The specification IS publically available to everyone interested. Please have a look at: http://www.knx.org/knx-standard/how-to-order/ Caused by many of the open source initiatives, the term open is in the last years mistaken for for free. But it is not the same.

Alternatively, you can also buy the EN and ISO standards: EN50090-1, EN50090-2-1, EN50090-2-2, EN50090-3-2, EN50090-4-1, EN50090-4-2, EN50090-5-1, EN50090-5-2, EN50090-5-3, EN50090-7-1, EN50090-4-3, EN50090-3-3 or EN1321-1 and EN13321-2 or ISO/IEC 14543-2-1, ISO/IEC 14543-3-3, ISO/IEC 14543-3-1, ISO/IEC 14543-3-2,ISO/IEC 14543-3-5,ISO/IEC 14543-3-6, ISO/IEC 14543-3-7, ISO/IEC 14543-3-4 and ISO 16484. Can it be any more open? And yes, open is not free.

But using the KNX technology is without any license conditions! Solely the use of the KNX quality and conformance logo is to be paid for.

Moreover, there are numerous books about KNX.

° The spec is NOT licensed (only the use of the logo is); it IS publically shared (but not for free) and from all standards (specifications that can claim being compliant with an ISO/IEC or EN standard)... it's one of the cheapest. No, these prices are targeted at commercial companies, selling tens of thousands of these products, and are not envisaging the individual developer or the free initiative, that is correct.

° But to be honest, the quality isn't all that great Can you give an example? The KNX specifications meet the structure and contents requirements of European and worldwide standardisation, so I'm wondering... PLEASE give any reference.

For clarity: I do work for KNX and I am happy supporting the private initiative to my best possibilities. (I'd get in troubles if I'd provide some papers here that need purchasing otherwise, so, I hope for your understanding.) But for me to be constructive to your project (as far as KNX): - I do have to react on incomplete and incorrect things; - I have to ask for foudning any statement. Further on, I have seen concerning other technologies things in this forum that are not correct or not complete, just expressed in a clever way, to which I - respecting any other technology, do not react.

You need to chill a little. ALL standards suck because they represent a compromise between experts. I haven't even read the KNX stuff but I'm sure I will say WTF at least once per page. So if you cut the rhetoric down a little, reading the really useful bits in your postings would be more enjoyable.

Hello Steven,

Looks like I hit a raw nerve with you :-) Sorry about that.

Steven De Bruyne wrote on Oct 14, 2008 13:38:
Can it be any more open?

Well, yes it can. KNX association could put it all up on a public website for everybody to read. Then it would be more open. Or KNX association could allow redistribution of the specifications. Then it would be more open. Open as in 'Open Source' meaning of the word (free redistribution and right to modify).

And mind you, Open Source doesn't equal 'free' (as in free beer) either. It means right to redistribute and modify.

Steven De Bruyne wrote on Oct 14, 2008 13:38:
The spec is NOT licensed (only the use of the logo is)

I'm finding it difficult to follow the logic of your argument here, unless you are arguing that the KNX handbook is a separate thing from the specification itself?

There is a license with the KNX Handbook that explicitly denies sharing of its contents unless under NDA. That's pretty restrictive.

You seem to try to dance around this fact by proclaiming that everything is open and publicly shareable when clearly it is not. Yes there is a price but even you pay that price you still can't freely share the information.

Steven De Bruyne wrote on Oct 14, 2008 13:38:
But to be honest, the quality isn't all that great Can you give an example?

Sure. I come from a software developer background with years of experience in reading specs and implementing some of them. Maybe I'm not the right audience (I don't look at it from the HW perspective) but I think you could pick pretty much any random Java EE specification and see how it is more readable and approachable to implementation. Or perhaps the Internet RFCs? Have you looked at the HTTP specification lately?

Oh and by the way, while some of those specs do come with a license, you are still free to download them online and read them. No money required ;-)

While Java EE is not the perfect example (there are faults there as well), I'd be excited to see the same level of accessibility in KNX specifications.

For instance, I'm struggling to find any clear description of the wire protocol in the KNX handbook, something that has been rather trivial to extract from other control protocols I've looked at. But the amount of text in KNX handbook is huge so maybe I'm just looking in the wrong place? I'd appreciate some pointers there.

Steven,

Thanks for pointing me into the right direction. This should help in going forward.

Jesus Arias wrote on Oct 14, 2008 17:38:
You need to become a member to get access to the whole specifications.

That is not correct. Everybody can get a copy of the specifications. This is not limited to members only. Compared to other technologies, or buying copies of EN or IEC standards, that's rather cheap.

If you want to become a member you have to pay (I don't know how much is it, but I think it's not cheap)

You have to become a member if you want your products to bear the KNX logo. The prices for this are listed here: http://www.knx.org/knx-members/joining-fees/

Result: very expensive products.

Please note as well that together wit compliance with KNX, KNX manufacturers have chosen a high product quality, concerning stack, applications, mechanics, electronics, etc. This, and the fact that many KNX devices are design devices, makes them indeed expensive.

Steven De Bruyne said: Please note as well that together wit compliance with KNX, KNX manufacturers have chosen a high product quality, concerning stack, applications, mechanics, electronics, etc. This, and the fact that many KNX devices are design devices, makes them indeed expensive.

Tell that to the client, and I talking with the experience in the 2nd largest KNX market, that's Spain.

KNX it's very good for large installations (and if you use it with other systems), but for houses or in the construction industry, the developer will not spend this huge amount of money.

I recognise that now there are, more and more, more interesting devices, but we need lower prices.

Dries Verbrugge said: Already one extra question, what can you do with ETS? For what i understand you can use it to configure all certified devices, and by configuring i think about assigning address en link the different components together. Is that correct? can you do more with that piece of software? I am just wondering if i will need it in the first stage. For what i understand, if i hard code the addresses in the devices i will not need that program?

Exactly, you can use to configure all certified devices. You can download each ETS file device in the web of the manufacturer. And you can configuring each device (of course, the configuring depend on the device features).

You can also create group addresses, where you use the object functions of the devices that you inserted before in your project.

But ETS allows more actions, like a what is happening in the bus right now, basically you can read what is happening in the bus, and it allows you find failures, and, the most important:

You can simulate whatever you want, I mean that you can write in every group address and then you can know what's happening.

Cheers!

I should also add, that I think there is a future for that. Imagine a embedded controller with ethernet/wifi, ipv6 and enough mem to run android and expose its data and programmatic interfaces that way. That would be killer. Using KNX as a first implementation gives you the advantage to ride an existing market of course.

Interestingly enough it is not the cost of hardware that is limiting the field. Nor the cost of the spec, although I hear it is expensive for hardware manufacturers. Steven do you have a percentage part of cost is represented by the KNX certification? on average? I suspect that cost is small overall. The cost is actually embedded in the installer community. To get this stuff deployed is expensive. The HW cost $10 to make, the final cost is $250.

It is possible to take the cost to $25, start pulling some magic with J2ME or Android, and still find a price point that is at half. Efficiency will be in tools and enable by a new generation of hardware actuators. The cost of the hardware is getting there. Remember that KNX/EIB is 20 years old.

Marc Fleury wrote on Oct 15, 2008 12:57:
I should also add, that I think there is a future for that. Imagine a embedded controller with ethernet/wifi, ipv6 and enough mem to run android and expose its data and programmatic interfaces that way. That would be killer. Using KNX as a first implementation gives you the advantage to ride an existing market of course.

That would certainly be a killer. However, note that the classic Ethernet-networks build a topology in a star around a hub or switch. That requires that every domotics device would have to be wired individually. That increases the amount of wiring. That is the reason why for instance KNX (but also other systems with a twisted pair) with its free physical topology, and powerline media (like X10, but also KNX PL) are so popular: the wiring is more simple.

I hear you thinking WiFi, but please note that this technology has not been designed for embedded controllers. WiFi radio cannot be used in combination with battery powered devices or energy harvesting technologies (like Z-Wave and certain KNX RF products). Note as well that the massive use of WiFi now already causes in areas with many WiFi-LANs so-called black holes. I assume that each technology will find its place in the future.

Interestingly enough it is not the cost of hardware that is limiting the field. Nor the cost of the spec, although I hear it is expensive for hardware manufacturers. Steven do you have a percentage part of cost is represented by the KNX certification? on average? I suspect that cost is small overall.

° About the cost of the specifications, we have talked before. Moreover, companies that are member or licensee (do not confuse with KNX Partner) get a copy for free.

° Registration costs can be found here: http://www.knx.org/knx-certification/of-products/ NOTE That's one-time a registration cost, not a license fee, that would increase by the number of nodes sold. So, for larger amounts of products, this gets negligable.

The cost is actually embedded in the installer community. To get this stuff deployed is expensive. The HW cost $10 to make, the final cost is $250. It is possible to take the cost to $25, start pulling some magic with J2ME or Android, and still find a price point that is at half. Efficiency will be in tools and enable by a new generation of hardware actuators. The cost of the hardware is getting there. Remember that KNX/EIB is 20 years old.

I do not think that it is possible to give a proper answer to these statements; price building is a crucial element in the development and sales of a product, obviously. I leave it to people with a diploma and position in that field to decide. I can only say that KNX manufacturers do not make a profit of 2400% (profit of $ 240 what costs $10).

Steven De Bruyne wrote on Oct 15, 2008 14:16:

Marc Fleury wrote on Oct 15, 2008 12:57:
I should also add, that I think there is a future for that. Imagine a embedded controller with ethernet/wifi, ipv6 and enough mem to run android and expose its data and programmatic interfaces that way. That would be killer. Using KNX as a first implementation gives you the advantage to ride an existing market of course.

That would certainly be a killer. However, note that the classic Ethernet-networks build a topology in a star around a hub or switch. That requires that every domotics device would have to be wired individually. That increases the amount of wiring. That is the reason why for instance KNX (but also other systems with a twisted pair) with its free physical topology, and powerline media (like X10, but also KNX PL) are so popular: the wiring is more simple.

Please, we are not knocking TP. Also I just visited a KNX house where the installer made a star topology, why? because it is simpler to touch the buttons on actuators all in one place. The result was a VERY EXPENSIVE cable layout that was un-necessary, I took pictures, it wasn't pretty. Dinking Ethernet technology as a packet delivery is technically naive.

I hear you thinking WiFi, but please note that this technology has not been designed for embedded controllers. WiFi radio cannot be used in combination with battery powered devices or energy harvesting technologies (like Z-Wave and certain KNX RF products). Note as well that the massive use of WiFi now already causes in areas with many WiFi-LANs so-called black holes. I assume that each technology will find its place in the future.

Actually we are thinking 802,15. This is very clearly the future for mid-level installations.

Interestingly enough it is not the cost of hardware that is limiting the field. Nor the cost of the spec, although I hear it is expensive for hardware manufacturers. Steven do you have a percentage part of cost is represented by the KNX certification? on average? I suspect that cost is small overall.

° About the cost of the specifications, we have talked before. Moreover, companies that are member or licensee (do not confuse with KNX Partner) get a copy for free. ° Registration costs can be found here: http://www.knx.org/knx-certification/of-products/ NOTE That's one-time a registration cost, not a license fee, that would increase by the number of nodes sold. So, for larger amounts of products, this gets negligable.

Actual numbers for nodes is where it is interesting. It cost OR $1000 to get the spec. But we don't manufacture hardware. The cost on hardware manufacturers is what is interesting.

The cost is actually embedded in the installer community. To get this stuff deployed is expensive. The HW cost $10 to make, the final cost is $250. It is possible to take the cost to $25, start pulling some magic with J2ME or Android, and still find a price point that is at half. Efficiency will be in tools and enable by a new generation of hardware actuators. The cost of the hardware is getting there. Remember that KNX/EIB is 20 years old.

I do not think that it is possible to give a proper answer to these statements; price building is a crucial element in the development and sales of a product, obviously. I leave it to people with a diploma and position in that field to decide. I can only say that KNX manufacturers do not make a profit of 2400% (profit of $ 240 what costs $10).

Please re-read my post. The cost is in the INSTALLER community, not the manufacturer. Also profit, at a very basic level, without a need for 'diploma', is a percentage of sales not cost, it is an ignorant statement. I don't blame you for not knowing I blame you for the way to tell us.

Steven, in a friendly way. You either contribute knowledge or you don't. Marketing your wares here and in the way you do is not welcomed. Take a chill pill or buzz off.

I have indicated in my last sentences, I was not giving marketing arguments. The arguments I have given referring to the use of Ethernet technology and certain Radio Frequency technologies are technical arguments; obviously, using a KNX Twisted Pair in a star topology does not make use of the benefits of the bus approach and results in an expensive ... darn, that's marketing?

I have indicated that I contribute knowledge to my possibility and to the extend that I am allowed to. I already chilled down some days ago and got back to the pure technics and assumed goal of the OpenRemote initiative. So, if there are technical questions, doubts ... concerning KNX technology, then please reply.

Steven De Bruyne wrote on Oct 16, 2008 12:38:
I already chilled down some days ago and got back to the pure technics and assumed goal of the OpenRemote initiative.

Oh good, you need chill some more then.... also reach some more for those 'technics'.

So, if there are technical questions, doubts ... concerning KNX technology, then please reply.

KNX is a no-brainer for an organization like ours. I am still unclear what the needs are for a software provider like OR to claim KNX compatibility. Think of OR as a Controller on a existing runtime. Something also tells me I will not be sending you emails for technical problems :)

While KNX is the most standardized actuator technology, it is an evolutionary dead-end unless it gives way to a more abstracted control language. Right now, the programming model is more limited by the physical constraints of what actuator hardware was 15 years ago than what UI and installers need to program.

In five years from now we can maybe build multilingual actuators that actually implement the KNX protocol, as legacy drop in replacement for aging KNX hardware but also supports the advanced Android based behavior for example. Chinese will rule these lands with $40 actuators. OR wants to help on the software front.

The cost on hardware manufacturers is what is interesting

Well, looking at the site there's an option at 1% of KNX revenues, min 1000 euros/year. Plus give or take 1000 euros per device for certification. They also requires ISO 9001 certification, not sure how much that would cost.

So, assuming you sell at least 1000, KNX costs less than 2% of a 250 euros device. But the manufacturer does not get 250 euros, probably something like 125 (and I am not sure which value counts for KNX). You can pretty much bet on 30% for the electrician and 30% for the distributor.

Sounds reasonable to me, but I may have missed something.

Exactly, you can use to configure all certified devices. You can download each ETS file device in the web of the manufacturer. And you can configuring each device (of course, the configuring depend on the device features).

You can also create group addresses, where you use the object functions of the devices that you inserted before in your project.

But ETS allows more actions, like a what is happening in the bus right now, basically you can read what is happening in the bus, and it allows you find failures, and, the most important:

You can simulate whatever you want, I mean that you can write in every group address and then you can know what's happening.

Cheers!