Honeywell Valencia problems

[DBMandrake]

Let me start by saying that I don't have any of the Honeywell products you have in my own system.

I've got Tower radiator valve bodies, which are fitted with Peggler I-Temp controllers.

Once I had fitted the I-Temps, I found that most of them faulted out at the initialisation/calibration phase of their installation and so everything was stopped in its tracks even though the two components were listed as 'compatible', according to the fact sheet.

After playing around with loosening the I-Temp from the body, I found I could force the calibration to complete, but you obviously can't leave things like that, so after a bit of trial and error measuring I decided to make some packing washers in the range of ~ 1.0mm to place between valve body and I-Temp.

Having a lathe, I was able to make mine from brass bar and part them off to correct thickness.
After fitting, all worked perfectly. Maybe you could do similar with fibre or plastic washers if you can't find a better fix?

Washer size: ID 23mm OD 28mm thickness to suit.

Thanks for the suggestion - the testing I did coincidentally suggested that 1mm was about the right amount. You need to make sure it can still firmly close the valve, with anything more than about 1.5mm it wasn't able to.

The problem I can see is that the ID needs to be really large as there is only a very narrow shoulder area where the washer could rest and still clear the large brass hex. So being able to find off the shelf washers that are a perfect fit seems unlikely. (And I'd prefer to use metal washers so there isn't any compression over time)

I'll keep it in mind though, I like the idea of a shim as I don't want to make any drastic non reversible changes such as filing the pin.

[blowlamp]

Thanks for the suggestion - the testing I did coincidentally suggested that 1mm was about the right amount. You need to make sure it can still firmly close the valve, with anything more than about 1.5mm it wasn't able to.

The problem I can see is that the ID needs to be really large as there is only a very narrow shoulder area where the washer could rest and still clear the large brass hex. So being able to find off the shelf washers that are a perfect fit seems unlikely. (And I'd prefer to use metal washers so there isn't any compression over time)

I'll keep it in mind though, I like the idea of a shim as I don't want to make any drastic non reversible changes such as filing the pin.

Maybe you could cut up some old credit cards to test the idea? I tried thick card with pretty good results and apart from the initial compression, they bedded down nice and firm with no further loosening - now replaced with brass washers for completeness though.


Martin.

[rotor]

Ooh the washers idea might be really good... I await your results DBMandrake.

[blowlamp]

Lockshield (balancing) valves are usually only just 'cracked open' by 1/4 - 1/2 a turn or so to allow sufficient water flow, which must equate to a lifting of less than 1mm of the valve off its seat - but more probably something like 0.5mm. It therefore figures that a Valencia valve should be adjusted similarly if it's to perform the same function as a lockshield valve.

This means the HR92 would only have a very small range of movement to fulfill its task of regulating the flow of water through the radiator, so rendering the majority of the upper part of its stroke redundant.


Martin

[DBMandrake]

Lockshield (balancing) valves are usually only just 'cracked open' by 1/4 - 1/2 a turn or so to allow sufficient water flow, which must equate to a lifting of less than 1mm of the valve off its seat - but more probably something like 0.5mm. It therefore figures that a Valencia valve should be adjusted similarly if it's to perform the same function as a lockshield valve.

Depends on the lockshield valve design I guess, but I typically find that those that need closing down a bit are set to around 1 turn out of 3 turns. If you're having to set them to 1/4 of a turn chances are the pump speed is too high! Compensating for a pump speed that is too high with most of the lockshields being closed right down will only lead to a large flow through the ABV...

This means the HR92 would only have a very small range of movement to fulfill its task of regulating the flow of water through the radiator, so rendering the majority of the upper part of its stroke redundant.

Which is precisely why I chose not to use the balancing insert but to balance using the lockshield valve I'm perfectly happy to count the turns on the lockshield if I'm removing a radiator and put it back where it was.

Like you I was concerned about the effective pin operating range of the HR92 being significantly truncated, causing a lot of deadband at the open end of its range. Not necessarily a huge problem with a mechanical TRV, but with an HR92 you have the added complication of the heat demand signal being sent to the controller which is based on a fixed relationship with the pin position. If the integrated balancer is screwed down too far then it could easily be that it already hits the balancer limit stop before it starts calling for heat!

On the contrary, the useful operating range of the HR92 pin movement should be maximised to give the best temperature control, which means winding the internal balancer mostly out (so that it's only acting as a bush to steady the pin, it can't be removed completely as it's needed as a bush) and balancing via lockshield, which is what I've done.

Regarding the discussion of shims earlier, I did some testing with an unused valve body and an unused HR92 to see how it calibrates in both stroke 0 and stroke 1 mode with the base unscrewed by different amounts to simulate different shim thicknesses.

And unfortunately the conclusion I reached is that they didn't really help with the calibration of the opening point relative to heat demand. The only benefit from using a shim on the order of 1mm seemed to be that in stroke 1 mode you could get the valve to open about 1mm further with a shim, however the calibration of at what reported pin position the valve started to flow didn't significantly change as the HR92 just adapted to the offset during the calibration process.

The only way to change this I found is to choose stroke 1 mode, which affects the operating range of the motor and therefore the scaling of the reported pin position.

In stroke 0 mode the valve doesn't start to allow water flow until about 55-75% indicated, but as it starts calling for heat from the boiler >30% this means if you use stroke 0 mode, as a room's temperature slowly drops below the set point it will start firing up the boiler long before the radiator gets any water flow, so you're just heating up your ABV loop if no other zones need heat...

On the other hand stroke 1 mode seems to calibrate such that water starts to flow at around 25% indicated - just before it starts calling for heat, so that by the time it starts calling for heat from the boiler the radiator is already able to flow slightly and start to warm up.

If another zone is calling for significantly more heat, 25% to close is not so low that it can't make a small correction in the closing direction to stop the flow, thus avoiding overshoots in a zone with a small heat demand if another zone comes online with a high heat demand.

This was a big problem for us before in the living room with our old sticky valve as in early evening the living room is the only zone at a high temperature (most others set back significantly) then the bedroom radiator for our son would come on full blast to warm his room changing the boiler duty cycle to 100% for a while, which would then cause a big overshoot in the living room of 1-1.5 degrees as the HR92 had to move a long way from the previous equilibrium to close the valve.

No sign of this sort of this early evening overshoot now and temperature control of the living room is excellent without any oscillations:



As well as the calibration of the opening point, stroke 0 also suffers from not opening the valve far enough, IMO. With the balancer screwed out (as shown in my photo) the valve has something like 4mm of travel off the seat - with the HR92 in stroke 0 mode it only lifts the valve about 1mm off its seat when at 100%, or about 1/4 of the potential valve opening....

In stroke 1 mode it's lifting it a bit more than 2mm off the seat, or about half the possible travel - still not as much as I'd like but a lot better. This is partly because there is a lot of compression in the rubber seal in the plunger that causes lost travel of the plunger lifting from the seat, as the first part of the travel is the rubber decompressing. (The bulldog valve seemed to have a harder rubber seal that didn't decompress as much)

I have a couple of radiators that need maximum possible flow in winter so I have no choice but to use stroke 1 mode to get as much opening as I can. It remains to be seen whether this 2mm lift above the seat is enough. In it's favour, the Valencia has a much bigger diameter hole in the side port than the bulldog valve, (I'm baffled as to why the bulldog has a very restrictive hole) also it has a much larger, tapered valve seat compared to the small diameter flat seat on the bulldog valve, so for a given valve seat gap it probably has a lot less flow resistance compared to the bulldog valve. (It's certainly a lot quieter)

So I'll just have to see how it performs in the depths of winter. If it needs more opening a 1mm shim together with stroke 1 mode should help.

So after all this testing my conclusions about the HR92 / Valencia combination are:

1) Shims don't help with the heat demand/valve flow calibration issue at all, but a shim of about 1mm could increase maximum valve opening somewhat used in combination with stroke 1 mode if you really need every last bit of flow for your radiator. A shim of significantly more than 1mm will prevent the valve closing properly so this is about the maximum thickness shim you could use.

2) Don't use the internal balancer - wind it most of the way out so that it is not restricting pin travel at all. It should be wound in just enough for the threads to catch so it can act as a support bush. Let the HR92 base itself be the pin limit stop.

3) Stroke 0 mode is almost unusable IMO. It calibrates the heat demand/valve opening relationship completely wrong, (possibly due to the soft rubber washer) resulting in calls for heat when the radiator can't flow, (wasted gas) severely restricting the maximum possible valve opening and therefore flow, (insufficient heat to the radiator) and compressing the entire effective operating range into the top end of the pin travel. (Poor temperature regulation)

On an Opentherm system it would also result in high flow temperatures even when the true need for heat is quite small, as the radiator wont even flow at all until the call for flow temperature is very high. This could explain some people seeing high opentherm temperatures all the time and the overshoots that may result from that. Even on a BDR91 you're going to see high duty cycles when it is not necessary.

Stroke 1 mode seems to work quite nicely with the Valencia - it opens the valve as much as possible when needed, (short of adding a shim) calibrates the opening point vs heat demand quite well, and so this is what I would recommend. I'm using stroke 1 on all radiators, and overall I'm happy with how they're performing in this mode after very disappointing first impressions in stroke 0 mode. I guess I'll just have to suck up the increased battery drain in the interests of actually having the system working properly...

If you have an HR90 then my advice would be don't use a Valencia, as there is no way to adjust the stroke setting and it's default is similar to stroke 0 on an HR92...

[blowlamp]

It seems logical to me to have valve bodies that have hard rubber washers that give low hysteresis to the TRV (HR92 in this case) to work with. It's counter productive to have a large, undetermined % of the movement wasted in decompressing a soft washer.

The other point is that a small radiator will need a tiny adjustment of the pin to maintain a steady room temperature, in comparision to a large radiator. I think it would help if Evohome asked for radiator size information at installation time, so a suitable scaling value could be applied to valve movement. Though maybe this is taken care of during the learning period.

I believe the purpose of the lockshield is to allow just enough hot water to flow through the radiator that lets it heat the room under the most severe weather conditions. Flow needs to be restricted within individual radiators to ensure hot water is directed to all other radiators that may be calling for heat, without leaving any starved for water.



Martin.

[DBMandrake]

It seems logical to me to have valve bodies that have hard rubber washers that give low hysteresis to the TRV (HR92 in this case) to work with. It's counter productive to have a large, undetermined % of the movement wasted in decompressing a soft washer.

I tend to agree. With the Valencia in Stroke 1 mode 25% of the available calibrated pin travel is used compressing the washer and not varying the water flow, I consider that acceptable as there needs to be some compression to ensure that when it fully closes it really does close water tight despite any dimensional variations, and also calls for heat start at about 30% pin movement.

However in Stoke 0 mode well over 60% of the calibrated pin travel is spent before the washer even starts to lift off the seat and allow water flow. This seems totally wrong to me, and this was not the case on the bulldog valves which where approximately 30% travel in Stroke 0 mode before starting to flow and maybe 15% in Stroke 1 mode. (In this case Stroke 0 was more suitable)

The HR92's calibration process determines the closed position empirically - when you first fit the head to the valve and lock it, the cycle mode that runs basically winds the valve pin down until a set amount of resistance is "felt" by the motor, no doubt determined by measuring the current drawn by the motor.

This is then considered to be the "closed" or 0% reference for pin movement. Obviously as you compress a washer it starts off easy then more and more force is required to compress it further. Because it's compressible there isn't a sudden increase in required force to detect "closed" so there has to be some arbitrary threshold of force chosen that is considered closed. This threshold differs between Stroke 0 and 1, with 1 applying more force during calibration.

The problem is, the more compressible the washer is the more pin travel there will be between the calibrated 0% position based on force and the washer actually beginning to lift off the seat. The softer the washer the more lost travel. So built into the calibration routine of the HR92 is an assumption of washer compressibility to help it estimate at what point the water will start to flow. The reality is an HR92 has no idea at what point water is really starting to flow, so this is one of the challenges it has to work with.

As an HR92 starts calling for heat >30% pin travel its pretty clear that the built in design assumption is that 30% of the calibrated pin travel is lost to washer compression and is where it's expecting the valve to start allowing water to flow, so any large deviation from this will have side effects on the ability to control the radiator temperature and call for heat at the correct time.

For whatever reason, Honeywell's own flagship valve body doesn't seem to line up with the design expectation of the HR92's calibration routine - at least in Stroke 0 mode.

Stroke 1 mode changes several things...

1) Increased torque when closing - this affects the calibration of the 0% closed position - it will be pushed down a bit harder and further when closed than Stroke 0.
2) Increased pin travel between 0% and 100% - in Stroke 1 mode it will turn the full 2 1/4 revolutions that the black wheel is typically capable of, in stroke 0 mode it will only turn about 1 1/4 turns from the fully closed position, remembering that the fully closed position will also be further up due to the different calibration. So essentially Stroke 1 will move further in both directions compared to Stroke 0.
3) The motor turns faster.

The reason why the opening threshold calibration changes is simply a matter of scaling - 0 to 100% pin position in Stroke 1 is more revolutions of the wheel and more pin travel, so the fixed amount of travel consumed by the washer compression becomes a smaller overall percentage of the total available movement, which brings the opening threshold down in percentage, while the percentage opening to heat demand curve remains the same.

For best compatibility with a variety of valves more than the two Stroke settings we have now would have been ideal as there is a big jump between the two settings, but I guess they didn't want to complicate things too much.

The other point is that a small radiator will need a tiny adjustment of the pin to maintain a steady room temperature, in comparision to a large radiator. I think it would help if Evohome asked for radiator size information at installation time, so a suitable scaling value could be applied to valve movement. Though maybe this is taken care of during the learning period.

Actually I think it's the reverse - a large radiator, for a given room, needs a smaller more precise adjustment to keep a steady temperature, because the amount of heat it can put into the room, and quickly, from opening the valve too much by mistake is a lot more than a smaller radiator. A larger radiator has a larger thermal mass/storage so the penalty for getting it wrong is much greater because even when you realise the mistake and close the valve the thermal mass will put a lot of heat into the room after the water flow stops, causing an overshoot. A small radiator will give a smaller overshoot as a penalty for a mistake.

If we assume that an HR92 is a self tuning PID controller (which I believe it is, by graphing and observing its behaviour closely over the last 2+ years) then I would hope that it can self tune the proportional coefficient. It definitely tunes the differential coefficient heuristically as when you make a large change to the response time of a room (such as covering a radiator with a towel, or removing it) you can see the step response overshoot or undershoot the first time after that but then very quickly - after only one or two cycles of set point changes it will adapt and start hitting it's targets again. So it definitely learns and adapts to factors like room response time and tendency to over or undershoot.

I think the biggest problem with making small accurate adjustments is "stiction" of the valve body itself, and that's one of the main reasons I replaced my old valve bodies. The pins in them had corroded internally, and there were clear signs of stiction in their operation.

After comparing with the Valencia the old valves had a larger diameter pin with a metal to metal sliding surface (large static friction) vs a small diameter pin with a plastic sleeve on the Valencia (low static friction) and another unexpected difference is that the return spring on the Valencia is a LOT stronger than the bulldogs ever were. So strong in fact that when you remove the HR92 from the base the black wheel actually spins itself instantly to the fully up position - something that never happened on the old valves.

On the old valves the combination of increased friction and a weak return spring would introduce some unwanted hysteresis into the control of the pin position by the HR92. It would have to turn the motor a significant amount before the pin would actually move, especially in the unwinding direction, as the spring was not strong enough to immediately overcome the friction when the HR92's wheel moved away from the pin slightly.

This added hysteresis would cause oscillations in the room temperature as it was unable to make a sufficiently small adjustment to the valve position, so it would keep overshooting one way then the other. This was my theory before changing the valve bodies and I'm pleased to report that it seems to have been experimentally confirmed.

The combination of lower movement friction and much stronger return spring probably eliminates the hysteresis that was there in pin movement before - now the tiniest movement of the motor in the HR92 translates directly to a tiny movement of the pin.

The result is that my rooms that originally did not overshoot 2 years ago but started overshooting as the bulldog valves deteriorated, are once again not overshoot at all, and maintaining a very steady temperature, and those rooms that due to HR92 sensor placement can't avoid a little bit of overshoot (like the kitchen radiator under a counter top) are overshooting less than half as much as they were before. So clearly there is finer control available to the HR92 with the new valve bodies...

[DBMandrake]

I believe the purpose of the lockshield is to allow just enough hot water to flow through the radiator that lets it heat the room under the most severe weather conditions. Flow needs to be restricted within individual radiators to ensure hot water is directed to all other radiators that may be calling for heat, without leaving any starved for water.

I have a different view on the purpose of balancing with lockshield valves than many. Many say that it's all about getting the 11C (or 20C these days) drop across radiators for efficiency. But as I don't have a condensing boiler that doesn't apply to me.

In my opinion the purpose of balancing, condensing boiler or not is to equalise the flow resistance of every radiator branch so that no radiators can steal flow from others if they are all on at once. Depending on the length and diameter of runs to each radiator and the radiators themselves, they will all have a different flow resistance, when multiple radiators are open that means that those with low flow resistance will steal a lot more flow than those with high flow resistance.

The difference can be enough that with all radiators on a couple of radiators close to the boiler can get hot quickly when far away radiators may hardly get hot at all as they are starved of flow due to the water preferring to follow the easiest paths.

So with this as a goal you would start with the radiator that already has the most flow resistance (long or skinny pipes etc) and leave that lockshield fully open, that becomes your baseline. Then you would adjust the ones with lower flow resistance to increase their flow resistance to match.

In theory once balanced this way, on a cold start all radiators should heat up at about the same rate with no radiator starved or significantly straggling - the water is equally happy to flow into all the different branches at about the same rate, and no matter what combination of radiators are open, they all get their fair share.

This will introduce some temperature drop across the radiators of course but the key thing here is that if the drop is not sufficient (not at least 11C or whatever you choose) then turn down the pump speed/pressure, do not close the lockshields any further. There should be at least one radiator in the house whose lockshield valve is fully open otherwise you are just making the pump work harder for no reason.

Another issue with adjusting them purely by measuring temperature drop is that temperature drop depends on heat loss, but for a given water flow rate a longer/large radiator or one with better fins is clearly going to lose more heat to the room, so going by temperature drop alone you would tend to open the lockshield more, which may starve a small, further away radiator.

Just my opinion though...

[paulockenden]

The thing is, despite all this talk about pin travel, valve positions etc., for many of us the HR92+Valencia combination works well.

It's important not to ignore that.

P.

[DBMandrake]

The thing is, despite all this talk about pin travel, valve positions etc., for many of us the HR92+Valencia combination works well.

It's important not to ignore that.

Well there's working, and then there's working optimally. I didn't say it doesn't work at all in Stroke 0 with a Valencia, just that it's a long way from optimal. And I had the same issues on all 7 operational radiators, so I doubt it's a faulty valve or HR92.

Depending on how closely you monitor your system and what your zone schedules are like you may not necessarily notice obvious problems, especially if you were a non-technical house owner.

For example calling for heat long before the radiator opens probably wouldn't be noticed if there are always multiple zones configured to be on at once. I suspect I'm in the minority having a schedule that has a single zone on for significant periods of the day.

And unless I got up to go look at the boiler relay light or check in the boiler cupboard I may not realise that the boiler was churning away during a period when only that one zone should be on and the radiator was cold to the touch... I do actually have the boiler relay on the kitchen wall near the fridge, so I am generally somewhat aware of whether the boiler is running or not, so if I see it on when I don't think it should be I tend to investigate.

I imagine that many people have their BDR91's hidden away in a cupboard with their boiler so would not be aware it was on without specifically looking in the cupboard, and Honeywell steadfastly refuse to add any sort of heat demand information to the home screen of the Evotouch that might alert people to unexpected boiler firing. I wonder why...

Then you only have to look at the symptoms that can result from the issue I've discovered to see if other people have independently reported the same issues prior to me:

1) Zones calling for heat from the boiler while the radiators remain cold - check. I replied to a thread about this not that long ago, suggested Stroke 1 mode to the poster and surprise surprise, it fixed the problem for them. I don't recall if they had Valencia's or some other valves but I'll see if I can find the thread to check. As this calibration issue probably depends mainly on pin length and compressibility of the rubber washer it's feasible there are valves other than the Valencia that could suffer from the same issue in Stroke 0 mode.

2) Unusually high OpenTherm boiler flow temperatures requested in situations where radiators are not very hot and not much heat is needed - check. I remember reading multiple threads about this in the last few months. I'll see if I can dig them up to investigate further.

3) Excessive interaction between different zones, in particular when a 2nd zone comes online at 100% when the first zone was maintaining a constant temperature with a low heat output - check. I was having this problem myself and have certainly seen it reported many times.

So consider this thread to be a public service announcement of what I've discovered, and that potentially some systems are not running nearly as well as they could be, particularly in regards to unnecessarily high flow temperatures and calls for heat to the boiler when a radiator is not flowing. It's applicable specifically to the Valencia, but its quite possible there are other 3rd party valves that are similarly affected.

Those with a more technical interest in their system may want to investigate further to see if it is working optimally or not.