Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

"Single conductor", noisy serial comms with opto-i

Status
Not open for further replies.

eem2am

Banned
Joined
Jun 22, 2008
Messages
1,179
Helped
37
Reputation
74
Reaction score
24
Trophy points
1,318
Activity points
0
serial communication opto

Hello,

Please can you help me to correct where I went wrong with this interview question concerning serial data transmission over some distance, with a single comms line (eg not twisted pair) ?

There is also opto-isolation of the sender from the receiver.

The interviewer did not say exactly how far the data was being sent….but he said that sender and receiver had “some distance” between them………-since they were talking about electronic systems aboard auto-mobiles, I think its safe to assume that he meant that the signal was being sent from one part of a car or truck to another part of the same car/truck.

The interviewer also stated that there were bad electrical noise problems in the environment.
The interviewer also stated that signal isolation was needed….so I suggested using pulse-transformers, but the interviewer said I must use opto-isolation.

Anyway, I stated that I would want to use shielded twisted pair or co-axial cable…
-But he insisted that we had to consider a SINGLE comms line for the serial data…..

…….this was the diagram that we drew together of the system…….i basically drew the opto-coupler bit, since he asked me to do that………..

***DIAGRAM OF “SINGLE LINE” SERIAL COMMS WITH OPTO ISOLATION

s46ue8.jpg


I appreciate that the above diagram may look a bit unusual to readers?…….
-but this is the diagram that the interviewer asked me to consider……..
-During the interview, I presumed that the DPST switch was just like a representative way of showing that it can send pulses.

Anyway………..

The interviewer referred me to the above diagram, -and told me that it was a very electrically noisy environment.
-The interviewer then asked me how I would deal with such noise.
-I told him that I would filter it out.
-He then asked me to demonstrate how I would filter the noise out.

……so I drew in the capacitor as follows…………

***DIAGRAM OF “SINGLE LINE” SERIAL COMMS WITH OPTO ISOLATION
-Include Filter cap

2z7o9ch.jpg



Its only now that I realise that my placement of the cap in the above position was incredibly stupid.
-Little wonder that the interview was terminated soon after this.

Would you agree that the cap should have been placed across the following position ?…….


***DIAGRAM OF “SINGLE LINE” SERIAL COMMS WITH OPTO ISOLATION
-Include Filter cap in better (?) position.

1h7afc.jpg


Also, have you ever heard of this kind of “single-line” communication?……….i thought that ‘go’ and ‘return’ should be as close as possible?….eg, as in twisted pair?

Do you believe that I should have filtered the noise with a common-mode-choke ?

Or do you know how I should have gone about filtering the noise?
 

wire (single conductor) electronic symbols

Well you are right with the twisted pair as in single wire there would be huge problem of CM noise.

for filtering i would prefer to use ferrite bead along with the capacitor if twisted pair is not be used
 

using a single conductor as a voltage source

we had to consider a SINGLE comms line for the serial data
Without defining the return path properties, the question can't be answered. Actually, a single line electrical circuit doesn't exist. You may ignore the return path, but it's present anyway.

In automotive electronics, you have of course some kind of "single line" connections, using the chassis as return path. But you are able to analyse e.g. loop inductance, resistance, possible DC and AC interferences, if necessary.
 

derive environment rc command

Such kind on device is needed not for communication (data transfert) but to send commands or a "consigne" to a control system (very small information to send). Thats why it is one wire and unidirectional, and there is no return .

FvM said:
Actually, a single line electrical circuit doesn't exist. You may ignore the return path, but it's present anyway.

It exist but require a certain protocol in order to avoid interference.
Google "one wire protocol"

Cheers,
Advares.
 

single unidirectional wire protocol

Thats why it is one wire and unidirectional, and there is no return.
Don't confuse data flow and electrical current flow respectively voltage reference. Also the said "one wire" protocol depends on a common ground return path (and the protocol is bidirectional, by the way).
 

switch drop go and return conductor

FvM said:
Thats why it is one wire and unidirectional, and there is no return.
Don't confuse data flow and electrical current flow respectively voltage reference. Also the said "one wire" protocol depends on a common ground return path (and the protocol is bidirectional, by the way).

I didn't get what you meant by "Actually, a single line electrical circuit doesn't exist" thats why my reply was general.

Concerning your last remark grounding is not necessary if the sender and receivers are feed from the same source.
 

single wire no return

By answering that general, the answer get's simply wrong to my opinion. Of course a common supply provides a common ground as well. Please consider two battery operated devices and how a true "single wire" connection between the two can work.

P.S.: The circuit shown above has a common ground symbol, and thus is actually a two-wire connection. The problem is when you try to discuss the circuit behaviour, e.g. with interferences present without defining the properties of the ground connection.
 

sender and receiver diagram for r/c cars

Sure.
I experienced my self the problem when we have two independant batteries. Consider a simple one wire serial communication between 2 Laptops. In that case you need 2 wires one for data and the other for the ground. Now suppose you wanna do the same connection with two desktop PC. In that case only one wire is needed coz the two PCs have the same ground (from the common power supply). This is the case of our discussion coz in cars there is a general power supply which is the alternator.
 

serial comms noise solutions

Reviewing the above discussion, I stated that in automotive applications the chassis is used as return. You stated however "there is no return". A this point, I felt a need to clarify that the communication can't work without a return.

This may sound completely obvious, but I have learned that it isn't in edaboard discussions.

Apart from previous misunderstandings, it's clear now, that we are talking about a configuration with a single signal wire and the chassis as common ground. I had basically two points related to this configuration:
- Single wire is a marketing slogan but not really describing the circuit
- If you want to know more about it, you should precisely analyse the ground return loop

P.S.: I must consider, that I didn't yet try to answer the questions in the original post. I apologize for the ignorance.

- As already clarified by several contributions, the shown configuration is typical for some of the (slower) communication channels found in a car.

- Clearly, strong interferences must be expected. Because of the "long distance" return path, there's no clear difference between common mode and differential interferences in this case. No common mode choke can help against it. In theory, a current signal source would give better interference supression than a voltage source, e.g. the 5V in your example. But for practical reasons, high level voltage sources (5-12 V) are more suitable.

- Regarding the discussed filter means, your first RC solution isn't that bad at all. The second one isn't clearly better, a RCR (RC low-pass with separate LED current limiting resistor) would be preferable to my opinion. Furthermore, a defined threshold receiver, at best a schmitt-trigger should follow the low pass. You can't do much more.

- I also would like to notice, that no clear motivation for the opctical isolation is given. Actually, it isn't directly related to the said interference problem. But it may be meaningful for practical reasons not shown in the present circuit. However, for the interference suppresion discussion, the LED can represent the receiver, with or without a succeeding isolating coupler.
 

Re: "Single conductor", noisy serial comms with op

FvM said:
This may sound completely obvious, but I have learned that it isn't in edaboard discussions.

Hi all,
Hi FvM.

I dont remember have said in my last replies that internal car's communication have no return. I've just said that the return is not necessary in such ciruits (eem2am's circuit) and that's different. I know that the the chassi as well some parts of of the "carcasse" are used as conductors in the car's electric circuits. Certainly there are other reasons for using the chassis as a returm that I ignor for the moment. Serial communication between two desktop PC using a single wire is possible, but it's better to use a return if the PC are distants coz without a return the ground path between the two PCs is too long compared to the data path which can affect the performance of the communication.


Let me ask you this question:
Are wires used in cars consired as long wires ?

@eem2am reply is correct. I would have given the same if I was at your place.
 

Re: "Single conductor", noisy serial comms with op

I was referring to this statement
Such kind on device is needed not for communication (data transfert) but to send commands or a "consigne" to a control system (very small information to send). Thats why it is one wire and unidirectional, and there is no return .

But there are no open questions related to this point, I think. It's basically a different usage of the term return. I understand it in a general way as the electrical return path, that's obviously necessary in any electrical circuit and thus always present, while you apparently mean a dedicated return wire, that can be omitted if a common ground is present.

I used the phrasing "long distance" return path to describe a conductor loop, that involves a considerable enclosed area susceptible to interferences, possibly without an exactly defined location of the return. It isn't theoretically demanding but should be understandable in an intuitive way, I think.
 

Re: "Single conductor", noisy serial comms with op

FvM said:
you apparently mean a dedicated return wire, that can be omitted if a common ground is present.

Yes, that's exactelly what I mean :D
 

FvM:-

Regarding the discussed filter means, your first RC solution isn't that bad at all

FvM, my first RC filter solution (in the second diagram down in the first post above) was really bad, and i am sure that you actually mean that my corrected solution (third diagram down) is better.

putting a filter cap across the opto diode like i did above (second diagram down) was terrible of me...

.....Any noise current will just go through the opto diode instead of through the cap...........in any case, a noise current will put a noise voltage in the series resistor and this will create noise in the opto-diode, so my second diagram (above) was a disaster.

FvM.....
Because of the "long distance" return path, there's no clear difference between common mode and differential interferences in this case. No common mode choke can help against it.

........Bit i thought all electrical noise was either differential_mode or common_mode ? ......so i thought there would always be a difference between common_mode and differntial_mode noise ?

dipnirvana:-

Well you are right with the twisted pair as in single wire there would be huge problem of CM noise.
for filtering i would prefer to use ferrite bead along with the capacitor if twisted pair is not be used

The thing is, doesn't a ferrite bead acting as a common-mode-choke need to be put around the go and return conductor ?.......and in this case, it looks like the return conductor is going to be the car's chassis as above posts mention. (the interviewer did not state what the return conductor would be)


...so i would have thought a ferrtie bead was impractical here ?


as far as the type of noise is concerned...

dipnirvana:-
Well you are right with the twisted pair as in single wire there would be huge problem of CM noise.


I tend to agree with you..........the above "single_wire" comms set-up (the return being the chassis) would surely cause common-mode noise and not differential mode noise ?
......so am i right in thinking that a common-mode-choke would be needed ?

Added after 1 minutes:

also, do you think that not only would single wire comms induce common mode noise into other circuitry.......

...but it would be very susceptible to common mode noise itself ?
 

Re: "Single conductor", noisy serial comms with op

Your pretty limited if that’s all you have to work with. In an extremely noisy environment the best solution is dedicated returns star grounding, shielding and twisted pair. You can also get optos with logic gates integrated with Schmitt trigger but they are pricey.

What you could try is this from Agilent.

32zj39z.png

1exg5v.png


Forgive my cut and paste I’ve done enough typing for one day.

The full Agilent literature is here.The above was taken from page 25 of 82.


https://icecube.wisc.edu/~kitamura/NK/Datasheets/misc/5988-4082EN designers guid.pdf
 

Re: "Single conductor", noisy serial comms with op

I actually meant the circuit with the capacitor across the LED with "not that bad at all". It's basically a question of capacitor dimensioning. The third version is much worse to my opinion, cause it uses the undefined wire inductance as a filter and most likely causes resonances. As said, an RCR circuit would be better, but the second is clearly better than the third, to my opinion.

I simply keep my statement, that it's meaningles to distinguish between common mode and differential interferences (or "noise") in this case. To make a difference, the circuit must have three nodes, a differential input and an additional ground node. You may analyse your circuit as a differential receiver with the negative input connected to ground, then you actually get Vcm = Vdiff.

I already made a statement on common-mode chokes and thought, there ineffectiveness would be obvious here. dipnirvana suggested a ferrite bead as a (differential) filter, which basically can work. But most likely, a RC filter with a much lower bandwidth than achievable with a ferrite bead is required.

Finally, the contribution by max012 is interesting, but unfortunately unrelated to the present problem.
 

FvM

"I actually meant the circuit with the capacitor across the LED with "not that bad at all". It's basically a question of capacitor dimensioning. The third version is much worse to my opinion, cause it uses the undefined wire inductance as a filter and most likely causes resonances. As said, an RCR circuit would be better, but the second is clearly better than the third, to my opinion. "

........The thing is, the second diagram has the opto diode basically shorting out the cap, since a diode is basically a "short circuit with 0.7V across it"

so i am not understanding how this is ok?
 

As you can see from a datasheet as well as derive from semiconductor theory, a diode is no short circuit but has a well defined dynamic impedance and you can calculate a RC time constant. But cause the time constant is different in on- and of- state, the RCR circuit would be better, as said.
 

Re: "Single conductor", noisy serial comms with op

Max0412 has a good approach. Putting 2 resistors instead of one will cause the same voltage drop on them and respective ports of LED. Therefore voltage across LED will not change due to CM noise currents. Good work.

PS. it is well explained in Avago (former HP/Agilent) docs as pasted already.
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top