4 wire control cable, How does the control work? Mini split heat pump.

psdonald42

Hello,

First time poster.

I have searched the web for this information and have come up short.
I am a controls guy and get to integrate lots of cool discrete units.

On a mini split heat pump the indoor units "talk" to the outdoor units via the 4 wire control cable.
The four wires are, ground, AC, AC and Control.

So one wire for the control to turn it on and off and ask for heat or cool.
Does anyone know if this is level based ac, serial coms, or something else.
I could go in and test the wire but I'm not able to access that right now.

I did control one system by mimicking the IR remote and learning all the settings and then sending the IR signals to each unit from a PLC. This worked for 4 years but has too many moving parts. Simple is better.

Any thoughts?

Peter

Jamie Hall

Comment went to la la land. Try again here.

Some of the most recent and sophisticated systems may use a serial bus, similar in concept to the CAN bus used in recent cars.

However, as a general rule, in our trade one is playing with 24 VAC as the control power, and the controlling device will either be switching a line up to 24 volts or dropping it to ground -- or in many older devices simply closing a circuit between two of the wires without any regard for the voltage on the wires at all.

EBEBRATT-Ed

@psdonald42

Your correct 2 wires AC power, 1 wire ground, the other wire starts and stops. I believe the 3rd wire in that case is AC 120 volt.

Some units depending on the brand require an ac circuit to the indoor unit and another ac circuit to the outdoor unit with a communication wire between the units

Some units the ac power goes to the outdoor unit and then a 4 wire (thwn) or (control tray) (TypeTC cable) is used between the units which it sounds like is the case you are dealing with

ratio

Many (most??) mini splits communicate via a signal imposed on a conductor that is at line voltage to ground. It's not like a standard condenser, energize the control conductor to call for cooling, it's actually bidirectional communication.

I saw somewhere on the internets where someone had reverse engineered the Mitsubishi physical layer & was working on interpreting the data, but it hadn't been updated in years.

Some mfgrs have an adapter that will take standard Y, G, etc. calls & communicate them to the condenser to allow a mini split condenser to operate a standard evap coil (some changes to the coil are necessary).

Jamie Hall

A minor comment, if I might -- if one or more of the wires to which folks are referring is 120 VAC, and the rest are signal or 24 VAC, they cannot be in the same cable or conduit. Nor can the 120 VAC and a 24 VAC line share a ground.

I believe (someone check the latest code, which I don't have) that 24 VAC hot, ground, and switched signal lines can share a conduit with a "USB" or CAN type serial bus or ethernet bus, and the latter three can have a 5 volt DC reference and ground in the same cable.

Personally I wouldn't do it. If I were running a high speed digital line, I'd want it separated physically from anything much else, and at least twisted pairs if not shielded and twisted pairs.

psdonald42

Good information people.

I will attempt to measure and decipher the 4th rail when i get home. The power to the wall unit is i'm pretty sure ground, N, Hot and signal.
Now the outdoor unit is variable speed and slows and speeds up according to delta from setpoint, and can also be heat or cool, so something sophisticated is going on.

I can think a a few ways of doing this without coms. How about 24VDC signal, 0 to 2V off. 2 to 12 V cool with 12 being full cool, 12 to 14v off 14 to 24v heat. 24V full heat. Of course it will be nothing like that.

Any guesses?

Peter

HVACNUT

 You get continuing  swinging AC voltage between 90v and 240v between terminals 1 and 3.
I don't know if that helps or can be used to control another source. 

ratio

I suspect that all the different brands speak a different dialect, if not language; but here's a start:
https://github.com/LenShustek/M-NET-Sniffer
I've installed two units that actually do 'communicate' like a standard split (Y call for cooling), a Carrier 4 ton mini split, and some other smaller one, I forget the brand. Everything else that I've worked with is communicating, referenced to a hot conductor (which puts it straight on Sparky's plate, no matter how much he argues), or via RS485 (VRF/VRV systems, the big uncle of mini splits).

pecmsg

Jamie Hall said:

A minor comment, if I might -- if one or more of the wires to which folks are referring is 120 VAC, and the rest are signal or 24 VAC, they cannot be in the same cable or conduit. Nor can the 120 VAC and a 24 VAC line share a ground.

I believe (someone check the latest code, which I don't have) that 24 VAC hot, ground, and switched signal lines can share a conduit with a "USB" or CAN type serial bus or ethernet bus, and the latter three can have a 5 volt DC reference and ground in the same cable.

Personally I wouldn't do it. If I were running a high speed digital line, I'd want it separated physically from anything much else, and at least twisted pairs if not shielded and twisted pairs.

If all the wires are the same size and type of insulation then Yes they can be in the same conduit, chase, Harness.

Fujitsu's I run 18/4 cable.

ratio

@psdonald42, this's been tickling at me for a while, I finally realized what it was—Mitsubishi has BACnet/Modbus (RS485) & Lon interfaces available, no need to reinvent the wheel, see http://mylinkdrive.com/USA/Controls

psdonald42

I checked the Signals for my Diakin indoor unit.
4 wires. Black, white, red, ground.

Black to White 240VAC
Ground
Red to White Pulsing AC signal of up to 50VAC. Looks like it transmits data every 2 seconds.
I'm just using a multimeter so I cannot determine any better information than that. I would need an oscilloscope for more.
I tried various settings but it just looks the same.

ratio

You'll almost certainly need a 'scope to see what's on the wire. It's serial communications, mark & space (that'll need decoded into commands/replies), not a simple on/off (like a standard furnace) nor even a level-based one like your comment up-thread.

MLJ

Have not troubleshot a mini-split in a while, the four-wire are used for communication like what was said...but the strange thing is some are a d/c voltage like a 190 volts d/c... that's all I remember

Was putting one in for a school that was in a union strong city, nothing wrong with that I was union. The electrician insisted it was his job to wire it up, even the wiring between the outdoor unit and indoor.

He ran 12 ga wire per city code he said. The unit would not work was getting a communication error.

Called Mitushubishi to figure out what was going on, they said the 12 ga wire was too big.
Told this to the electrician, of course, he was not happy. He took out the 12 ga put in 14 ga the mini-split worked fine.

ratio

@psdonald42, take a look at https://coolautomation.com/products/coolplug-coolinkhub/

EBEBRATT-Ed

Putting in smaller wire couldn't possibly fix the problem. If it did the electrician had something connected wrong

ratio

I've heard they're sensitive to wire gauge, as well as stranded vs solid, as well as splices in the conductors. They must've gotten better, because I find Sparky feeding the indoor units with #10/3 solid MC from time to time without any issues (that occur in the first year).

archibald tuttle

rebooting this thread because it seems kinda close to what i'm looking for and some informed speculation on my question is already here.

Silly me, i figured you look at the installation manual and it will specify and range of code approved field wiring but no. Here is typical lingo from an LG install manual:

"All communication / connection (power) cable from the outdoor unit to the indoor unit is field supplied and must be a minimum of four-conductor, 14 AWG, stranded, shielded or unshielded (if shielded, it must be grounded to the chassis of the outdoor unit only), and must comply with applicable local and national codes. "

I cannot find a NEC section that deals with Minisplit/heat pump raceways that combine refrigerant lines and cables. Diversitech said i'm the only person who ever asked about the wire coating temperature requirements esp. vis-a-vis raceway fill (of course they don't even admit that what they are making is a raceway and were like deer in the headlights when I pointed to generalized raceway specifications in the NEC, so we already aren't speaking the same language). I seem to be the only one who asks about a lot of stuff . . . drives my wife crazy which is why I come here with my questions :-).

Of course, if you go to the supply counter and say you want the wire that goes between the indoor and outdoor unit, I get handed 14/4 stranded thhn which has 90C (and this being Amuurica I wish they would make those like 194 F but instead its FU and I have to double it, take off 10% and add 32, which I guess helps stave off alzheimers . . . now what was i saying . . .) but @EBEBRATT-Ed mentions 14/4 THWN which is 75C (167 F) wired a also calling it generically TC (Tray Cable) wire (which also leads one to ask if there is a difference between a tray and a raceway. I believe the NEC says a cable tray is a system for supporting either cables or raceways. I also see trays sometimes referred to as having a divider to isolate power and communication wires).

And LG clearly crosses over communication and [power] connection per the language from the install manual quoted above, but that doesn't mean they cross voltages excepting that I think some of the comm and/or connection work is done by effectively varying the voltage by pulsing or who knows, if one looks at the meter measures mentioned in this thread. Now I don't think this would be considering mixing voltages as in running 24V signal in the same cable but some control function besides just on and off is exercised with a modest number of conductors, vs. the good ole days when you would have had more conductors that hit different points in the motor winding to get different fan speeds. Regardless of control strategies that might vary voltage I would think that 240V outdoor units that have no neutral wire are feeding up to 240 to the indoor unit and ought not to be referred to as 120 except in the case where the outdoor unit is native 120v. Of course none of this addresses whether it is AC or DC that is transmitted on this cable.

In any event, I'm assuming the amperages they list in the install manual are the max that must be sustained even if the effective amperage is sometimes/often less. Which gets me to another question, which is, I simply can't wrap my mind around the notion that a wire could be too thick of a gauge. It might be a waste of copper but I don't see how it couldn't work unless the control strategy is based on taking advantage of a known rather than measured resistance in the wiring? And depending on how specifically some return voltage drop or whatever surrogate was measured even a known gauge is going to exhbit different properties based on length of run.

I can try to transfer the ampacity from the LG tables to the NEC table 310-16 but, in this case, i'm being offered a roll of RVV wire which isn't even in the tables. Maybe I should take that as a sign it isn't considered suitable for raceways, but It is 70C 158F) rated wire (which isn't even a temp in the table so there is an interpolation problem , or I could just conservatively go with the 60C 140F ampacity). It is 'soft sheath cable' and I imagine, although I can find virtually no uniform definition, that it's principal use is as temporary cord, but if it meets the temp requirements and doesn't overfill the raceway i'm not sure that it would be inappropriate for install other than meeting up with the inspector who says that ain't the right stuff . . . ?

well, anyone who read that novel deserves the forebearance merit badge.

brian

ratio

Johnstone & I'm sure others sell 'mini split cable', which seems to be 14/4 type TC, which is cable tray cable. IIRC the Code accepts it outdoors; & indoors in raceway or cable tray. The first time I saw Sparky use it, he got dinged for no raceway indoors, but lately he hasn't had any problem with just taping it to the line set (possibly because the inspector hasn't noticed).

Kinda sounds like you're overthinking things. They've been using mini splits in other parts of the world for ages, most|all of the subtle bugs should've been worked out by now.

archibald tuttle

Right, the TC 14/4 is available THHNat the counter i deal but maybe THWN is ok also .... my question distills to , what makes TC wire TC wire. What temperature rating, heat transfer , flexibility, and  assumptions on raceway fill, etc. If i have half price option for roll that will handle my next dozen jobs or more that is a  somewhat parallel alternative that could meet specs of that wire available at the counter, should i say screw it and pay double?

Jamie Hall

Actually, I can imagine that in some cases the different self-capacitance of larger gauge wire might be enough to defeat the control signal -- effectively short it out -- since the chances are excellent that the control signal is a bidirectional digital signal (off and on) in packets, and operating at a very high frequency. You will definitely need an oscilloscope to see what is happening. Start payimg some attention to some of the better YouTube automotive channels (suggestions: Raiman Ray, Pine Hollow Auto Diagnostics, South Main Auto) to get some idea as to how these things work. They are most emphatically NOT just on or off.

archibald tuttle

super interesting thought and i adverted to that in thinking about a strategy that is meant to take advantage of resistance. i guess what I don't get is why this is a preferable strategy to the option of running separate primary voltage (240 or 120) to the indoor unit and then using signal wire between the units to control fan speed and component monitoring, the historical methodology for air conditioner operation.

i also can understand sensitivity but if it is going to be so sensitve that going up to 12 gauge is problematic, wouldn't a really short run of 14 have the same problem. Maybe the unit where @MLJ thought upsizing the wire had possible caused the problem, or at least it didn't work with 12 gauge but worked with 14 gauge was a really short run?

This kind of control strategy seems a little too clever by half. But maybe there is some great advantage gained besides just engineers self-satisfied with having made it work.

Probably in the end its being again too clever by half to think of measuring the resistance of the installed wiring and if it is too little adding a resistor?

GW

Mitsu says do not run 12 gg on the control wiring.

Jamie Hall

It's not a problem of resistance, @archibald tuttle . It's capacitance. The control signal isn't a slowly changing voltage at all -- it's a stream of "bits" (ones or zeros) which stand for numbers or commands -- or questions. The frequency is very high -- into the megahertz range at least in some automotive applications.

EBEBRATT-Ed

TC is Tray Cable. It is approved for use in tray cable and in a raceway like conduit.

However, there are a few exceptions that allow its use in residential, and it can only be used for two installations that I know of:

For minisplit control

for emergency generator control.

Inside the building it can be run with the same rules that apply to Romex.

I know many use it for generators and minisplits in commercial work but I do not think it is approved for that... I will have to reread the code

ratio

Most (all) of the insulation I see nowadays is multi-rated: THHN THWN THHW MTM, all on the same wire. I suspect modern plastics outpaced the insulation classifications some time ago.

If you're interested in exactly how Mitsubishi xmits, take a look at the links I posted way back at the beginning, at least one of them includes a schematic for a receiver along with a quite understandable description.

I've seen some pretty f'd up wiring for mini splits, including the aforementioned #10 solid MC, #10 stranded THHN, a mix of #12 MC & stranded, 14/4 TC, & once (I think) a single conductor from S3 coupled with S1 & S2 coming from the feeder j-box. That got changed. Everything else worked. It's gotten a lot better as mini splits became more common, I haven't had Sparky scratch his head & look at me funny when I ask him if he's wired one before in a long time.

EBEBRATT-Ed

The only hanger now is the disconnect issue. Every motor driven appliance needs one in site withing 50'. Nobody puts them in much for the indoor section except I used to. Some inspectors look the other way

Jamie Hall

Some makes run the power to the indoor sections from the outdoor unit, and thus is controlled by the disconnect for that.

ch4man

think of an AC sine wave, the peaks are at 120 volt potential. between the peaks its at 0 volts. this happens every 120th of a second. it is at this exact moment the data wire can send its digital signal on a hot line.

more or less.

this is how a Mitsubishi factory training rep described this concept to me

ratio

EBEBRATT-Ed said:

The only hanger now is the disconnect issue. Every motor driven appliance needs one in site withing 50'. Nobody puts them in much for the indoor section except I used to. Some inspectors look the other way

If I'm sure that the inspector will insist that a disconnect be placed in sight of the indoor unit, I suggest to Sparky that he bring the outdoor unit feeders nearby & break those rather than the conductors coming from the outdoor unit. There will never be a case when I need the indoor unit off while the outdoor unit stays on. Also, this trick will use a (comparatively) cheap two pole switch, as opposed to a three pole switch or manual motor starter.

Google 'zero crossing control' for details on how Mitsubishi & oters talk on the power line.

archibald tuttle

@Jamie Hall if you keep beating the dark of night long enough, dawn breaks over Marblehead.

You're stretching my theoretical understanding. If I read your reference to capacitance in concert with zero crossing control references, I'm a little confused by the mitsubishi reps suggestion that signaling is at the 0 in the sine wave but that may be a convenience to what I understand to be syncing control to 0 in the sine wave.

Best as I can tell the zero crossing control isn't about signaling when voltage hits zero (although maybe that's a thing of a different name . . . ?) but timing current delivery by solid state relays (although some slip of the finger or autocorrect says "solid state replay") to the AC wave not possible with slower analogy relays (or pulsed DC that imitates and AC wave but only on one side of the horizontal zero volt x-axis).

By switching in time with the 0 point controlled by some other component that can select for taking part of each wave you control the input (and my guess is apparent voltage although the various sources I studied don't say that, i'm inferring it from the truncated waves depicted. I got two sources that focus on two different aspects/applications. The first analogizes this process to a lighting dimmer and can adjust the application of power to the load by milliseconds following the zero point. The second would suggest that method is contraindicated as it causes a voltage spike by connecting the load at a point when the voltage is high.

However that first example the switch seems to open on its own when the voltage hits zero so in order to take part of the wave rather than the whole thing it has to close part way through the wave above (or below) 0. And maybe the switch dropping out at zero given the speed of the control strategy is just an easy way to engineer it.

The second source suggests this method is most simple for controlling resistive loads but can also be used on inductive loads (motor control) but is more complicated to use for capacitance loads.

There is the hint, one would imagine, if the wire itself is presenting significant capacitance that it could be a problem for the control strategy.

albeit in these 'classic' applications presented, i'm not sure why the capacitance wouldn't just act to extend that application of voltage beyond the switched regime so you wouldn't have as precise a control but it would still work . . .?

so much for yours truly inductive (sorry) understanding of what is going on.

so i still wonder:

1. is this just way cheaper, more reliable, more precise, more electrically efficient or otherwise advantageous than using some kind of variable transformer/resistor to control motor speed?

2. what is the tech expected to do about any of this? do we carry oscilloscopes and then compare output on the load wire to some anticipated pattern or give up on instrumented diagnosis and just keep R&R bits on a best guess basis until it works which is where I'm stuck on auto mechanics until I go study the automotive oscilloscope videos suggested by @""Jamie Hall"

3. short of starting over from ohm and farrad calculation (not that there is anything wrong with that and maybe I just need to try to get the zen of these algorithms) is there a way to predict what wire sizes, formats and lengths would actually manifest problems or unacceptable inaccuracies.

4. finally, the actual application based question that brought me to the wall in first place on this question: the wire i'm looking at is 12/4 but highly stranded (i'm trying to get the stand count for 12 but a spec for simliar "RVV" 16 gauge multi conductor cable is each conductor has 45 strands (and is promoted inter alia as "speaker cable") whereas the THHN spec if for 7 strands. At first blush this increases the resistance and slightly lessense the capacitance per length because the actual copper crossection for any given gauge is less in a stranded conductor although maybe the bunch is actually slightly larger than the gauge and you actually still have the cross section of the copper. On the other hand I always had in the back my mind that conductance is increased in stranded cable because the current actually travels on the outside of the copper and there is much more outside circumference of copper in a stranded set up vs. single strand.
And then there is the question of the twists of conductors which you often see, if not at the frequency of twists in signaling cable still relevant to total length of copper vs. nominal length of cable.

This cable has specs similar to extension cord wiring (300V insulation rating) highly flexible and UV resistance coating. It might be similar to a PLTC-"power limited tray cable" rated although I don't think it carries the rating. I know there are some crush resistance requirements for TC and PLTC because it is approved to run out of raceway as well–as long as supported every 6 feet, if i'm reading right. And you certainly see it used that way for the 'last mile' to outdoor compressor units often.

Leading me back to wondering if there is a list somewhere of the collected specs for TC wire.

more than nuff said, as usual

brian

brian

Jamie Hall

Honestly, @archibald tuttle , I'm not that familiar with the communication protocols used in heat pumps and the like. In almost all automotive systems today, though, the communication is all digital, with the information or instructions being carried by a serial bus as a string of "1"s and "0"s, sometimes referenced to 5 volts and sometimes to 12 volts. The 1 is commonly a wide pulse, the 0 is a narrow one. The frequency of pulses is very high -- as I say, often in the megahertz range. The information is organized into packets -- a lead sequence saying something like "this is the temperature or the coolant" followed by the value. The beauty of it is that one wire pair can carry a huge number of signals from a variety of devices -- and just as well deliver instructions to a huge variety of devices ("hey you there at the purge valve, turn it off") and is almost completely insensitive to lower frequency electronic noise and, to a remarkable extent, variations in the actual voltages.

Sometimes the pulse train is variable frequency and fixed pulse duration (often used for rotating devices, to detect the rpm), but it is the same principle -- and that pulse train is converted into a digital signal by one of the computers.

Analogue outputs, such as how fast a device should turn or what pressure a device should maintain, are created by a computer analyzing the digital instructions and converting them to analogue signals. Similarly, analogue inputs, such as a pressure or temperature, are converted close to the device to a digital signal and transmitted along the bus that way.

In a heat pump, I can easily imagine the same principle -- with one wire pair carrying instructions to the remote unit and information back from it to the central unit -- and computers at each end converting those instructions into analogue outputs if need (for instance, for fan control). That is often done by various SCR or Triac circuits, which do use the zero crossing timing of an AC current as a reference point.

You might also be interested in looking at how Class D audio amplifiers work. Fascinating gadgets.

ethicalpaul

I'd be surprised if the minisplit comms were in the megahertz range just because it's all unshielded wire running next to other wires, right? Plus there is no need for that much bandwidth...these signals are like "turn stuff on", "turn stuff off", not streaming HBO Max