I've got sort of a simple question. Here's my setup. My cable modem runs a CAT 5 cable into my WRT54G router, a CAT5 cable comes out of the router and into my desktop.. I've also got high gain antennas set up for wireless. This works fine when my laptop is running windows.. however I recently decided to start using linux again on the laptop. I supposedly could use 'ndiswrapper' to get my WUSB54G Wireless USB adapter to work in linux.. but that's pretty much over my head. I was thinking I could buy another WRT54G router and set it up in my room, and run a CAT5 cable to the laptop from there... but I'm a bit confused and thinking that might not be the cheapeast way of doing that. Is there another piece of hardware I could buy instead?
Sorry. I misread your question. It appears you want to use a WRT54G as a client adapter. CAT5 from laptop to this WRT54G. Wireless back to the main WRT54G.
Well, that can be done but not with the stock firmware. Wireless access points simply do not talk to other wireless access points. However, if you install alternative firmware both WRT54G boxes, some new features appear that will make it works.
formatting link
WDS repeater. You can run your CAT5 to the 2nd WRT54G, setup WDS and the 2nd WRT54G acts as an access point and wireless bridge simultaneously.
Client mode. For this you don't need to change anything on the main WRT54G. The client mode makes the 2nd WRT54G act like a client bridge. I think this is what you were asking.
Watch out for WRT54G v5. It cannot be used for either method.
With the stock firmware, no. But load up dd-wrt firmware and you can certainly use a WRT54G as a client device. Sort of overkill though seeing as how a WAP54G is usually a bit less expensive.
It sure looks that way for the worthless stock firmware. I scratched my head for a while trying to figure out how to turn off all the "intelligent" nonsense the wrt54g tried to do. All I wanted from the unit was to bridge packets from the incoming ethernet to the airwaves and back. It didn't seem possible from the stupid gui hand-holdish screens they provide to run the thing.
Running openwrt on it was a breath of fresh air. Configuring it as a pure bridge with all the firewall, dhcp, nat crap turned off was no problem.
Linksys might do well to preload the v5 linux version of the WRT-54GL with openwrt.
As to the OP, if he wants simple/cheap/fast why not a nice piece of cat5 cable?
Its not that hard. I did it a while back and it was lots of fun.
formatting link
Expect to get lots of fiberglass and spider webs on you if you live in California and route through the attic. That's only a temporary inconvenience though and you get a really nice 1gig-capable ethernet out of the deal.
As Jeff said, you don't need any additional hardware to get it talking. Just ethernet your one of the WRT54G LAN ports to your laptop's ethernet port.
It seems silly to run CAT5 to a laptop, that wants to be mobile, just because of the OS though. Is your laptop's wireless hardware built-in or are your using a PC Card? What are the specs and/or model of the wireless hardware? If there aren't any native *NIX drivers for it, did you check to see if it works with the Win drivers using ndiswrapper?
Ndiswrapper is well documented:
formatting link
Is your wireless hardware listed there by others as working under ndiswrapper? Even if not, give it a shot yourself. The installation guide is simply a step-by-step instruction:
formatting link
(I'm using ndiswrapper on a few computers.)
If it doesn't work under ndiswrapper, you have an excuse to buy a new piece of wireless hardware that does natively support GNU/Linux. Most likely your installation medium for your GNU/Linux flavor already has drivers for wireless PC Cards. Cisco Aironet cards are well supported and can be found relatively cheaply. I picked up an Aironet card off Ebay a while back for use with a Jornada 720 H/PC and found it to also work well with my laptop under Linux.
Another idea might be using a ethernet-wireless bridge; simply plugged into your laptop's ethernet port. They are both small and relatively cheap these days, but the real advantage is that you don't have to mess with wireless drivers at all. Everything is built-in, making them very versatile. Anything that has an ethernet port can be made wireless by simply just plugging them in. No driver setup, no rebooting (Win) etc, necessary. I use mine all the time for quick wireless connectivety for friends with laptops wanting to use my wireless network (they don't need to have the WPA passphrase since it's in the bridge), connecting game consoles, connecting printers, and getting funky older stuff like an older Sparc I have, up and talking.
This is admittedly a diversion, but methinks still interesting.
The problem with ethernet and laptops is the connector. The RJ45 is a PITA to insert and remove. My idea for solving this problem is to build a magneticly retained ethernet connector. An RJ45 plug, with an array of 4 gold contacts gets plugged into the laptop. A similar connector with gold spring loaded contacts connects to these pins. A pair of magnets retain the connection and deal with alignment issues. If you want to unplug an run, just yank on the cable. It doesn't solve the portability problem, but my experience is that even the most portable laptops tend to get used repeatedly in the same physical locations.
Note that Apple has a magneticly retained power connector on their new laptops. If it works for power, ethernet should be a no brainer.
The magnetic attachment would have been my first guess too a few months ago. I no longer think you need to do that. A simple rj-45 plug with the retainer tab snapped off seems to work fine as a quick disconnect. The friction from the 8 wire springs hold the rj-45 plug in tightly enough untill you give it a tug. We've been using a plug like that for months and it works just fine.
If you don't have any broken rj45's you can wrap some electrical tape around the latching tab and the body, holding it in the unlock position permanently.
Putting any metal near the wires of an unshielded cable is going to cause reflections. Putting small magnets on the side of the connector shouldn't be that bad. Those little ~3/16" rare-earth magnets Radio Shack sells are pretty nice. They have quite substantial holding power for something so small. I was very surprised at how hard it was to unstick the stack of blister packs from one another.
The very small magnets used will not cause any problems. If properly designed, the combination of the magnet and it's metal "keeper" should not have any external magnetic field leakage. I have some nifty ring magnets, that are magnetized north on the I.D. and south on the O.D. that should work just fine. There's no field in the middle.
Just for fun, I did some large file copies between two ethernet ports on the home computah[1]. I dropped all the magnets I could find on top of the connecting cable. As long as I didn't move any of the magnets, there were no transmission errors as reported by: netstat -es However, when I moved a magnet, I started to see occasional errors. Apparently the ethernet xformer or receiver does not like the impressed DC. The tiny ring magnet had no effect, moving or stationary.
I've had problems with magnetics, but in a rather odd way. It was on an early gigabit ethernet system, where we were using existing CAT5 (not CAT5e) cable. The cables were very carefully and solidly bundled in a gorgeous structured wiring rack and cable trough. 3Scum SuperStack model something with gigabit backbones. SNMP showed about
10% errors. Fluke cable certifier said everything was perfect for
100mbit/sec. However, we didn't have a gigabit certifier so we weren't sure about the remaining 2 pairs of wires. I tried other cables in the bundle with made it worse. So, I ran 300ft of CAT5 down the hallway, down the stairs, and eventually to the equipment closet. That worked. Hmmmm.... something wrong with the cable.
After several hours of crimping new connectors and tinkering with the tester, I arbitrarily decided to rip up the gorgeous structured wiring bundle and isolate the individual backhaul cable. Bingo. That worked. However, I wasn't sure if it was proximity to the other cables in the bundle or proximity to something else. Eventually, I had just the one cable running INSIDE the C channel rail of the equipment rack. The errors were back. Move the cable outside of the C channel, and the errors were gone. Huh?
Eventually, I discovered that the steel rack was seriously magnetized. I have no idea how it happened. De-gaussing was not an option, so I stupidly volunteered to replace the steel rack with an aluminum version. Ever try moving equipment between racks without turning anything off or crashing anything? It can be done, but I don't recommend doing it. Anyway, the aluminum rack worked fine and the errors were gone.
I don't completely understand the failure mechanism. My best guess (might be wrong) is that the magnet acted as a "line stretcher" or created a delay line out of the CAT5. That could only happen with a long magnet such as the 6ft rack frame. No need for moving magnet to create this problem. Since the cable run was fairly close to the official maximum for gigabit over copper (100 meters), I'm guessing that the increased delays caused timing problems.
[1] Because I was too lazy to get my laptop out of the truck. I have to ports for two different IP blocks so that I can sniff both sides of a router.
Interesting. Probably figures - I /have/ seen problems running AC power cables too close to networking, but you have to try really hard to make a serious issue out of it.
Drivel. The NFPA National Electrical Code requires AC power cables and signal cables to be in seperate conduits and about 3" (not sure of exact distance) apart. Same with structured wiring installations but with different specs (i.e. no conduit required).
Mixing ethernet and AC power will cause problems if the cables run parallel for any distance. The twisted CAT5 pairs and differential ethernet transceivers are quite good at avoiding the effects of common mode DC and garbage on the ethernet pairs. The ethernet xformer also has quite a bit of loss at 60Hz. What it can't handle is differential DC and/or garbage. The way to create that problem is to put the ethernet twisted pair very close to the source of the garbage. One wire from the pair will be slightly closers to the garbage source than the other. The difference in garbage pickup is a differential voltage which the ethernet transceiver has to deal with. Depending on the amplitude, it usually does quite well, but can saturate the transformer or overload the ethernet receiver. Either will create packet corruption or loss. About the only way to create this situation is to bundle the AC power and ethernet cables together over some distance. Nobody does that intentionally.
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.