Buffalo WHR-HP G54

"Adair Witner" hath wroth:

I'll assume you do NOT have any RF test equipment.

  1. Use IPerf to get performance statistics at various ranges, power levels, and through varies number of walls. Try it with the speed set to "auto" and also locked to the highest available speed (54Mbits/sec). Also, experiment with various wireless settings. Try to roughly duplicate the IxChariot found on:

Do the test with one laptop, where nothing changes during the test. Don't try to optimize anything or squeeze the last bit of range out of the laptop. Just run the test for a comparison and record the conditions including: 1. Make and Model 2. Firmware and exact version 3. Power setting 4. Range 5. Number of walls 6. Iperf Upload Mbits/sec 7. Iperf Download Mbits/sec 8. Indicated connection speed on client. 9. Any other changes (antenna, 802.11b compatibility, other settings, orientation, position of moon).

Note that I did NOT include signal level and SNR. While these would be interesting, I think they will just confuse the issue. The basic question is "how far and how fast"? An Excel spreadsheet would be handy.

  1. 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. Also try it with various settings such as 802.11b compatibility, frame burst mode, Turbot-G, etc. Leave the settings on the laptop alone.

Keep it simple. Try to change only one thing at a time.

Reply to
Jeff Liebermann
Loading thread data ...

But will that let the router communicate with a client at a longer range, or at a faster speed, or both? I thought I had seen reports in the forum of people getting a lot farther on their "walk tests" using an HP. Would the extra receive gain produce that result?

The data sheets may not, but the operator's manuals do. Here's an excerpt from my HP manual. I don't have it now, but my memory is that the WHR-G54S was exactly the same, both as to transmit power and receive sensitivity:

But as you say, the FCC numbers do show the difference.

Well, I'll continue to watch things with great interest. I should also say that I've been very impressed with the HP's radio performance. I've tried taking the laptop into the back yard, which means the signal is going through half a dozen walls, and everything still works quite well.

Reply to
Peabody

After reading Jeff's post, I realize that I should've kept my mouth shut about testing. It's so tempting to put some sort of faith in those signal meters....

Reply to
seaweedsteve

Peabody hath wroth:

Good, but slightly ambiguous question.

There are exactly 3 parameters that are involved.

  1. Range
  2. Speed (mbits/sec)
  3. BER or PER (bit error rate or packet error rate) You can trade any one of these for the others. For example, if you lock the speed to 54Mbits/sec and start walking, the BER will increase with range until it craps out. However, if you let the speed remain at "auto", the speed will drop as you walk, keeping the BER constant (at about 8% PER), until it totally craps out. You can go much farther if the speed is allowed to drop but of course, much slower.

Notice that I didn't throw signal level into the previous explanation. That's because it's a fixed function of the BER or PER. It's a family of curves that relate signal level, Signal to Noise level, and BER/PER. If you know one, you can derive the others. I could replace the BER/PER in the above list of 3 parameters with SNR (or signal and noise levels) and get the same results.

Yes, it will. It's a symmetry problem. Let's try it with a simple WHR-HP-G54 and laptop system. To keep things simple, let's use the numbers for 54MBits/sec and lock the speed at 54Mbits/sec.

WHR-HP-G54 Laptop TX power +24dBm +15dBm Antenna gain +2dBi 0dBi RX sens -66dBm -66dBm

I'm assuming identical receivers at this point, which is not a very wonderful assumption. I'm trying to use realistic numbers but the receive sensitivity is a best guess.

The gain from the WHR-HP-G54 to the laptop is: 24 + 2 + 0 - 66 = -40 dB and in the other direction is: 15 + 0 + 2 - 66 = -49 dB Therefore, there is 9 dB more system gain in one direction. During the walk test, the receiver on the WHR-HP-G54 will crap out before the one on the laptop.

So, if the receiver sensitivity (not the gain) on the WHR-HP-G54 were to improve by some mythical amount, the range would increase because the WHR-HP-G54 would hear the laptop furthur, while the laptop would continue to hear the WHR-HP-G54 over the same distance. This is the alligator effect.

The problem is that it's highly unlikely that tinkering with the receive pre-amplifier or onboard 2nd pre-amp is going to produce a 9dB improvement in receive sensitivity. Certainly GaAsFET front end amplifiers will help by perhaps 2-3dB, but not 9dB. I've been recording the receive sensitivities of various devices and vendors and find that the spread is rather large among those that bother to measure the sensitivity, but not enough to assume that the WHR-HP-G54 is among the better units.

Not even close to finished:

Frankly, what I'm seeing is widespread guesswork among vendors on receiver sensitivity. For example, I just traversed the Ubiquiti web pile grabbing rx sensitivity numbers from the spec sheets. What I found is that all 2.4Ghz devices seem to have exactly the same sensitivity at all speeds and corresponds exactly to the published specs for the Atheros chipset. Yeah, right. Never mind the losses through the maze of connectors inside the box before the signal gets to the Atheros chipset.

That agrees with the FCC data for the WHR-G54S.

Nice of them to specify at what speeds those numbers were conjured. Extra credit if they specify which procedure was used and at what BER/PER they were tested.

That's a measure of how much multipath and reflections the chipset can handle.

You had me shopping on eBay for RF test equipment last night. I managed to resist the temptation.

Reply to
Jeff Liebermann

"seaweedsteve" hath wroth:

Grumble.... Just lock the speed at 54Mbits/sec and take a walk until it craps out. That's easy enough. My guess is you'll be lucky to go more than about 20ft using the stock antennas.

Reply to
Jeff Liebermann

And I would assume that in the case of a laptop with built-in Wi-Fi, the laptop's transmitter would typically be the limiting factor. And by the way, where is the antenna in such a laptop? Is it in the lid?

I guess the point that I was trying to make is that the HP could actually have better connectivity range without necessarily having more xmit power, if the receive sensitivity was materially better than average. Your FCC numbers show that it does have more xmit power, but as the example you gave demonstrates, that doesn't necessarily improve performance if the receive function is what craps out first. 9dB would indeed be a whole bunch, but whatever lesser amount they are getting seems to work pretty well, at least with the stock firmware.

But then the question arises as to whether it might be better sometimes to have LESS receiver sensitivity. If there are other transmitters in the area, could you actually be worse off picking up more of them. I'll try to answer my own question by saying no, the relative levels of the incoming signals (the "signal" and the "noise") are the same regardless of sensitivity of the receiver. Well, ok, it was just a guess. But do you find in practice that it may help to reduce sensitivity in some cases?

Yeah, but the manuals for the WHR-G54S and the HP both have the same specs. In fact, I think the only difference in the entire Specifications sections was the power supply voltage. But now it's clear that the manuals mean nothing.

Don't do anything rash. :-) I have the advantage that I wouldn't know what to do with the test equipment, so it's easy for me to resist. I just rely on my basic concept of RF stuff - it's magic. Doesn't even use smoke, or mirrors. Well, maybe a reflector, but otherwise it just magic.

Reply to
Peabody

Yep. If the access point transmitted at the more typical power level of +15dBm, then the system would be roughly symmetrical.

Most of the new laptops have two antennas around the LCD frame, usually at the top. Some are on the sides. However, there are a few that have the antenna in the base and at least one with the antenna on the hinge:

Yech.

That's certainly possible but not proven.

Nope. Only if there were a strong signal out of band issue that was causing blocking or intermod mixing. That's unlikely for the typical home user.

Yep. That's the way it works. If you increase the gain of the receiver, you amplify both the signal and the noise equally. The resultant signal to noise ratio is the same. More crudely, an increase in gain does NOT necessarily mean an increase in sensitivity.

Sorta. There are many cases where too much sensitivity is a waste of effort. For example, in HF communications, the atmospheric noise in the 2 MHz thru about 16 MHz region is sufficiently strong that a high sensitivity receiver is a waste of effort. Better to keep the sensitivity low and use the receivers dynamic range to raise the overload point so that it can handle nearby strong stations better. The same principle might apply to 2.4GHz if one was building an outdoor wireless access point that might have to tolerate strong interference and overload signals. However, the trend is toward more and more sensitivity and simply not worry about overload issues.

Now you know why I don't read manuals. I would tend to trust sources that have actually performed the measurments (FCC ID data, various reviewers, independent labs, etc). Everybody lies, but that's ok because nobody reads the test reports and manuals.

You can do that to a point. Eventually the mechanical analogies and gedankenexperiments break down and you have to do the math or drag out the test equipment. You can't see RF. That's what the test equipment is for. It's the eyes of the RF blind techs and engineers. I'm still tempted.

Reply to
Jeff Liebermann

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.