Differential controller for domestic hot water

Good morning, Adrian. Thanks for a good problem description - they're never too long.

Before anything else, a "theoretical" question. What happens if your diverter valve goes on for a few minutes, but fails to bring down the water temperature enough to reset the controller? In that case, you would lose the effect of the control system. The timer is done, it won't turn back on until the temp once again goes below the setpoint, which might be when you reset the system or next spring. It would be better to mount the sensors in such a way that it will normally take a couple of minutes with the diverter on to bring down the differential temperature. In a sound control system, this kind of thing would be taken into account. The sensor placement decision would also take into account the thermal lag of the temperature sensors.

For water temp controllers, precision RTDs are most frequently used. These have the advantages of simplicity (one passive component) and low temperature drift/accumulated error over time. A second, less popular choice is the IC temp-to-current or temp-to-voltage sensors, such as the Analog Devices AD590 (1uA/degree K) or the National Semiconductor LM35 (10mV/degree C). But when you add in the increased complexity, the requirement for independent calibration for all but the tightest spec (and most expensive) parts, and the fact that these are sold as ICs which would have to be packaged in watertight enclosures so as to be insertable in water, they are less desirable in general. Thermocouples are also used, but they tend to have less absolute accuracy over a narrow temp range, drift more over time, and require additional circuitry for cold junction compensation. Bimetallic switches are too crude for this application, and are generally not used. The change in output voltage for a silicon diode over such a small temp range would be too small to reliably read at a distance, and would be swamped by electrical noise.

Differential temperature controllers are already manufactured, and would almost certainly be less expensive in terms of total cost, and more reliable than a "home brew" solution. A quick Google shows

as a solar water heater differential temp controller with 10K ohm thermistor sensors. These are also sold by this outfit, mounted in 4" immersible probes with 1/4" NPT fittings. The controller is $143.20 USD and the sensors are $46.11 USD each (2 req'd). Something like this would be a recommended solution.

But let's say you didn't have $240 USD to spend, and were determined to "home brew" a solution. You might start with the "differential thermometer" circuit found on p. 8 of the old Intersil data sheet for the AD590:

(using half of a dual single supply op amp in place of the 741) and use the other op amp as a comparator with hysteresis to measure the temp (10mV/degree C = +40mV on, +20mV off). The output of that comparator would drive a transistor which would turn on a relay to run your diverter. You can obtain the "low grade" AD590s (the premium ones are rather dear), because the offset pot in the apps circuit allows you to trim for zero output with both sensors at the same temp. It might make more sense, considering your small differential temp range requirement, to use a 100K feedback resistor instead of 10K to get 100mV/degree C.

You could conceivably do the whole thing with two AD590 sensors ($3.46 ea. from mouser.com), an LM358 (be sure to use a 10K pulldown resistor at the output of the first op amp to get close to 0VDC output), and a 12VDC unregulated wall wart, 12VDC relay, a couple of trimmer pots and various junkbox components and miscellaneous parts. Of course, you still have to find a way to seal off the AD590 in a temperature probe in such a way that it won't come in contact with the water, but will still be responsive to changes in temp. I'd suggest using high temp 5-minute epoxy (regular gets soft at 80C) to affix the top of the TO-52 can to the inside of a plastic tube of some kind.

I hope this has been of help. If you have any questions, feel free to email.

Good luck Chris

Good morning, Adrian. Thanks for a good problem description - they're never too long.

Before anything else, a "theoretical" question. What happens if your diverter valve goes on for a few minutes, but fails to bring down the water temperature enough to reset the controller? In that case, you would lose the effect of the control system. The timer is done, it won't turn back on until the temp once again goes below the setpoint and rises above it again, which might be when you reset the system or next spring. It would be better to mount the sensors in such a way that it will normally take a couple of minutes with the diverter on to bring down the differential temperature. In a sound control system, this kind of thing would be taken into account. The sensor placement decision would also take into account the thermal lag of the temperature sensors.

For water temp controllers, precision thermistors are most frequently used. These have the advantages of simplicity (one passive component) and low temperature drift/accumulated error over time. A second, less popular choice is the IC temp-to-current or temp-to-voltage sensors, such as the Analog Devices AD590 (1uA/degree K) or the National Semiconductor LM35 (10mV/degree C). But when you add in the increased complexity, the requirement for independent calibration for all but the tightest spec (and most expensive) parts, and the fact that these are sold as ICs which would have to be packaged in watertight enclosures so as to be insertable in water, they are less desirable in general. Thermocouples are also used, but they tend to have less absolute accuracy over a narrow temp range, drift more over time, and require additional circuitry for cold junction compensation. Bimetallic switches are too crude for this application, and are generally not used. The change in output voltage for a silicon diode over such a small temp range would be too small to reliably read at a distance, and would be swamped by electrical noise.

Differential temperature controllers are already manufactured, and would almost certainly be less expensive in terms of total cost, and more reliable than a "home brew" solution. A quick Google shows

as a solar water heater differential temp controller with 10K ohm thermistor sensors. These are also sold by this outfit, mounted in 4" immersible probes with 1/4" NPT fittings. The controller is $143.20 USD and the sensors are $46.11 USD each (2 req'd). Something like this would be a recommended solution.

But let's say you didn't have $240 USD to spend, and were determined to "home brew" a solution. You might start with the "differential thermometer" circuit found on p. 8 of the old Intersil data sheet for the AD590:

(using half of a dual single supply op amp in place of the 741) and use the other op amp as a comparator with hysteresis to measure the temp (10mV/degree C = +40mV on, +20mV off). The output of that comparator would drive a transistor which would turn on a relay to run your diverter. You can obtain the "low grade" AD590s (the premium ones are rather dear), because the offset pot in the apps circuit allows you to trim for zero output with both sensors at the same temp. It might make more sense, considering your small differential temp range requirement, to use a 100K feedback resistor instead of 10K to get 100mV/degree C.

You could conceivably do the whole thing with two AD590 sensors ($3.46 ea. from mouser.com), an LM358 (be sure to use a 10K pulldown resistor at the output of the first op amp to get close to 0VDC output), and a 12VDC unregulated wall wart, 12VDC relay, a couple of trimmer pots and various junkbox components and miscellaneous parts. Of course, you still have to find a way to seal off the AD590 in a temperature probe in such a way that it won't come in contact with the water, but will still be responsive to changes in temp. I'd suggest using high temp 5-minute epoxy (regular gets soft at 80C) to affix the top of the TO-52 can to the inside of a plastic tube of some kind.

I hope this has been of help. If you have any questions, feel free to email.

Good luck Chris

OK

Ummm - hadn't thought of that....... Maybe the timer idea isn't such a good one - perhaps a certain amount of 'thermal inertia' in the sensors would help. (I know that thermal inertia doesn't exist - I was meaning that, if the sensor on the circulation pipe had a big enough heat capacity, then it'd take time to warm / cool and this would have the effect of 'smoothing' any short-term peaks and troughs in the temperature....?)

OK - the controller can be within a foot or so of the sensors - but I guess it's a potentially noisy environment (electrically-speaking).... Packaging-wise - I'm going to clamp one sensor to the outside of the water tank (the solar heating sensors are already in proper 'pockets' inserted into the tank - but the new sensors will have to make do with a good contact to the copper of the tank. The other sensor will be clamped to the 22mm water pipe - so again, no waterproofing required.

I did see a source which quoted the silicon diode sensor as 2mV / degree C - which is five times less than the LM35's. I don't think that absolute accuracy is a problem - after all - it's a differential reading that I'm looking for....

Right on both counts - especially the first one !

Hmm - that looks interesting..... got to rush off now but have saved the link and will take a longer look later.....

Many thanks - very helpful

Adrian Suffolk UK

I like your thoughts on this. It sounds like a good way to economize on heat. I would forget the timer. It's going to cause a few problems.

Consider

If the boiler water temperature is warmer than the tank temperature and the cabin temperature is at or above setpoint, then divert heat to the hot water tank.

This gives you an issue if both the house and the hot water tank are too hot. If that is an issue then consider two divert valves: valve 1 chooses whether to heat the house, valve 2 chooses whether to heat the water.

If the house is too warm, divert the water from the house circuit to the heat reclaim circuit, else diver to the house circuit.

If the reclaim circuit temperature is warmer than the tank, divert through the tank, else recalculate to the boiler directly (or pick another heat sink).

If you use modulating valves (not just on/off, but able to position in between) then you could play with a bit more advanced controls and split the streams proportionately to your heating needs. You could use half the heat to warm the house if that's what was needed.

Michael

Hi Michael

Thanks for the reply - coments inline

I think you're right.....

There's a subtle difference between the boiler I am using, and a 'conventional' gas or oil-fired boiler. It's kind of difficult to turn on & off - and there is no 'central heating thermostat' as such.

So..... each of the rooms in the bungalow has a thermostatic radiator valve - so, in the event of the woodburner being allowed to run at maximum, each of the rooms will get to its preset temperature and then the trv will cut off.

However, if this situation (woodburner run at maximum) took place the big room in which the woodburner is situated would become unconfortably hot - so it tends to be allowed to 'idle' along - rather than running at maximum....

But either way - it can't be easily turned on & off, like a conventional boiler.

Until the water in the boiler hits a present temperature (about 40 -

50 C), the circulating pump remains 'off' - controlled by a simple thermostat on the pipework. In the summer months the domestic hot water is heated to a reasonable level by the solar heating system (and topped-up by an electric immersion as necessary) - but in the dead of winter it struck me that a useful amount of preheating could be gained from the woodburner, with a corresponding saving in electricity costs.

So.... we have a heat source which is 'always on', and only regulatable/controllable in terms of increasing/decreasing the draft and adding more wood.

I was planning to give the water heating priority - as, when the woodburner is started, there is direct heat output from the boiler to the main room in the house, and no great need for heat to the other rooms with radiators. At this time the spare heat could be dumped into the hot water tank - and once that had heated up the hot water could then be pumped around the radiator circuit.

I know this sounds fairly crude - but the heat source itself is also fairly crude

Thanks again Adrian Suffolk UK

A good source of solenoid valves is the ones they use in residential clothes washing machines. You should have trouble picking up used ones at any appliance shop that takes used ones in 'trade'. They often just throw them out.

Walter.

In this case, the system I described could be modified a bit. It would work as described already. Set up a header from the fireplace. Use one diverter valve per room plus one for the water tank. Arranged these in series, with a tee between each one. The first valve will divert water to room 1 if the room is too cold. The water will return through the tee. The second valve will do similarly for room 2. This will continue until on the tank, if the heated water temperature is greater than the tank temperature, then the water is diverted to the tank.

Your system would then warm the house up with something of a priority system. As the lower number rooms warmed up, then the higher number rooms would get more heat. At the end, any excess heat goes to the tank. If no diverter valve were set, then the water would just recirculate back to the heater.

If you wanted to save electricity, you could make it a little more complicated by adding a relay system so that if any diverter valve was opened, and the fireplace was on, that the pump would start up. It would also need a setting so that if the water temperature in the coils was warmer than the tank, the pump would start. Just remembering to turn the pump on would probably be just as good.

I'm assuming that you can find some relays that will be able to use two temperatures. Perhaps an instrumentation guru can explain how to connect two thermocouples to get the temperature difference between two points. For the precision you require, mounting a thermocouple on the outside of a pipe would woud be sufficient. The in tank thermocouple should be in a probe (a piece of empty pipe) immersed in the tank.

If push came to shove, and you have trouble with the hardware for the logic on the tank, then if you just dumped the excess heat to the tank as long as the fire is on, you'll probably do ok.

Michael

Hmmm - sounds a bit complicated. No doubt it's the right solution in a larger building - but the thermostatic valves on each radiator give a similar level of control and don't require any electrical energy or expensive diverter valves to achieve this.

As I said above - there is a 'stat on the piperwork from the boiler that senses the circulating water temperature, and only runs the pump if the water temp is > 40 degrees C. Apparently this also extends the life of the boiler, as pumping cooler water through the boiler can increase the likelhood of corrosion.....

I expect there is an 'amplified relay' available (probably expensive!)

- but there are also simple circuits with just an op-amp and various types of sensor. The pipe sensor will be clamped onto the outside of the pipe (possibly by making a mount from another piece of the same copper pipe, cut in half lengthwise to make a c-shaped piece which can then have a sensor epoxied or clamped to it.

The 'tank' sensor should probably be inside the tank - but it's not going to be . Too much hassle to drain the whole system down and get a water-tight bush into the tank wall. Instead I'm going to remove some of the foam insulation from the outside of the tank and fix the sensor to a metal disk which can be in thermal contact with the metal tank. Won't be as precise as a proper probe - but it will be a lot less hassle!

Hmm - we'll see. The whole thing works (for water heating) on some fairly fine differentials. The most we expect to get from the boiler is 55 - 60 C - so it'll only really be a pre-heat for the hot water tank - but any energy that can come from the woodburner is going to save money on the electricity bill - so it's worth a try.

Time to get the soldering iron out and have a play !

Many thanks Adrian Suffolk UK

Thanks for the suggestion. I've used washing machine valves before, they control the water flow to the irrigation in our polytunnel. 'Proper' solenoid valves are horrendously expensive - the washing machine valves are only a few UK pounds when new - and (so far) I've had no problems..... they run a bit warm, but nothing to worry about.

However, in this case I already have the right valve - its just a matter of switching it . As I said elsewhere, I think it's gettting close to the time when I need to start building something and see how it works......

Many thanks Adrian Suffolk UK

For me, the most fascinating thing about your project is this: given that now in England all domestic central heating systems are subject to Building Regulations Control, what will you have to do to you explain the workings of your system to your local Building Inspector, and get him to approve it and allow you to use it? Is he sympathetic?

It should be possible; but...: Good luck!

Kelv> Hi Walter

Well - I didn't know that..... you learn something new every day When did this latest regulation get introduced ?

More to the point - does he know about it ?? I guess 'It was like that when I moved in, guv' would be a good starting point ?

Actually - I'm on reasonable terms with the local Building Inspector - as he haunted the place while we built on a utility room and a workshop. Because the drain to our septic tank goes under the utility room, we had to dig down to undisturbed soil (about 8 ft down!) in order to build a structural 'bridge' of brick & concrete to protect the drain. Much digging - good thing the 'soil' here is mostly sand.

He was also very helpful with the workshop - which comes within ten foot or so of our fresh-water well. Successive 'guesses' by our architect led to the workshop ending up with steel reinforced foundations that were over 2ft thick at the edges and 18" thick in the centre. The Building regs man pointed out that, if the workshop were an independant structure, rather than being attached to the garage, then it instantly became outside of the Regs, and we could do what we liked. So it's now a timber-framed structure, sitting on 2ft square x

2" paving slabs on properly compacted ground with concrete & everything - and it's fine.....

Do the new regs apply to 'proper' c/h systems (as in gas, oil etc) ? This one, as far as radiators & pipework is concerned, was installed before we bought the house - all I've done is to finish the installation and add an improved woodburner / boiler....

Thanks for the info Adrian Suffolk UK

2002 AIR

Just so long as you don't change anything!

You would want to look at: The Building Act 1984 and building regulations, Part L1 - Conservation of fuel and power in dwellings (2002 edition). It's scope knows few bounds: covering circulating systems, boilers, central-heating controls; even what type of windows you are allowed to have!

Oh! and if you're contemplating doing any electrical work after January 2005, you might want to look also (before maybe deciding to emigrate) at:

Kelvin B. Hales Kelvin Hales Associates Limited Consulting Process Control Engineers E-mail: snipped-for-privacy@khace.com Web:

They *are* 'proper' solenoid valves. You are right, they are not really intended for long term continuous energization. If you have a problem, the solution is to put them on fractional power once they've pulled in. A very simple way is to use two relay contacts. The first connects the power to a circuit that includes a series diode. The second contact simultaneously bypasses the diode. After a brief moment, the second contact opens. With luck you will have enough current to hold the coil in. If not, a resistor selected by trial and error will do the trick.

By the way, a typical 120 VAC solenoid will run perfectly well on 24 VDC. This is helpful if your electronics can't handle AC. The diode trick won't work, of course.

Walter.

That's easy -- connect two T/Cs back to back. The resulting signal will be proportional to the difference. A simple amplifier circuit can be built out of about two transistors. The output drives a small relay. There's no adjustment in this simple circuit but it can be arranged to close the contact when one T/C is warmer than the other by some amount. That's all your really need. To find out what the magic amount is, you'll need a pair of thermometers.

The devices used to detect flame failure in domestic gas furnaces and water heaters are T/Cs. Replacements can be bought cheaply. You can get them even cheaper from junk water heaters.

By the way, the 'pocket' for measuring the temperature inside a pipe or vessel is called a thermowell (TW). I'll bet you can cobble together a simple one using brass plumbing fittings, copper water pipe and solder.

Walter.

HI Walter Thanks for the comments - mine are inline....

Yes - sounds like the smae principle as the circuit I breadboarded last night - except I used a couple of 1N400x si diodes forward biased, and an op-amp.

Spent a few happy minutes alternately freezing / warming each diode and watching the led light. Sad how little things amuse, isn't it ??

Anyway - if the forward voltage drop is indeed 2mV / degree C (as I read somewhere) then this little circuit will adjust to switch from about 2 to 10 degrees differential - which will do nicely.

I thought of 'slugging' the diodes by connecting a uF or so of capacitance across them - just to stop the thing from jittering - haven't tried that mod yet.....

OK - I'll think about it, if the diode plan doesn't work - thanks! I suppose a thermocouple might come in a nicer 'packaging' than a bare diode - but I'll see how it goes....

I bet I could - but fitting another probe to the domestic hot water tank would involve draining it down, and much fiddling / faffing about

- if I can make it work on the basis of external contact rather than actual immersed probes then I'll be happier...... I think the existing 'thermowell' in the tank might have been fitted before the top was fitted on the tank - not sure it could be done 'from the outside'.....

An earlier plan was to monitor the actual differential temperature between the boiler water feeding into the tank heat exchanger, and the same water coming back out. The solar water controller on the same tank does something clever with 3 sensors - one on the tank, one on the solar collector and a third one in the water going back to the collector, having been through the heat exchanger in the tank... but I don't really want to complicate the issue

Thanks again Adrian

Sorry - I was using 'proper' in the 'Cornish' sense - as in 'proper job' - which translates as 'excellent, ideal for the purpose'.......

They do get warm - but the one I'm using at the moment has lasted all year, and is cooled, to some extent, by the water flowing through it, and the fact that it's mounted in free air.

Oh - that's a thought - I'll watch for long-term reliabilty and remember your advice if I seem to have a problem. The watering system that the valve is controlling consists of very low pressure drip pipe - so to reduce the heating a little, and to improve the watering, I set the valve to open many times a day (simple mains-operated timer with the little 'pips' that you push in or out to control the on/off periods. As winter approaches I'll need less watering anyway, and the ambient temp will drop - so I probably won't have a problem with the solenoid overheating...

Oh - I never thought of that. We're 240V AC over this side of the pond - but I guess it might be interesting to see how low a DC voltage will run the thing. I've got a spare in the cupboard - I'll have a play (when get an idel moment!)

One of those unusually busy days today - dog walking, shopping, see 2 clients, out for evening meal and then to the theatre.... so not much time for playing with electronics today!

Thanks again for the thoughts Adrian Suffolk UK

Walter,

Thanks, I thought that it would be ok to do hook them together, but I wasn't sure if there were some other magic to be considered and at the time I wrote that, I was a little too close to asleep to want to take a guess.

Michael