M-Bus Water Meter Reading Discrepancies

There are three water meters in the house:

  • One on the incoming Cold water supply
  • One on the cold water supply to the bottom of the hot water cylinder, which measures the Hot water consumption (and needs to be subtracted from the incoming cold meter reading to leave the cold water usage)
  • One on the separate cold water supply to the potential-future-rainwater-harvesting pipework
    • This third one is actually fed from the incoming mains supply outside the house but might in future be connected to a rainwater harvesting system
    • Within the house, the pipework connected to this meter feeds the toilets, the washing machine and the outside taps (i.e. all the appliances which are permitted to be fed with harvested rainwater)

These meters are in addition to the water company’s ‘revenue’ meter and could be considered excessive, but:

  • The water company’s meter is about 600 metres away from the house so there’s a lot of underground pipework that could potentially leak (and which runs through fields which get ploughed) and it’s good to be able to cross-check the meter readings in case any leaks develop
  • It’s useful to be able to compare the Hot water usage with the Cold water usage – and to see how much of the water usage could potentially be displaced by rainwater harvesting
    • This is especially pertinent as I firm-up the specification for the Outbuildings, to be constructed as part of “Phase 2” of the building work, and compare the cost of a rainwater harvesting system with the potential savings

The three internal meters have M-Bus readers fitted and are automatically read every minute – along with all the other M-Bus readers in the house. (Reading so often is completely over-the-top for Water meters, but not for e.g. Electricity meters, and it’s easiest just to read all the meters on every read-cycle than to mess about reading different meters at different intervals.) Then, every month, I manually record the monthly meter readings (as reported by M-Bus) into a spreadsheet to look at a summary-level view and monitor trends in usage.

Over the past couple of months I’ve noticed the ‘rainwater’ meter showing much higher readings than before, which seemed odd. On the 1st of May, the usage was up a bit; on the 1st of June the usage was up a lot – roughly 10 x the usage in February, for example. Initially I suspected a leak but on checking further it was clear the meter reading wasn’t changing unless there was obvious water usage.

Then I realised what was happening: it’s not that the meter is now reporting ‘high‘ – it’s that the meter has previously been reporting ‘low‘. The dials on the meter are showing 114 m3 but the M-Bus adaptor is reporting 85 m3. Plotting a graph of the meter readings over time (that I don’t normally do) makes it clear that something went wrong around the end of April 2022 that got fixed around the middle of April 2024.

‘Rainwater’ Water Meter readings, early 2022 through to mid 2024

I know what I did to ‘fix’ it – I’d been checking the details of the meter to be able to match it in the new building and removed and re-fitted the M-Bus adaptor. I don’t know what I did to ‘break’ it – but given that some readings were still coming through I presume the adaptor was knocked slightly out of position. (If it had stopped reading completely I would have spotted it.)

The meters are Itron Aquadis models with Cyble M-Bus adaptors that slide into some slots on the front of the meters – but they don’t clip into place hence they’re reliant on friction (or gravity) holding them in position, and they can get knocked.

Now the problem is that the internal registers on the M-Bus adaptor don’t match the dials on the meter and there are no buttons or display on the M-Bus adaptor to make any adjustments – so it’s a case of using the Cyble M-Bus software that I originally used to set the M-Bus IDs to re-set the reading from the dials.

This was easy enough – once I’d worked out the Cyble M-Bus software needs to run as Administrator. Changing the reading is as simple as clicking on the digits and entering the new reading.

Screenshot of Cyble MBus v1.4 software

I’d forgotten these adaptors have the option to report against a Leakage threshold value; what’s not clear is how the detection of a ‘leak’ gets reported – there no obvious ‘leakage’ field in the XML data record that comes back to the M-Bus reader.

Sony STR-DE685 AV Receiver Repair

While this topic isn’t directly self-build-related, it’s at least house-adjacent. Embedded systems which rely on electronics tend to age more quickly than the buildings they are embedded in and where it’s possible to extend the life of the electronics by making simple repairs that’s A Good Thing.

I’ve been using a Sony STR-DE685 AV Receiver for about 20 years. These days I don’t use its FM input or multi-channel features but it’s been working very nicely as the stereo amplifier in the living room, taking a digital audio input via a coaxial cable from a Logitech Squeezebox ‘Classic’ music player and using the receiver’s built-in DAC to convert that to analogue for the speakers. The digital input is happy with 44.1, 48 or 96 kHz sample rates so there’s no need to down-sample high-resolution audio media, and the signal stays ‘digital’ for as long as possible. The audio quality is excellent.

Over the past couple of years, I’ve noticed the right channel becoming quieter than the left. Previously it’s been possible to mitigate this by adjusting the left-right balance but now that’s at the maximum setting (8 dB) and the problem is getting worse.

A bit of searching on the Internet turned up a couple of YouTube videos of people repairing similar issues on slightly older variants of the STR-DE685. The best video I came across is this one, for the STR-DE485:

Essentially, the issue is that there are some electrolytic capacitors mounted close to various heat-generating semiconductors that gently ‘bake’ the capacitors over time. These capacitors then fail to operate to their specification and impact the audio gain control behaviour. Typically, all channels get quieter and it’s common for one channel to get quieter than the others – so exactly the symptoms I’m seeing.

I’ve experienced similar issues with other electronic equipment previously. The most common types of electrolytic capacitor have a limited lifetime at high temperature (typically 3,000 hours at 85 C) so any equipment that gets hot is prone to capacitor failure. The good news is that new capacitors are cheap and they’re generally quite easy to replace.

Where there are only a few capacitors on a circuit board and they’ve all been exposed to the same temperature profile the best plan is to replace them all (which is what I’ve done previously). Where there are lots of them and where some have been hot while others have been (relatively) cool it’s better to be more selective and only replace the ones which are showing signs of failure. From the YouTube video I learned that the best way to assess the health of an electrolytic capacitor while it’s still connected in-circuit is to use an Equivalent Series Resistance test meter. Since I didn’t already have one of those I decided to buy one (because I predict I will have to do the same sort of testing on other equipment in the future). It quickly became apparent that one of the leading candidates is designed and manufactured by a small company located just 35 miles away – Peak Electronic Design Ltd in Buxton; specifically their Peak Atlas ESR Gold ESR70. Their online order form has a specific option for ‘Collect from Peak’ and I don’t need much of an excuse for a trip to the Peak District, so that was an easy decision to make.

In conjunction with measuring ESR to detect faulty capacitors, I decided to be a bit more methodical about checking by how much the speaker outputs were affected (and, hopefully, corrected). For this specific fault it would have been possible to use a ‘diagnostic’ digital audio file (pure sine wave at a suitable frequency) and to measure the output waveforms on each channel when playing that, but using a signal generator seemed a better way to go, in case it proved necessary to inject the signal other than via the digital input.

I already had a second-hand (ex Royal Navy, made by Fluke/Philips) oscilloscope which did a good job of confirming a ‘clean’ sine wave and measuring the output voltage (once I’d reminded myself how to use the oscilloscope well enough to take measurements from it). I was surprised by how bad the voltage differential was across the two channels: at the maximum (8 dB) balance offset the Left channel was twice the voltage of the Right channel, and with the balance offset removed the Left channel was five times the voltage of the Right.

The unit isn’t too bad to get into – by removing the right screws and connectors the main PCB comes away with the back panel, providing easy access to the solder joints underneath.

The ESR meter confirmed three 100 uF, 10 V electrolytic capacitors near to the pre-amplifier ICs were effectively open-circuit and three 4.7 uF, 50V electrolytic capacitors a little further away were suspiciously high on ESR. I bought replacements for both types, choosing 105 C-rated Panasonic-branded components (£1.12 and £1.16 for 5 of each, respectively, from RS Online). I replaced all six but after testing the 4.7 uF ones once they were unsoldered they weren’t too far out of specification and I don’t think they were causing any issues.

With everything back together, the channel balance was fully restored: overall more voltage (volume) and the difference between the two channels was too small to measure, confirmed when playing audio.

So overall a great result. It’s good to save a mostly-fine unit from the scrap heap for just a couple of pound’s worth of parts and a couple of hours of work.