Duh... yes. I forgot that most wireless routers have built in 4 port switches. The switches (actually multi-port bridges) will take care of bridging function. I'm not sure how it would work with more than two WDS routers, but that's not an issue here. I still smell something is wrong, but my previous guess was not it. Thanks.
I'll give WDS a try as soon as I have a chance and post back to the newsgroup with the results.
I read a couple of messages about LMR-100A fitting better the connectors than LMR-195, due to the cable size. Is that correct?
I downloaded the catalog, but couldn't find that adapter, anyway I think your guess is pretty good.
Thanks for the further explanation, I got it now. Since I am going to put the antenna next to the AP, I'll get the R-TNC male to N female adapter, no additional pigtail needed for me.
That sounds reasonable. It looks like Stella Doradus uses RG-213/u (60 cm) cable for the antenna pigtail:
Is it a good one? BTW, what does the 'u' stand for?
Thanks again.
I got chills when I saw the picture of the plate being drilled without being clamped and just being hand held. My friend did *major* damage to his hand doing that while using an seemingly innocent small drill press when the workpiece caught and spun.
And such a nice picture of the same metal plate _clamped_ to the drill press on the previous page.
I did the opposite of your friend. A 1/2" handheld drill motor spun me around when it caught in a heavy steel plate. No long term damage, but I was young and invincible then.
Well, the connectors and cable diameters for LMR-100A and LMR-195 are completely different. Hopefully, someone didn't just mix up the crimp connectors and crimp tools.
RG-213/u is double shielded RG-8/u. For a 1ft run, it's fine.
RG means "radio guide" as in something that guides radio waves to whever they're going. The various numbers were originally the page numbers of some lost military catalog of coax cables. The /A /B /U are the different types of jackets. I think /u means non-contaminating jacket. Like any other significant numbering system, the military dumped the whole RG mess in about 1945 in favour of the C17 numbering system which nobody uses.
Bah. For a really fun time, try running a gasoline engine powered post hole digger. If you want a fun carousel ride, you've found the right contraption. Stick the auger into a rock and around you go. These do have a "dead mans switch" on the handles, but I figured out that one would be dead by the time it works.
Yep. Broadcom BCM4712KPB
That's the sensitivity of the chip, measured at the input to the chip and not the sensitivity measured after going through a diversity switch IC, across some lossy G10 circuit board, through a lossy coax cable, and into 1/2 of a lossy R-TNC pair. Use the sensitivity of the final product, not that of the chip.
Try again, but this time do NOT use "-3" for the coax losses. The spread sheet wants it without the minus sign. Use just "3". Remind me to yell at YDI about that.
No. It's an artifact of the extra minus sign.
Just to be sure, I cramming it into a more complex calculator at:
I don't know exactly where the broadcom chipset overloads. Older bipolar chips overload at about -30dBm. However, I don't have numbers on the newer Broadcom BiCMOS chips thanks to Broadcom controlling the distribution of their data sheets. If I had a WRT54G box to play with, I could put it on the bench and run a quick (rough) test.
Yeah. This is a guess, this is only a guess. A similar BiCMOS chip has a 3rd order intercept point (IP3) of -6dBm. Subtracting my guess of the conversion gain and extrapolating the curves, that *ROGHLY* yields an overload point about about -30dBm.
Yep. However, if you have an accurate attenuator, you can just plug the antennas from two boxes into each other. attenuator loss = +15dBm tx power - (-30dBm overload point) = 45dB So, try a 40dB attenuator and see if you have problems.
Using what for antennas? If you get to close, you end up with a near-field calculation instead of a far-field. I don't wanna grind the numbers. Use the calculator at:
Trees are bad. Go for the 24dBi antennas.
84mw = +19.2dBm. The data sheet power +18dBm. Close enough. Why would you want to decrease it? Note that I used +15dBm for my power output, not +19dBm. This is from personal experience with other radios that never seem to deliver rated power. The test results on the FCC web pile show a test power output of +16dbm for 802.11b and +15dbm for 802.11g. See section 5.6 at:I dunno.
Methinks that's a bad idea. Your performance is determined by the signal to noise ratio of the received signal. If you crank down the xmit signal to any large degree, you won't have the required signal to noise ratio necessary to maintain a link, much less get decent thruput.
No.
It looks like the WRT54GS uses the BroadCom 4712 chipset:
The datasheet is at the following URL:
The sensitivity looks a little better than the D-Link one:
I'm suspicious when someone uses dB instead of the correct dBm. The numbers look real enough to be useful as they are "only" 2-3dB worse than my stated numbers. That's within the accuracy of a BER (bit error rate tester) and my ancient and uncalibrated test equipment.
Why would you want to reduce the TX power from +15dBm to +1dBm? A
15dB drop in tx power is the equivalent of a 31 times drop in power level. Stay with the 15dBm tx power.There are others online that will work. The complex one above is what I use when I have to deal with marginal systems running at the very edge of the performance envelope. It's also a good learning tool. If you can understand EVERY item on the page, you're well on your way to understanding wireless propogation. (The rest is magic and witchcraft).
Other fade margin calculators:
BiCMOS and SiGe should be somewhat better than bipolar. GaAs should be spectacularly better than bipolar. However, there are many ways to screw things up. Most of the modern chipsets emphasize low power consumption. At some point, the quest for low power consumption can ruin the ability of the front end to handle overload. Modern designs are also direct conversion receivers, which are quite different from the older superhetrodyne style of double conversion receivers. Again, the simplicity of direct conversion is paid for in overload handling. If you're into this problem, CommDesign magazine has something on the topic, but I don't wanna dig for it now.
Well, if you insist on reducing your power 30 times, sure. As I said before, I do NOT think that overload will be a problem, especially since you introduced a few trees in the way in the last message. Trees are very lossy at 2.4Ghz and should reduce your signal to tolerable levels.
They're easy enough to find. The can also be made from cheap lossy coax cable. As for expertise, Learn by Destroying(tm). Just calculate how far you would need to be seperated to get -30dBm at the receiver input. That you can do.
Yeah. Figure about 2.5dBi. I know that some data sheets claim 4dBi, but my models show much less.
Big difference. In the near field, signal strength does NOT follow square law. Normally in far field double the distance results in 1/4 the signal strength. In near field it's roughly linear. None of the online calculators do near field.
Yeah, someone posted some really lousy spectrum analyzer photos of what the spectra looks like when one does that. They left off the grid markings and reference levels so I couldn't tell what was happening or whether they met FCC emission requirements. Methinks I'll pass on commenting more on such hacking until I determine if the transmit spectra is really tolerable, or if it goes non-linear and sprays trash all over the band.
About 20dB over the receiver sensitivity. That's your fade margin as calculated on the various online calculators. Note that the sensitivity changes with connection speed. Use your -65dBm for
802.11g at 54Mbits/sec. Add 20db resulting in -45dBm receive signal.
I found some info about the WRT54G, the WRT54GS should be close:
Okay, I'll do that. I still wonder what is the minimum TX power to get a good signal, considering the signal to noise radio thing you told me ...
I tried to search their web site, but no luck ...
They are oak trees: they loose foliage in the fall/winter and the attenuation will be much less than in spring/summer.
Can you estimate the approximate distance between the two LinkSys boxes to get a -30 dBm signal at the receiver input using the bundled antennas?
Can I damage the boxes if I put them too close?
Do you know if the WRT54GS has any feature to read the signal level at the receiver input? I don't think you can do that from the web interface, maybe it has some hidden command in the CLI?
Okay, then I would need a receive signal between -45 dBm (stable 54 MBps link) and -30 dBm (estimated overload point).
According to the calculations, using 24 dBi antennas and 15 dBm transmit power, the signal will be in that interval, then substract a few dBm for the trees, yeah, I should be okay ...
I am going to order the 24 dBi parabolic antennas. I'll let you know if the link works when I set it up ...
Thanks a lot for your advice.
Actually, that's a PDF of a PowerPoint presentation. Buried in the middle are some useful numbers. The BCM2050 chip (802.11g) has a 3rd order intercept point (IIP3) of -16dBm at maximum gain and +4dBm and minimum gain. You're going to be running near the middle to upper part of the range, which puts you where I previous guessed, at -6dBm IIP3 or better. That's roughly -30dBm or higher overload. Probably higher.
Sorta. I think you're eggagerating the effects of too much signal. It isn't likely to be a problem, especially with trees in the way to reduce the signal furthur. In addition, losses tend to be higher than calculated, and antenna gains tend to be less than the manufacturer specified.
Nope. I have a WRT54Gv1.1 sitting on my desk, next to my BEFW11S4v4 and a WAP11v1.1. Antennas are almost touching. I'm playing with some intereference tests. No fire, no smoke, no problems, and but plenty of interaction.
I went to the dump and recycling station yesterday and couldn't resist scavenging some electronics out of the e-waste pile. Included was a fairly dirty WRT54Gv1.1 that didn't power up. After cleaning, and replacing a diode, capacitor, and power connector, and upgrading the firmware, it works nicely. That's the way I like my wireless routers, for free.
I found a PDF document about the Broadcom radio chipset specifications and tests:
Does it help to get a better estimation of the overload point for the LinkSys WRT54GS radio receiver?
Can I permanently damage the LinkSys boxes if I put them too close each other to find out the actual overload point?
"Learn by Destroying" is pretty cool, but these LinkSys boxes are pretty expensive in Italy, I would prefer to not "brick" them for now! :-)
Thanks a lot.
Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.