Sontex heat meter accuracy over time

How are people finding the accuracy of Sontex heat meters, especially over time, as compared to ASHP flow meters?

In theory, the Sontex meter comes pre-calibrated and should be trusted much more than any built-in flow meters in ASHP, but in my experience, the flow rate reported by the Sontex meter isn’t always as accurate as you’d assume.

I had a “Sontex Superstatic 449” heat meter installed along with Valiant AroTHERM just over a year ago as part of a PasivLiving (government-subsidised) monitoring system. But when I looked into the low COP’s reported in the dashboard I found that the Sontex meter was reporting 20% less flow rate than the ASHP provided information. This in turn impacted the energy output, COP and also my RHI repayments.

As part of my first annual ASHP service, I asked my installer to replace the Sontex flow meter body and immediately the flow rates matched the Vaillant flow rate exactly. That said, by the end of the same day the Sontex meter was reporting -5% less than the ASHP and a week later is showing -7%.

The flow meter body is installed according to the installation instructions vertically on the return inside the house.

In theory, ultrasonic flow meters should not be impacted by dirt, but I’m struggling to get consistent figures that can be relied on from the Sontext meter I have installed and wondered if anyone has experienced anything similar or has any advice?


1 Like

Hello @dfeist that’s interesting. I only have the one Sontex 531 on mine, no parallel meter to compare with. @glyn.hudson has a Sontex 789 and then also the Sika VFS that comes with his Samsung heat pump. The flow rates from both of those appear to be within 1% and there’s no visible change in that since it was installed last September.

I am comparing it with the ASHP flow-rate reports. Initially, I assumed the Vaillant flow rate must not be accurate. Others, from what I’ve seen on Twitter also trust heat-meter over their ASHP figures.

But given a large difference, I looked in more detail and saw that if I added 20% to the heat-meter reporting flow rate then power output and COP all started matching the spec sheets perfectly.

The flow rate of the new flowmeter body was within 1% here too initially, but within a day it was reporting 5% less! I can’t easily clean/replace it without draining the heating circuit and I don’t want to pay our installer every few months to look at it.

Sontex are not ultrasonic. They use pressure fluctuations from vortex shedding to measure flowrate.

The only weaknesses that I know if are:

  • air in system / sensors

  • dirt in system / sensors

  • flow disturbances that will affect flow through the sensor (mostly on larger sizes that explicitly require straight lengths

  • vibrations that cause spurious pressure fluctuations (you can get a successor unit to clock up volume with it say dry on the desk if the vibrations are right)

I’ll cut up an old one so you can see the guts when I first a chance.


I’ve got a Sontex 789 heat meter and a Samsung heat pump with a Sika flow meter, I’m logging data from both. I’ve found they match very closely. The Sontex has a slightly more steady output:

Over the past month the Sontex recorded an average of 8.9L/min which the Sika reported 8L/min.

If there was any discrepancy I would be more included to believe the Sontex since the Sika is a vortex flow meter which is prone to being more effected by air bubbles than the fluid oscillation principal that the Sontex uses.

Temperature sensing is a different story, I’ve found significant inaccuracies in the onboard temperature sensing in my Samsung heat pump compared to the Sontex which uses wetted pocket temperature sensors. This results in the heatpumps inbuilt monitoring reporting 30% less heat delivered than the Sontex. The Sontex has Class 2 accuracy guarantee while the manufacture’s inbuilt monitoring has no accuracy guarantee, therefore I’m more inclined to believe the Sontex.

1 Like

Hard to take an average on your system I think as the samsung does not always report the flow rate at the same time? there’s that strange delay on the data reporting?

Good point, can be a 60s between the two (Sika reported via Samsung modbus), this will be messing up the average:

Better to look at the average during steady state running which less than 2% difference.

1 Like

@marko You are right, I wrongly assumed they were ultrasonic!

Air in the system is unlikely because the flow meter is installed vertically, the system supposedly has the necessary valves to allow air out, and I have run the purge cycle for 1hr.

Dirt is a possibility I guess. The only filter I know of (which is incorporated in the return shut-off valve) is after the flow meter, rather than before.

There is a 150-200mm straight pipe before the flow meter, and the flow meter is DN20. I believe this meets the installation requirements. There are no vibrations. The flow rates we are talking about are in the range of 800-1200 l/hr.

Any thoughts on how I could/should adjust the installation to avoid these issues?

Dirt in the system is not good news for any components, have you got a strainer and magnetic filter fitted? Assuming the installer cleaned these out during the service? If so your system should now be dirt free, maybe it wasn’t properly flushed after the initial installation? The Sontex should be able to handle dirt better than other ultrasonic meters, but I’m sure there are limits to this! Does the old meter body the installer removed look dirty inside?

Yes, the installation has both a strainer and magnetic filter, although the strainer is after the flow meter (between the flowmeter and the monoblock), rather than before it.

The strainer was cleaned as part of the service, unsure about the magnetic filter. Initially, the new flow meter body reported exactly the same as the ASHP.

The old meter body did have some dirt, but I didn’t see it personally so not sure how much. It’s possible that the system wasn’t fully flushed after the initial installation, although I would have assumed the installer would have done this.

I do actually have another Sontex heat meter in the system (on the UFH loop) and by comparison, this has had no issues at all from what I can tell. The only difference with this meter is:

  • It’s installed horizontally and not vertically.
  • It’s qp1.5 and not qp2.5
  • It’s after the buffer. not on primary return.

Our set up has a Sontex heat meter (as part of Passivsystems MMSP) and we also have a circulation pump speed % from the NIBEuplink data feeds. Obviously the NIBE is just a % of the max speed at which it thinks the circulation pump can run. Nevertheless the 2 parameters are clearly in reasonable agreement in relative terms.


“Temperature sensing is a different story, I’ve found significant inaccuracies in the onboard temperature sensing in my Samsung heat pump compared to the Sontex which uses wetted pocket temperature sensors. This results in the heatpumps inbuilt monitoring reporting 30% less heat delivered than the Sontex.”


That discrepancy almost entirely due to the working fluid configuration not the temperatures. The heat pump is coded to assume glycol mix; which has a lower heat capacity than water.

“The Sontex has Class 2 accuracy guarantee while the manufacture’s inbuilt monitoring has no accuracy guarantee, therefore I’m more inclined to believe the Sontex.”

I’d say not quite.

The Sontex is type approved to MID Class 2. In the real world it will only be as accurate as the installation. For pure water systems it will indeed be good. For glycol mixtures it’s anybody’s guess as to how closely the actual mix matches the working fluid configured on the heat meter calculator. There is no “guarantee” of accuracy though.


I’m not a Sontex employee or rep so can’t add much more than I have already.

If the old meter body “did have some dirt” that is definitely a problem for your system in general. (there shouldn’t be any)

I would waegr that it’s probably also a problem for the Sontex meter in particular as these rely on vortex-shedding, vortex-shedding relies on sharp edges and strong reflections amongst other things, and dirt in our meter is probably gong to soften all those / reduce vortex shedding frequency / cause under-reporting of flow.

If it were mine I would isolate, pull the thing to bits (you can dismantle the flow sensing element on these), and physically inspect for muck in the pressure ports to rule that out.

Here’s what one looks like inside: (the little holes are the pressure ports to the pressure transducers that are looking to sense the vortices as these are shed from the assembly designed to create/reflect these - see Sontex Superstatic 749 Fluidic Oscillator Heat Meter - YouTube for the flow patterns that crate pressure changes that the meter picks up)

Vs an ultrasonic meter of the “external transducer and mirror” style: (signal shoots across the pipe, hits the mirror, along the pipe, hits the mirror, then to the other ultrasonic transducer) Very important on these that the transducers are on the SIDE of a pipe if mounted horizontally not on the top (air) or bottom (dirt). There’s really nowhere for any dirt to settle here.

Or an ultrasonic meter of the “inline transducer” style: (signal shoots along the pipe between the two transducers) Orientation is less critical for this style and again there’s really nowhere for any dirt to settle here.


What makes you say this @glyn.hudson ?

If anything I think they’re the most sensitive type of meter to dirt.

Air less so. The vortex shedding and pressure sensing elements can handle some entrained air.

Dirt (sediment) I’d say there’s plenty more scope for mischief for these non-ultrasonic meters than the ultrasonic types.

Photos in post above.

1 Like

Hello @marko those are really interesting cut through pictures and insights on dirt vs air, thanks for sharing!

We did initially think that this was the case, so I will try and explain why we think the error is due to the surface mounting of the temperature sensors, or perhaps an issue with one of the Samsung sensors.

Taking a snapshot of the data in a steady state period:

Sontex flow temperature: 35.23°C
Sontex return temperature: 32.39°C
Sontex deltaT: 2.84K
Sontex flow rate: 15.74 L/min
Sontex heat output: 3097W  
3097 / ((15.74/60)*2.84) = 4157 J/kg.K

Samsung flow temperature: 34.20°C
Samsung return temperature: 32.10°C
Samsung deltaT: 2.1K
Samsung flow rate: 15.66 L/min
Calculated heat output based on 4157 J/kg.K
4157×2.1×(15.66÷60) = 2278 W
Calculated heat output based on 20% glycol 4020 J/kg.K
4020×2.1×(15.66÷60) = 2203 W

The samsung in this example is reporting 1- (2203/3097) = 30% lower than the sontex, exactly in line with the differences Glyn is seeing from the energy reports on the samsung display. If the samsung was measuring the same DT as the sontex but had a calibration for 20% propylene glycol at 4020 J/kg.K that would only present a 3.8% error.

Glyn also has a couple of DS18B20 temperature sensors attached carefully to the surface of the pipes:

DS18B20 flow temperature: 34.32°C
DS18B20 flow temperature: 31.60°C
DS18B20 deltaT: 2.72K

A couple of other spot comparisons show the following DT differences:

Sontex DT: 34.82−31.7 = 3.12K
Samsung DT: 33.81−31.36 = 2.45K
DS18B20 DT: 33.9−30.90 = 3.00K

Sontex DT: 48.65−44.00 = 4.65K
Samsung DT: 47.26−43.38 = 3.88K
DS18B20 DT: 47.07−42.56 = 4.51K

The samsung DT seems to be around 0.7K lower than the other two DT’s, that’s going to vary a bit of course but seems to be fairly consistent and appears to explain the 30% discrepancy that Glyn is seeing


No worries on cut throughs. Sometimes it’s easier to take the 4.5" electric spanner to something than to describe it in words. :slight_smile:

(the day job is heat metering and much time is spent on the various ways the real world differs from lab conditions - should write a cheat sheet at some point as time allows)

Thanks for the explanation on the Samsung. I think you’re closer to right than I am on this one. :slight_smile:

Differences in heat capacity are not in fact large enough to explain what you’re seeing.

Watch the units though. 15.66 litres/minute is not 15.66 kg/second. You ought to correct for density when converting volume at a temperature to a mass flowrate.

Perhaps the error caused by their measuring temperatures on the refrigerant side is also more significant when operating at deltaTs near the bottom of the expected range. (nits probably designed to operate at 5 or 8K dT rather than 3K?)

FWIW @dfeist - I’ll do a proper version in due course but perhaps this gives some better idea:

Unless your Sontex meter is mounted with the flow going DOWN I can’t really see anywhere for the muck to accumulate. If they’re mounted horizontally (with the pressure sensor on top) or vertically (flow going UP) there shouldn’t be anywhere that muck can sit.

If it is mounted with the flow going DOWN then you could be accumulating muck in the chamber that the oscillations are supposed to occur in. This won’t be visible with the top remove though. You’d have to cut the thing open to see it. (or rinse in clean fluid and measure the sediment later)

Ultrasonic types are:

“-” straight transit

“|_|” reflector-ed straight transit

“V” or “W” reflected transit (not shown in the clip above)

The “|_|” “V” or “W” types MUST be mounted with the ultrasonic sensing running horizontally NOT with the sensors/reflectors at the top or bottom of the pipework. If you ALWAYS mount ultrasonic types with the sensor on the side of the pipe when they’re mounted horizontally you’ll be safe.

The “V” “W” or “Z” types are probably the most resistant to muck and bullets in operation. You can happily pour a jug of frozen peas through a DN25 or larger “V” or “W” type flow sensor without issue. (the system strainers / heat pump heat exchangers will complain long before the flow sensor does)

Here you can see one of the sensors (foreground, shiny piece first visible in the tube) and the reflector (next shiny piece) that form the top of a “W” sensor:

(an Axioma in DN25)

The smaller flow sensors are often “-” type or the “|_|” type as you can assume flow through the orifice in the flow sensor body is more uniform. The “V” and “W” types are exposed to more of the flow “profile” as it were / less affected by variations in fluid velocity at lower flowrates.

1 Like

Another doc on flow sensor orientation and positioning with regards disturbances; this time from Kamstrup but relevant to other vendors using similar ultrasonic tech:

The challenge of correct flow sensor installation with respect to flow disturbances.pdf (517.7 KB)

The primary Sontex meter which has this issue does indeed point downwards. I have another Sontex meter on the UFH circuit in a horizontal position that hasn’t suffered from the same issue, so it looks like this might be the issue.

The thing is, Sontex installation guidelines explicitly say that vertical installation (either way up) is fine, so in theory, the installation is correct and the installer has done things correctly.

Looks like the only way I’m going to get a consistently correct meter reading (and correct RHI payments) is to either put a strainer before the meter or change the orientation. Right?

Or clean the system if it is indeed dirt. Sontex instructions will say should be filled with water not dirt.

I’d pull the meter and look myself. We’re still guessing here!