Those are tech from like 20+ years ago. The hit on efficiency is pretty large that manufactures would not use a wasteful defrost timer. The software knows when to defrost based on inputs such as the outdoor fan load increase, a drop in outlet temperature and condenser temperature.

    Did you measure your duct temperatures yet at max capacity?

    1. tdechant | | #14

      He wasn’t able to evacuate and recharge the system because he didn’t have 410a on the truck, but he did do a capacity check when it was about 30° out—it was working up to spec.

      1. Expert Member
        BILL WICHERS | | #15

        You can check if the refrigerant charge is correct by carefully measuring operating temperatures too, you don’t have to weigh it. You need a competent tech though.

        Regular techs should have “wands” that basically sniff out refrigerant leaks. They’re basically refrigerant gas detectors on sticks. Medium and large leaks will set those wands off and make quick work of a leak check.

        Smaller leaks are tricky. If you think you have a small leak, ask the tech to put some of the UV leak detection dye in the system. Wait a few days or a week and then use a UV lamp to look for the florescent stain that indicates a refrigerant leak when there is dye in the system. Don’t wait too long to check since the dye eventually degrades.

        Classic leaky spots are flare fittings, places where refrigerant lines pass through a clamp, and brazed connections near bends entering fittings or devices. I’ve seen issues with TXV valves too, but I don’t know if your unit would have one of those.

        Bill

        1. tdechant | | #17

          Thanks, Bill. That's great info. He did a cursory sniff test for leaks at the usual spots, but didn't find any. After he did a capacity test, he didn't think there was a high possibility of leakage. That said, if we continue to have problems, they'll check again.

          Temps are back up outside and the system is performing great again. It was around 17 this morning, so to bring it back up to temp faster, we put the set point around 6-7˚ above our actual target. That helped a lot—the air coming off the evaporator coils went up 10-15˚ in just a few minutes. I'm beginning to suspect there's something wonky with the way the MHK1 thermostats are programmed. I understand that heat pumps work better slow and low, but when it dipped 10+ degrees below the set point, it wasn't showing any urgency.

          1. bfw577 | | #18

            I still think you should measure the duct outlet temperatures. Mitsubishi has all the service and technical manuals online available to anyone. It most likely has the supply air temperatures listed and you can compare.

  10. tdechant | | #19

    Quick update: We've been able to maintain good temps when it gets down to 10˚ F at night by setting the thermostats significantly higher than the desired set point and placing the mini-split fan on high. When it gets that cold, we set the mini-split (which serves about 750 sq ft or so of cathedral ceiling space) to 85˚. The ducted unit for the bedrooms we set at 75-78˚ with the fan on auto. Together, that keeps things pretty comfortable (about 66-68˚).

    Why the MHK1 thermostats don't work as actual thermostats to work is beyond me. I'm going to call Mitsubishi to see if they have any answers.

    1. joshdurston | | #20

      There is a setting in the MHK1 setup to choose whether the temperature is sensed at the remote stat or the built in return air sensor. Perhaps you're measuring locally still?

      1. tdechant | | #21

        Good question. I'll have to look into that. Could be the issue with the mini-split, though it's not placed at the top of the cathedral wall (rather it's at the bottom slope of the ceiling about 6-7' off the ground). Regardless, something to investigate. Thanks.

Log in or create an account to post an answer.

Community

Recent Questions and Replies

  • |
  • |
  • |
  • |