Wireless network range calculations

Can I think aloud and ask you knowledgeable people your advice?

A couple of years ago I installed a wireless ethernet link using a pair of
Cisco BR342 with 100ft antenna cables and 13.5dB Yagi antennae at each end -
configured as a point-to-point bridge. The required range was 450 metres,
and performance was very good. One antenna is 2 metres above a flat roof,
the other at roof level on a wall bracket - but generally 7 to 10 metres
above the intervening ground level.

A customer now wants a similar solution for 150metre range, but much
cheaper. So I installed a pair of D-Link DWL900AP+ with 30ft antenna cables
and 8.5dB patch antennae at each end. The antennae are 5 metres above
ground level.

The D-Link product gives no figures for receiver sensitivity, just 15dBm
(i.e. 30mW) power output. There is a suggestion that the product will
achieve a range of 400 metres, but no indication of which antenna would be
required to achieve this.

With the D-Link configuration I can achieve reliable communication over 60
metres, but at 150 metres communication is virtually nil - pinging one unit
from the other gives about a 1% success rate. Curiously, receiver in
location 1 could reliably receive packets from transmitter at location 2,
but receiver 2 generally did not receive anything from transmitter 1 - but
with the same antennae and feeders I would have thought the perfomance would
bave been symmetrical. This was repeatable with a different pair of
DWL900AP+ so I'm very confused..

I have therefore replaced the antennae with 18dB gain products at each end -
and the link is now 100% OK.

Is this simply because the D-Link product is just much less sensitive?

Cisco provide a nice spreadsheet range calculator. I have tried my own sums
from first principles, but get figures which suggest a range only about a
quarter of that indicated by the Cisco calculator - so I my reasoning must
be wrong somewhere. Can anybody suggest a basis for a reliable calculation?

Has anybody any experience with the D-Link product?

Further - D-Link suggest a maximum data rate of 22 mbits/sec. Bench tests
over about 5 metre range achieved about 300kbytes/sec - well under what I
would have expected. I can't remember what we achieved with the Cisco units
but I think it was something like 800 kbytes/sec, which would be consistent
with 11 mbits/sec over the radio link, with some management overhead.

All (polite) comments welcome ...

Thanks

--
Graham

"Graham" <(E-Mail Removed)> wrote in message
news:3f0706c7$0$11386$(E-Mail Removed) ...
> Can I think aloud and ask you knowledgeable people your advice?
>
> A couple of years ago I installed a wireless ethernet link using a pair of
> Cisco BR342 with 100ft antenna cables and 13.5dB Yagi antennae at each
end -
> configured as a point-to-point bridge. The required range was 450 metres,
> and performance was very good. One antenna is 2 metres above a flat roof,
> the other at roof level on a wall bracket - but generally 7 to 10 metres
> above the intervening ground level.
>
> A customer now wants a similar solution for 150metre range, but much
> cheaper. So I installed a pair of D-Link DWL900AP+ with 30ft antenna
cables
> and 8.5dB patch antennae at each end. The antennae are 5 metres above
> ground level.
>
> The D-Link product gives no figures for receiver sensitivity, just 15dBm
> (i.e. 30mW) power output. There is a suggestion that the product will
> achieve a range of 400 metres, but no indication of which antenna would be
> required to achieve this.
>
> With the D-Link configuration I can achieve reliable communication over 60
> metres, but at 150 metres communication is virtually nil - pinging one
unit
> from the other gives about a 1% success rate. Curiously, receiver in
> location 1 could reliably receive packets from transmitter at location 2,
> but receiver 2 generally did not receive anything from transmitter 1 - but
> with the same antennae and feeders I would have thought the perfomance
would
> bave been symmetrical. This was repeatable with a different pair of
> DWL900AP+ so I'm very confused..

This last part above could be an indication of what is going on. You state
that RX #2 could not receive, from #1, yet #1 could receive from #2. Being
that the receiver sensitivity of the product is unknown, I would expect it
to receive at least -90dBm for adequate performance, more likely the
sensitivity is around -98dBm. But that is not important, what is important
is that RX#2 cannot receive, yet RX#1 can. This can be that the position of
the RX#2 is in what could be considered a high noise floor area, hence
requiring more power to receive than RX#1. The fact that you were able to
use higher gain antennas and make the link work kind of supports this as
well. Maybe if you have time, and if it is not to difficult, move RX#2, to a
different location and see if using the original antennas work.

>
> I have therefore replaced the antennae with 18dB gain products at each
end -
> and the link is now 100% OK.
>
> Is this simply because the D-Link product is just much less sensitive?

You mentioned cost eariler, so yes the receiver sense may be a lot weaker
than the Cisco product.
>
> Cisco provide a nice spreadsheet range calculator. I have tried my own
sums
> from first principles, but get figures which suggest a range only about a
> quarter of that indicated by the Cisco calculator - so I my reasoning must
> be wrong somewhere. Can anybody suggest a basis for a reliable
calculation?

To really do that you would really need to know what the RX sense was. You
can figure your path loss though, as you were given the TX power.
The path loss would be something like this.

Start with your Tx power subtract your antenna cable line loss, add your
antenna gain, subtract your air interface medium loss, add the gain of teh
RX antenna, and subtract the loss of the receiving cable from antenna to
receiver.
That will tell you what you will see on the rx end.

Check out this page here.
http://www.swisswireless.org/wlan_calc_en.html

>
> Has anybody any experience with the D-Link product?
>
> Further - D-Link suggest a maximum data rate of 22 mbits/sec. Bench tests
> over about 5 metre range achieved about 300kbytes/sec - well under what I
> would have expected. I can't remember what we achieved with the Cisco
units
> but I think it was something like 800 kbytes/sec, which would be
consistent
> with 11 mbits/sec over the radio link, with some management overhead.
>
> All (polite) comments welcome ...
>
> Thanks
>
> --
> Graham
>
>
>

On Sun, 6 Jul 2003, Allan Butler wrote:
> Graham wrote:
> > Can I think aloud and ask you knowledgeable people your advice?
> >
> > A couple of years ago I installed a wireless ethernet link using a pair of
> > Cisco BR342 with 100ft antenna cables and 13.5dB Yagi antennae at each end
> > -
> > configured as a point-to-point bridge. The required range was 450 metres,
> > and performance was very good. One antenna is 2 metres above a flat roof,
> > the other at roof level on a wall bracket - but generally 7 to 10 metres
> > above the intervening ground level.
> >
> > A customer now wants a similar solution for 150metre range, but much
> > cheaper. So I installed a pair of D-Link DWL900AP+ with 30ft antenna
> > cables
> > and 8.5dB patch antennae at each end. The antennae are 5 metres above
> > ground level.
> >
> > The D-Link product gives no figures for receiver sensitivity, just 15dBm
> > (i.e. 30mW) power output. There is a suggestion that the product will
> > achieve a range of 400 metres, but no indication of which antenna would be
> > required to achieve this.

You are correct - no RX sensitivity stated. However, one can INFER a minimum
necessary sensitivity based on a "network" of two such cards because one knows
the transmitter power and maximum distance claimed. Using that, I note that
these products seem to fit the standard sensitivity of -78dB for 11MBps (at
least the DWL-520's and DWL-650's did, which are also +15dBm [or -15dB]).
Note that the sensitivity computed above is without regard to "real world"
factors such as atmospheric absorption (temperature, humidity, dust, etc.) and
such - it's purely the reverse-square of the distance formula.

Also, if your access point has a DIPOLE antenna, then your patch antenna isn't
8.5dB in gain above that. Patch antenna gain is measured usually in dBi, of
which a dipole is 2.15dBi, not zero. Yagi antenna gain is often measured in
dBd (and a dipole is 0 dBd by definition).

> > With the D-Link configuration I can achieve reliable communication over 60
> > metres, but at 150 metres communication is virtually nil - pinging one
> > unit
> > from the other gives about a 1% success rate. Curiously, receiver in
> > location 1 could reliably receive packets from transmitter at location 2,
> > but receiver 2 generally did not receive anything from transmitter 1 - but
> > with the same antennae and feeders I would have thought the perfomance
> > would
> > bave been symmetrical. This was repeatable with a different pair of
> > DWL900AP+ so I'm very confused..
>
> It is possible one of the antennas is not operating properly.
> it could be that the internal connections are broken and the antenna is
> exhibiting a high VSWR. It would have been interesting to try this
> particular link with just swapping the two antennas.

Or other equipment that is known to work.

> As someone else suggested though, there could be some excessive
> noise getting into the one location and with the higher gain antenna you
> are getting around the noise problem.

Always a possibility.

> What type of coax cable are you using to make the connection to
> the antenna? It is very important at these frequencies to have a high
> quality coax due to signal loss.

Even 1/2" hardline (e.g Andrew heliax) may lose about 5.5dB over 100' at 2.45GHz.

> > I have therefore replaced the antennae with 18dB gain products at each end
> > - and the link is now 100% OK.
> >
> > Is this simply because the D-Link product is just much less sensitive?
> >
> > Cisco provide a nice spreadsheet range calculator. I have tried my own
> > sums from first principles, but get figures which suggest a range only
> > about a quarter of that indicated by the Cisco calculator - so I my
> > reasoning must
> > be wrong somewhere. Can anybody suggest a basis for a reliable
> > calculation?
> >
> > Has anybody any experience with the D-Link product?
> >
> > Further - D-Link suggest a maximum data rate of 22 mbits/sec. Bench tests
> > over about 5 metre range achieved about 300kbytes/sec - well under what I
> > would have expected. I can't remember what we achieved with the Cisco
> > units but I think it was something like 800 kbytes/sec, which would be
> > consistent with 11 mbits/sec over the radio link, with some management
> > overhead.
> >
> > All (polite) comments welcome ...
> >
> > Thanks

"Allan Butler" <(E-Mail Removed)> wrote in message
news:2CKNa.113582$R73.12819@sccrnsc04...
> Graham wrote:

[snip]

> > have been symmetrical. This was repeatable with a different pair of
> > DWL900AP+ so I'm very confused..
>
> It is possible one of the antennas is not operating properly.
> it could be that the internal connections are broken and the antenna is
> exhibiting a high VSWR. It would have been interesting to try this
> particular link with just swapping the two antennas.

Surely this would affect transmit and receive equally - it would not account
for the apparent difference in performance form one end to the other ...

> As someone else suggested though, there could be some excessive
> noise getting into the one location and with the higher gain antenna you
> are getting around the noise problem.
>
> What type of coax cable are you using to make the connection to
> the antenna? It is very important at these frequencies to have a high
> quality coax due to signal loss.

I bought the D-Link ANT24-CB09N which appears to be about half a dB per
metre...

--
Graham

"D. Stussy" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed) g...
> On Sun, 6 Jul 2003, Allan Butler wrote:
> > Graham wrote:
> > > Can I think aloud and ask you knowledgeable people your advice?
> > >
> > > A couple of years ago I installed a wireless ethernet link using a
pair of
> > > Cisco BR342 with 100ft antenna cables and 13.5dB Yagi antennae at each
end
> > > -
> > > configured as a point-to-point bridge. The required range was 450
metres,
> > > and performance was very good. One antenna is 2 metres above a flat
roof,
> > > the other at roof level on a wall bracket - but generally 7 to 10
metres
> > > above the intervening ground level.
> > >
> > > A customer now wants a similar solution for 150metre range, but much
> > > cheaper. So I installed a pair of D-Link DWL900AP+ with 30ft antenna
> > > cables
> > > and 8.5dB patch antennae at each end. The antennae are 5 metres above
> > > ground level.
> > >
> > > The D-Link product gives no figures for receiver sensitivity, just
15dBm
> > > (i.e. 30mW) power output. There is a suggestion that the product will
> > > achieve a range of 400 metres, but no indication of which antenna
would be
> > > required to achieve this.
>
> You are correct - no RX sensitivity stated. However, one can INFER a
minimum
> necessary sensitivity based on a "network" of two such cards because one
knows
> the transmitter power and maximum distance claimed. Using that, I note
that
> these products seem to fit the standard sensitivity of -78dB for 11MBps
(at
> least the DWL-520's and DWL-650's did, which are also +15dBm [or -15dB]).
> Note that the sensitivity computed above is without regard to "real world"
> factors such as atmospheric absorption (temperature, humidity, dust, etc.)
and
> such - it's purely the reverse-square of the distance formula.
>
> Also, if your access point has a DIPOLE antenna, then your patch antenna
isn't
> 8.5dB in gain above that. Patch antenna gain is measured usually in dBi,
of
> which a dipole is 2.15dBi, not zero. Yagi antenna gain is often measured
in
> dBd (and a dipole is 0 dBd by definition).

Interesting. I followed the calculations at
http://www.swisswireless.org/wlan_calc_en.html
and concluded that for the link to fail as described the receiver
sensitivity would have to be -78.5 dB (but this allowed a 10dB margin, so
given that the link actually fails the receiver sensitivity appears to be
more like -68.5 dB. By comparison the Cisco AIR-BR350 receiver is -85dB for
11Mbits/sec operation.

--
Graham

On Mon, 7 Jul 2003, Graham wrote:
> "D. Stussy" <(E-Mail Removed)> wrote in message
> news:(E-Mail Removed) g...
> > On Sun, 6 Jul 2003, Allan Butler wrote:
> > > Graham wrote:
> > > > Can I think aloud and ask you knowledgeable people your advice?
> > > >
> > > > A couple of years ago I installed a wireless ethernet link using a
> pair of
> > > > Cisco BR342 with 100ft antenna cables and 13.5dB Yagi antennae at each
> end
> > > > -
> > > > configured as a point-to-point bridge. The required range was 450
> metres,
> > > > and performance was very good. One antenna is 2 metres above a flat
> roof,
> > > > the other at roof level on a wall bracket - but generally 7 to 10
> metres
> > > > above the intervening ground level.
> > > >
> > > > A customer now wants a similar solution for 150metre range, but much
> > > > cheaper. So I installed a pair of D-Link DWL900AP+ with 30ft antenna
> > > > cables
> > > > and 8.5dB patch antennae at each end. The antennae are 5 metres above
> > > > ground level.
> > > >
> > > > The D-Link product gives no figures for receiver sensitivity, just
> 15dBm
> > > > (i.e. 30mW) power output. There is a suggestion that the product will
> > > > achieve a range of 400 metres, but no indication of which antenna
> would be
> > > > required to achieve this.
> >
> > You are correct - no RX sensitivity stated. However, one can INFER a
> minimum
> > necessary sensitivity based on a "network" of two such cards because one
> knows
> > the transmitter power and maximum distance claimed. Using that, I note
> that
> > these products seem to fit the standard sensitivity of -78dB for 11MBps
> (at
> > least the DWL-520's and DWL-650's did, which are also +15dBm [or -15dB]).
> > Note that the sensitivity computed above is without regard to "real world"
> > factors such as atmospheric absorption (temperature, humidity, dust, etc.)
> and
> > such - it's purely the reverse-square of the distance formula.
> >
> > Also, if your access point has a DIPOLE antenna, then your patch antenna
> isn't
> > 8.5dB in gain above that. Patch antenna gain is measured usually in dBi,
> of
> > which a dipole is 2.15dBi, not zero. Yagi antenna gain is often measured
> in
> > dBd (and a dipole is 0 dBd by definition).
>
> Interesting. I followed the calculations at
> http://www.swisswireless.org/wlan_calc_en.html
> and concluded that for the link to fail as described the receiver
> sensitivity would have to be -78.5 dB (but this allowed a 10dB margin, so
> given that the link actually fails the receiver sensitivity appears to be
> more like -68.5 dB. By comparison the Cisco AIR-BR350 receiver is -85dB for
> 11Mbits/sec operation.

You got -78.5dB. I got -78dB. It's not like we're NASA/JPL trying to
communicate out past Saturn, so isn't that good enough? :-)

I doubt that the Cisco has -85dB for 11Mbit/s, but instead is between 2Mbit/s
and 5.5Mbit/sec for that RX sensitivity. Maybe you accidentally factored in
antenna gain?

[snip]

> > Interesting. I followed the calculations at
> > http://www.swisswireless.org/wlan_calc_en.html
> > and concluded that for the link to fail as described the receiver
> > sensitivity would have to be -78.5 dB (but this allowed a 10dB margin,
so
> > given that the link actually fails the receiver sensitivity appears to
be
> > more like -68.5 dB. By comparison the Cisco AIR-BR350 receiver is -85dB
for
> > 11Mbits/sec operation.
>
> You got -78.5dB. I got -78dB. It's not like we're NASA/JPL trying to
> communicate out past Saturn, so isn't that good enough? :-)

My point being that -78.5 dB assumes a 10dB margin, which ought to ensure
reliable communications (approaching 100% ??), so the fact that I see
typically 1% suuccess would suggest that all the 10dB margin has been used
up, and the actual receiver sensitivity is more likely to be -68dB ...

> I doubt that the Cisco has -85dB for 11Mbit/s, but instead is between
> 2Mbit/s and 5.5Mbit/sec for that RX sensitivity. Maybe you
> accidentally factored in antenna gain?

Er, no. See:

http://www.cisco.com/en/US/products/..._data_sheet091
86a008008883c.html

where it is clear that:

Receive Sensitivity
1 Mbps: -94 dBm
2 Mbps: -91 dBm
5.5 Mbps: -89 dBm
11 Mbps: -85 dBm


--
Graham

On Mon, 7 Jul 2003 19:04:00 +0100, Graham <(E-Mail Removed)> wrote:
>
> Interesting. I followed the calculations at
> http://www.swisswireless.org/wlan_calc_en.html
> and concluded that for the link to fail as described the receiver
> sensitivity would have to be -78.5 dB (but this allowed a 10dB margin, so
> given that the link actually fails the receiver sensitivity appears to be
> more like -68.5 dB. By comparison the Cisco AIR-BR350 receiver is -85dB for
> 11Mbits/sec operation.

Well unless it's defective the D-link couldn't possibly be that bad so the
problem is probably elseware.

Have you:

Set the D-Links to use the external antenna port?

Tried swapping out the cables?

Eliminated noise as a possible culprit?

These two antenna really do have LOS right? You're not trying to shoot
through something?

--
Ray