Stories of subpar engineering

I've set up this page to summarize a few issues I've run into with the Seltron Promatic D20, a weather-adjusted heating controller. I hope it will help other people diagnose and prevent similar problems. Given what follows, I recommend NOT buying this product, and if it's any testimony to the quality of Seltron products and the company's customer support, I can't recommend buying other Seltron products either.

I have bought this controller in 2013 (at a time where it seems the company was already phasing it out - something I didn't know then) because it fit the bill for my needs (including support for boiler-integrated d.h.w. tank) and was quite affordable. Unfortunately, as the following will illustrate, it turned out to be an example of a good idea disserved by poor execution. It all became obvious when the heating pipes exploded last winter because of frost, something that was never supposed to happen if the device had performed correctly. I'm not rulling out mistakes on the user side, but I have done quite a bit of investigation around these issues so I am led to believe that they are inherent to the system itself. Seltron even acknowledged some of them.

Unmet expectations

I'll list here some expectations that turned out to not be verified, the consequences and a proposed fix. These all apply to firmware version 2.1r8. The selected hydraulic scheme is 401 (single mixing circuit with circulating pump, motorized 4-way valve, integrated d.h.w. tank). The regulation controls the circulating pump (UWP), the burner (LB1/B1), and the 4-way mixing valve (M-/M+). Attached sensors are:

Importantly, there is no room sensor (RF). This shouldn't be a problem for a well designed regulation: they can operate based solely on outdoor temperature and a few parameters (water-temperature law - aka heating curve, and building inertia constant).

Frost protection

Assumption

When the heating is "turned off", frost protection will protect both the boiler and the heating circuit.

This is probably the most basic expectation one can have for such a device, unfortunately it turned out not to be met (hence the blown pipes).

Actual behavior

The burner does turn on when the frost protection outside temperature threshold is reached, but the circulating pump is not turned on, which has two very important consequences:

  1. Water isn't flowing in the circuit, which facilitates frost
  2. The VF/RLF sensors aren't picking up thermal change (and are actually "contaminated" by the boiler room temperature - where they are typically installed)

Consequences

The heating circuit (heating pipes and radiators) is not protected against frost and catastrophic failure will happen in very low temperatures.

Solution

At the very least, when frost protection is engaged (below the desired temperature threshold), the circulating pump should be activated. This will ensure water flow in the pipes (already reducing the risk of freeze) and it will enable the VF/RLF sensors to read the actual water temperature, instead of the boiler room temperature. This should suffice to ensure adequate frost protection for the installation, assuming the regulation does open the mixing valve (which is far from certain, see the next point).

Note: Seltron suggested that this problem could be "fixed" by adding a room sensor. That is unfortunately not an option (if it had ever been one, I would have added it in the first place), and in any case not a satifactory solution for a device supposedly designed to operate correctly without such a sensor.

Minimum mixing circuit temperature

Assumption

The Promatic D20 has a setting (S2.6) to select a minimum temperature for the mixing circuit. The factory default sets it at 20°C, and I left it unchanged. I assumed that when the frost protection would engage (or when the heating was turned on), the water temperature in the mixing circuit would not go below this setting.

Actual behavior

The regulation sets a mixing circuit target temperature of 4°C when heating is turned off, including when frost protection conditions are met.

Consequences

This setting is not honored during frost protection, which means this additional failsafe is actually useless.

Solution

The firmware should honor the setting in all situations where the mixing circuit is active, including during frost protection.

Gap in timing settings

Assumption

There shouldn't be a gap between timing settings that overlap day boundaries (i.e. that cross the midnight mark). For instance, it should be possible to set a continuous "Comfort" time range from 8AM till 1AM the next day, without any interruption in the heating program at midnight.

Actual behavior

The time range selection is only possible in increments of 5 minutes, and it is not possible to select 24:00 (midnight) as the ending time for a given time range: the maximum is 23:55.

Consequences

To achieve the aforementioned behavior (Comfort from 8AM till 1AM), one might be tempted to set a time range from 8:00 till 23:55 and then from 0:00 till 1:00 on the next day. Unfortunately, this causes a 5mn gap during which the regulation switches to "eco" (night) mode. When the next period starts at midnight, it triggers a transition from eco to comfort mode. Now, if the heating boost setting is enabled (via S2.3 and S2.4), this transition triggers the boost period, and since there never really was a cool down in the first place (because the eco mode lasted for only 5 minutes), and because we don't have a room sensor, this overheats the place and results in an unnecessary overconsumption of fuel.

Solution

The firmware should enable continuous timing ranges that overlap midnight.

Note: Apparently this was partially fixed in a later firmware revision, but upgrading is non-trivial (see below).

Transition to lower heating

Assumption

In many such regulation systems, when the regulation transitions from a high heating state to a lower one (e.g. from Comfort to Eco, or from Comfort to Frost Protection), it will leave the output circuit(s) circulating pump(s) running for a few more minutes to enable thermal dissipation from the boiler (and avoid boiler overheating). In the case of a mixing circuit, this requires the mixing valve to keep its last position while the cooldown happens, otherwise (if the valve closes) the circulating pump is not circulating water through the boiler.

Actual behavior

While the Promatic D20 does have a provision for leaving the circulating pump active for a while after the transition has begun, it immediately starts closing the mixing valve.

Consequences

This is especially bad when the transition begins immediately after the boiler has been heating up: due to the boiler's own inertia, heat builds up and cannot be evacuated in the mixing circuit that has closed already. Meanwhile, the circulating pump operates for nothing (and wastes electricity doing so).

Solution

The firwmare should retain the mixing valve position when the transition begins and maintain it during the cooldown grace period.

Building time constant

Assumption

In order to operate efficiently, especially without a room sensor, a regulation typically needs to have some information about the building temperature inertia. This is known as the building time constant. To give a quick visual representation, a house made of paper walls will not have the same inertia vis-a-vis temperature changes (both internal and external) as a house made of heavy, thick insulated walls.

While this parameter is essential for proper operation without a room sensor to monitor indoor temperature changes, it is also helpful even when such a sensor is available: it enables the regulation system to anticipate indoor temperature changes when it transitions from one mode to another (typically to anticipate the night mode so that when the night mode time range begins, the temperature will have just started moving down already).

Actual behavior

The Promatic D20 doesn't allow setting the building time constant.

Consequences

It proved really hard to adjust the regulation settings to obtain a reasonnably adequate operation in the (old) building it was installed in.

Solution

The firmware should allow for the building time constant to be set and should take it into account for calculations.

Aftermarket support

While Seltron has been relatively responsive when I contacted them about these issues, they offered very little actual solutions. As the list shows, most of these problems could be fixed by a firmware upgrade.

They did offer to update my device to the last available firmware (which fixes the overheating due to boost trigger after the 5mn switch to eco between 23:55 and midnight), but they kept repeating that the product development had stopped a couple years ago and that they wouldn't update it anymore. They even suggested that the newer WDC20/WDX20 models were more powerful versions of the D20 (which I interpreted as a suggestion for me to replace my recently bought D20).

Then again, that update:

Things would have been a lot easier if Seltron had made a provision for on-site upgrade (e.g. via RS232 if not USB).

I mentioned at the beginning that the product had been phased out (apparently it had already been so when I bought it, as they told me later on), and I have to point out that I did reach out to Seltron before making the purchase, asking them information about this controller. They didn't mention it was a deprecated product then, and it was still being sold by their local distributor.

They also have entirely removed all information about this controller from their english website now, including documentation and user manuals. This is really bad practice for devices that will typically remain in use for many years after they have been phased out: these are not the kind of devices people upgrade yearly.

Overall, this shows a fairly low level of care for aftermarket support: suggesting that the only fix to clearly identified (and relatively easy to fix) problems is to entirely replace the device with a newer one because "development of this product has stopped" is simply not realistic nor does it show much concern about consumer satisfaction, durability or even sustainable development.

Conclusion

These are some of the most significant problems I encountered with this product. In the end, I had to replace this device due to the broken/missing features (as described above), and due to the insufficient support from the company.

I currently cannot recommend Seltron's products and I have replaced this controller with a custom-designed one which I document here.