Experiences with OTGW

This Forum is about the Opentherm gateway (OTG) from Schelte

Moderator: hvxl

Experiences with OTGW

Postby gda01 » Thu Sep 21, 2017 9:05 pm

Experiences with OTGW in winter season 2016-2017

In winter season 2016-17 we used the OpenTerm Gateway (OTGW) in our weekend home. This post provides a report on our experiences.

We use our weekend home, as the name says, mainly during the weekends. We needed a facility to remotely switch on the heater a few hours before our arrival, so that the home would be warm when we arrive. During the week in the winter season the home may cool down to 7 C.

Some energy providers offer smart thermostats with connection to internet which offer these remote control facililties. We did not opt for this solution as it requires subscription fees. Further on all remote control traffic is routed via server of the providers. As we prefer open standards, we opted for the OTGW controlled by a Raspberry Pi.

Our technical configuration: OTGW (v.4.2.5) with OT monitor (v. 4.3) on RPI model 2B with Jessie. Apache2.4 server reverse proxy with TSL and client certificate. Thermostat Honeywell Chonoterm Touch Modulation. Heater Vaillant hrPRO VHR NL CW 3/3 (with additional OpenTherm PC board).

We installed the OTGW and OT monitor on the RPI in the standard configuration and got it working after some initial hickups:
- The PIC as provided was loaded with the diagnose software and not with the OTGW software,
- The client certificate as created with the OT monitor is not accepted anymore by current browsers. We created with OpenSSL a local Certificate Authority and server and client certicates. We got the server certificate installed with the OT monitor. But we did not succeed installing the client certificate with OT Monitor.

We created a housing for the OTGW plus RPI and installed this in the heater closet. October 2017 we set it operational.

The configuration was not that stable as desired. Occurring problems:
1. RPI lost communication with OTGW
2. Surprising behaviour of the OTGW
3. Thermostat lost communication with the heater.

1. Serial communication between RPI and OTGW is sometimes lost.

One reason of losing the serial communication may have been the hardware connection between RPI and OTGW. In the first setup this was a pc board bridge containing the voltage divider for the 5V Tx of the OTGW to the 3.3V Rx of the RPI. There was some mechnical tension on this bridge that might have affected the electrical connectivity of the connectors. Is replaced now by a flexible cable. This issue seems to be resolved.

Another reason is that the PIC sometimes hangs. There is no clue yet for this hanging. This problem happens once in a few weeks. After a power reset the communication starts running again. This problem prevents us sometimes to switch on the temperature when we go to the weekend home. Our plan is to make a OTGW hardware remote resettable via the RPI.

A reason for the PIC hanging might be EMC interference with the pump motor of the heater. We added earthing to the housing of the OTGW-RPI, and added a line filter in the power line.

The RPI appeared remarkebly stable.

2. Surprising behavior of the OTGW

We installed an outside temperature sensor (DS18B200) connected to the GPIO. This works ok. The outside temperature is shown on the thermostat, as well as on the remote web page. However since we installed the outside temperature sensor, for unknown reason , the Setback function of the OTGW is triggered now and then. The setback issues a constant temperature override (CT) command. There is no clue what triggers the setback function. We suspect there is a relation with the outside temperature sensor. As work around we set this setback temperature to 6C to prevent the weekend home to heat up during weekdays.

3. Thermostat loses communication with the heater.

This is the most serious problem. It happenend twice that the thermostat lost contact with the heater while the heater was on. This kept the heater burning for a week, by which the temperature in the weekend home rose to almost 40C.

The communication between thermostat and heater - with OTGW in between - appears to be temperature dependent. This shows up in error 3 errors on the heater messages. We had a case when opening the door of the heater closet the error 3s disappeared (the temperature in the closet lowered), and when closing the door, appearing again (after the temperature in the closet rose again).

This temperature dependency of the OTGW is really an issue. The temperature in the weekend home may vary between 7C and 20C. When the heater starts burning to keep the temperature in the house at 7C (while freezing outside), the temperature in the heater closet may rise quickly from a 10C to a 25C.

Errors 3 can be solved by changing the voltage reference setting of the PIC. But when the communication between OTGW and RPI is lost also, you are remotely blind on the OTGW.

For the upcoming heating season, still to investigate what components of the OTGW make this temperature sensivity. Opto couplers are know temperature sensitive, but that is in linear operations. Is it our specific OTGW exemplar that is temperature sensitive, or is it in the design ? We have now a second exemplar of the OTGW and plan to do some measurements and tests on temperature sensivity. We also will have look at the temperature characteristics of the components being used, and eventually add temperature compensation circuitery.

Remote control

To apply a client certificate, and to open a remote channel for a hardware reset under RPI control we installed Apache as reverse proxy to the OT monitor. This required us to migrate on the RPI to Jessie OS, which has Apache 2.4, which is necessary for reverse proxying websockets. We add an always on relay in the power line of the OTGW, which we can interrupt as hardware reset, with a web call to the RPI.

With these enhancements we hope to get better control on the OTGW in the upcoming heating season.
Starting Member
Starting Member
Posts: 4
Joined: October 2016

Re: Experiences with OTGW

Postby gda01 » Sat Sep 29, 2018 8:29 pm

Update - Experiences with OTGW in winter season 2017-2018
We use the Open Term Gateway (OTGW) in our weekend home. The previous post reported a number of problems during the winter season 2016-2017. This post gives an update of the winter season 2017-2018.

Our technical configuration: OTGW (v.4.2.5) with OT monitor (v. 4.3) on RPI model 2B with Jessie. Apache2.4 server reverse proxy with TSL and client certificate. Thermostat Honeywell Chonoterm Touch Modulation. Heater Vaillant hrPRO VHR NL CW 3/3 (with additional OpenTherm PC board).

In the winter season 2016-2017 the problems were:
1. RPI lost communication with OTGW
2. Thermostat lost communication with the heater.

This resulted twice in heating the weekend house till about 40 C.

To determine whether the issues were on device instance level or on design level, we ordered and assembled a new OTGW.

With this new OTGW installed, the reported issues did not appear in the winter season 2017-2018. So the issues happened on the device instance level, not on the design level. May be an unhappy combination of components.

We checked the temperature sensivity of the opto-coupler by heating and cooling the opto-coupler IC. Yes we observed a temperature dependency with the opto-couplers. But the way the opto-couplers are being configured, this temperature dependency should not be an issue. We did not create a test setup on temperature sensivity of the thermostat interface providing input to the PIC comparator input. As this setup was more difficult to create and the new OTGW appeared stable.

Because the new OTGW was stable, there was no need to invoke a remote OTGW hardware reset via de relay we built into the OTGW power supply line.

Summarizing: after a 2016-17 heating season with many OTGW issues, the 2017-18 heating season with a new OTGW showed stable with no issues.
Starting Member
Starting Member
Posts: 4
Joined: October 2016

Return to Opentherm Gateway Forum

Who is online

Users browsing this forum: No registered users and 1 guest