Pondering a ducted attic minisplit air handler to replace the bedroom wall units
Hey everybody, we had an old cat infested 1200sf house in the Boston area gut renovated and have been heating and cooling exclusively with a multi-zone Mitsubishi mini split system, and while it has generally been working well, after two winters and one summer I’m thinking of making some changes. Hoping you can guide me through it.
Outside is a single 30k BTU condenser. The downstairs is about 750sf and is an open floor plan with a living room that has a cathedral ceiling, and has one 18k BTU wall unit. In each of the two small bedrooms upstairs is a 6k BTU wall unit. I have wireless remotes as well. The house is pretty tight from the entire envelope having been spray foamed with open cell. There’s a 5 kW solar array on the roof, and it actually supplied 75% of the first year’s electricity, but last year was a mild winter, this one was very cold so we’ll see.
My primary issue is that the bedrooms get too hot in the winter and too cold (and humid) in the summer because (1) the fans on these small units never turn completely off (2) I think the refrigerant continues to flow to all units even if the thermostat for the unit is satisfied (and even when the zone is powered off), and (3) the condenser and the downstairs unit are cranking away because the downstairs is bigger and leakier (e.g. opening the patio door to let the dog out). Basically, the bedrooms take no time to get to temperature, but then keep going because their fans don’t turn off and refrigerant is still flowing. At 5 am it’s 80 degrees in the winter or super cold and clammy in the summer. Also, the bedroom units make annoying cracking noises in the winter from thermal expansion/contraction after defrost cycles, even when the zone is powered off (see my point #2 above). My typical solution is to suffer through hardly running the bedroom units at all, or running them for a few hours before we go to bed and then turning them off. The downstairs is fine for the most part other than the heat being not very well distributed.
I’m thinking that replacing both of the upstairs units with a single 9k ducted air handler in the attic (which is conditioned due to roof rafters having been insulated) will help because (1) the lower BTU capacity should match the load better (2) my understanding from some searches is that the fan on this ducted unit will turn off when not calling for heat/cooling (3) pulling the return from the hallway between the bedrooms may keep them from short-cycling, and (4) hopefully less expansion creaking due to hopefully less cheap plastic for a unit not meant to be seen, and any creaking will be at the other end of a duct rather than in the room.
Other ideas include a small whole house dehumidifier in the attic, either in addition to the above, or possibly alone if I just want to solve the humidity problem. Or, supposing I kept the bedroom wall units but instead put them on a separate condenser? Another idea I’d love to discuss unrelated to the above is installing hydronic radiant floor heat under just the first floor, since these splits seem to get exponentially less efficient when it gets really cold, the solar array is not very big (small roof), and the single head unit on the first floor provides uneven heat given that the basement is unheated.
Am I thinking about this correctly? Do my observations make sense? Did anyone read this far? I could call the company that designed and installed my system, but I’d rather do some research first. Any recommended local consultants? Thanks in advance!
GBA Detail Library
A collection of one thousand construction details organized by climate and house part
Replies
User 698etc.,
First of all, can you tell us your name?
I don't advise the installation of radiant floor heat. Your heating system is already oversized.
I think that your plan to install a ducted minisplit for the two bedrooms (zoned separately from the great room) makes sense -- it's too bad the original installer didn't think of it. That leaves the system in the great room oversized, of course, but it will probably provide adequate comfort.
A whole-house dehumidifier is almost always a bit of a sledgehammer solution to swatting flies, even in places far more humid than Boston.
Calculating the actual room loads is the first step in finding the optimal solution. Swapping out a pair of half-ton heads for a single 9K mini-duct unit might be the right thing, or it might be a total sidestep. The entire problem may be tied to the minimum modulted output of the compressor, not the heads. Without knowing the actual model numbers it's hard to make more specific guesses, but even with that information it's all just a WAG until the load numbers are nailed down with some confidence.
Hi, thanks for the prompt replies. First, my name is Bob. First time posting, didn't realize it didn't put my name up. The units are MXZ3C30NAHZ, MSZ-GE18NA-8, and MSZ-GE06NA-8. The designer/installer is a local fairly large HVAC company that probably has done a lot of installs, but probably in houses where the splits supplement a conventional boiler or furnace system. Also, at the time we designed the system, I thought the plan was to blow insulation on the attic floor rather than spray the rafters. Also, I think we just presumed that ducts were going to be inefficient, and that individual control of the bedrooms would be important to us.
I know he calculated the loads, but he didn't share them with me. I could ask him for it, but I'd love to just calculate it myself. The house might have actually ended up a little different from when he calculated them. Can you recommend a reasonably priced or free program to calculate the loads? Possibly a rough estimate on the bedrooms can be arrived at based on their sizes at 14x10x8 and 14x8x8, one vinyl window each. The walls are the original 2x4's from 100 years ago, filled with open cell. We didn't go nuts aiming for net zero here, no money and no space to do that.
The radiant floor in the great room I was hoping to use to supplement the system when it gets really cold, as a backup if the splits were to ever shut down, to improve comfort overall, and to intentionally leak some of that heat into the unfinished basement which is otherwise unheated. Generally though, the single air handler in the great room provides sufficient heating and cooling without the humidity issues.
Thanks again and please let me know if you have any more questions.
> pulling the return from the hallway between the bedrooms
Be sure to review how many CFM you are moving and the pressures involved.
> the refrigerant continues to flow to all units even if the thermostat for the unit is satisfied
The manual confirms that this occurs. Over-sized and doesn't shut off is a bad combination.
Thanks Jon R, where is the Mitsubishi manual that says that refrigerant flows continuously? Would like to read up on the technical specs and operation. I would have hoped that the compressor would have valves to regulate flow. I asked my installer about these things but their replies gave me the impression that he didn't know.
It was in one of the Q&A sections (sorry, can't seem to find it now). No idea why it would do this.
While the fan in the indoor unit is reasonably variable, I didn't see any indication that it varies speed automatically. This makes humidity control cumbersome.
Load is not a function of floor sizes, it's about exterior surface areas & the construction of those exterior assemblies. A pretty good (but not perfect ) easy to use online freebie tool is loadcalc:
http://loadcalc.net/
It tends to overshoot reality, sometimes by quite a bit, especially if you aren't being aggressive on all of the R-value & U-factor assumptions. For instance, if it's balloon framed full-dimension 2x4s the center-cavity R-value with half-pound open cell foam is R15, but the lower framing fraction of balloon framing delivers a lower U-factor than R15 in a typical 3.5" deep 2x4 at more contemporary framing fractions. You'll actually get closer to reality if you select 2x6 R19 for the walls when using that tool if that is your construction type. Even then it usually overestimates the load by more than 10% when compared to a properly aggressive Manual-J using professional tools.
A typical tight 2x4 framed 1200' house with clear glass (NOT low-E ) double panes or clear storms over wood sash double hungs with at least R20 in the attic and a tight but uninsulated basement will usually have a heat load between 15- 20,000 BTU/hr @ +10F outdoors (a typical Boston area 99% outside design temp:
https://articles.extension.org/sites/default/files/7.%20Outdoor_Design_Conditions_508.pdf ) and +70F indooors. The load at +47F will be less than 7500 BTU/hr, maybe even less than 6000 BTU/hr.
The minimum heating output of the 3C30NAHZ @ +47F is 7200 BTU/hr, the capacity at +10F is over 28,000 BTU/hr.
http://meus1.mylinkdrive.com/files/MXZ-3C30NAHZ_Submittal.pdf
Installing non-hyper heating GE18 and GE06 heads to a hyper-heating -NAHZ head works functionally, but doesn't necessarily take advantage of the hyper-heating capacity. Getting to the real answer on that will probably take a deeper dive into the engineering manuals, but they probably SHOULD have been FE18 & FE06 if installed five years ago, or FH18 & FH06 to be able to use the full capacity & function.
When tied to it's own compressor the GE18 can throttle back to as little as 3500 BTU/hr @ 47F:
http://meus1.mylinkdrive.com/files/MSZ-GE18NA-9_MUZ-GE18NA-1%20Submittal.pdf
Just a guess, but as the main zone is approaching it's setpoint it's likely that the GE18 head is calling for less and less refrigerant, which forces the system to deliver remainder of the 7200 minimum output BTU/hr to the other heads avoid short cycling the compressor. Replacing a pair of G06s with a single KD09 would not fix that- the bedrooms would be still taking the same amount of excess heat.
Replacing the GE06s with individual FH06 minisplits on their own compressors would allow the to modulate as low as 1600 BTU/hr out @ 47F (each, 3200 BTU/hr total), and turn fully off when cycling. Replacing them with or a KD09 on it's own KA09 compressor would also work, with a minimum modulation of 3800 BTU/hr @ 47F, also with the ability to turn fully off when cycling.
https://ductlessamerica.com/pdf/SEZ-KD09NA4_SUZ-KA09NA_Submittal.pdf
http://meus1.mylinkdrive.com/files/MSZ-FH06NA_MUZ-FH06NA_ProductDataSheet.pdf
Without the other zones to dump excess into the 3C30 & GE18 would be cycling more, operating at lower overall efficiency than a GE18 on a dedicated 1.5 ton compressor, but it wouldn't have the overshoot/undershoot problems on the bedroom zones.
Being a large local HVAC company that has installed many systems doesn't really mean much, and it's highly likely that the installer really DOESN'T know how they work.
When a co-worker was soliciting proposals for his 1700' circa 1920 antique in Arlington MA proposals were all over the place most of them pretty bad, with gross oversizing and even less appropriate equipment than what you have. After running independent load calculations and specifying the equipment himself it STILL ended up oversized, but it was three separate FH-series mini-splits, with a ludicrously oversized FH09 serving the kitchen (at the insistence of his spouse), an FH12 in the main first floor living area (appropriately sized for the zone), and separate (but oversized) FH12 in the upstairs hall, directed at the (usually open) doors of a pair of bedrooms. The upstairs could have (and perhaps should have) been a KD09/KA09 but it was hard to find a contractor willing and competent to install it. The downstairs could have been a Fujitsu 18RLFCD in the basement too, but again, finding a contractor willing & competent to do it was the problem.
The contractor who got that job was NOT one of the bigger HVAC companies, but 4-5 person company specializing in mini-splits. The contractor willing to listen and was able to answer most of the questions without blinking or obfuscation. The reps from the bigger companies were almost universally clueless sales-droids blowing smoke, who would push back when questioned with higher quotes. When asked to bid on the equipment specified by the homeowner one of them came back with an number nearly twice what the winning bidder quoted. The winning bid was not lowballed, and was not the lowest bid- there were others in that range, from contractors with longer time schedules or potential questions on competence.
Thanks so much Dana that's very helpful, and consistent with what we've been experiencing.
Turns out my units are GL, not GE. Not sure if there's a difference. He quoted the GE's before the GL were available, this was back in late 2015. He at first said that the FH's were not meant for multi-zone application so weren't available for me, and said that the efficiency comes from the compressor not the head units. He also later offered to upgrade to the FH's though (for more $$$). I say that he's from a large HVAC company in part to say that they probably don't know what they are doing. Should have asked here first obviously.
Ah so the compressor is actually directing refrigerant selectively then. But when the GL18 approaches its setpoint, the GL06's are already well past their setpoints. Why does the GL18 "win" then? Or does that refrigerant just go everywhere? Also, I think the other problem is that the GL06 fans never turn completely off, so that's one reason the heat keeps going up in the bedrooms. The KD09 apparently will turn its fan off when the setpoint is reached, and since it would physically be in the attic, wouldn't the bedrooms not heat up excessively even if the compressor is dumping refrigerant on the KD09?
> super cold and clammy in the summer
> the GE18 can throttle back to as little as 3500 BTU/hr
But with a high SHR under conditions that need low SHR (more % latent removal). AC comfort depends on both (sensible and latent) - BTU/hr alone isn't a good metric.
Since all of the control algorithms are proprietary, and vary from model to model (and even from year to year within a particular model) it may not be possible to fully explain the behavior.
But at the bottom line, this sort of thing is a MAJOR drawback to multi-splits relative to single head/cassette mini-splits.
The hyper-heating (but not the efficiency) is solely a function of the compressor type, but most of the control "smarts" reside in the heads/cassettes. Even if it works functionally, there is no guarantee that a GE or GL head will run the compressor optimally for either capacity or efficiency. They don't market GE or GE heads paired with FE or FH series single zone hyper heating compressors, and there's probably a reason for that. While the MXZ-__C__NAHZ series are "compatible" from a basic control point of view with GE & GL heads, compatibility is different from being optimal.
Then again, in the operational performance tables they don't specify the models of the indoor heads, only the sizes:
http://meus1.mylinkdrive.com/files/M-Series_MXZ_Operational_Performance.pdf
On page 90 they specify any 6K + 6K + 18K combination will deliver 5800 BTU/hr per 6K head, 17,000 BTU/hr for the 18K head, but those are probably AHRI nominal @ +17F, not the max capacity, which is likely to vary by the serie of head.
I'm a bit surprised that the GL06 heads don't stop the fans (or refrigerant) completely when married to an -NAHZ multi-split. I would think that a rational control algorithm would do that, even if it forced the GL18 to run higher than it's minimum speed whenever the other zones were set to "OFF".
Have you buried yourself in the installation & operation manuals on both the 3C30NAHZ compressor and the GL heads? (I haven't.) On p.EN 9 (p10 in PDF pagination ) for the GL heads it mentions "• A small amount of refrigerant continues to flow into the indoor unit even though it is not operating." but doesn't indicate that the blower would still be running. The blower might be set to run continuously (is a true OFF even an option for the GL heads?) , which could be 100% of your problem. To mitigate the clammy factor, run all three GLxxNA heads in DRY mode during the cooling season and set the GL06 heads to only the lowest fan speed, not "AUTO". That should be less clammy, even if it doesn't fix the temperature setpoint undershoot factor.
Jon R: All Mitsubishi heads/cassettes have a "DRY" mode to enhance the latent cooling fraction (at the necessary expense of some SEER efficiency.)
And I was CLEARLY addressing the HEATING mode ( modulation range at 47F) not the cooling modulation (which can only drop back to 3700 BTU/hr for the GE18). The GL18 has even less modulation range than the GE18, with 5400 BTU/hr min @ +47, 5800 BTU/hr cooling:
https://nonul.mylinkdrive.com/files/MSZ-GL18NA-U1_MUZ-GL18NA-U1_ProductDataSheet.pdf
Thanks Dana, the GL units do turn their fans off when the mode is set to OFF, but they do always run at a low speed when they are set to heat or cool even after the set point has been reached. I've read in forums that this is in part so that the temperature sensor on the unit can continue to sense room temperature. However, in spite of my having MHK1's on two of the three units, and having them configured to sense temperature on the remote itself, the fans still run. I've also learned from the MHK1 installation manual that there is an installer setup options 125 and 127 "Thermal OFF Fan Operation" for heating and cooling mode respectively. These would be exactly what I need, but neither of these functions are available, I presume they are not possible on these GL units, or possibly all wall units. My hope, and I'd love to find out, is that these options are available on the ducted KD09 unit. Who would know this for sure?
I do now see the note in the installation manual that says a small amount of refrigerant continues to flow. I've read most of the operation, installation, and some of the service manuals and haven't been able to find much more of interest.
If the KD09's fan will shut off, possibly an option which salvages most of what I have is to replace the two GL06's with a single KD09, and possibly put one of those GL06's in the currently unheated basement which I want to partially finish. I'd be happy for any excess heating or cooling to be dumped into the basement.
I've been on the loadcalc.net website but need to research details of my building construction better in order for it to be more accurate, and some of the terminology I don't understand. Thanks again everybody, I'm learning a lot and hope to have a better solution in place once it starts to get hot and muggy. Also, any recommendations of installers or consultants in the Boston area would be most welcome.