Page 2 of 2 FirstFirst 12
Results 11 to 18 of 18

Thread: Evohome: living room getting warmer when warming other room!!??

  1. #11
    Automated Home Legend
    Join Date
    Sep 2014
    Location
    Scotland
    Posts
    2,361

    Default



    The living room has three panels mounted in a bay window and a DTS92 as sensor on another wall. Temperature regulation is excellent and there is no overshoot at all.



    The dining room has two radiators - a long radiator under a very large window and a small one on the opposing wall. The HR92 on the long one (exterior wall) is the sensor for the zone. Very minimal overshoot here. +/- 0.5C which is typical when sensing from an HR92. I plan to fit a DTS92 in this room soon as the HR92 is sometimes obscured by toys and naturally doesn't control the room temperature as well in those situations!



    The study has a long panel under the window and measures the temperature at the HR92. This room also suffers from obstructions near the HR92 sometimes and the radiator itself is a bit sick. Overshoot is minimal however. I'm considering fitting a DTS92 in this room as well because it often under reports the temperature of the room when there are obstructions near the HR92.



    The main bedroom has a somewhat oversized radiator under a loft window, with a DTS92 at the head of the bed on another wall. This room is subject to the window being open different amounts (or not at all) on different nights so the system has to adapt to this. This room had severe overshoot in the mornings with my old TRV valve bodies but as you can see is now OK despite the somewhat oversized radiator.

    Continued in next post due to 5 image limit (!)
    Last edited by DBMandrake; 4th January 2020 at 02:44 PM.

  2. #12
    Automated Home Legend
    Join Date
    Sep 2014
    Location
    Scotland
    Posts
    2,361

    Default



    This is our son's room and also has a somewhat oversized radiator under a loft window and a DTS92 on another wall near his bed. The room is an awkward L shape so a bit of a challenge to regulate the temperature of, however the system does a pretty good job, usually keeping it within +/- 1C as reported at the DTS92, and it does not overshoot significantly when other zones come on in the morning. (Perhaps 0.5C at the most)

    I believe the +/- 1C variations are due to this being the only radiator on most nights, causing the heat demand to fall below the minimum 10% (1 minute on time) the Evotouch requires to call for heat from the boiler. As a result the temperature sometimes has to fall significantly below the set point before the boiler comes back on. I suspect that Opentherm control would avoid this issue, so if I ever replace my geriatric boiler I'll be trying to get one with Opentherm for this and other reasons.

    Hopefully that gives an example of real world data on a system that is not experiencing the overshoots you see.

    Going to try the valve stroke setting on the my one valencia valve, as I thought would not need it setting
    Have a look at my valencia thread - it's a long read but despite others disagreeing I found that I had to use Stroke 1 for satisfactory results, and also not use the integrated balancing insert - if you try to balance the radiator with the integrated insert you add a large deadband to the HR92's control range as you are basically restricting how far the pin can come out when you wind the balancer down. I found it much better to wind the balancer out until it was nearly loose, then balance the radiators the traditional way with the lockshield valves, allowing the HR92 to maintain a full control range.

    Also trying to throttle some of the troublesome zones.

    I will keep you posted, and thanks for your input
    Is your system well balanced in terms of lockshield valves etc ? If not it might be worth having a look at that.


    Quote Originally Posted by frankmalia View Post
    Checked the honeywell valencia trv and valve was 13% position and definetly passing water across it.

    Set valve stroke to Full 1 and disconnected TRV and Reconnected to calibrate again.

    So about the less than 30% being closed this one only closes fully at 0% I think.
    I'm surprised your HR92 was reporting 13% and still flowing on a Valencia. On mine in Stroke 0 mode they start flowing at 40-60% (!) and in Stroke 1 mode about 30%.
    Last edited by DBMandrake; 4th January 2020 at 02:48 PM.

  3. #13
    Automated Home Sr Member
    Join Date
    Apr 2019
    Posts
    63

    Default

    DBMandrake
    Another great response and had a play with Valencia after reading your other thread.

    Ok, correction to my statement my Valencia is opening some where above 17%(close) and definitely at 43%(open), could not get valve to go to 30ish% with temperature conditions and setpoints. Played today when measuring pipe temperatures. To detect hot water flow. Unfortunately the Valencia valve radiator sounds like its flowing when it was not, never realised, this is the only radiators that makes noise like this when not flowing, must be coming from below pipework.
    Your correct some of my older radiators, flow below 30% valve position, one flowing at 11% position, checked with temp gauges and touch.
    Some of these, yeah I am seeing the +1.5 on these ones.

    Looking at you nice graphs, what do you use for that?
    I’m a little less concerned as you are getting overshoots. More than 1.5 on your graphs sometimes. Like you said the room temp is ok, if you measure it with separate temp gauge away from HR92 - but I use the only constant reading I have which is the HR92.
    My hallway is one of the better rooms for minimal overshoots, like yours. Large space for heat to dissipate, plus flow of heat upstairs.

    Couple of the rooms have obstacles close that may have any effect. But I think like you say it’s the TRVs open below 30% spool position.
    The fact that Honeywell do this is crap because the system is meant to be able work on other TRV valves, they have a large compatibility list. System should be 0 %, valve rammed closed. Then anything else above this considered a calibrated and learned open position.
    Again this still a great system that is better than others. But the fact you buy a system that is sold like its has this potential is not good. 8 of TRVs are new ish not Honeywell(different manufactures, drayton/bulldog etc) 1 is a Valencia. The 30% close/open thing is not good, if dodgy on certain valves.

    Anyway I have the kitchen and a couple of bedrooms to look at changing valves for Valencia for a little project in a couple of months time.

    Thanks for you help and info.

    Frank

  4. #14
    Automated Home Legend
    Join Date
    Sep 2014
    Location
    Scotland
    Posts
    2,361

    Default

    Quote Originally Posted by frankmalia View Post
    DBMandrake
    Ok, correction to my statement my Valencia is opening some where above 17%(close) and definitely at 43%(open), could not get valve to go to 30ish% with temperature conditions and setpoints.
    To get closer figures it just takes a bit of patience, waiting while it makes gradual changes in response to room temperature changes and then rechecking option 10.

    I checked my Living room radiator yesterday and confirmed that the valve was flowing slightly at 37% valve position indicated and was warm, and at 27% it was definitely completely stopped, so I'm happy with that. That was also one of the radiators which had problems with overshoots due to not shutting down the flow properly with my old valve bodies.
    Played today when measuring pipe temperatures. To detect hot water flow. Unfortunately the Valencia valve radiator sounds like its flowing when it was not, never realised, this is the only radiators that makes noise like this when not flowing, must be coming from below pipework.
    I have the same issue in our bedroom - our radiator branches off the lines that go to our son's bedroom and as his radiator is set to a higher night time temperature it tends to run during a winter night when ours does not - ours hisses slightly and sounds like it running but the radiator is stone cold and the room above the set point so the noise must be transmitted from the pipes behind the walls to the radiator panel...
    Your correct some of my older radiators, flow below 30% valve position, one flowing at 11% position, checked with temp gauges and touch.
    Some of these, yeah I am seeing the +1.5 on these ones.
    Still flowing at 11% is definitely going to cause overshoots, no doubt about it, that's even worse than the ones I was originally having problems with. You would want to see the threshold between 25-35% for optimal operation and minimum interaction between zones. If you can solve that you should be able to solve or greatly mitigate the problem. Did you try enabling Stroke 1 mode and letting the valve re-calibrate yet ? If the valve is a bit stiffer than average (strong spring tension on the pin) the HR92 may have trouble closing it and may not calibrate well.

    The calibration procedure winds the pin down until a certain amount of force is reached (I presume by measuring the current drawn by the DC motor) and this point is considered to be 0%, with 100% counting back a certain number of turns from there. This builds in some assumption about the amount of force a valve takes to sufficiently compress the sealing washer, and therefore how far the pin will move back out before the washer un-compresses and water starts to flow past the washer.

    If the spring in the valve is unusually strong (or the valve is starting to seize slightly due to corrosion of the pin, like mine were) they require more force to operate, this can cause the calibrate routine to find the fully down position incorrectly as when it thinks it is fully compressing the washer it is really just finding high resistance from the spring or the pin itself. This means the washer is barely compressed at "0%" and it only needs to move say 10-15% before it starts flowing. In extreme cases it may not be able to close the valve at all and it may leak all the time.

    One thing Stroke 1 mode does is increase the force exerted when finding the closed position during the calibration, as a result 0% will be pushing down harder on the pin and compressing the washer more, therefore the opening point where it starts to flow will occur at a higher numeric percentage, and it will more reliably calibrate on a stiff valve. If the reason for the stiffness is friction from a corroded pin this may still cause erratic calibration problems where it initially calibrates correctly but over time the calibration goes out as it keeps trying to re-calibrate during use. I think this is what was happening on my old valves.
    Looking at you nice graphs, what do you use for that?
    They're from Linux based graphing software Munin using the Evohome-munin plugin.

    I run it on a Raspberry Pi. The outdoor temperature is normally taken from an online service (weather.com I think) however I've modified the script to receive my outdoor weather station using an RTL-USB receiver to get more accurate and realtime outdoor temperatures.
    I’m a little less concerned as you are getting overshoots. More than 1.5 on your graphs sometimes. Like you said the room temp is ok, if you measure it with separate temp gauge away from HR92 - but I use the only constant reading I have which is the HR92.
    My hallway is one of the better rooms for minimal overshoots, like yours. Large space for heat to dissipate, plus flow of heat upstairs.
    In the rooms I am seeing big overshoots in (kitchen and bathroom) this is a result of the placement of the HR92 and not a true representation of the room temperature. If I was using a remote DTS92 sensor in these zones I don't think I would see an issue there.

    When an HR92 is the temperature sensor its important not to assume that overshoots or fluctuations necessarily show changes in the true room temperature as localised heating from the radiator can cause larger swings that don't really exist in the room. If you see large overshoots with a remote sensor then yes, the overshoot is real.
    Couple of the rooms have obstacles close that may have any effect. But I think like you say it’s the TRVs open below 30% spool position.
    The fact that Honeywell do this is crap because the system is meant to be able work on other TRV valves, they have a large compatibility list. System should be 0 %, valve rammed closed. Then anything else above this considered a calibrated and learned open position.
    Only shutting at 0% would not work very well for proportional control. Although water shouldn't be flowing below 30%, the range from 0-30% position is still part of the proportional band from a control perspective. So if the room is slightly over temperature it may drop to 25%, if its significantly over it may drop to 20% and so on. How much below 30% it is allows it to track how far it is from equilibrium.

  5. #15
    Automated Home Legend
    Join Date
    Sep 2014
    Location
    Scotland
    Posts
    2,361

    Default

    Again this still a great system that is better than others. But the fact you buy a system that is sold like its has this potential is not good. 8 of TRVs are new ish not Honeywell(different manufactures, drayton/bulldog etc) 1 is a Valencia. The 30% close/open thing is not good, if dodgy on certain valves.
    To be fair to Honeywell, after first encountering these problems and studying them and how the system works I've thought a lot about the issues of trying to proportionally control the water flow of a radiator and juggle that with sending a heat demand to the boiler, and concluded it's a very difficult problem to solve, and on two fronts.

    One is the interaction between zones. On the one hand you obviously don't want the boiler at maximum flow temperature all the time with a valve barely open to put out a little bit of heat from a single bedroom radiator on a cool night, as that would be hugely inefficient with 99% of the flow going back through the ABV and a lot of heat loss from the boiler casing and nearby piping etc.

    So under a low demand situation you want a large(ish) water flow through the radiator (and little or none through the ABV) but a low flow temperature so that the total heat into the room is low. In this circumstance you modulate the heat into the room by varying the flow temperature (directly by OpenTherm or indirectly via TPI) and not so much by water flow.

    The Evohome system naturally does this because it takes a fairly high valve position to generate a significant heat demand - even a heat demand of only 20% requires a valve position of 55% - well beyond the 30% where the valve should start to flow water.

    However when any zone needs a high demand the flow temperature needs to rise to meet the demand of the most demanding zone, and at that point zones with low demand must then close their valves down significantly to prevent an overshoot and modulate their heat output by controlling the water flow only.

    The HR92's are unfortunately not aware that some other zone has requested a higher flow temperature, so their first clue that something has happened is when the room temperature starts to overshoot and they will start to close a bit. This can take up to 4 minutes and in that time a lot of latent heat can be stored in a large panel radiator that will be radiated even after the valve closes. If the valve calibration is good the overshoot will typically be only about 0.5C though.

    So one theoretical improvement that could be made to a multi-zone system is for the controller to also send the aggregate heat demand that is going to the boiler back to all the individual radiator controllers.

    That way an HR92 could be "forewarned" that the flow temperature is about to shoot up and roughly by how much, and by using a lookup table it could probably preemptively throttle back its valve position in anticipation to prevent an overshoot even happening at all. And by comparing it's own heat demand to the aggregate one it could tell whether it was currently the "controlling" zone or whether some other zone was controlling the heat demand (by having a higher demand) and make decisions based on that.

    Unfortunately I strongly doubt there would be any way to retrofit that into the Rameses protocol nor the current HR92's, (which don't even have over the air firmware update ability) so we would be talking about an all new system. However there is a second problem with this preemptive throttling - for it work well the HR92 would need to know precisely at what position the valve starts flowing, otherwise its preemptive action could just as easily lead to the room dipping below the setpoint instead of overshooting...

    And that comes to the second problem - calibration of the valve. How does it figure out at exactly what point any given valve body starts flowing water ? The short answer is it can't... it doesn't have enough information to deduce that. All it has available is a force vs distance calibration which it uses to try to find the fully closed position of the valve, and a temperature sensor.

    Could it slowly increase the opening and wait for the temperature to rise, then open it a bit more and wait again and then guess that is the opening point when it sees a rise ? Maybe, with some clever heuristics... but it could equally get very confused by changes in heat demand from other zones or someone closing a door etc...

    Could it measure the temperature of the valve body itself from the water flowing through it ? Maybe - that solves the problem of unrelated air temperature changes but you still have the zone interaction problem, and now you get a very different result depending on whether the HR92 is on the flow or return side as the flow pipe will heat quickly but the return pipe can be delayed many minutes on a large radiator.

    Apart from the preemptive messaging of system wide heat demand back to TRV's I think Honeywell is already doing as much as they could given that they are trying to be compatible with existing TRV valve bodies.

    To go the next step up you would need to incorporate a calibrated water valve as part of the TRV itself instead of them being separate. In that you could potentially put both a flow rate sensor and a temperature sensor and that would give the TRV controller a LOT more information to go on. It would know exactly at what point the valve opened and what temperature of water was flowing through. (Although it would probably need a setting to tell it whether it was on flow or return side so it could adjust its algorithms)

    The question is, is all this extra over engineering of a fully integrated TRV/valve body worth it for slightly better control ? I'd say probably not, and the market wouldn't go for it if it meant a drain down to replace the valve bodies on every installation...

  6. #16
    Automated Home Sr Member
    Join Date
    Apr 2019
    Posts
    63

    Default

    Understand your point. Its all good and interesting and gives me something tinker with. Work with proportional control at work, big industrial shit. For me it just needs a small change, a variable in the menu. But if the HR92s are not update able. Then I think we are stuck with it as there maybe is only so much the touch screen can do.

    Got four valves I want to change out and also want add a towel radiator hr92. £16 for standard valve bodies is great but £80 towel radiator one is steep

    I'll let you know. Will be in 2 months now, as I am busy.

    Don't suppose you got a write up for how you log stuff with your graphs.

    Got a raspberry running a music so I can talk to Alexa and play my mp3 music for free. instead of paying for amazon music, was wondering how difficult the setup you have is. Bear in mind I am a copy only techy type guy when it comes to this Raspberry and python stuff.

    Thanks

    Frank

  7. #17
    Automated Home Legend
    Join Date
    Sep 2014
    Location
    Scotland
    Posts
    2,361

    Default

    Quote Originally Posted by frankmalia View Post
    Understand your point. Its all good and interesting and gives me something tinker with. Work with proportional control at work, big industrial shit. For me it just needs a small change, a variable in the menu. But if the HR92s are not update able. Then I think we are stuck with it as there maybe is only so much the touch screen can do.
    I agree that an adjustment in the menu to cope better with the different characteristics of different valves would be helpful, although technically there already is one - the Stroke setting. Unfortunately it only has two settings, 0 or 1, so won't account for all valve types, and there will be some that are incompatible no matter what due to the pin length being out of range, being too hard to push down etc...
    Don't suppose you got a write up for how you log stuff with your graphs.

    Got a raspberry running a music so I can talk to Alexa and play my mp3 music for free. instead of paying for amazon music, was wondering how difficult the setup you have is. Bear in mind I am a copy only techy type guy when it comes to this Raspberry and python stuff.
    Don't really have anything no, however it's not too hard if you've done anything with Linux.

    Munin can be installed from Raspbian using apt-get. You also need to install watchforstock's evohomeclient libraries (python bindings for the Evohome API) which can by done using python pip. There are some basic instructions on how to set it up on the Evohome-munin github repo I linked to earlier.

  8. #18
    Automated Home Sr Member
    Join Date
    Apr 2019
    Posts
    63

    Default

    Had a play about with balancing radiators. Got all of them with about a 10'C drop across valves with boiler running 61'C.

    Then left system couple of days. Found a couple of rooms could not keep temperature setpoint, so cracked lock shield a bit more open 1/8 turn every day until the temperature raised to S.P. and behaved better. Overshoots have dramatically reduced and control is better.

    As always there is still overshoots when there is large heat demand you still get a overshoots, but there not as bad.

    Let you know when I change some valves

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •