Hot water failure detector

Thread Starter

Mrdouble

Joined Aug 13, 2012
107
My gas water heater burner unit fails from time to time and just requires a power cycle and then it's fine. Thing is, it's downstairs and only gives led flashing sequence as notification. As you probably guessed, we only find out it's failed when we get hit with cold water in the shower.

So I'm trying to design a circuit that detects any one of the (7?) LEDs and inturn turns on a buzzer so we can hear it up stairs.

My idea so far is to velcro a PCB that has multiple phototransistors in parallel that trigger a speaker/buzzer if a light is detected.

Maybe someone else has a better idea, I'm just brainstorming right now. Screenshot_20201124-134402_Chrome.jpg
 

jpanhalt

Joined Jan 18, 2008
11,087
Here's a link to the Service Manual, in case you don't have it.
http://waterheatertimer.org/pdf/Intellivent-service-manual.pdf

There is also a supplement that I have uploaded here. It looks to me like those 6 LED's provide a 6-bit data output (see supplement). I am not sure all are off when the symptom you describe happens. Is there is a specific bit pattern for the failure you are seeing? What is it?

Maybe something simpler will work. Let the wife always shower first? ;)
 

Attachments

Thread Starter

Mrdouble

Joined Aug 13, 2012
107
The manual says it requires a power cycle when kicks out when it doesn't light via glow rod as it is not a LockOut fault.

So does anyone have any suggestions on how to monitor lights?

Btw, the wife usually is the one to get bit by cold water :)
 

ronsimpson

Joined Oct 7, 2019
2,986
I had cold water this week to.
I am playing with a "sonoff TH16". It measures temperature & humidity and includes a power switch. It talks wireless and needs a network to work right. It can do many jobs but I can set it up to email my phone when the temperature is too low. Then I can turn off then back on the power from the phone. I do not know enough to combine different functions but it might be able to do a 1 minute power cycle when the temp is too low.
 

jpanhalt

Joined Jan 18, 2008
11,087
The manual says it requires a power cycle when kicks out when it doesn't light via glow rod as it is not a LockOut fault.
Which "it" are you referring to. There are several different codes, e.g.,
1606282969464.png

Is that what you are referring to or is "it" something else?

If all you want to do is determine whether a specific LED is lit, you could use an optical sensor as you suggest, but I find such approaches that duplicate what your eye would do are often more difficult than anticipated. I would consider a low-voltage relay or optocoupler in line with or in place of the led. That will give an actual switch function directly for your signal. You can add the LED function back once you have the switching mechanism.
 
Last edited:

Alec_t

Joined Sep 17, 2013
14,280
I haven't checked the manual. Are the electronics isolated from the mains? If not, be very careful with any circuit modifications.
An alternative approach to light sensing would be to have a thermistor attached to any pipework/metalwork which gets hot when the burner is working correctly.
 

Tonyr1084

Joined Sep 24, 2015
7,852
I have something similar; a single LED that flashes once about every three seconds. It doesn't have a power source connected to it so I'm guessing it's using some sort of thermopile to fire the LED every few seconds.

I used to have the same problem, the pilot going out. I'm assuming you have a pilot too. My experience with electronic ignition is that when temperature demands heat the electronics makes the burner come on. In my case, it's older school thermostatically controlled. When the water temperature drops below the set point the burner comes on. It depends on the pilot being lit to ignite the gas. If the pilot goes off the gas burner can not come on. (hopefully)

The issue I had was with the flue pipe. On the roof was one of those Rice Patty Hat types (like this) that prevent rain from falling down the flue pipe. But on very windy days it can cause a vacuum and draw excessive amounts of air up the pipe. The rush of air when severe enough would blow the pilot out and shut down the water heater. The way I fixed that was to change the cap similar to one of these. Can't find the exact match to what I have. Since then I haven't had a problem with the pilot going out.
 

jpanhalt

Joined Jan 18, 2008
11,087
That furnace doesn't have a pilot. Its control uses an igniter with each cycle as many (all?) energy efficient ones now do.

On my furnace, Weil/Mclain brand, if the igniter is bad, it cycles and eventually errors. In 10 years, I have had to replace the igniter once. I don't know when it was last replaced before I bought the house. There are lots of other safety features with start sequence that can go bad, including dead batteries in the individual room thermostats.

Failed ignition seems to be a fairly common problem with that unit based on what Google found. So the problem may not be easily solved with a new igniter.
 

Tonyr1084

Joined Sep 24, 2015
7,852
@jpanhalt; we're talking about a water heater, not a furnace. Water heaters with electronic ignition require electrical wiring. My water heater has no wiring except the sparker that you have to press to light the pilot light. High efficiency water heaters definitely use electrical power to light and control the burn and can extract over 90% of the heat generated by the burner. I say over 90 because I don't know the exact numbers. But I would not be surprised at 97% heat extraction. My forced air furnace is a 98% efficiency furnace and has plastic flue gas extraction. PVC.

Here's one possible solution: Since the hot water pipe is always warm or hot why not sense the temperature. If the heater has shut down you'll get an indication simply by detecting a drop in water temperature. At present my hot water pipe at the very top of the water heater is 100˚F. Being that it is exposed to ambient temperatures, if the water temperature falls inside the tank due to a lack of heating then the pipe will likely cool. Using a temperature sensor and an alarm you can signal when the water temp has dropped to (oh, lets say) 95˚F and trigger the alarm.

Another approach would be to put a photo transistor directly over the LED you want to detect. If it's like mine, flashing once every three seconds you can build a circuit where the flash of the LED resets a timer. The timer will detect periods of longer than 30 seconds without a flash and can trigger an alarm. Since mine would reset every three seconds, I would know in less than one minute of failure.
 

jpanhalt

Joined Jan 18, 2008
11,087
@Tonyr1084
The TS refers to power and LED's. That indicates to me that the control is powered electrically. Moreover, the manual (first page after the picture page) and troubleshooting guide make that very clear.

My furnace heats both domestic hot water DHW) and the home. Specifically, the hot water heater is relatively small volume and gets quite cool (almost ambient) when there are long periods of no demand. Turn on the HW and the 105,000 btu furnace gets it warm within a very few minutes. During the day, because of occasional needs, the water tank stays warmer. In fact, although the furnace itself is a Weil-Mclain, hot water control is White-Rodgers.
 

strantor

Joined Oct 3, 2010
6,782
I haven't checked the manual. Are the electronics isolated from the mains? If not, be very careful with any circuit modifications.
An alternative approach to light sensing would be to have a thermistor attached to any pipework/metalwork which gets hot when the burner is working correctly.
TS has the right idea, attaching some sort of external thing to observe the LED status, not tampering with things that could electrocute or explode the house.

Personally I would suggest getting a professional to troubleshoot and fix the cause of the issue. That is, if getting the issue fixed is the primary goal. If the primary goal is to build something cool then by all means, this sounds cool enough. I would use phototransistors, photodiodes, or photoresistors; any of them could work. Just Google "photo______ detector circuit." I would use an arduino just for fun, open up a world of possibilities. You can program it to send you a text in the event the heater quits.
 

jpanhalt

Joined Jan 18, 2008
11,087
1) Page 8 manual: It is mains powered.
2) Page 5 (pdf#):
1606327937776.png
3) Look at the error codes. The leftmost led is always on. I have not wasted the time to map all temperatures to evey code. but if you consider just two codes, open ignitor (101011) and ignition failure (100110) and normal operation temperature, say 125 to 145 (001100, 000100, 000110, 000110, , 000010, 000011), you will need to monitor more than one LED. Replacing each LED with an optocoupler would make that easy. Then, an arduino or any other MCU can read it.

One concern about monitoring a single LED is that there are likely to be false alarms, and more important, with periods of disuse, the temperature may drop well below any chosen temperature -- one of the "benefits" of energy saving. Also, since a failure may occur during normal sleeping hours, the TS may not want to be awakened at 2:30 AM. An MCU solution could delay the awakening until a set time.

I agree about fixing the root cause, but when I Googled this problem, it appears to be common with that controller and is not due to failure of any specific part.
 

Tonyr1084

Joined Sep 24, 2015
7,852
That indicates to me that the control is powered electrically.
My water heater has a flashing LED but has no external power. I don't know if there's a battery inside the control box, I think not. Will say it again, my water heater has a flashing LED but has no external power.

Yes, I see now that the TS water heater has external power. I didn't know that before.
 

Tonyr1084

Joined Sep 24, 2015
7,852
After having paid more attention to the details, if you can get an alarm you can get a power interruptor to interrupt power for 30 seconds to a minute then restore power and voila! You have your power cycle to restart.
 

jpanhalt

Joined Jan 18, 2008
11,087
My water heater has a flashing LED but has no external power. I don't know if there's a battery inside the control box, I think not. Will say it again, my water heater has a flashing LED but has no external power.

Yes, I see now that the TS water heater has external power. I didn't know that before.
This thread is NOT about your hot water heater.

The link to the manual was posted LONG before your first post.
 

strantor

Joined Oct 3, 2010
6,782
1) Page 8 manual: It is mains powered.
2) Page 5 (pdf#):
View attachment 223372
3) Look at the error codes. The leftmost led is always on. I have not wasted the time to map all temperatures to evey code. but if you consider just two codes, open ignitor (101011) and ignition failure (100110) and normal operation temperature, say 125 to 145 (001100, 000100, 000110, 000110, , 000010, 000011), you will need to monitor more than one LED. Replacing each LED with an optocoupler would make that easy. Then, an arduino or any other MCU can read it.

One concern about monitoring a single LED is that there are likely to be false alarms, and more important, with periods of disuse, the temperature may drop well below any chosen temperature -- one of the "benefits" of energy saving. Also, since a failure may occur during normal sleeping hours, the TS may not want to be awakened at 2:30 AM. An MCU solution could delay the awakening until a set time.

I agree about fixing the root cause, but when I Googled this problem, it appears to be common with that controller and is not due to failure of any specific part.
I think he planning on using more than one photo(device); one per LED was my impression. And not opening or tampering with the device was my impression. I've rolled my own photocoupler in the past by putting a LED and a photodiode together end-to-end inside a heat shrink tube. Without opening the case, just glueing/taping photo-somethings to the case, it might present a small challenge to detect whether an LED is actually on, or if the one next to it is on - there will be some "cross-talk." So it may make more sense to use photoresistors connected to a level detector. This would be super simple to do inside an arduino using analog inputs and empirically derived values to differentiate between LED-ON and LED-(neighbor)-ON (OFF).

Using an arduino or similar uC, it would be trivial to program in recognition of all these binary states and unique actions for each/any of interest.
 
Top