Why is there a minimum spacing?

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

DHP wrote:

(snip)

> Actually the fundamental at 10MHz is totally suppressed and *all* the
> energy is put into the sidebands. 10Mbps of random data needs a
> minimum bandwidth of 5MHz (in both sidebands). A slowly changing
> sequence just brings the sidebands in, it doesn't reduce their
> amplitude.

Many systems, such as most modems, use a scrambler to randomize
the bit stream. Otherwise the data may not be all that random.

For random data, yes, the 10MHz is zero, but consider ftping
a file full of zeros. Maybe the background for an uncompressed
image file, for example.

-- glen

 

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

glen herrmannsfeldt <gah@ugcs.caltech.edu> said

>DHP wrote:
>
>(snip)
>
>> Actually the fundamental at 10MHz is totally suppressed and *all* the
>> energy is put into the sidebands. 10Mbps of random data needs a
>> minimum bandwidth of 5MHz (in both sidebands). A slowly changing
>> sequence just brings the sidebands in, it doesn't reduce their
>> amplitude.
>
>Many systems, such as most modems, use a scrambler to randomize
>the bit stream. Otherwise the data may not be all that random.
>
>For random data, yes, the 10MHz is zero, but consider ftping
>a file full of zeros. Maybe the background for an uncompressed
>image file, for example.

Not many networks are dedicated exclusively to large pure black
bitmaps.

 

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

DHP wrote:

(snip, I wrote)

>>For random data, yes, the 10MHz is zero, but consider ftping
>>a file full of zeros. Maybe the background for an uncompressed
>>image file, for example.

> Not many networks are dedicated exclusively to large pure black
> bitmaps.

Round trip on the cable is about 42 bits, so 42 zeros in a row
is enough to do it.

-- glen

 

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

the 500 meters part of 10BASE5 actually relates to the Slot time of
Ethernet... Or the maximum time taken to realize a cdollision has
occurred... 2165 nanseconds.

With regards to the spacing, when you look at the impedance variation
of a vampire tap and associated capacitive coupling of the MAU, you get
some level of reflection. If these elements are introduced and managed
at the odd integrals of the quarter wavelength, they cancel each other
out. Just like to 17.5, 70.7, and 177 M lengths specified for
mismatches of cable segments when mixing cables from different
maufacturer and lot numbers.

Dougie!

 

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

Rich Seifert <usenet@richseifert.com.invalid> said

>In article <cvtci1pats9jpsfk29l92r1ilm48qh2cbl@4ax.com>,
> DHP <me@privacy.net> wrote:
>
>> With Manchester encoding, even a stream of zeros is actually a square
>> wave.
>>
>
>It *could* be a square wave, but we intentionally slew-rate limited the
>signal impressed on the coaxial cable; it has a 25 ns nominal rise/fall
>time. This both reduces the effect of tap reflections and reduces the
>EMI generated by the signal.

Or, to be perverse, it limits the bandwidth

I agree about it being simpler in the time domain though, because we
are really talking about narrow pulses - the derivative of the edges.

>Let's see if I remember the numbers correctly:
>
>The nominal voltage resulting from a single transmitter on the coaxial
>cable is ~2V p-p. The current in the capacitive tap (which gets
>reflected into the coaxial cable) can be calculated as:
>
>I = C dv/dt
>
>dv/dt is 2V / 25 ns, or 80 MV/s (that's MegaVolts per second)
>C is 4 pf worst-case, so the current is 4 pf * 80 MV/s = 320 uA.
>
>The impedance seen by the capacitor is 25 ohms (it effectively sees two
>50 ohm cables in parallel, one in each direction away from the tap).
>Thus the "noise" voltage generated by a single tap is 25 * 320 uA = 8 mV
>
>8 milliVolts by itself would not be a problem, however, if we had the
>worst-case situation of all 100 transceivers lumped together, we would
>have 800 mV of signal, which would blow away our required 5:1 signal to
>noise ratio. So the idea is to make sure that as few of these 8 mV
>spikes (they only last for 25 nS, while the voltage on the cable is in
>transition) add up in phase. Also, by creating a minimum cable length of
>250 m for those 100 taps (i.e., by spacing them by at least 2.5 m), we
>are sure to get a fair amount of attenuation, at least for the taps that
>are farthest away.
>
>The simulations simply tried a zillion variations on numbers of
>transceivers, placement along the cable, spacing requirements, and data
>patterns to determine if there were any pathological situations where
>the noise exceeded the budget allowance.

Interesting, because with 2.5m taps, a 25ns pulse produces
(practically) non-overlapping reflections. So if you're trying to
break the system you don't gain anything by clustering them. That
means the amplitude is that of a single reflection per half-bit of
line, 11.7m. So the worst possible case is about 42 reflections or
328mV. What's that, about -10dB in a 1V system? But as the noise isn't
random it can *never* go any higher so, ignoring other noise, it
should be possible to set threshol voltages that avoid errors
altogether. But only just.

> By the way, this took a HUGE
>amount of computer power, at least by the standards of the time. I
>managed to distribute the simulation runs across dozens of VAXen (780s,
>the only ones in existence at the time) all around the world, using
>DEC's private network. I used the idle compute power of just about every
>machine in those time zones where the normal work day was over. It was a
>rather ambitious task for its day.

 

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

glen herrmannsfeldt <gah@ugcs.caltech.edu> said

>DHP wrote:
>
>(snip, I wrote)
>
>>>For random data, yes, the 10MHz is zero, but consider ftping
>>>a file full of zeros. Maybe the background for an uncompressed
>>>image file, for example.
>
>> Not many networks are dedicated exclusively to large pure black
>> bitmaps.
>
>Round trip on the cable is about 42 bits, so 42 zeros in a row
>is enough to do it.

Well, put it this way then, not many networks are dedicated to pure
never-ending jam.

 

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

> Various illuminating prompts and comments,
> the latter particularly by Rich Seifert,

Thanks all for your efforts, it is always a pleasure to read such
material.

I have never actually worked with "thick" Ethernet however I had
assumed that the regular spacing was to cause some effect
rather that to prevent some effect. - I hope that makes sense.
Very interesting.

And apparently he is becomming a Lawyer!!!

 

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

In article <usenet-7C29BA.16020613092005@news.isp.giganews.com>,
Rich Seifert <usenet@richseifert.com.invalid> wrote:
:By the way, this took a HUGE
:amount of computer power, at least by the standards of the time. I
:managed to distribute the simulation runs across dozens of VAXen (780s,
:the only ones in existence at the time) all around the world, using
EC's private network.

The 750 was officially introduced in October 1980, 25 years ago next
month. In an earlier posting, you indicated your thesis was
"some 25 years ago". Coincidence??
--
"I will speculate that [...] applications [...] could actually see a
performance boost for most users by going dual-core [...] because it
is running the adware and spyware that [...] are otherwise slowing
down the single CPU that user has today" -- Herb Sutter

 

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

Dougie! wrote:

> the 500 meters part of 10BASE5 actually relates to the Slot time of
> Ethernet... Or the maximum time taken to realize a cdollision has
> occurred... 2165 nanseconds.

No, repeaters are allowed between segments.

The 500m limit is mostly due to cable attenuation and the
ability to properly do collision detect over the length of
the cable. Collisions are detected by the DC voltage on
the cable. The threshold must be more than the maximum for a
single nearby transmitter, and, for receive more detection,
less than two at the far end.

The slot time allows for repeaters and FOIRL (fiber) links,
at least traditionally.

-- glen

 

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

In article <dg855b$ski$1@canopus.cc.umanitoba.ca>,
roberson@ibd.nrc-cnrc.gc.ca (Walter Roberson) wrote:

>
> The 750 was officially introduced in October 1980, 25 years ago next
> month. In an earlier posting, you indicated your thesis was
> "some 25 years ago". Coincidence??

Not really, except to the extent that: (1) I was doing my graduate work
at WPI in 1976-79; (2) I was working at DEC at the time (and going to
school at night); (3) My job at DEC focused on the 10 Mb/s Ethernet
design and the writing of the original DIX specifications; and (4) DEC
was developing the VAX product line at the same time.

Before going to the Ethernet project, I did some signal integrity work
on the VAX 11/780 (code-named Star). While the VAX 750 (code-named
Comet) and VAX 730 (code-named Nebula) were being designed in the same
building where I worked (an abandoned and converted shopping mall in
Tewksbury, MA), I didn't have much to do with them. My thesis involved
the development of methodology for analyzing signal behavior on long
transmission lines, which came directly from my Ethernet work.


--
Rich Seifert Networks and Communications Consulting
21885 Bear Creek Way
(408) 395-5700 Los Gatos, CA 95033
(408) 228-0803 FAX

Send replies to: usenet at richseifert dot com

 

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

Rich Seifert wrote:

> I
> managed to distribute the simulation runs across dozens of VAXen (780s,
> the only ones in existence at the time) all around the world, using
> DEC's private network.

I used to support some VAX 11/780s. That was where I first saw ethernet.
However, the first lan I worked on was part of a Collins 8500C system,
which used time division multiplexing on a ring. A device would transmit
individual bytes in it's time slot and the receiving device would listen to
that time slot. IIRC, the high speed "TDX" loop ran at 8 Mb/s and the low
speed "TDM" loop was 2 Mb/s. Each device had relays, which were used to
connect it to one of two rings. There was also a loop sync box, that would
emit a "chirp", whenever a device joined or left a ring. This was in the
late '70s, though (IIRC), the low speed TDM loop originated in the '60s.

 

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

Rich Seifert wrote:

> By the way, this took a HUGE
> amount of computer power, at least by the standards of the time. I
> managed to distribute the simulation runs across dozens of VAXen (780s,
> the only ones in existence at the time) all around the world, using
> DEC's private network.

I seem to recall, that when the Intel 386 CPU was announced, it was claimed
to be as powerful as the VAX 11/780. I remember the VAX systems we had,
included some 16 MB memory boards, that had stacked (one chip welded on top
of another) chips, to obtain that amount of memory.

 

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

Dougie! wrote:

> the 500 meters part of 10BASE5 actually relates to the Slot time of
> Ethernet... Or the maximum time taken to realize a cdollision has
> occurred... 2165 nanseconds.

Actually, it's due to signaling limitations. That's why the maximum
distance is different for 10base5, 10base2 and 10baseT. Due to different
cable characteristics, they can support the signal for different distances.
Note that the velocity factor is similar for all three, so timing is not
dependent on cable type. As for the collision distance, a collision has to
be detected within 512 bit times. At 10 Mb/s, that's 51.2 uS (round to 50
uS for convenience). Now, at the speed of light, that signal can travel 15
Km. Since there is also the return trip, the maximum length would be 7.5
Km. Allowing for the cable velocity factor, you're down to somewhere
around 6 Km, which greatly exceeds the distance for any copper segment.
Of course, with full duplex connections, there are no collisions to worry
about.

 

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

"Dougie!" <dougie.stevenson@gmail.com> said

>the 500 meters part of 10BASE5 actually relates to the Slot time of
>Ethernet... Or the maximum time taken to realize a cdollision has
>occurred... 2165 nanseconds.
>
>With regards to the spacing, when you look at the impedance variation
>of a vampire tap and associated capacitive coupling of the MAU, you get
>some level of reflection. If these elements are introduced and managed
>at the odd integrals of the quarter wavelength, they cancel each other
>out. Just like to 17.5, 70.7, and 177 M lengths specified for
>mismatches of cable segments when mixing cables from different
>maufacturer and lot numbers.

Sure. But that's lengths of cable where one may reasonably expect
small mismatches in the resistive impedance and it makes sense to try
to cancel some of the reflections. I'd be interested to know the exact
reasoning, since successive bits need bear no relationship to each
other, so what cancels for a 1xx1 pattern will reinforce for a 1xx0
pattern.

That aside, my query was about the spacing between taps which is
specified as multiples of 2.5m but also referred to as a minimum of
2.5m. Taps produce short pulse reflections, occupying about 5m of
cable and it is much easier to think of a lot of short pulses being
reflected than to try to use frequency domain concepts like wavelength
on what is a wide-band signal.

The multiples vs minimum question keeps cropping up but from what Rich
has said I infer there's no technical benefit of sticking to multiples
rather than minimums, in fact random spacing is probably better.
However, if you do so, you *will* meet a minimum spacing criterion
automatically as well as having a simple and practical way of being
absolutely sure you're doing it right.

According to Rich, who did the work for the standard, it's all about
reflections. I've seen several people earnestly assuring each other
that it's to do with triggering collision detection but that would
appear to be an urban myth. It seems that even the minimum spacing is
a really an "on average" criterion. However, to put that into a
standard, you'd have to say something "a maximum of 7 taps in any
10m, a maximum of10 taps in any 20m and a maximum of 12 taps in any
30m, each maximum being taken over the entire length of the [segment]"
etc. That would be far too complex and I can understand why they
decided on a simple foolproof method.

 

[Guest]

Archived from groups: comp.dcom.lans.ethernet (More info?)

Henry <me@privacy.net> said

>I wonder if anyone can give a definitive answer as to why there is a
>minimum spacing specified on (some) ethernet cable. The thick stuff
>with markers every 2.5m for example which is 1 bit of delay at 10 MHz.
[No it's not -oops!]
>
>There is some mention of it on various web sites but the reasons for
>it are not stated. Maximum lengths etc are simple enough to
>understand: you need to be sure that collisions are not late. The only
>reason I can think of for specifying a minimum distance is to maximise
>the effect of a collision when two MAUs start transmitting at the same
>time. Only I can't see that it would. They won't actually start
>together. If they're waiting for the line to become free, the last
>data going past them will make sure one starts after the other. So the
>second will start up at the eaxct moment the first's one's data
>arrives. So it will experience a zero time-difference collision. The
>first one will have a two bit difference. Even if there's an advantage
>in that - which I don't understand -it assumes exactly one 2.5m
>section of cable. But the 2.5m is only a minimum: the spec doesn't
>require exact multiplesof 2.5m over hundreds of metres! So I'm racking
>my brains as to why it was ever specified at all.

<...>

Thanks everyone for the ensuing discussion and thanks especially to
you, Rich, for giving the definitive facts behind the case.