USing wireless modem over cable

I want to extend my LAN to a facility about a mile away. I cant run anthing
new but there are twisted pairs and a 1 inch Heliax cable. I could use a
dial-up connection but this is d efintely not my first choice. I have tried
using the existing coax in a thicknet connection but this didnt work. I was
thinking about using some wireless equipment we had in a novel way. Instead
of using it with an antenna I intend to make direct connections via the
cable. The cable has about 64 db of loss at 2.4 Ghz. Im not sure of the
sensitivity of the receivers but I measured the tx out at about +20 Dbm.
ITs not just being able to hook it up and try it. The connectors for the
Heliax are very expensive and I have to know with some reasonable certainty
if it will work before I buy them

Jimmie D wrote:
> I want to extend my LAN to a facility about a mile away. I cant run anthing
> new but there are twisted pairs and a 1 inch Heliax cable. I could use a
> dial-up connection but this is d efintely not my first choice. I have tried
> using the existing coax in a thicknet connection but this didnt work. I was
> thinking about using some wireless equipment we had in a novel way. Instead
> of using it with an antenna I intend to make direct connections via the
> cable. The cable has about 64 db of loss at 2.4 Ghz. Im not sure of the
> sensitivity of the receivers but I measured the tx out at about +20 Dbm.
> ITs not just being able to hook it up and try it. The connectors for the
> Heliax are very expensive and I have to know with some reasonable certainty
> if it will work before I buy them
>
Love the idea. I wouldn't be so sure of connecting a cable on both sides
though (common mode, lightning etc if the cable is between different
buildings), and I can only vaguely imagine what reflections in the cable
would be like. But what about abusing the twisted pairs as huge antennae
from both sides?

J

"jack" <(E-Mail Removed)> wrote in message
news:s7i1o3-(E-Mail Removed)...
> Jimmie D wrote:
>> I want to extend my LAN to a facility about a mile away. I cant run
>> anthing
>> new but there are twisted pairs and a 1 inch Heliax cable. I could use a
>> dial-up connection but this is d efintely not my first choice. I have
>> tried
>> using the existing coax in a thicknet connection but this didnt work. I
>> was
>> thinking about using some wireless equipment we had in a novel way.
>> Instead
>> of using it with an antenna I intend to make direct connections via the
>> cable. The cable has about 64 db of loss at 2.4 Ghz. Im not sure of the
>> sensitivity of the receivers but I measured the tx out at about +20 Dbm.
>> ITs not just being able to hook it up and try it. The connectors for the
>> Heliax are very expensive and I have to know with some reasonable
>> certainty
>> if it will work before I buy them
>>
> Love the idea. I wouldn't be so sure of connecting a cable on both sides
> though (common mode, lightning etc if the cable is between different
> buildings), and I can only vaguely imagine what reflections in the cable
> would be like. But what about abusing the twisted pairs as huge antennae
> from both sides?
>
> J

I hadnt thought of that but it shouldnt be hard to build an isolation
transformer . I have tried connecting two wireless nics together antenna
port to antenna port with a 60db pad in between. This worked grreat. Next
step will be building a 2.4 Ghz UNUN isolation transformer.

On Thu, 6 Jul 2006, in the Usenet newsgroup comp.os.linux.networking, in article
<Fnhrg.84136$(E-Mail Removed)>, Jimmie D wrote:

>I want to extend my LAN to a facility about a mile away.

And the company doesn't want to spend any money.

>I cant run anthing new but there are twisted pairs and a 1 inch Heliax cable.

I'm assuming the stuff is buried, rather than running on catenary (overhead
pole line). Have you discussed this with the company's insurance carrier?
If you're somewhere in the US South, you should be aware of lightning
problems, never mind ground loops.

>I have tried using the existing coax in a thicknet connection but this
>didnt work.

What happened? Did the transceivers catch fire? The original 10Base5
Ethernet was limited to 500 meters of a cable _similar_to_ RG-8/U but
allowed two repeaters between the furthest host, and a point-to-point
link which gave a total distance of 2500 meters (8200 feet). That total
length (at a Vp of about 0.7c) sets the time delay called a Collision
Domain. The maximum cable attenuation between a pair of transceivers was
8.5 dB at 10 MHz _AND_ 6.0 dB at 5 MHz. I don't have a spec sheet handy
(and am to lazy to grab it from Andrew), but 1 inch Heliax is probably
going to have a loss of roughly 1.2 dB per thousand feet at 10 MHz, and
0.9 dB per thousand feet at 5 MHz - so a mile is going to be within spec.
As I recall, the Vp is about 0.7 as well, so you're within collision
domain and loss specs - it should have worked assuming you can find the
appropriate vampire tap ;-) (Actually, transition to a 3 meter stub of
Belden 89880 or equal and put the transceivers and terminators on that.)

>I was thinking about using some wireless equipment we had in a novel way.
>Instead of using it with an antenna I intend to make direct connections
>via the cable. The cable has about 64 db of loss at 2.4 Ghz. Im not sure
>of the sensitivity of the receivers but I measured the tx out at about +20
>Dbm.

I take it that you don't have clear line of site and a couple of good sized
dishes to run those radios wireless as they were intended.

I'll take your word for the cable loss although that seems a bit low. The
receiver sensitivity is a function of the bandwidth. Point your news tool
at alt.internet.wireless (or better, hit google.groups.com and search
that newsgroup) and you'll see posts from Jeff Liebermann - and he answers
this type of question on a regular basis.

Receiver Sensitivity for DI-624 (and most other similar radios):
* 54Mbps OFDM, 10% PER, -68dBm)
* 48Mbps OFDM, 10% PER, -68dBm)
* 36Mbps OFDM, 10% PER, -75dBm)
* 24Mbps OFDM, 10% PER, -79dBm)
* 18Mbps OFDM, 10% PER, -82dBm)
* 12Mbps OFDM, 10% PER, -84dBm)
* 11Mbps CCK, 8% PER, -82dBm)

So, +20 dBm in, cable loss of 64 dB - there should be -44 dBm (probably
less) coming out the hose at the far end. The 'PER' term is Packet Error
Rate - meaning that signal level is about the minimum usable at that speed.

>ITs not just being able to hook it up and try it. The connectors for the
>Heliax are very expensive and I have to know with some reasonable certainty
>if it will work before I buy them

Again - alt.internet.wireless. You should already have type N connectors
on the end of the cable (and be pushing nitrogen or at least dry air into
the cable to prevent water intrusion), and you may have noticed that Heliax
isn't very flexible. Transition from the Heliax to something more flexible
yet still low loss (example LMR400) for the connection to the sets.

I don't see an obvious reason this wouldn't work, assuming you have an
undamaged cable and the connectors were fabricated/attached correctly
and tested at 2.4 GHz. Test the cable loss. If the cable hasn't been
pressurized, and has been outside for a year or two, you probably have
a corrosion problem internally, and in that case your screwed.

Now, the local fire marshal and your insurance carrier's risk department
might be horrified at this, but I'd simply isolate the cable at both ends
(or better, ground the cable at the building entrance at ONE END ONLY),
and power the radios via 5 KV rated isolation transformers. Undoubtedly,
that coax has been there for "a while" and hasn't caused a fire yet.
However there is that risk, so it's your responsibility, not mine.

Old guy

Moe Trin wrote:
> On Thu, 6 Jul 2006, in the Usenet newsgroup comp.os.linux.networking, in article
> <Fnhrg.84136$(E-Mail Removed)>, Jimmie D wrote:
>
> >I want to extend my LAN to a facility about a mile away.
>
> And the company doesn't want to spend any money.
>
> >I cant run anthing new but there are twisted pairs and a 1 inch Heliax cable.
>
> I'm assuming the stuff is buried, rather than running on catenary (overhead
> pole line). Have you discussed this with the company's insurance carrier?
> If you're somewhere in the US South, you should be aware of lightning
> problems, never mind ground loops.
>
> >I have tried using the existing coax in a thicknet connection but this
> >didnt work.
>
> What happened? Did the transceivers catch fire? The original 10Base5
> Ethernet was limited to 500 meters of a cable _similar_to_ RG-8/U but
> allowed two repeaters between the furthest host, and a point-to-point
> link which gave a total distance of 2500 meters (8200 feet). That total
> length (at a Vp of about 0.7c) sets the time delay called a Collision
> Domain. The maximum cable attenuation between a pair of transceivers was
> 8.5 dB at 10 MHz _AND_ 6.0 dB at 5 MHz. I don't have a spec sheet handy
> (and am to lazy to grab it from Andrew), but 1 inch Heliax is probably
> going to have a loss of roughly 1.2 dB per thousand feet at 10 MHz, and
> 0.9 dB per thousand feet at 5 MHz - so a mile is going to be within spec.
> As I recall, the Vp is about 0.7 as well, so you're within collision
> domain and loss specs - it should have worked assuming you can find the
> appropriate vampire tap ;-) (Actually, transition to a 3 meter stub of
> Belden 89880 or equal and put the transceivers and terminators on that.)
>
> >I was thinking about using some wireless equipment we had in a novel way.
> >Instead of using it with an antenna I intend to make direct connections
> >via the cable. The cable has about 64 db of loss at 2.4 Ghz. Im not sure
> >of the sensitivity of the receivers but I measured the tx out at about +20
> >Dbm.
>
> I take it that you don't have clear line of site and a couple of good sized
> dishes to run those radios wireless as they were intended.
>
> I'll take your word for the cable loss although that seems a bit low. The
> receiver sensitivity is a function of the bandwidth. Point your news tool
> at alt.internet.wireless (or better, hit google.groups.com and search
> that newsgroup) and you'll see posts from Jeff Liebermann - and he answers
> this type of question on a regular basis.
>
> Receiver Sensitivity for DI-624 (and most other similar radios):
> * 54Mbps OFDM, 10% PER, -68dBm)
> * 48Mbps OFDM, 10% PER, -68dBm)
> * 36Mbps OFDM, 10% PER, -75dBm)
> * 24Mbps OFDM, 10% PER, -79dBm)
> * 18Mbps OFDM, 10% PER, -82dBm)
> * 12Mbps OFDM, 10% PER, -84dBm)
> * 11Mbps CCK, 8% PER, -82dBm)
>
> So, +20 dBm in, cable loss of 64 dB - there should be -44 dBm (probably
> less) coming out the hose at the far end. The 'PER' term is Packet Error
> Rate - meaning that signal level is about the minimum usable at that speed.
>
> >ITs not just being able to hook it up and try it. The connectors for the
> >Heliax are very expensive and I have to know with some reasonable certainty
> >if it will work before I buy them
>
> Again - alt.internet.wireless. You should already have type N connectors
> on the end of the cable (and be pushing nitrogen or at least dry air into
> the cable to prevent water intrusion), and you may have noticed that Heliax
> isn't very flexible. Transition from the Heliax to something more flexible
> yet still low loss (example LMR400) for the connection to the sets.
>
> I don't see an obvious reason this wouldn't work, assuming you have an
> undamaged cable and the connectors were fabricated/attached correctly
> and tested at 2.4 GHz. Test the cable loss. If the cable hasn't been
> pressurized, and has been outside for a year or two, you probably have
> a corrosion problem internally, and in that case your screwed.
>
> Now, the local fire marshal and your insurance carrier's risk department
> might be horrified at this, but I'd simply isolate the cable at both ends
> (or better, ground the cable at the building entrance at ONE END ONLY),
> and power the radios via 5 KV rated isolation transformers. Undoubtedly,
> that coax has been there for "a while" and hasn't caused a fire yet.
> However there is that risk, so it's your responsibility, not mine.
>
> Old guy

Thanks, checked the cable loss again today and you are right it is more
than -64db , more like -75db. I think this should still work. Im
thinking a 1:1 UNUN isolation transformer on each end of the cable
should give me the ground loop isolation I need. You did bring my
attention to details I had previously not noticed. The unused pairs on
of 2 50 pair and 1 25 pair cable are attached to ground on BOTH ends.
It apparently has been this way for years.
The buildings are also bonded together with a #4 bare copper ground
wire buried along with the cables.

"Moe Trin" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> On Thu, 6 Jul 2006, in the Usenet newsgroup comp.os.linux.networking, in
> article
> <Fnhrg.84136$(E-Mail Removed)>, Jimmie D wrote:
>
>>I want to extend my LAN to a facility about a mile away.
>
> And the company doesn't want to spend any money.

They consider it a personal convience we dont really need.


>
>>I cant run anthing new but there are twisted pairs and a 1 inch Heliax
>>cable.
>
> I'm assuming the stuff is buried, rather than running on catenary
> (overhead
> pole line). Have you discussed this with the company's insurance carrier?
> If you're somewhere in the US South, you should be aware of lightning
> problems, never mind ground loops.
>
>>I have tried using the existing coax in a thicknet connection but this
>>didnt work.
>
> What happened? Did the transceivers catch fire?
I never really verified that the thicknet equipment was working as normal.
I do like this better if I can get it to work. There may have been some
cockpit errors when trying to get the thicknet to work.

The original 10Base5
> Ethernet was limited to 500 meters of a cable _similar_to_ RG-8/U but
> allowed two repeaters between the furthest host, and a point-to-point
> link which gave a total distance of 2500 meters (8200 feet). That total
> length (at a Vp of about 0.7c) sets the time delay called a Collision
> Domain. The maximum cable attenuation between a pair of transceivers was
> 8.5 dB at 10 MHz _AND_ 6.0 dB at 5 MHz. I don't have a spec sheet handy
> (and am to lazy to grab it from Andrew), but 1 inch Heliax is probably
> going to have a loss of roughly 1.2 dB per thousand feet at 10 MHz, and
> 0.9 dB per thousand feet at 5 MHz - so a mile is going to be within spec.
> As I recall, the Vp is about 0.7 as well, so you're within collision
> domain and loss specs - it should have worked assuming you can find the
> appropriate vampire tap ;-) (Actually, transition to a 3 meter stub of
> Belden 89880 or equal and put the transceivers and terminators on that.)
>
>>I was thinking about using some wireless equipment we had in a novel way.
>>Instead of using it with an antenna I intend to make direct connections
>>via the cable. The cable has about 64 db of loss at 2.4 Ghz. Im not sure
>>of the sensitivity of the receivers but I measured the tx out at about
>>+20
>>Dbm.
>
> I take it that you don't have clear line of site and a couple of good
> sized
> dishes to run those radios wireless as they were intended.
>
> I'll take your word for the cable loss although that seems a bit low. The
> receiver sensitivity is a function of the bandwidth. Point your news tool
> at alt.internet.wireless (or better, hit google.groups.com and search
> that newsgroup) and you'll see posts from Jeff Liebermann - and he answers
> this type of question on a regular basis.
>
> Receiver Sensitivity for DI-624 (and most other similar radios):
> * 54Mbps OFDM, 10% PER, -68dBm)
> * 48Mbps OFDM, 10% PER, -68dBm)
> * 36Mbps OFDM, 10% PER, -75dBm)
> * 24Mbps OFDM, 10% PER, -79dBm)
> * 18Mbps OFDM, 10% PER, -82dBm)
> * 12Mbps OFDM, 10% PER, -84dBm)
> * 11Mbps CCK, 8% PER, -82dBm)
>
> So, +20 dBm in, cable loss of 64 dB - there should be -44 dBm (probably
> less) coming out the hose at the far end. The 'PER' term is Packet Error
> Rate - meaning that signal level is about the minimum usable at that
> speed.
>
>>ITs not just being able to hook it up and try it. The connectors for the
>>Heliax are very expensive and I have to know with some reasonable
>>certainty
>>if it will work before I buy them
>
> Again - alt.internet.wireless. You should already have type N connectors
> on the end of the cable (and be pushing nitrogen or at least dry air into
> the cable to prevent water intrusion), and you may have noticed that
> Heliax
> isn't very flexible. Transition from the Heliax to something more flexible
> yet still low loss (example LMR400) for the connection to the sets.
>
> I don't see an obvious reason this wouldn't work, assuming you have an
> undamaged cable and the connectors were fabricated/attached correctly
> and tested at 2.4 GHz. Test the cable loss. If the cable hasn't been
> pressurized, and has been outside for a year or two, you probably have
> a corrosion problem internally, and in that case your screwed.
>
> Now, the local fire marshal and your insurance carrier's risk department
> might be horrified at this, but I'd simply isolate the cable at both ends
> (or better, ground the cable at the building entrance at ONE END ONLY),
> and power the radios via 5 KV rated isolation transformers. Undoubtedly,
> that coax has been there for "a while" and hasn't caused a fire yet.
> However there is that risk, so it's your responsibility, not mine.
>
> Old guy

On 7 Jul 2006, in the Usenet newsgroup comp.os.linux.networking, in article
<(E-Mail Removed) .com>, (E-Mail Removed)
wrote:

>Thanks, checked the cable loss again today and you are right it is more
>than -64db , more like -75db. I think this should still work.

That's a bit more like it. Yeah, you have 13 dB of fade margin for the
wide-band link (which would be marginal over the air - reliability of
roughly 92%), up to 29 dB for the 12Mbps OFDM link.

>Im thinking a 1:1 UNUN isolation transformer on each end of the cable
>should give me the ground loop isolation I need.

That ought to be fun. If I were really concerned, I'd take the Ethernet
outputs and run them through a media converter to fiber and back. The
normal isolation in Ethernet ought to be enough to avoid ground loops
provided the link power source (AC line) is isolated on the "ungrounded"
end.

>You did bring my attention to details I had previously not noticed. The
>unused pairs on of 2 50 pair and 1 25 pair cable are attached to ground
>on BOTH ends. It apparently has been this way for years.

Well, if no one has noticed (including Thor, the god of thunder and
lightning), you're either lucky, or there won't likely be a problem.

>The buildings are also bonded together with a #4 bare copper ground
>wire buried along with the cables.

Depending how many years - I wonder how bad the corrosion is. Lessee,
a mile of #4AWG is about 1.25 ohms. If it's un-insulated and in contact
with the ground throughout, that's probably the equivalent of one heck
of a good ground rod. I normally wouldn't like to use that, depending
on real ground rods to dissipate any lightning induced currents. Note
that "ground" is not "ground". If you installed two 30 foot long rods
about ten feet apart, and then took your trusty ohm meter and measured
the resistance between the two, you'd see something up to 25 ohms
(obviously depending on your ground conductivity).

Old guy

On Sat, 8 Jul 2006, in the Usenet newsgroup comp.os.linux.networking, in article
<GeNrg.44318$(E-Mail Removed)>, Jimmie D wrote:
>
>"Moe Trin" <(E-Mail Removed)> wrote

>> What happened? Did the transceivers catch fire?

>I never really verified that the thicknet equipment was working as normal.
>I do like this better if I can get it to work. There may have been some
>cockpit errors when trying to get the thicknet to work.

Not many people still using thicknet. The advantage is that there is
isolation between cable and AUI connection - the spec requires an isolation
impedance of greater than 250KOhm measured at 60Hz, and a minimum breakdown
voltage of 270 Vrms. In practice, it's often a MegOhm, and 700 Vrms or
better.

The _disadvantage_ is that thicknet is 10 MHz half duplex, and you'd have
problems pushing more than 4 Megabit in a party-line mode (though you
might be able to peak out at 9.6 Megabit for short periods sending nearly
back-to-back packets till you run out of TCP window size and need that ACK
to send more).

Thicknet depends on "decent" VSWRs on the cable. It's not specified as such,
but the design maximum was 1.08:1. You can get that much with a poorly
assembled connector, or a ding in the coax. A question would be what was
the error numbers shown in /sbin/ifconfig at both ends. Note particularly
any 'Carrier' errors which is a sign of cable problems. 'Frame' errors are
probably either noise or signal level problems.

Looking at your revised 2.4 GHz cable loss, I'd still prefer the wireless
solution on a bandwidth issue. Even if you don't need the bandwidth now,
Murphy says you will need it later.

Old guy

"Moe Trin" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> On Sat, 8 Jul 2006, in the Usenet newsgroup comp.os.linux.networking, in
> article
> <GeNrg.44318$(E-Mail Removed)>, Jimmie D wrote:
>>
>>"Moe Trin" <(E-Mail Removed)> wrote
>
>>> What happened? Did the transceivers catch fire?
>
>>I never really verified that the thicknet equipment was working as normal.
>>I do like this better if I can get it to work. There may have been some
>>cockpit errors when trying to get the thicknet to work.
>
> Not many people still using thicknet. The advantage is that there is
> isolation between cable and AUI connection - the spec requires an
> isolation
> impedance of greater than 250KOhm measured at 60Hz, and a minimum
> breakdown
> voltage of 270 Vrms. In practice, it's often a MegOhm, and 700 Vrms or
> better.
>
> The _disadvantage_ is that thicknet is 10 MHz half duplex, and you'd have
> problems pushing more than 4 Megabit in a party-line mode (though you
> might be able to peak out at 9.6 Megabit for short periods sending nearly
> back-to-back packets till you run out of TCP window size and need that ACK
> to send more).
>
> Thicknet depends on "decent" VSWRs on the cable. It's not specified as
> such,
> but the design maximum was 1.08:1. You can get that much with a poorly
> assembled connector, or a ding in the coax. A question would be what was
> the error numbers shown in /sbin/ifconfig at both ends. Note particularly
> any 'Carrier' errors which is a sign of cable problems. 'Frame' errors are
> probably either noise or signal level problems.
>
> Looking at your revised 2.4 GHz cable loss, I'd still prefer the wireless
> solution on a bandwidth issue. Even if you don't need the bandwidth now,
> Murphy says you will need it later.
>
> Old guy

I tried to get the Thicknet stuff to work today and couldnt so I think that
was the origional problem. I t had been used to send data about 300 ft and
was replaced with CAT 5 a couple of years ago. We have 3 computers out at
the site that has to be brought in for software upgrades and thats a real
pain. I Megged the cable out today and got infinety at 1000 volts, its even
larger than 1 inch and its 75 ohm cable and there are 2 coax cables. I also
didcoverd that much cable can hold quite a charge, OUCH.

On Mon, 10 Jul 2006, in the Usenet newsgroup comp.os.linux.networking, in
article <kXmsg.113013$(E-Mail Removed)>, Jimmie D wrote:

>I tried to get the Thicknet stuff to work today and couldnt so I think that
>was the origional problem. I t had been used to send data about 300 ft and
>was replaced with CAT 5 a couple of years ago.

I thought you said that the cable was a mile long? 300 feet??? What
was the error indication other than it didn't work. What did /sbin/ifconfig
show?

>I Megged the cable out today and got infinety at 1000 volts, its even
>larger than 1 inch and its 75 ohm cable and there are 2 coax cables. I also
>didcoverd that much cable can hold quite a charge, OUCH.

Given that the cable probably has a nominal capacitance of 30 picofarads
per foot for 50 Ohm coax (roughly 20 pF for 75 Ohm coax), that's to be
expected. Are you still talking about the Heliax? I know that they build
that stuff in 75 as well as 50 Ohm, but it's generally used for cable
television distribution. Both Ethernet and the wireless radios want 50
Ohm, not 75. I've heard of people using RG-11/U in place of the specified
50 Ohm cable similar to RG-8/U for thicknet, but they also terminated the
line with 75 Ohm loads rather than the specified 49.9 Ohms, and they only
had two systems attached. The vampire tap had to be assembled carefully
to avoid crushing the non-standard sized cable while making sure that the
stinger is in solid contact with the smaller center conductor. As I have
mentioned, 10Base5 Ethernet is fairly picky about VSWR, with the design
requirements of less than 1.08:1.

Using 75 Ohm at 2.4 GHz isn't as big of a crime, as that is a 1.5:1 VSWR,
and many of the common antennas are at least that bad. It's going to cost
an additional decibel (or so) of loss, but you probably won't miss that.
The 14 dB return loss shouldn't damage the transmitter.

Old guy