m...@sushi.com wrote:
> On Jan 8, 9:47 am, Paul Keinanen <keina...@sci.fi> wrote:
>> On Fri, 8 Jan 2010 23:17:28 +1100, "David L. Jones"
>>
>> <altz...@gmail.com> wrote:
>>
>>> Standard differential RS485 for example is simple to implement and capable
>>> of more than 1km at low enough data rates.
>> Standard RS-422/485 hardware should be capable of well over 200 kbit/s
>> at 500 m. In addition, some manufacturers have transceivers that are
>> slave rate limited to 250 kbit/s, thus reducing problems with
>> reflections and sensitivity to external narrow EMI peaks. The rise and
>> fall times are in the order of 1 us, so quite accurate synchronization
>> could be achieved.
>>
>> At such large distances, I very much doubt that the quite limited
>> RS-422/485 common mode voltage range (-7/+12 V) would be sufficient in
>> practice, especially considering lightning induced voltages, so the
>> connection should use galvanic isolation  at least at the other end.
>> Typically the transceiver would be powered by a power supply and
>> optoisolators would be used on the TTL side to connect to the rest of
>> the station.
>>
>> However, if isolation is required, why not use some DC-free coding
>> such as Manchester coding and use small signal transformers (as in
>> Ethernet) to isolate the reference levels from each other and from the
>> line ?
> 
> This is all good advice. You probably want the transformer anyway
> since the grounds will not match at such distances.


Put put 100K or so across so the whole cable doesn't statically run away 
from you. A totally floating cable out in the open could cause one of 
the transformers to go ... tzk ... *POCK* ... li'l blue sparkle in the dark.

-- 
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

On Fri, 08 Jan 2010 13:49:31 -0800, Joerg <i...@invalid.invalid>
wrote:

>m...@sushi.com wrote:
>> On Jan 8, 9:47 am, Paul Keinanen <keina...@sci.fi> wrote:
>>> On Fri, 8 Jan 2010 23:17:28 +1100, "David L. Jones"
>>>
>>> <altz...@gmail.com> wrote:
>>>
>>>> Standard differential RS485 for example is simple to implement and capable
>>>> of more than 1km at low enough data rates.
>>> Standard RS-422/485 hardware should be capable of well over 200 kbit/s
>>> at 500 m. In addition, some manufacturers have transceivers that are
>>> slave rate limited to 250 kbit/s, thus reducing problems with
>>> reflections and sensitivity to external narrow EMI peaks. The rise and
>>> fall times are in the order of 1 us, so quite accurate synchronization
>>> could be achieved.
>>>
>>> At such large distances, I very much doubt that the quite limited
>>> RS-422/485 common mode voltage range (-7/+12 V) would be sufficient in
>>> practice, especially considering lightning induced voltages, so the
>>> connection should use galvanic isolation  at least at the other end.
>>> Typically the transceiver would be powered by a power supply and
>>> optoisolators would be used on the TTL side to connect to the rest of
>>> the station.
>>>
>>> However, if isolation is required, why not use some DC-free coding
>>> such as Manchester coding and use small signal transformers (as in
>>> Ethernet) to isolate the reference levels from each other and from the
>>> line ?
>> 
>> This is all good advice. You probably want the transformer anyway
>> since the grounds will not match at such distances.
>
>
>Put put 100K or so across so the whole cable doesn't statically run away 
>from you. A totally floating cable out in the open could cause one of 
>the transformers to go ... tzk ... *POCK* ... li'l blue sparkle in the dark.

Ground one end.  Tie cable ground at other end to local ground via
100K.  I've seen 60VAC between the "grounds" of two adjacent high
rises :-(
		
                                        ...Jim Thompson
-- 
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
             
I love to cook with wine.     Sometimes I even put it in the food.

On 8 Jan., 11:27, "pimpom" <pim...@invalid.invalid> wrote:
> Prelim: If anyone remembers me from previous posts, please note
> that constraints imposed by my location dictate that I use, as
> far as possible, only long-established standard components in my
> projects.
>
> I want to send two independent unidirectional pulsed signals over
> a distance of about 500 meters and I'm considering my options.
> The signals will be single triggering pulses with long irregular
> intervals in between pulses. The primary requirement is that
> interference should not cause false triggering, and the two
> channels must not interfere with each other.
>
> I've considered optical fiber cable and RF, but am currently
> thinking that CAT5 cable might offer the best compromise of
> simplicity, cost and reliability. I'm always open to alternative
> suggestions though. Where I'd like to ask for your inputs is:
>
> 1. Is my scheme at all feasible with fairly simple circuitry?
>
> 2. I've never made a study of the electronics that drives
> computer network lines, and I can't do it now because my ISP is
> working in sporadic bursts (I had to wait >1 hr to post this). I
> could drive each twisted pair as balanced differential lines
> instead of using one wire as ground. Is this likely to be
> necessary?
>
> 3. What factors do I need to know and consider?

rs485 supposed to run  ~megabits at at that range


-Lasse

On Fri, 08 Jan 2010 20:07:40 +0000, Baron
<b...@linuxmaniac.nospam.net> wrote:

>
>What about the transformers from dead modems !

These are audio transformers (0,3-3 kHz) designed for line side
impedances in the order of 600 ohms.

It was not clear how high frequencies were to be carried, so the
losses may be larger than expected and a complex waveform may be
distorted in such transformers.

Also the receiving end of the CAT5 line should have a terminating
resistance about 100 ohms, so the nominal 600 ohm is not so nice, thus
the secondary side of the transformer should be loaded with 1/6
impedance compared to the original load impedance in order to present
a 100 ohm load on the CAT5 line. Operating the telephone transformer
so much out of the design specifications can cause some other
problems.
  

On Fri, 08 Jan 2010 18:42:39 GMT, Jan Panteltje
<p...@yahoo.com> wrote:

>On a sunny day (Fri, 8 Jan 2010 23:33:14 +0530) it happened "pimpom"
><p...@invalid.invalid> wrote in <hi7s19$359$1...@news.albasani.net>:
>
>>I've asked an associate to construct the circuit below, solely as 
>>a temporary test setup. Please comment.
>>http://img687.imageshack.us/img687/1074/cat5txrx.png 
>
>The 10k pull one way is weak, compared to the pull the otehr way by the transistors.

In addition, in the idle state, there is only two 100k resistor to
ground and hence totally 220 kohm loop resistance between the receiver
transistor inputs.  Thus, any external differential noise could cause
false triggering. 

To make the situation worse, in the idle state, the differential stage
is in the middle of the transition region, hence, any noise gets
amplified and can cause a false triggering.

Use hysteresis on the receiver or at least bias the input stage so
that strongly pulled into the idle state. 

Removing the capacitors and the 100 k resistors from the transmitter
will improve the situation considerably. R2 and R7 should also be
reduced significantly. 

If short pulses (<100 us) a 100 ohm termination resistor should be
inserted between the receiver inputs and driver resistor values
updated accordingly.
 

Paul Keinanen Inscribed thus:

> On Fri, 08 Jan 2010 20:07:40 +0000, Baron
> <b...@linuxmaniac.nospam.net> wrote:
> 
>>
>>What about the transformers from dead modems !
> 
> These are audio transformers (0,3-3 kHz) designed for line side
> impedances in the order of 600 ohms.

I thought POTS lines were a lot lower than 600 ohms.

> It was not clear how high frequencies were to be carried, so the
> losses may be larger than expected and a complex waveform may be
> distorted in such transformers.

How then did they cope with 56Kbs ?

> Also the receiving end of the CAT5 line should have a terminating
> resistance about 100 ohms, so the nominal 600 ohm is not so nice, thus
> the secondary side of the transformer should be loaded with 1/6
> impedance compared to the original load impedance in order to present
> a 100 ohm load on the CAT5 line. Operating the telephone transformer
> so much out of the design specifications can cause some other
> problems.

-- 
Best Regards:
                     Baron.

On Sat, 09 Jan 2010 11:31:44 +0000, Baron
<b...@linuxmaniac.nospam.net> wrote:

>Paul Keinanen Inscribed thus:
>
>> On Fri, 08 Jan 2010 20:07:40 +0000, Baron
>> <b...@linuxmaniac.nospam.net> wrote:
>> 
>>>
>>>What about the transformers from dead modems !
>> 
>> These are audio transformers (0,3-3 kHz) designed for line side
>> impedances in the order of 600 ohms.
>
>I thought POTS lines were a lot lower than 600 ohms.

It is quite possible that current telephone cables have much lower
impedance and even open wire connections might not actually go that
high even on audio frequencies, but in the early years of the
telephone industry, that was the assumption and since those days the
telephone industry has been operated at the 600 ohm impedance levels.
The 0 dBm = 1 mW is also used in audio industry as a reference level,
since 1 mW is 0.775 V into 600 ohms.

Some early electricity experimenters even assumed (based on the
maximum power transfer theory) that the electric generator resistance
should be the same as the total load resistance and they used very
thin wires in the early generators. This of course dissipated half of
the power in the generator windings :-) and only less than half of the
generator mechanical input power would be delivered to the load.

>
>> It was not clear how high frequencies were to be carried, so the
>> losses may be larger than expected and a complex waveform may be
>> distorted in such transformers.
>
>How then did they cope with 56Kbs ?

Multilevel modulation. In 56 kbit/s up to 7 bit could be transferred
in each symbol (cycle) and only from the exchange to the end user (8
kHz sampling rate and 7 bits/sample). Of course, the practical
throughput was less than 7 bits/symbol.

Anyway, phone line modes above 2400 bit/s include a frequency
equalizer to correct any frequency and phase distortion created by the
lines and audio transformers. To set up this frequency equalizer, the
training session (with a known bit sequence) was performed before the
actual data could be transferred.

Paul Keinanen Inscribed thus:

> On Sat, 09 Jan 2010 11:31:44 +0000, Baron
> <b...@linuxmaniac.nospam.net> wrote:
> 
>>Paul Keinanen Inscribed thus:
>>
>>> On Fri, 08 Jan 2010 20:07:40 +0000, Baron
>>> <b...@linuxmaniac.nospam.net> wrote:
>>> 
>>>>
>>>>What about the transformers from dead modems !
>>> 
>>> These are audio transformers (0,3-3 kHz) designed for line side
>>> impedances in the order of 600 ohms.
>>
>>I thought POTS lines were a lot lower than 600 ohms.
> 
> It is quite possible that current telephone cables have much lower
> impedance and even open wire connections might not actually go that
> high even on audio frequencies, but in the early years of the
> telephone industry, that was the assumption and since those days the
> telephone industry has been operated at the 600 ohm impedance levels.
> The 0 dBm = 1 mW is also used in audio industry as a reference level,
> since 1 mW is 0.775 V into 600 ohms.

Yes I was aware of that reference. 

> Some early electricity experimenters even assumed (based on the
> maximum power transfer theory) that the electric generator resistance
> should be the same as the total load resistance and they used very
> thin wires in the early generators. This of course dissipated half of
> the power in the generator windings :-) and only less than half of the
> generator mechanical input power would be delivered to the load.
>
>>>
>>> It was not clear how high frequencies were to be carried, so the
>>> losses may be larger than expected and a complex waveform may be
>>> distorted in such transformers.
>>
>>How then did they cope with 56Kbs ?
> 
> Multilevel modulation. In 56 kbit/s up to 7 bit could be transferred
> in each symbol (cycle) and only from the exchange to the end user (8
> kHz sampling rate and 7 bits/sample). Of course, the practical
> throughput was less than 7 bits/symbol.

Ah.  Is that why the data exchange rate was asymmetrical, 56/33 Kbs. 

> Anyway, phone line modes above 2400 bit/s include a frequency
> equalizer to correct any frequency and phase distortion created by the
> lines and audio transformers. To set up this frequency equalizer, the
> training session (with a known bit sequence) was performed before the
> actual data could be transferred.

Thanks for filling in the blanks.

-- 
Best Regards:
                     Baron.

On Fri, 8 Jan 2010 15:57:40 +0530, "pimpom" <p...@invalid.invalid>
wrote:

>Prelim: If anyone remembers me from previous posts, please note 
>that constraints imposed by my location dictate that I use, as 
>far as possible, only long-established standard components in my 
>projects.
>
>I want to send two independent unidirectional pulsed signals over 
>a distance of about 500 meters and I'm considering my options. 
>The signals will be single triggering pulses with long irregular 
>intervals in between pulses. The primary requirement is that 
>interference should not cause false triggering, and the two 
>channels must not interfere with each other.
>
>I've considered optical fiber cable and RF, but am currently 
>thinking that CAT5 cable might offer the best compromise of 
>simplicity, cost and reliability. I'm always open to alternative 
>suggestions though. Where I'd like to ask for your inputs is:
>
>1. Is my scheme at all feasible with fairly simple circuitry?
>
>2. I've never made a study of the electronics that drives 
>computer network lines, and I can't do it now because my ISP is 
>working in sporadic bursts (I had to wait >1 hr to post this). I 
>could drive each twisted pair as balanced differential lines 
>instead of using one wire as ground. Is this likely to be 
>necessary?
>
>3. What factors do I need to know and consider? 

We use RS485 over CAT-5 to 1500' at 450kbps, or so.  The data will go
even further but we run out of power after that (a Power Over Ethernet
sort of setup).  We do transformer couple the data (the DC gets tied
to the center taps) and do a 100ns pre-emphasis on the transmit side.

On Jan 8, 1:49=A0pm, Joerg <inva...@invalid.invalid> wrote:
> m...@sushi.com wrote:
> > On Jan 8, 9:47 am, Paul Keinanen <keina...@sci.fi> wrote:
> >> On Fri, 8 Jan 2010 23:17:28 +1100, "David L. Jones"
>
> >> <altz...@gmail.com> wrote:
>
> >>> Standard differential RS485 for example is simple to implement and ca=
pable
> >>> of more than 1km at low enough data rates.
> >> Standard RS-422/485 hardware should be capable of well over 200 kbit/s
> >> at 500 m. In addition, some manufacturers have transceivers that are
> >> slave rate limited to 250 kbit/s, thus reducing problems with
> >> reflections and sensitivity to external narrow EMI peaks. The rise and
> >> fall times are in the order of 1 us, so quite accurate synchronization
> >> could be achieved.
>
> >> At such large distances, I very much doubt that the quite limited
> >> RS-422/485 common mode voltage range (-7/+12 V) would be sufficient in
> >> practice, especially considering lightning induced voltages, so the
> >> connection should use galvanic isolation =A0at least at the other end.
> >> Typically the transceiver would be powered by a power supply and
> >> optoisolators would be used on the TTL side to connect to the rest of
> >> the station.
>
> >> However, if isolation is required, why not use some DC-free coding
> >> such as Manchester coding and use small signal transformers (as in
> >> Ethernet) to isolate the reference levels from each other and from the
> >> line ?
>
> > This is all good advice. You probably want the transformer anyway
> > since the grounds will not match at such distances.
>
> Put put 100K or so across so the whole cable doesn't statically run away
> from you. A totally floating cable out in the open could cause one of
> the transformers to go ... tzk ... *POCK* ... li'l blue sparkle in the da=
rk.
>
> --
> Regards, Joerg
>
> http://www.analogconsultants.com/
>
> "gmail" domain blocked because of excessive spam.
> Use another domain or send PM.

I suppose I should take my own advice and always try to design so I
can state the voltage at any node (i.e. I don't like cap dividers),
but there are all sorts of transformer to transformer applications out
there that don't seem to blow up. Video over twisted pair is a good
example.

This 100k across the cable doesn't make sense.  The transformer puts a
low impedance at DC across the cable. Did you mean to ground?