Fluctuating temp during shower

[DustinH]

My wife and I just had our old water heater replaced with a Takagi
T-K4U-OS-N. When showering, I noticed the system tends to go from hot to luke warm to cool and then back to hot again. Also, when we initially ran a bath for our daughter, the water started out hot then just went completely cold and wouldn't come back. Now, this morning, my wife took a shower and the water went from hot to completely cold and then back to hot.

I've been doing research and want to see if I'm headed in the right direction. I set the water temp down to 115 from 120 as my first step (we live in California so it's not terribly cold but, the ground water in the morning is currently being measured at 60 degrees). I figure that could help with flow. According to the spec sheet from Takagi, a 55 degree rise means it can do about 5.7gpm. Which should be plenty to run our 2.9gpm shower (two heads) or our bath tub for our daughter (3.5gpm).

The other thing we're looking at is having PGE come out and upgrade our gas meter. It appears to be a 250CFH and we have quite a few gas appliances (dryer, stove, furnace + the new water heater). I'm guessing that's going to be way underpowered come this winter when we get the furnace going.

I'm also looking at having a recirculation pump installed but, not sure that's going to fix this particular problem. I saw in another thread here that this might also be caused by fluctuating water pressure from the city (we're on city water, not a well).

What do you guys think? Any help will be greatly appreciated. Sorry for the long winded post but, I wanted to get all the information on our system and location out so that you guys don't have to guess at it.

 

[Stuff]

If it is going completely cold with a good flow then it is shutting down. What does the display show? Any error codes?

190,000 BTU requires a good supply. What size pipe and how far from the meter is it?

 

[Sponsor]

 

[Dana]

If it goes completely cold during a tub fill the burner turned itself off, and it should have spit a code if it was malfunctioning rather than sensing that the flow had dropped too low. It's possible that the tub or shower mixer's anti-scald valve hangs up when the pressure difference between the hot and cold side is too high. With a tankless running at high flow there is a significant pressure drop across the unit, which sometimes interacts with mixer valves. As an experiment drop the temp on the TK4 to 108F-110F (or maybe even lower if that still results in a reasonable temperature tub-fill) with the mixer turned to full-hot. It may do just fine with the valve cranked full-hot from the get-go, or it may make the problem worse (higher pressure difference) it the mixers are design such that there is always at least some flow on the cold side.

Or there could be something wrong with the unit that it isn't self-sensing and giving an indication of the problem. With the wired remote controls it will usually spit out a number when it's mis-behaving, which you can look up in the manual to get a hint.

Is there some reason you aren't taking this up with the people who installed it?

Have you bucket tested the actual flow of the shower heads with a stop watch?

How fat is the gas plumbing between the TK4 and the meter? In most houses a 190K-BTU burner needs a 1-1/4" dedicated gas plumbing home-runned to the meter, with no branches to other appliances. Without it it may not be able to operate in the higher end of it's burning range without issues, and that is apparently where you're running it, given the dual gusher-shower heads. If it's 3/4 gas line the whole way (replacing a tank HW heater?) it needs to be under 30' away, measured to include the "equivalent length" of every ell or tee added to the raw linear length footage. It's best to tee-off to other appliances as close to the meter as possible, which allows the regulator to manage some of the fuel pressure fluctuations.

Hooking up way more burner than the gas meter is rated for without upgrading it is a code violation in my area. YMMV. Have you added up the BTU requirements of the furnace and other gas appliances yourself, or are you going to just trust whatever PG & E's tech tells you?

 

[DustinH]

The unit literally sits about 2 feet away from the gas meter. They installed it on the outside wall of the house where the gas meter is and it has it's own dedicated line. Admittedly, I don't know what size that line is, it looked to be about an inch so, I'll have to measure it to get accurate size. What I can see for hookup is an adapter from the gas meter to about a foot of black pipe and then a flexible yellow hose from the black pipe to the water heater (I'll add a photo later this evening when I get home).

I checked for errors on the wired remote display but, nothing when I got to the display. I'll have to try to see if I can make it error out and get a code readout.

Have not bucket tested the shower heads. I simply went by what the digital display showed while the tap was open.

We're in the process of getting PGE to come out and upgrade the meter. I've roughly added up the btu's

190K btus water heater
53K btus furnace (gathered from home inspection report when we bought the place)
20K btus stove/oven (rough guess on my part as I don't know the exact model of our stove)
22K btus dryer (again, rough guess based on googling GE gas dryers)

I believe we asked for a 400K btu meter upgrade but if there's something in the 300K range, I suspect that would work as well.

 

[Dana]

If it's literally 2 feet away from the meter it's probably not a gas pressure fluctuation problem unless the regulator on the meter is defective/worn.

The digital display on the remote only displays the flow through the water heater, not the flow at the shower head, and it's not exactly a precision meter. So, it's 2.9 gpm hot water through the TK4, when both heads are flowing, or is it 2x that much?

If 2x, the pressure drop will be significant, in which case you may be better off setting the temperature to 140F or higher, and mixing it down to 115F before the hot water distribution plumbing with a thermostatic mixing valve or tempering valve. You can test it without the mixing/tempering valve to see if it cures the symptom, but don't leave it at the high temperature until you've installed it, since the scald risk is pretty extreme.

The temperature at the shower head is usually no more than 108F (pretty hot), with 105-106F being where most people like it. With 60F cold water mixed with ~3 gpm of 120F, the water flowing out of the head at 105F would be about 4 gpm. To get the same 105F out of the shower head after setting the hot water temp down to 115F, the flow trough the TK4 would have to increase to about 3.2-3.3 gpm, which increases the pressure drop across the TK4. If you set it to 110F the flow through the hot side would need to be 3.5-3.6 gpm, for an even bigger pressure drop. If you bump the output temp up to 140F, you'll only have 2.2-2.3 gpm going through the tankless, for less pressure drop.

So, if dropping the output temp to 110F makes the problem noticeably worse, and bumping it to 140F makes it better it points to the mixer/anti-scald valve.

Tankless water heaters are quite tolerant of big in-to-out temperature differences, so the only down-side to bumping up the temperature and mixing it back down is having sufficient flow to keep it firing for low flow taps, say when washing hands at the bathroom sink, etc. Routinely running a tankless at the high end of it's gpm range prematurely wears out the flow sensor, and can even erode the internal plumbing (a common problem with some home-brew radiant floor heating systems people hack together with tankless water heaters.) The absolute maximum flow rate specified for the TK4 is 8 gpm, but lower flow/higher delta-T is usually going to be nicer to the unit.

In the mean time, if someone monitors the remote while another person is running the tub or shower you may be able to catch it spitting a code, then resetting itself. It's worth a try.

 

[DustinH]

It's 2.9gpm with both heads going, 2.2gpm with just the handheld head going.

I will say, after bumping the system down to 115 from 120, we were able to run a tub full of hot water for our daughter without any cold water sandwiching in.

The problem seems to come and go. I set the shower to just the handheld unit this morning after the wife got her cold water surprise and I didn't have any issues with fluctuating temperature.

The other things you've given me to look at, i'll have to do for homework tonight

 

[Dana]

If it's the type of handheld with a finger-operated valve, every time the flow stops, the tankless turns off and will have to re-light, which would result in major coldwater sandwich issues (not to mention wear & tear on the ignition systems, and low efficiency due to the short-cycling.)

Report back what you find, we'll see if it can be diagnosed any better.

 

[DustinH]

Posting an update. Got back from a camping trip, decided bath (for daughter) and showers were in order. Hot water ran out and long story short, I was able to get the error code this time. 121 - loss of flame. Hot water wouldn't come back so, I went out, took the cover off (noticed the cover was very very loosely left on), stared at the hi-limit switch with no real way to gauge if it's functioning correctly. After a couple of minutes I decided a reboot was in order (IT is what I do for a living). Hit the on/off switch, waited ten seconds, turned the unit back on. Went back in and checked, the hot water was flowing again, all seemed to be in order. I put the cover back on and tightened the screws appropriately. I also set the unit back to 120* water instead of 115* as it didn't seem to make a difference in the problem after all. Hot water appeared to work for the rest of the evening (and worked well).

So, questions, it can be fairly windy where we live and the unit is an outdoor unit. Could the cover being left on very loosely allow wind to get in and screw with the flame? Everything seems well protected in there but, it was just something I wondered.

How likely is it that the hi-limit switch would be faulty from the manufacturer (granted it's brand new and sometimes things get past QA but, is it common to see that happen right out of the box)?

I still haven't gotten a picture of the install yet but, I looked again at the gas hookup. About 18 inches of pipe to 18 inches of flexible yellow gas line connected to the unit. The pipe has to be at least 3/4", possibly an inch (will measure soon, promise). I looked in the manual and the manual said for less than 10' of pipe run, that 3/4" pipe will allow 364K BTUs. According to PGE our 275 CFH meter is capable of 380K BTUS. So, these numbers seem to fall in line but, we're still considering having them (PGE) come out and upgrade the unit to the next higher model. We just have to get the city to come out and inspect the install and certify that it was done correctly (according to PGE).

Also, I noticed the unit will not let me set the temp higher than 120*. Being in IT, I'm guessing there's probably a button combo to press to unlock that but, doesn't seem necessary at this point.

Right now, I'm figuring either wind was a larger factor in screwing with the flame than I thought it would be or, inconsistent gas pressure coming from the meter. Possibly a bad hi-limit switch as well. Again, all what I could gleam from the manual and troubleshooting.

 

[Dana]

It looks like setting to 140F present is a dipswitch-only proposition. See p.39. But if it's working just fine as-is after your diagnostic session and buttoning down the cover more tightly, let's not steal a defeat from the jaws of victory, eh?

It's conceivable that a gust of wind could take out the flame, but I'd be surprised.

 

[DustinH]

Yeh, I'm going to mess with it a little more now that we're not in dire need of being clean. I seem to have noticed that higher flow rates tend to cause the cut out. So I'll try the tub again tonight and see if that makes it behave similarly again (filled the tub for the daughter using the showerhead last night since she doesn't need that much tub water to get clean). It seems like under 3gpm is the magic number to keep the hot water flowing.

 

[Jadnashua]

You didn't say how large the actual gas supply pipe going to the unit is.

Also, 20K btu for a stove is pretty feeble, 2-3x that with everything running is a bit more normal, but you may not see that except when having a dinner party and all burners are cooking along with the oven.

 

[Dana]

You didn't say how large the actual gas supply pipe going to the unit is.

Also, 20K btu for a stove is pretty feeble, 2-3x that with everything running is a bit more normal, but you may not see that except when having a dinner party and all burners are cooking along with the oven.

Even if it's half-inch, if it's less than 5 equivalent- feet (all equivalent lenghts of ells included) away from the meter it should be OK, and if it's 3/4" it has capacity to spare:

Since the hook-up to the unit is always 3/4" it's unlikely somebody took the trouble to plumb it with anything smaller than 3/4".

 

[DustinH]

 

[DustinH]

So there are the pics of the install. Looks like 3/4" piping. According to the product manual from Takagi, piping under 10' at 3/4" should be able to handle 364K BTUs. So far, I noticed once, yesterday morning, the water temp fluctuated a tiny amount (went slightly less hot then returned to hot). So far, it seems to be holding up better. I'm going to be testing it again soon by trying to run hot water for the tub (that seems to be where it reliably cuts out).

 

[JRC3]

How big is that gas meter? It looks smaller than the 250cfh ones around here, not that that means anything. The pipe feeding the meter looks small. Idk, your supply might work at a higher psi or something. I've also heard of people having to adjust the regulator for tankless use.

My meter is in 150' in the back yard and feeds the house with 2". Inside I have my tankless ran 25-30' with 3/4", it's a 150k. It does fine even with the 90k furnace running.

I asked the guy from the gas company that inspected and turned the gas on for the house about the meter and tankless; He said the meters are actually good for about another 30% over their rating. If I were you I'd check into that supply line and regulator feeding the meter.

Also you might check the screen in the tankless were the gas feeds in, it might be clogged since there is no sediment trap installed. IDK, I'm no pro.

 

[Dana]

Could that have been installed any twistier? You need to account for the "equivalent feet" of every turn, and you have lots of 'em!

There are a total of three low-radius 90 degree ells, two branch tees and a longer 90 degree bend in the flex, and a ball valve.

Each ell is worth 4.4', so you have 13.2 feet of ells, and each tee is 5.3' for 10.6 feet of tees, maybe another foot for the bend, for a total of about 25' in fittings & bends, plus another 3-4 feet of raw pipe length. Call it 30 feet total.

At 30 equivalent-feet you're only good for ~130,000 BTU/hr, and you have 190,000 BTU/hr of burner. (I know Takagi's crummy chart says it would be good for 200,000 BTU, but...) Worse, rather than being a home-run to the meter it is sharing flow with the section of pipe & fittings back from the tee that feeds other appliances, introducing more fluctuations. The closer that tee is to the meter, the less shared pipe, and the less interaction it will have with other burners.

It's likely that the gas plumbing is indeed contributing to the problem at higher firing rates.

The galvanized tee is also a problem down the line, and not best practice, since platings can flake off and clog orifices. Floating crud is commonly mitigated with "drip leg" between the flex connection and the tankless, to let gravity trap any corrosion or flaked plating debris, a feature absent in your installation.

Here they put the flex between the drip leg and tankless, which is also fine.

Your installation would be quite a bit better if it were teed right out of the meter going straight up, with the stem of the tee feeding the other stuff, with a single ell over toward the tankless. Come straight down from the tankless to a drip-leg tee, with straight flex to the valve and ell above the tee at the meter, with the side-branch of the tee at the meter feeding the other stuff. That would add up to

drip-leg tee: 5.4'

meter tee: 2.4'

ell: 4.4'

straight length: 2'

Total: 14.2' (equivalent-feet)

That would usually make it just fine 3/4" pipe and any 190K burner.

If they are coming out to upgrade the meter soon, that would be the right time to fix it, but try to get the original installer do the work under warranty, and get rid of any galvanized fittings (really!).

 

[DustinH]

What to use instead of galvanized fittings? Also, I'm not sure if they will touch the connection coming out of the meter. That remains to be seen (and possibly why they did it the way they did it). Unfortunately, we're not fond of the installer so, we may pay someone else to come out and fix up the plumbing.

 

[Dana]

Black iron is the standard. You can paint the pipe if you like (as they did on the older plumbing), it's the potential for flaking plating and crud on the interior that matters, since particulates can clog gas-burner orifices.

In most places you own all the plumbing on your side of the meter, right up to the ell currently attached to yours. A licensed gas fitter should know what is and isn't yours to monkey with in your area. A GOOD gas-fitter would also run the equivalent length numbers and BTU capacity too, but I guess that wasn't the original guy, eh?