Adding an amp ....

snipped-for-privacy@lsi.com (Gordon Montgomery) hath wroth:

Not much. First, you will need to add *TWO* amplifiers, one at each end. Unless you are trying to deal with the loss of a long run of coax cable, one amp will probably not be sufficient.

The one detail that I don't know is the coax cable between the WHR-HP-G54 and the 24dBi dish. What are the coax types and lengths?

The effects of amplifiers can easily be calculated. See:

TX power +25.8 dBm (380mw from FCC data) TX coax loss ? dB TX ant gain +24 dBi Distance 9 miles RX ant gain +24 dBi RX coax loss ? dB RX sens -84 dBm (at 12Mbits/sec connection) Fade margin (aim for at least 20dB)

You to supply the missing coax numbers. Then plug the numbers into: while watching the fade margin. Adjust the tx power until it's about

20dB and you have the TX power necessary to maintain a minimal connection.

However, a bit of warning. The numbers generated above are very ideal. They will not get better, only worse. Antennas never seem to have the gain specified. The transmitters never seem to put out as much power as claimed. Receiver sensitivity is affected by just about everything. Interference and noise will seriously ruin a link. There's also Fresnel Zone clearance that needs to be considered:

For long links, I do a bit of additional testing. Fire up Kismet and see how many other systems are on the channel and the adjacent two channels. If there are any really strong signals, find another channel or move the antenna.

Thanks for the reply. I was worried about over driving the radios. As for the coax, both radios are mounted next to their antenna, connected with a pigtail. I don't see any writing, but the coax is at most, 5mm in diameter and probably about 60cm long. The antenna leads appear to be RG-8/U and about 50cm long. I was going to try just one amp. I ran this link for 5 years with some SMC2682W bridges with an amp on only one side, and had no problems until one of the bridges started locking up and rebooting for no reason.

Thanks again.

Gordon Montgomery Living Scriptures, Inc snipped-for-privacy@lsi.com (anti spam - replace lsi with livingscriptures) (801) 627-2000

snipped-for-privacy@lsi.com (Gordon Montgomery) hath wroth:

That's a real worry. Most amps have AGC (automagic gain control) which will adjust the levels. However, those are made for perhaps

50mw (+17dBm) and LOWER to compensate for coax cable losses. Higher power is always problematic. Fortunatly, the power is adjustable on the WHR-HP-G54 so you probably will not need an attenuator or long mess of lossy coax to get the levels down to where they need to be.

Is the small coax rather stiff? I'll use worst case and guess that the small stuff is RG58a/u garbage coax and that the combined coaxes and connectors have a total loss of about 4dB.

Grinding the numbers, TX power +25.8 dBm (380mw from FCC data) TX coax loss 4 dB TX ant gain +24 dBi Distance 9 miles RX ant gain +24 dBi RX coax loss 4 dB RX sens -84 dBm (at 12Mbits/sec connection) which yields a fade margin of 24.5dB. Anything over 20dB is considered usable. The link should work as it stands, without a power amplifier. Raising the tx power with a 1 watt power amplifier (maximum legal) will only increase the fade margin to about 28.7dB, which is better, but not sufficiently better than the current 25.8dB to justify the expense.

My guess(tm) is that something else is happening and that you're making some assumptions. It might be that the WHR-HP-54G is operating at reduced power output and needs to be adjusted. It might be some settings in the routers. It might be water in the coax cables. It might be a badly assembled antenna (I've seen this all too often with dish antennas). However, my guess(tm) is that it's interference along the path or some obstruction in the Fresnel Zone.

I recently troubleshot a 2 mile link that had worked just fine for about 2 years using old 802.11b radios. When I upgraded them to

802.11g, the link kept going up and down. After my usual bad guesses, I eventually disabled all the wiz bang turbo and super modulation features, disabled 802.11b compatibility, and locked the wireless speed to 12Mbits/sec. It then worked prefectly, even with over half the antenna covered carbon foam absorber. If the old radios worked, but the new radios do not, I would look for radio problems, not path problems.

Hint: The more numbers you supply, the easier it is to troubleshoot a problem.

I doubt it is RG58, isn't that what TV cables are usually made from? I dug up the receipt for the pigtails, and they say "195 series cable".

I've never noticed the power setting before in the setup screens. It looks like it just gives me percentages, and no where does it explain what the actual transmit power is. I'll have to do some digging over the weekend I suppose. ( Again, these are still running the stock Buffalo firmware. )

The new link does work, but if I run a continuous ping, I'll get 20% to

50% packet loss, sometimes. Sometimes, it is great. That is why I am assuming weak signal that is getting trod on sometimes. The two points actually have a small valley inbetween them, so I don't believe Fresnel Zone encroachment is a problem.

I believe I have disabled all the whiz bang features, but I'll double check those as well. I have these running in WDS only mode, because I just need the bridging component.

Thanks for your helpfulness. I'll be messing with random settings all week end. ( what...... isn't that how everyone trouble shoots?!? :)

Gordon Montgomery Living Scriptures, Inc snipped-for-privacy@lsi.com (anti spam - replace lsi with livingscriptures) (801) 627-2000

RG8 52 ohm cable is about half inch diameter, RG-58 is about quarter inch.

TV cables use RG59, 75 ohm, RG58 is 52 ohm.

I haven't played with very many consumer grade WiFi units, much less with third-party firmware, but from what I have read, you can push the power output levels up to perhaps 200 mW, but the Linksys users website says pushing it over 100 mW will eventually take out the RF power amplifier.

Adding a "booster", may it be a one-way transmitting power amplifier or a bi-directional amplifier, *generally* isn't the way to go if you use multipoint-to-point, as you run into power output limitations, but in your case for point-to-point there would be an advantage. On the receiver side, they typically have 13 dB of gain..and that means the signal as well as interference will get a 20x boost, so effectively your signal to noise ratio will be the same. But that 13 dB gain is misleading as you have to figure in the additional patch cable losses.

At nine miles (and assuming you have a clear Fresnel path), you could use a diversity like I use for my ten mile plus shots with a 2.4 GHz and 5.8 GHz link to maintain carrier class reliability.

snipped-for-privacy@gmail.com hath wroth:

That's true for the WRT54G v1 thru v4. 100mw is safe. 250mw is maximum. Everything in between is subject to anecdotal horror stories, including one of mine. One of my customers fried his WRT54Gv4 at 250mw, probably by cooking the PA.

That's not true for the Buffalo WHR-HP-G54, which is FCC type certified at 380mw for 802.11g and 280mw for 802.11b. See previous discussion at:

Ugly spectrum analyzer photos. Note how dirty the waveform looks at

250mw which is another reason to not turn it up to maximum.

Incidentally, the latest 2wire wireless routers at type certified at

400mw.

snipped-for-privacy@lsi.com (Gordon Montgomery) hath wroth:

That's LMR-195. 0.62dB/meter loss at 2.4GHz. Drop my 4dB coax losses down to about 2dB total. That will yield a fade margin of 28.5dB for the 9 mile link. If here are no obstructions or water in the coax, it should work just fine, even and higher speeds. You don't need an amplifier.

I have no idea. I never use the stock Buffalo firmware. As soon as I get a case of them in the office, I do production line BIOS transplants with DD-WRT v23 SP3.

For a clue of what's in DD-WRT, see the online emulator at:

Note that the default power out is 28mw. Methinks you have some room to work with on the power output.

Show me. Download fping 2.17 from:

and run: fping ip_address_of_router_at_other_end -c for about 100 lines and email the results to me at jeffl @ cruzio dot com. I wanna see the numbers. Make sure there's no other traffic going across the wireless when running the test. (I'm too lazy to write another disertation on how to interpret wireless ping results).

It's almost for sure getting stepped on. What you probably don't realize is that cranking up the power isn't going to help. Once another signal gets in between the packets (inter-symbol interference), you're data is trashed.

Also, it doesn't have to be wi-fi interference. See shopping list at:

See if any of these are possible culprits.

Perfect. No Fresnel Zone problems assuming the valley is deep enough. For 9 miles, you need at least 68 ft radius clearance around the line of sight at mid span. Do you have this much clearance?

WDS has some overhead, but not much. When you load DD-WRT, you might look at the "client bridged" mode.

Nope, at least not for me. Numbers and calcs first. Then I tinker. If I have time, I RTFM.

Tinker settings:

CTS protection mode: disabled Frame Burst: disabled Preamble: short (only if you have 802.11b compatibility mode off) Afterburner: off WMM (QoS) support: off

snipped-for-privacy@gmail.com hath wroth:

Nope. The impedance of the cable is dependent on the ratio of the OD of the center conductor, the ID of the shield, and the type of dielectric. Both 50 and 75 ohm cables come in all manner of different diameters ranging from 0.085" semi-rigid, to monsterous 4.5" heliax used for broadcast xmitters.

Sorta. Most TV cable is RG6/u. RG59/u is suitable for short patch cables and not much more due to higher losses than RG6/u. For less loss, but a larger diameter, there's RG11/u. Lots more types depending on size and application.

For 2.4Ghz, the LMR series of cables are quite popular because of low loss, decent connectors, and high quality.

Most common is LMR-400 which is .405 OD.

Some dish and panel manufacturers use cheezy RG8/u for pigtails between the feed and the coax connector. This is possible because although the RG8/u is rather lossy, there isn't very much coax involved, so the losses are minimal. I would prefer something that doesn't wick water into the braid and has 100% coverage, but cheap is cheap.

While the coax impedance is ratio defendant, the topic was about RG-8, RG-8, and RG-59...so my answer still would be correct, "RG-8 52 ohm cable is about half inch diameter, RG-58 is about quarter inch."

In the context of comparing RG-58 and RG-59, my answer is definitively still correct. However the standard is to not use RG59 for anything longer than a patch cable.

Done. I tried different channels until I found one that seemed to do better that the others. ( oddly enough it was 9, not the usual 1, 6 or 11 )

Definately plenty of clearance at mid span. I am a lousy judge of distance, but I would say at least 100 ft or more.

Gordon Montgomery Living Scriptures, Inc snipped-for-privacy@lsi.com (anti spam - replace lsi with livingscriptures) (801) 627-2000