Room thermostat - Honeywell DT90E - relay replacement (created my own version).

[ljwinkler]

Hi All,

Recently I have replaced my old, bimetallic thermostats with digital ones (DT90E).
The thermostat is battery operated (2x AA), it is very simple, just up/down temperature setting, no BT, no WiFi, no smartBS, etc.
However, it has a relay that makes a loud 'clunk' that spreads across a plaster board wall. It is really annoying during night
When I was playing with it on a bench it almost wasn't noticeable, a quiet 'click'.
I tried to lifting it off the wall, putting rubber seals around the screws (something like in a HDD) to minimalise the vibrations but it didn't help much. The switching noise still passes through the screws into the wall...
The relay is mounted in a way that the contacts move towards the wall. First I though about desoldering the relay and mounting it in a way that the contacts will move parallel to the wall but then... what about a solid state relay?

Would it work in my case?

The main central heating controller sends a live 240V signal to the thermostat and if it closes the contacts it just passes the voltage further to a valve (so it does not power anything as some thermostats do).
The relay is an Omron G6CU-2114P, 3V DC for the coil, 8A 250V AC.
The motorised valve takes 30mA with the initial surge in a range of 600mA that last a fraction of a second.

If I replaced the relay with an SSR that would just pass the 240V AC further down the line - will it work?

Thanks and regards,
LJ

[ConKbot]

The relay in that thermostat is a latching type with 1 winding. So a current pulse in one direction turns on the contact, in the other turns it off.   This makes it extremely low average current, as LCDs take very little power, a low-power micro programmed well is very low current too.  Thats why it can run off of 2 AA's for months, or even beyond a year.  This also means that its a no-go on using a SSR as they require constant power when on, and your batteries will be done in a few hundred hours.

Youre right about the screws still conducting the sound. Proper sound-isolating mounts attach the item to be isolated to the rubber, and the rubber the wall/frame/whatever. 

My immediate go-to would be something like this https://www.mcmaster.com/75315a32  neoprene foam is generally the 'right amount' of softness and damping for this kind of thing. Too firm and it still conducts the sound, too soft and it compresses and sags, but a thermostat isnt exactly heavy.  If you cant find double stick neoprene locally, I'd look at the plain foam core double stick tape, the stuff that's about 1mm thick.  You could also try to find a good quality weatherstripping tape material (not the super lightweight stuff thats piratically falling apart even as you're applying it) glue the non-adhesive side to the thermostat baseplate and stick it to the wall with the normal adhesive side.    For eventual removal later, use a piece of floss to cut though the foam and you'll be able to get to the tape to clean it up.

[ljwinkler]

Thanks ConKbot - I missed the part about the latching type... this explains everything. Can't use ssr.

I left the thermostat hanging on the wires - it was much quieter. The relay was still noticeable but the sound wasn't amplified by the empty wall.

I know what are you talking about with the firmness of a tape - I have bunch of them so I will try and error different ones.

Thanks a mill for such quick reply!

[ljwinkler]

So...
After testing different tapes and other options I have decided to make my own

Please see attached schematic.

My thermostat utilises ATMEGA328P running on internal 8MHz oscillator, Nokia5110 LCD module, DS18B20 temperature sensor and SKA10420 SSR. All powered from 2x AA batteries.
The power consumption in sleep mode is 200uA (which is consumed by the LCD), the SSR takes roughly 2mA (with 440R in series, more than 560R wasn't reliable). The thermostat will be in use for max. 4-5 months and I don't expect the heating be more than 5-6 hours a day during the coldest days. Energy budget suggests that 2xAA will be ok in this scenario.

From the left on schematic:
- JP1 -> battery connector + switch (so I can turn off the entire unit completely),
- JP2 -> a connector for USB2Serial adapter,
- JP4 -> Nokia5110 display module + manual backlight switch,
- S5 -> double pole switch to turn off mains (as I described earlier, the main controller provides 230V that the thermostat passes it on to the motorised valve) - I wanted to cut off both ends.
- S6 -> allows me to manually override the thermostat in a case of a failure
- the resistor network (yes, I know, it's an overkill but I wanted to limit the heat generated as much as possible) is a feedback loop to the microcontroller - if there is no signal from the central heating controller, there is no point of enabling SSR (and wasting battery energy).

The program works in a way:
- the UP/DOWN buttons increment temperature by 0.5C, the temperature is stored in EEPROM (and restored after boot/power-up),
- the room temperature is measured every 15s (I may change it to 30?),
- if the room temperature is LESS than set temperature (minus 1C as a hysteresis) AND 230V is present from the central heating, then enable SSR (and pass the 230V to the valve).
- if the room temperature is MORE than set temperature OR 230V is not present, then turn off the SSR,
- the uC sleeps all the time, waking every 15s to check temperature, conditions, etc. an goes back to sleep,
- the UP/DOWN buttons are connected to interrupt pins so these can be pressed at any time,
- the LCD shows: set temperature, current room temperature, battery voltage, percentage (from 2.4V to 3V+), battery icon (with 100 pixels so each pixel represents 1%) and a heating icon.

If you have any feedback/comments, please let me know. I haven't designed a PCB yet, it just sits on a breadboard right now while I'm testing it.

[MiDi]

Would guess an 8A SSR drop in replacement for a relay is hard to find - size and losses would be greater (latching type and price unconsidered).