Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
Design of a power supply for a non certified KNX bus
1,377 views
Skip to first unread message
pozz
May 5, 2016, 10:40:15 AM5/5/16
to
I like the idea of KNX TP1 bus: a two-wires half-duplex bidirectional
multi-master bus at 9600bps. The most important thing is data *and
power* are on the same couple of wires.
There are many KNX bit transceivers (for example, from On Semiconductor)
that integrates everything, from bus coupling to the DC/DC switching.
Add an MCU and you have a full node on the bus.
Now, I don't want to be KNX certified. I only want a true two-wires
(data and power) bus, where all the nodes will be developed by me.
I can use KNX bit transceivers without problems, but the big problem is
a suitable KNX power supply. It is a normal 30Vdc power supply with an
integrated choke that is needed to separate data and power.
What is this choke? Is it simply a single inductor in series of positive
rail? I think the KNX bus is differential, so the inductors should be
two, one in series to positive rail and one in series to the negative rail.
Am I wrong? Is the KNX choke much more complicated than two simple
inductors?
multi-master bus at 9600bps. The most important thing is data *and
power* are on the same couple of wires.
There are many KNX bit transceivers (for example, from On Semiconductor)
that integrates everything, from bus coupling to the DC/DC switching.
Add an MCU and you have a full node on the bus.
Now, I don't want to be KNX certified. I only want a true two-wires
(data and power) bus, where all the nodes will be developed by me.
I can use KNX bit transceivers without problems, but the big problem is
a suitable KNX power supply. It is a normal 30Vdc power supply with an
integrated choke that is needed to separate data and power.
What is this choke? Is it simply a single inductor in series of positive
rail? I think the KNX bus is differential, so the inductors should be
two, one in series to positive rail and one in series to the negative rail.
Am I wrong? Is the KNX choke much more complicated than two simple
inductors?
George Herold
May 5, 2016, 11:38:54 AM5/5/16
to
has a PS schematic.
(one inductor will be enough... current has to flow out one leg and back the other.)
George H.
Martin Riddle
May 5, 2016, 10:51:41 PM5/5/16
to
value used. I gather it's a couple of uH from looking at the DIN rail
chokes from Gira.
Cheers
pozz
May 6, 2016, 4:15:48 AM5/6/16
to
Il 05/05/2016 17:38, George Herold ha scritto:
> [...]
download link. You can register to MyKnx (www.myknx.org) and make an
order for KNX specifications. The price of the product (specifications)
is zero. After applying the order, you will be able to download the KNX
specifications.
The specifications are very big, because KNX is a very complete standard
for building automation and specify almost everything, from phisical
medium (DC bus, powerline, RF, IP) to application layer, addressing,
fragments, and so on.
I'm interested only in the physical and data-link layer named KNX TP1
(TP stands for twisted pair). TP1 is descripted in Chapter 3_2_2, but
here there isn't a good description of power supply unit (with its choke).
Some info about the PSU for TP1 is descripted in Chapter 9_2. I uploaded
two images that you can see here:
http://imgur.com/a/IqKf0
I'm not an expert of power supply units and transformers and, as I wrote
in my original post, I don't need to be KNX certified, because I'm going
to develop a proprietary solution inspired from KNX TP1 standard,
resuing ready-to-use KNX transceivers (ON Semiconductors makes nice
products).
The big problem is the choke. It is a transformers with three windings.
What is the mean of the right winding/circuit with BC327 transistor?
Is it necessary?
As you can see, there are two windings/inductor: one on positive and one
on negative rail (in parallel with a resistor). Are they coupled? Why
there are two and not only one on a single rail?
KNX devices are very common in Europe, so I think some transformers
manufacturers has already designed and produced a suitable KNX choke,
but I couldn't find any.
Any help?
> [...]
> I'm not sure what you are talking about. Perhaps post a link that
> has a PS schematic.
>
> (one inductor will be enough... current has to flow out one leg and back the other.)
KNX (www.knx.org) specifications are open, but there isn't a direct
> has a PS schematic.
>
> (one inductor will be enough... current has to flow out one leg and back the other.)
download link. You can register to MyKnx (www.myknx.org) and make an
order for KNX specifications. The price of the product (specifications)
is zero. After applying the order, you will be able to download the KNX
specifications.
The specifications are very big, because KNX is a very complete standard
for building automation and specify almost everything, from phisical
medium (DC bus, powerline, RF, IP) to application layer, addressing,
fragments, and so on.
I'm interested only in the physical and data-link layer named KNX TP1
(TP stands for twisted pair). TP1 is descripted in Chapter 3_2_2, but
here there isn't a good description of power supply unit (with its choke).
Some info about the PSU for TP1 is descripted in Chapter 9_2. I uploaded
two images that you can see here:
http://imgur.com/a/IqKf0
I'm not an expert of power supply units and transformers and, as I wrote
in my original post, I don't need to be KNX certified, because I'm going
to develop a proprietary solution inspired from KNX TP1 standard,
resuing ready-to-use KNX transceivers (ON Semiconductors makes nice
products).
The big problem is the choke. It is a transformers with three windings.
What is the mean of the right winding/circuit with BC327 transistor?
Is it necessary?
As you can see, there are two windings/inductor: one on positive and one
on negative rail (in parallel with a resistor). Are they coupled? Why
there are two and not only one on a single rail?
KNX devices are very common in Europe, so I think some transformers
manufacturers has already designed and produced a suitable KNX choke,
but I couldn't find any.
Any help?
piglet
May 6, 2016, 6:20:34 AM5/6/16
to
On 06/05/2016 09:15, pozz wrote:
> The big problem is the choke. It is a transformers with three windings.
> What is the mean of the right winding/circuit with BC327 transistor?
> Is it necessary?
>
> As you can see, there are two windings/inductor: one on positive and one
> on negative rail (in parallel with a resistor). Are they coupled? Why
> there are two and not only one on a single rail?
>
The windings are all coupled and act as a big "differential mode" choke
> The big problem is the choke. It is a transformers with three windings.
> What is the mean of the right winding/circuit with BC327 transistor?
> Is it necessary?
>
> As you can see, there are two windings/inductor: one on positive and one
> on negative rail (in parallel with a resistor). Are they coupled? Why
> there are two and not only one on a single rail?
>
to partly isolate the data signalling from the power supply. My guess is
they chose symmetrical chokes in positive and negative lines to improve
balance to ground. If your intended version is only on a short cable
with just a few devices (as opposed to running round whole buildings) or
is free of ground connections then you might get away with a single choke.
I'm still thinking about the clamp circuit (winding 1 -2) but my guess
is that it prevents overshoot under some kind of event?
How is the data signalling done? What voltage/current levels and speeds,
you mentioned 9600 bps but is that modulated onto some higher frequency
carrier?
piglet
pozz
May 6, 2016, 6:52:27 AM5/6/16
to
Il 06/05/2016 12:20, piglet ha scritto:
> On 06/05/2016 09:15, pozz wrote:
>> The big problem is the choke. It is a transformers with three windings.
>> What is the mean of the right winding/circuit with BC327 transistor?
>> Is it necessary?
>>
>> As you can see, there are two windings/inductor: one on positive and one
>> on negative rail (in parallel with a resistor). Are they coupled? Why
>> there are two and not only one on a single rail?
>>
>
> The windings are all coupled and act as a big "differential mode" choke
> to partly isolate the data signalling from the power supply. My guess is
> they chose symmetrical chokes in positive and negative lines to improve
> balance to ground. If your intended version is only on a short cable
> with just a few devices (as opposed to running round whole buildings) or
> is free of ground connections then you might get away with a single choke.
One or two chokes is not a problem... the big problem is the transformer
> On 06/05/2016 09:15, pozz wrote:
>> The big problem is the choke. It is a transformers with three windings.
>> What is the mean of the right winding/circuit with BC327 transistor?
>> Is it necessary?
>>
>> As you can see, there are two windings/inductor: one on positive and one
>> on negative rail (in parallel with a resistor). Are they coupled? Why
>> there are two and not only one on a single rail?
>>
>
> The windings are all coupled and act as a big "differential mode" choke
> to partly isolate the data signalling from the power supply. My guess is
> they chose symmetrical chokes in positive and negative lines to improve
> balance to ground. If your intended version is only on a short cable
> with just a few devices (as opposed to running round whole buildings) or
> is free of ground connections then you might get away with a single choke.
(third winding).
> I'm still thinking about the clamp circuit (winding 1 -2) but my guess
> is that it prevents overshoot under some kind of event?
winding will be able to avoid overshoot on primary?
> How is the data signalling done? What voltage/current levels and speeds,
> you mentioned 9600 bps but is that modulated onto some higher frequency
> carrier?
Something is written in KNX transcervers datasheets, for example here
(page 16): http://www.onsemi.com/pub_link/Collateral/NCN5120-D.PDF
The device that wants to transmit should hear if the bus is idle
(CSMA/CA). If it is idle, it can start transmitting the telegram, a
sequence of 0s and 1s.
Only 0s can be seen on the bus, because 1s don't change the idle
condition of the bus.
The device that wants to transmit 0 puts a suitable load on the bus that
draw as much current as need to decrease the voltage on the bus of 6-9V
respect the idle value (typical 29Vdc). This step should be maintened
for 35us (active period), followed by an equalization pulse. The total
bit time is 104us (9600bps).
Of course the voltage step is created thanks to the choke in the power
supply.
piglet
May 6, 2016, 8:30:22 AM5/6/16
to
On 06/05/2016 11:52, pozz wrote:
>
> I'm not able to understand that section of circuit. How the secondary
> winding will be able to avoid overshoot on primary?
>
Here is my attempt at an explanation, which could be all wrong:
>
> I'm not able to understand that section of circuit. How the secondary
> winding will be able to avoid overshoot on primary?
>
The 100R resistor and 220nF capacitor are a snubber. During data
signalling current flows through winding 3-4 (and 5-6) and the dot end
is more positive, this happens too in winding 1-2 where the dot end
becomes more positive and makes the diode conduct and the capacitor
charge up. When the signal current ceases the voltage in winding 1-2
reverses and the diode is non-conducting and the transistor is turned on
to discharge or reset the capacitor. However the two 47R resistors
across the primaries also do plenty some damping and I do not yet
understand the need for the extra assymmetric snubber.
My guess is you could probably forget all that in a minimalist version
and just shunt a 5mH inductor with a 100R resistor and if you are
worried about bus overshoot put a 47V zener across the data bus?
piglet
Robert Loos
May 11, 2016, 5:42:29 AM5/11/16
to
> What is this choke? Is it simply a single inductor in series of positive
> rail? I think the KNX bus is differential, so the inductors should be
> two, one in series to positive rail and one in series to the negative rail.
>
> Am I wrong? Is the KNX choke much more complicated than two simple
> inductors?
Find a reverse engineered schematic here:
https://www.mikrocontroller.net/attachment/preview/149073/page_snapshots/001.png
You may also read the whole thread, unfortunately it is in German:
https://www.mikrocontroller.net/topic/264362
There seems also a patent involved:
https://www.google.com/patents/EP2802100A1?cl=en
The circuit around the transistor is to limit bus voltage when sending
(pulling the bus voltage lower for 6V iirc) stops.
If you want to google for it, try EIB instead of KNX. EIB is the
original name of KNX but it was not wisely chosen since it limits the
market to Europe (European Installation Bus) :-)
Cheers
Robert
b.kho...@gmail.com
Apr 25, 2017, 4:35:46 PM4/25/17
to
Thanks for your answer.I couldn't download schematic.Could you please upload it somewhere else?
Regards
Behdad
Robert Loos
Apr 26, 2017, 2:08:04 AM4/26/17
to
Am 25.04.2017 um 22:35 schrieb b.kho...@gmail.com:
...
try this link:
https://www.mikrocontroller.net/attachment/149073/6181.pdf
the other seems to be broken. Does not work for me, too.
Cheers
Robert
...
>
> Hi Robert
> Thanks for your answer.I couldn't download schematic.Could you please upload it somewhere else?
Hello Behdad,
> Hi Robert
> Thanks for your answer.I couldn't download schematic.Could you please upload it somewhere else?
try this link:
https://www.mikrocontroller.net/attachment/149073/6181.pdf
the other seems to be broken. Does not work for me, too.
Cheers
Robert
b.kho...@gmail.com
Apr 27, 2017, 2:24:14 AM4/27/17
to
0 new messages