Go Down

Topic: Isolating arduino from 24-30V signal (Read 327 times) previous topic - next topic

zaleopard

Hi

I want to connect my arduino directly to KNX network (read only), data signal drops like 0,45V when '0' is transmited (from what is it at transmiting '1'

But broblem is that gnd might not be same level at arduino's gnd and Im asking can I just isolate them with big condensators (KNX is selv voltage, so its not dangerous, its just I dont want fry my arduino or computers bus

I was planning something like that, where analog IN would try to stay at 2,5V so 0 bit would drop it to 2,05V and so on.
is it enought to secure devices? 




zaleopard

#1
Oct 27, 2020, 05:50 am Last Edit: Oct 27, 2020, 05:52 am by zaleopard
I do have one already done with siemens bus module, but I dont see reason to use it if im just reading data from bus

"scope" image is like https://www.researchgate.net/figure/Measured-waveform-of-the-data-communication-protocol-for-KNX-TP1-bus_fig1_308966692

or from wikipedia: https://commons.wikimedia.org/wiki/File:Electronique_bus_eib.png

I did remember that signal drops 0,45V but it was -4-5V 
Anyways, I do think it should be readable directly with arduino (ok, need to have zener to limit input to -0,6 to 5,6V area


raschemmel

Never seen anything about KNX here but maybe
you'll get lucky.
What exactly are looking for ?

zaleopard

#3
Oct 27, 2020, 06:31 am Last Edit: Oct 27, 2020, 06:41 am by zaleopard
Part of my thesis looks KNX if you can / if you should use DIY components for building tablet control to KNX, and how about  if you could you do most of the network just with Siemens bus modules (25€/piece) and cheap logic (Arduino/raspberry).

And if I could drop Siemens module away, I could have really dummy devices that can read data from KNX (dummy devices could do stuff like controlling lights for the plants, temperature displays and it wont really matter if your tablet control can't see the status of that device)

zaleopard


Never seen anything about KNX here but maybe
you'll get lucky.
What exactly are looking for ?
I do admit my first post wasnt really clear, I thoght that KNX would have been well knows, after all its most used (home)automation system here. 
But I honestly thought that fig 1. in https://www.researchgate.net/figure/Measured-waveform-of-the-data-communication-protocol-for-KNX-TP1-bus_fig1_308966692 would be self-explanatory about the signal.

raschemmel

#5
Oct 27, 2020, 05:40 pm Last Edit: Oct 27, 2020, 06:37 pm by raschemmel
Quote
But I honestly thought that fig 1. in https://www.researchgate.net/figure/Measured-waveform-of-the-data-communication-protocol-for-KNX-TP1-bus_fig1_308966692 would be self-explanatory about the signal.
Do you realize how unusual it is for a forum post to go almost 24 hours with no takers ?
It's starting to look like your assumptions were wrong. I can tell you for a fact that a an op amp window comparator can detect such voltages but why you are asking us when you have Siemens modules is beyond me. You still haven't posted any links to them so it's no surprise nobody wants anything to do with this post.  

Quote
I was planning something like that, where analog IN would try to stay at 2,5V so 0 bit would drop it to 2,05V and so on. Is it enought to secure devices?
How exactly did you come up with that schematic ? Do you have electronic circuit design experience you didn't tell us about ? I'm just wondering where you came up with that.

Quote
But broblem is that gnd might not be same level at arduino's gnd and Im asking can I just isolate them with big condensators (KNX is selv voltage, so its not dangerous, its just I dont want fry my arduino or computers bus
No.

An op amp window detector can detect any voltage range signal and convert it to a 5V signal but it's probably
more work than you want to do.
"https://www.google.com/search?client=firefox-b-1-d&q=op+amp+window+comparator"




As you can see from the waveform , the pulses go high only when the input is INSIDE the window.
If you look closely you'll see the one op amp has Vref connected to V+ while the other has a Vref connected
to V-, so they behave  opposite, with the first one generating a pulse when the input is increasing and the
second one generating a pulse when the input is decreasing but since they are open collector , the two pulses
generated by different op amps appear in the same waveform. There is no limit to how many window detectors
you can bus together with open collector outputs so you could have 15 different pulses, each corresponding
to a different event in time or space , all appearing in the same waveform, instead of having 15 sensors going
to 15 uProcessor inputs and having to process the input. Since the pulses are consecutive, that means that if
they were driving a counter, the second counter output would go high on the second pulse and so forth.
You can go crazy with this stuff, if you know how. In fact, I may already be crazy...

zaleopard

#6
Oct 27, 2020, 08:27 pm Last Edit: Oct 27, 2020, 08:41 pm by zaleopard
Sole reason is money, its same reason why I don't have Metso automaton system at my home, I would try some cheaper ones. I don't even have Siemens Logo at home so I try to do with Arduino.

This is the module I used to convert rs232 to KNX, really easy to use: https://mall.industry.siemens.com/mall/en/WW/Catalog/Product/5WG1117-2AB12

another viable option *might be*  https://fi.rsdelivers.com/product/stmicroelectronics/stknxtr/stmicroelectronics-miniature-knx-transceiver-with/1750745  but I didn't have time to order & test yet. that is cheap enought to produce.

I did small app to visualise knx messages, so siemens does work nicely, but would like to try other options too:


So How did I tought to have analog input pin to float around 2,5 with resistor divider.
I think its logical if you would have smaller signal, like audio.

lets say you use 2x 1k ohm resistors as divider, then you take audio to the middle point thru 1uF condessator (at 9,6kHz we could count it as 17ohm inpedance, so audio go nicely to analogport, even its voltage divided with resistors, that way audio middle point would be nicely at 2.5V

and after that point, you can connect it to op amps +pin and with resistor to output and add another voltage divider to - pin, so you get follower op amp connection, so it isnt so wobble, its important if your signal is weak (ph sensors might have >100Mohm impedance, if you sneezebadly it will measure wrong, so you had to add follower really close to sensor.

 (and if you have dualvoltage system with opamps, you can use it to align average audio to 0V).

that was just logical start point. Im total nyyb about these things.

yea, op amp might work, some of them works without dualvoltages, in my memory just stay that input shouldnt be too much below opamps negative operating voltage. but I had to admit I didn't look at it

raschemmel

The waveform you posted doesn't have 5V signals so why are you assuming that half of the input signal would
be 2.5V ?
Show me the input signal you are trying to convert that is 5V.

zaleopard

#8
Oct 27, 2020, 08:43 pm Last Edit: Oct 27, 2020, 08:48 pm by zaleopard
The waveform you posted doesn't have 5V signals so why are you assuming that half of the input signal would
be 2.5V ?
Show me the input signal you are trying to convert that is 5V.
But I already corrected myself at post "I did remember that signal drops 0,45V but it was -4-5V "  I remembered that signal (ripple over operating voltage) was about 1-2V peak to peak so it was nice idea to center it to 2,5V

edit: forgot to say, if you look class a -amplifiers, most of them use kind of that connection.

raschemmel

#9
Oct 27, 2020, 08:55 pm Last Edit: Oct 27, 2020, 08:57 pm by raschemmel
The waveform you posted doesn't have 5V signals so why are you assuming that half of the input signal would
be 2.5V ?
Show me the input signal you are trying to convert that is 5V.

If you do the math and create a voltage divider that reduces 45Vdc to 4.5Vdc , then you can use an analog input .
The arduino sample rate is about 10000 samples a second I think and the waveform you references shows a period
of 1 mS with three signals so you could take 10 samples in 1 mS (10000/1000=10) , which would not give you very
good resolution. An op amp comparator processing the same 4.5V reduced signal would have no problem detecting 3 signals in 1 mS. You could have a separate comparator for the high signals and one for the low signals.

Quote
that signal drops 0,45V but it was -4-5V
Where ? I don't see any drop in the signals. The lowest one is 22V.
I don't know where you're getting the 4.5V from.
Show me a waveform with a 4.5V signal.

zaleopard

#10
Oct 27, 2020, 09:05 pm Last Edit: Oct 27, 2020, 09:09 pm by zaleopard
Well, in that picture?
ok, drop in signal is 6V, and that spike up is really hard to measure, network size matters on that. and I might not have that good scope. (that is not my image)

I hope you understand that signal and dc offset is different thing and capacitor removes dc component?


raschemmel

Quote
Well, in that picture?
ok, drop in signal is 6V, and that spike up is really hard to measure, network size matters on that. and I might not have that good scope. (that is not my image)
The point is , you say one thing but show us something else.
At this point all we have is a hypothetical voltage with no proof.
Draw this hypothetical waveform you say exists with an amplitude of 4.5V.

As far as measuring those spikes, as I said, a custom comparator could easily measure them.


zaleopard

Well , I did sau 4-5v drop not 4,5 drop, but I dont think that is broblem, I had only analog scope and spike isnt that clean

specs says: about dc component 
In KNX TP the bus cable supplies all bus devices with both data and power. The rated voltage of the bus system is 24 V, while the voltage provided by the power supplies is 30 V. The bus devices work without error at voltages between 21 V and 30 V, so a tolerance range of 9 V is available to compensate for voltage
drops in the cable, and contact resistance. 


and about the chip producer says about transmiting:
"The active pulse is generated by the transmitter. Ideally, the drop has a rectangular shape, a depth between 6 V and 9 V and a duration of 35 us. Each active pulse is followed by an equalization phase of 69 us typical duration, which consists in an overshoot of the bus voltage above the DC level, followed by an exponential decay"



So, ol in theory it could be 9v below DC component, good to know. but it doesnt really help atm.

zaleopard

and I do know how to do it with op, it would be easy if I would have 1-40V power and opamp on that, I could just compare it to averade DC component -4V.

But as I dont have that kind of power and I dont want to take it from KNX line. (even its ok. solution, just need alot extra components)

zaleopard

Nah, I finished it already.

And if there isn't answers in 48h, then there isn't anyone who would have done it earlier and know it. I dont except anyone to do research for me after all. 
Sadly there was only ppl who point out useless points, it would have been a lot more usefull if you wouldnt answer at all so I wouldn't have to spend time for reading it ^^

Go Up