Buffalo WHR-HP G54

"Adair Witner" hath wroth:

For long and skinny, put the wireless router at one end of the building, and use a reflector to direct most of the power down the length of the building:

Yep. I like DD-WRT.

I read somewhere that DD-WRT will automatically recognize the differences between the WRT54G and WHR-HP-G54 transmitters and adjust the power ranges accordingly. I'm too lazy to verify if this is true.

One dumb question is just how high power is the Buffalo WHR-HP-G54 transmitter? The data sheets and web pages offer no numbers:

So, we go to the FCC ID web pile at:

where the test reports shows on page 29: 802.11b +24.5dBm (282mw) 802.11g +25.8dBm (380mw) You could run those power levels and be the first to dominate the neighborhood. Of course, your ranges will not be symmetrical. The WHR-HP-G54 will transmit over a much larger area than it can receive from a typical +15dBm (30mw) laptop radio client. What's a little more intereference and pollution on an already crowed band?

Hint: Run as little xmit power as necessary. +20dBm (100mw) should be more than enough. Any more is probably no better than jamming.

See the list of wireless options at:

I usually change: Transmit rate Auto -> whatever speed is appropriate for backhaul. AP Isolation Off -> On (isolate clients to prevent attacks) Preamble Long -> Auto (unless you have very slow clients)

Make sure the defaults: Frame Burst Disable Afterburner Off are set if you do *NOT* have any clients that need these features.

On the basic settings page at:

there's some issue about which mode to use. If you have 802.11b clients that need to connect, set it to "Mixed" mode. If you have only 802.11g clients, use "G-only".

There are a huge number of settings that I setup and/or change on the: Administration -> Management Administration -> Services pages. I can't offer any advise on setting up these pages without a clue as to what you're doing.

Actually there phone/data room is close the the center of the building so it will likely end up there.. hopefully with good coverage.

That's what I read also but I did not check it before I sold it off to a friend (not customer with longish building). So I just bought a WHR-HP-G54 and a WHR-G54S to do some harder testing on and try and find out for myself exactly which one of these models is the better router... or...whatever...

I agree about not running more power than is necessary, I feel that 96mW is plenty for most applications because even doubling power from there is only a 3db increase which prolly wont account for much and doubling again (256) is close to the max the radio will output. I guess what I was really looking for was if the HP has the built in amp what the best setting was so as not to over drive it's input.

As of right now all any of my Linksys and Buffalo routers are doing is acting as a routers, firewall and ap's for small home LANs. We are building a 802.11 network to link around town with and are looking for devices that will allow people to access the network.. be it with these boxes or something else such as

is a very nice bridge but still needs a router behind it.. Less money in two of the linksys or buffalo boxes if someone needs to save cash.

Thanks for the info so far.

Adair

On Sun, 11 Feb 2007 01:48:02 GMT, "Adair Witner" wrote in :

In that case, assuming you only have one floor to cover, you may want a high-gain omni antenna; e.g., Buffalo 6.5 dBi High Gain Omni Antenna WLE-HG-NDR.

"Adair Witner" hath wroth:

I'm not sure that's going to be such a great idea. The typical phone room is usually full of electronics, some of which -might- generate enough RFI to cause receive problems. I've only seen this once, so it's unlikely, but still should be considered if there are problems. The wires found on the walls of phone rooms make a truly impressive RF shield. Think of it as a partial RF screen room. The doors on most office building equipment rooms are metal (fire code requirement). The walls have foil backed insulation (fire code requirement). The floors and ceiling are usually poured concrete, which is rather impervious to RF.

Basically, you have to get the signal into the hallway. I've only done one PITA phone room installation. The building owner and planning department would not allow me to install the access point in the hallways. Even though I quoted chapter and verse from the NEC on low voltage wiring, they were adamant. However, they would allow me to install additional metal conduit from the phone room to the hallway to route a non-powered antenna. So, the access point ended up in the phone room closet with about 10ft of LMR-240 to a downward facing patch antenna in the hallway.

Beware of selling to friends. They have the irritating habit of returning with questions, more questions, and even more questions.

Dunno. I've only used the WHR-HP-G54.

Rules-of-thumb: 6 dB is 4 times the coverage area. 12 dB is 4 times the range. 18 dB is 8 times the range. or just: range = 10^(dB/20)

Also: 3 dB is double the coverage area (for omni antenna). 6 dB is 4 times the coverage area. or just: coverage = 10^(dB/10)

However, that's for antennas, which offer a gain boost in both transmit and receive. That's not the case with just increasing the transmit power, which only works in one direction. If you increase the power at both ends, you'll see a range increase. However, just increasing the power at one end doesn't necessarily improve the usable range if the return signal is marginal.

That's a problem with the WRT54G/GS series, where the internal power amplifier doesn't stay particularly linear with increased drive. See lousy spectrum analyzer photos at:

I keep wanting to make better photos, but never simultaneously have the time, the inclination, or the test equipment. Sigh. Anyway, everything 200mw and above start to look fairly bad with lots of locally generated intermod products. I would say 100mw is safe with

150mw possible. There have also been persistent rumors of WRT54G radios blowing up running at full (250mw) power.

The WHR-HP-54G apparently has a bigger power amplifier that has been tested to perhaps 400mw. I don't think you have to worry about overdriving it. However, I am worried that the settings in DD-WRT don't correspond to the actual power output.

No. What I meant by what you're doing is what services (DHCP, HTTP server, DNS relay, QoS, VPN, SNMP, etc) you're planning on using. No two installations tend to be the same and these setting will vary radically and there are not "standard" settings. Just go down the list, see what you need, and enable/configure those.

Ok, so you're building a WISP (wireless ISP). Equipment that is suitable for high reliability, high traffic, and a large number of users is quite different from the common commodity wireless router. For example, you will probably want to use access points instead of routers. Your router will probably be a Cisco something that can handle the volume and the abuse. You'll need some form of traffic manager to prevent bandwidth hogging. You'll need to implement logging, monitoring, and abuse detection.

Equipment will be the least of your problems. The real problems will be how to manage and bill the users, how to deal with problems (i.e. support), and how to guarantee reasonably reliable service in a very crowded and polluted frequency band. Some light reading:

Basically, you need to have *ALL* the features and facilities of a wire line ISP, with the added bonus of a marginally reliable method of delivery. Are you sure you want to do this?

You make a good point and maybe that's the way it is in Cali but this is not going to be the case with this phone room. I am a telephone tech for a company that does voice and data cabling both copper and fiber. We do phones, security, cameras, nursecall systems as well as other random stuff that as well as being an amateur radio operator makes me very aware of where to place equipment.

Thats ok I don't mind helping my friends. most of them pay me for my time unless I don't feel like charging them.

Good stuff to keep in mind however I read that the buffalo has a better receiver over say the wrt's so i'm just trying to match them up accordingly.

Nice pictures.

Ditto, but I read along with the power amp comes a preamp and or better tweakable receiver.

You make some pretty big assumptions about what we plan to do and actually all of them are wrong. The 802.11 network is comprised of people who are mostly ham radio operators that are interested in doing something different than just talking on there radio. Our group did a presentation at one of the local ham club meetings and people were fasinated in almost every aspect of the system.

Computers and the internet have drawn alot of people away from the radio and are trying to do our part to bring them back and renew some form of interest. Almost everyone nowa days has a broadband internet connection and some form of small wireless network in there home and this network is a way for all of us to learn something and get people active as well as possible provide community service with the system as it grows. Our goal is to offer services over the WLAN such as, intranet website and forums, FTP site, APRS and Packet traffic (between real nodes) and whatever else anyone wants to include such as IP cameras and just about anything else that can be converted and sent in an ethernet packet. We do not plan to sell internet service to anyone.

Adair - KD5DYP

On Sun, 11 Feb 2007 20:08:08 GMT, "Adair Witner" wrote in :

I seriously doubt there is a significant difference. All low-end Wi-Fi devices use a limited selection of standard chipsets, all of which deliver roughly comparable radio performance. Antenna(s) are what matters most.

I totally agree but coming from a back ground of radio I know that everyone has a different idea about how to do the same things and some people do it better IMHO.

My friend that I sold my WHR-HP-G54 to had been through two different wireless routers and was unable to keep his G4 Mac connect through two walls of his house.. After doing some inspecting with the WiSpy and tweaking on the router (which was also done days before the buffalo swap) he is able to maintaine a 100% solid connection to the router now which is sitting in the same spot as all the others have been.. So is it the better antenna on the buffalo vs the netgear and whichever other router he had or could the maybe slightly better receiver and more transmitter power be helping a little? Our next step if he was still having trouble was to install another antenna on the router.

I'm a big fan of using good antenna's but even a good antenna on a crappy radio wont make it work that much better.

Adair

Here in our installation, for our base station, I went from a WRT54G to a WHR-G54S to a WHR-HP-G54.

The Buffalos are on DD-WRT, the original Linksys v5 was stock FW. Although I failed to do thorough comparison testing, especially with the Linksys, I can make a few comments:

First I had the linksys v5 going there, but decided to go with the Buffalo WHR-G54S and DD-WRT. I sold the Linksys.

I was running the WHR-G54S on DD-WRT V23sp1 at 84mw transmit. All was well. Then I subbed in the WHR-HP-G54 on DD-WRTv23sp2 and started testing. Everything else, including the external antenna rig, remained the same, and I tested it on the same afternoon.

With the HP set at 28mw transmit, people noted that they were getting weaker signals than with the non-HP Buffalo at 84mw. One client with great reception went from 5 bars to 4, another marginal went from 2 bars to 1.

I cranked up the power on the HP to 84mw and everybody had the same signal strength showing on their clients.

Conclusion: 84 on the HP (with v23sp2) gives you about what 84 on the non-HP gives. The linksys I had at the base was weaker still, but it had stock firmware. Why is 84mw on the HP working like 84mw on the non-HP ? Well, after a thorough discussion on the DD-WRT forums.... nobody knows !

"Brainslayer" (the DD-WRT maestro) isn't talking. See:

or see topic: "Buffalo WHR-G54S vs WHR-HP-G54S"

Be sure to read all the way through as there are various different conclusions made along the way. The best takeaway I got from this thread is: use V23SP2 and go with 84-100 but it might be better to use SP1 and go with 10mw. They changed something from SP1 to SP2 but they're not telling. My guess is that they don't exactly know how the change affected output and are waiting for the users to sort it out.

I can add that the other day I cranked the HP up to 150 mw when I connected some distant clients. It barely improved (as expected) their received signal, but then, an hour later, when it was getting baked in the sun, it stopped working. (I forget the details, sorry) Upon inspection, it was hotter than usual. I cooled it off, rebooted and immediately knocked it back down to 84. Worked fine again. I don't know what really happened. I'm not trying higher power again ( over 100mw) on SP2.

I'm planning to put in sp1 sometime (at 10mw) and see how it works.

OK. On the RECEIVE side, the HP really shines and I can see a greatly increased signal quality and level as shown in the DD-WRT status. I'm basically getting 6-10 dB improvement on the receive end with the HP over the non-HP. Funny, but it seems that the HP is more a "rabbit" than an "alligator".

Somebody said the HP actually has a different Broadcom chip than the non HP - it's an "E" (for enhanced?)

Quote: "The BCM5352E also features Broadcom's new BroadRange technology, a standards-compliant hardware enhancement that extends the range of 54g- based wireless devices. The technology uses advanced signal processing techniques to provide the industry's best receive sensitivity, enabling Wi-Fi users to maintain high-speed wireless connections at up to 50% further from an access point.

BroadRange technology uses advanced digital signal processing techniques to provide the highest receive sensitivity of any 802.11g chipset on the market. "

Somebody else said: "The unit has actually 2 amps. There is one on the Tx line, and one on the Rx line. That is clearly visible on the schematics and photos found in the forum. The amplifier, that was affected by the boardflags was the Rx amp. And the amplifier affected by the 10mW setting is the Tx amp. (It even makes sense if you really think about it). The Tx amp is Anadigics AWL6153 "

All this came off the DD-WRT forum. I don't know if they are correct.

I now have another Linksys WRT-54G V2.2 running DD-WRT about 400 meters away from the Buffalo HP. It's working as a repeater in WDS. The linksys seems weaker on the receive side. Maybe on the transmit side too. Both are set at 84 mw transmit. On the Buffalo HP side, I see a 19% signal quality, or -78 signal. On the linksys side, DD-WRT reports -88 signal at 8% quality.

Now, one might assume that the Linksys has a stronger signal and that's why it shows stronger on the Buffalo HP, but based on all the other indications, I see it as another sign that the HP is 10 db more sensitive than the standard Broadcom device.

In sum, from my experience in the field here: Set the HP at 10mw with v23sp1 or 84mw with v23sp2. Don't expect more power than the regular, but expect much better sensitivity. Don't know why that is for me. Maybe 10mw with sp1 is better and I need to change versions !

Steve

"Adair Witner" hath wroth:

You're lucky. Most of the phone rooms I've dealt with are in the most RF disgusting locations. This one isn't an RF screen room but is fairly typical of what I get to deal with. Behind the plywood is something I've only seen once, foil backed drywall. The floor is concrete and the ceiling is foil backed fiberglass matting. The door is solid core wood which passes RF just fine. However, the foil wasn't the big problem. This phone room is on the 2nd floor and overlooks a mess of other 2.4Ghz users. That includes a headset distributor that has an unshielded test bench. Way too much interference.

Before:

After:

Ok. I won't lecture you on RF transmission through various materials.

I have friends and I have customers. The difference is that the customers pay me.

Where did you read that? I'm curious if whomever came up with that line actually did any measurements (and how they did the measurements). Measuring RF receive sensitivity for 802.11 is both a mess and expensive. The required BER (bit error rate or packet error rate) testers are expensive. Very few even bother to specify how they measured it, and to what BER reference point.

I've been playing with DLinks numbers from their data sheets lately. Judging by the distribution, I think they actually measure the sensitivity and post the results. See:

The 5dB variation among various models is actually fairly large but typical of what I've seen for setup, calibration, and cable errors.

Both the WHR-HP-G54 and the WRT54G/GS/GL use the Broadcom BCM2050 RF chip in all models. This is clue on the architecture:

Hmmm... 4dB system noise figure.

Note the rx sensitivity curves on Pg 28 and 35. Basically, they're the same for all boards that use this chip with some issues that may cause the sensitivity to deteriorate:

1. The numbers may not include the losses through the TR switch, BPF (band pass filter) or diversity switch. (I can't tell as there are no test conditions specified).
2. The numbers do not include deterioration caused by digital noise pickup from the baseband, router, and switch chips on the same circuit board.
3. The numbers do not include G10/FR4 circuit board losses, which can be substantial.

Barf. You can't see the graticule so I can't overlay the FCC spectrum mask to see if it will still pass Part 15.247 certification. The upper grid line is not visibly calibrated so I can't check if it's actually putting out the power claimed.

Well, I'll break open the box and take a look (when I have time). It's possible. I can't tell from the type certification pictures.

Mostly hams? Get them licensed.

Like ARRL's (HSMM) High Speed MultiMedia Radio or Hinternet? If you do it on anything other than 2.4GHz or 5.7GHz, I'm all for it. That's improving the state of the art, experimentation, etc. Otherwise it's just another Wi-Fi WLAN. In other words, build something, not just buy it off the shelf.

Also, of the dozen or so over-powered access points I've tracked down,

2 of them were owned by hams claiming that their ham license entitles them to trash all unlicensed operation.

Winlink, APRS, PSK31, synchronous CW, STV, etc all use computers.

Sounds like a plan. Have fun and good luck.

I wish that the new google groups had preview.

Try again on the DDWRT forum link:

"seaweedsteve" hath wroth:

Unfortunately, the two photos of the HP and non-HP versions cannot offer much help. There's a shield can over the RF section and a head sink covering the router chip on the non-HP.

That's the router and ethernet switch chip as a "system on a chip". No RF involved.

The chip also varies depending on WRT54G/GS/GL firmware version. See table at bottom of:

Most of the current WRT54G/GS/GL models use the BCM5352 chip.

Look at the WHR-HP-G54 photo at:

Notice the BCH2050 chip labeled IC13 in the general lower left area. That's the radio section. There is an obvious TX power amplifier labeled IC3 (large square thing to left and down from the BCH2050). There appears to be an RX MMIC? RF amplifier as TR2, but I'm not sure. I guess I have to crack open mine and take some photos.

Ah. I see that the v2.2 WRT has the same pair of chips as the HP: the

5353E and the 2050.

So that is obviously not what makes the HP the more sensitive one....

Great. It would be nice to understand why it shows such an improved signal level - as measured by DD-WRTs signal metering.

When will somebody sort out the amp thing in the HP and DD-WRT? I would've expected to see more transmit power from the HP than the non- HP in the field. But maybe cranking up the main chip in the unamplified router has the same effect, in a dirtier or hotter way...

Seeing as I seem to have a "rabbit", I'd like to give it more voice, you could say, get the transmit back up to receive level. I wish I had tested the HP with Buffalo FW before flashing it, to get a reference level. Steve

Hmmm, I thought HP meant higher power output (with built-in amp.) Anyhow my household coverage improved with this router from Linksys one.

Indeed. I have a new HP with the stock firmware 1.40. I've been debating with myself about flashing DD-WRT, and have been following the DD-WRT forum pretty carefully for quite a while. So far as I can tell:

1. Nobody has ever formally tested the HP with stock firmware versus any DD-WRT version to see if DD-WRT controls either transmit or receive power the same way as the stock, or specifically whether any version of DD-WRT achieves the same transmit power in the HP as the stock firmware.

1. Nobody has ever demonstrated that the HP has more transmit power than the regular WHR, or than the Linksys WRT. Several have demonstrated a clear difference in receive sensitivity in the HP with the right board flags set in DD-WRT. Claims of extended range for the HP on the Buffalotech website do not necessarily mean it has more transmit power - indeed, the manuals for the HP and the regular both show the same transmit power in the specs.

2. Nobody knows which version of DD-WRT actually works best with the HP.

I consider this whole situation to be a complete can of worms right now. Fortunately, with stock firmware my modem seems to be working quite well right now, so long as I don't get carried away with bittorrent stuff. Power-wise, my laptop always shows maximum signal strength and connection speed. So I'm gonna leave it as is until everything is sorted out.

Or maybe flash Tomato instead. :-)

Post back when you test this.

This is good info to have and exactly the type of discussion I was hoping to see out our people who have actually used them so thanks again. I will have WHR-HP-G54 and a WHR-G54S here in a few days that along with my spair WRT54GL will do some testing before and after flashing with DD-WRT.

Adair

I plan on doing some testing with stock firmware when I get both buffalo routers in hand in the next few days.. Anyone want to reccomend some testing methods? I have two WRT54GL routers both with DD-WRT 23 SP2 and a few wireless cards with different antennas. I can monitor signal levels from several points using different radios and antennas. I can create WDS links from all three routers to the other linksys. etc..

Adair

I agree, of course. I will add that it's a fairly happy can of worms, no real problems, it works well enough. Just less understanding of our controls than some of us like. It's still amazing that we get this. I paid $40 for my the HP and it's so much more than $40 or $60 will get in other realms. Hell, it's the cost of ONE auto tire. A plastic toyota TRIM piece costs more... Amazing tools at negligable cost.

Anyway, we want more info, but I don't regret putting dd-wrt on the buffalo HP. I just regret testing it first.

I imagine the real radio dudes will have something to say, but obviously you want setups that are as much the same as possible. Control over variations. Luckily the signal strength is relative; as long as the measurement is steady and repeatable, than it's useful for comparison to me. Also, outdoor range is what I care about and will give me the best idea.

Maybe this: WRT54GL in DDWRT as your base station. Maybe a high gain antenna, maybe stock. Clear outdoor LOS, clear fresnel. Test all three other routers (WRT, WHR, WHR-HP at fixed distances and compare realtime signal readings on the signal meter in the base station status window. Swap out the client ends in the same time period (afternoon- or evening?) to avoid too much climate variation.

For test distances choose several or maybe just find the limit of the weakest router/FW setup and measure what the betters can get. Or measure the max distance for each to connect at 12or 18 Mbps. That's what I wanna know, anyway. Othersmay want to know how close each has to be to get a connection at 54,128 or a certain thruput.

Measurement. The base has DD-WRT, so you can use it's signal meter to measure the TRANSMIT power of all permutations of client. Transmit power is easiest.

RECEPTION: For the clients that have DD-WRT, you can compare their signal readings to see how sensitivity varies as you swap them out in the connection to the controlled base station.

Maybe Jeff will have something to say about Bit Error Rates and such.

For testing the Buffalos or Linksys reception w stock FW, um....Well, measuring and comparing reception for each hardware rig across various firmwares, may be the hardest part.

Back to transmission: another transmit power test for each router hw/ fw rig would be to use Netstumbler with a proxim or some other fully compatible card and, keeping your laptop fixed at a given distance, swap out each router/FW config as a base station. Save and compare the data from NetStumbler for each.

WDS. Security. Should you be in WDS or just cliet? With or without security? With WPA-TKIP the minimum for how we actually use it....

I look forward to hearing about what you are getting.

Peabody hath wroth:

True. More specifically, the manner in which most users are doing their testing is IMHO rather useless. For example, almost all users rate their improvements based on increases in the signal strength on either the client or using Netstumbler. That's fine but the real indication of improvements is in an increase in S/N ratio and a corresponding increase in thruput. Many such measurements are made without any traffic moving, which tends to reflect the last reading instead of the current conditions.

If proper equipment and measuring techniques are not available, then I can recommend a simple alternative. Fix the wireless speed in the access point to 54 Mbits/sec. Start playing some streaming video or initiate a download from a local server (or get some traffic moving). Start walking away with the laptop until the download craps out. It will be quite sudden and obvious as there will not be any compensation due to slowing down the wireless connection rate when the speed is nailed to 54 mbits/sec. Then, change whatever your testing (new AP, better antennas, etc) and take another walk. The longer range wins.

The FCC test results for the WHR-HP-G54 rate the unit at: 802.11b +24.5dBm (282mw) 802.11g +25.8dBm (380mw) Actually, the 802.11g output is somewhat lower at the higher OFDM speeds but that also applies to most 802.11g devices.

Typical output from the WRT54G is +17dBm (50mw). That's about 8dB less than the Buffalo HP. That's 2.5 times improvement in range, which is where Buffalo gets it's advertised 3x improvement in range. range = 10^(dB/20)

However, you're correct that nobody has bothered to put a watts guesser or spectrum analyzer on the output to verify the numbers.

No, they've demonstrated an increase in receive *GAIN* by enabling an RF amplifier inside the BDM-2050 chip. This amplifier is normally disabled when using an external amplifier such as the one that appears to be on the WHR-HP-G54 circuit board. Having the 2nd amplifier running will not improve the sensitivity as the sensitivity is almost totally determined by the first RF amp. It will increase the signal and noise equally (plus adding it's own noise) resulting in zero net gain in sensitivity. However, it will reduce the dynamic range by lowering the overload point, and will really look impressive on Netstumbler looking only at the signal part of the S/N ratio.

Huh? Neither data sheet show anything about power output. Zero, zilch, nothing. Only the 3x, 8x and other marketing hype. See for yourself.

However, the FCC data is much more accurate. See the test reports at:

The WHR-G54S shows at TX power output of: 802.11b +19.1dBm (80mw) 802.11g +16.0dBm (40mw) That's considerably less than the WHR-HP-G54.

Agreed. Also, nobody seems to know how to test what works best.

Bah. All I can say is that my 3 Buffalo WHR-HP-G54 routers works as well or the same as those with the stock firmware. However, I'll admit that I haven't done any real testing other than the usual empirical guesswork. Just my general impression.

Ummmm... well, ok. Good luck.