I hate to rain on the parade, but the test that WTS performed that shows the increadible TX powers is wrong. The Praxsym PM model test meters can only measure up to ~20dBm of forward power. If you want to measure more than ~20 dBm, then you have to use an attenuator, and set a corresponding power offset on the meter, to get the forward power under 20 dBm (WTS has mentioned this in thread). However, if you watch his YouTube video you see that the 10dB attenuator is between the power meter and the antenna, not between the power meter and the Buffalo router. This means that the offset has been set to at least 10 dB on the meter, but there is NOT 10dB of attenuation between the transmitter and the meter. This means that the measurements are wrong by a factor of at least 10 (10dB in power). (this also means that the SWR of 1.2 is wrong... this is the SWR into the 10dB attenuator, which should be ~1.2 even if the antenna is not even connected)
Joined: 01 Aug 2006 Posts: 296 Location: Waldo Florida(trailer trash)
Posted: Mon Oct 08, 2007 20:35 Post subject:
Since you are technically informed can you tell me the power output of the WHR-HP-g54 running v-24 at 28, 50 & 80 mW tx out. Thanks, Bill _________________ DSL from Windstream
I feed a friend 400 meters through trees to Wi Fi station taped to his window from my bullet hp feeding 15db pringles type antenna
Joined: 06 Jun 2006 Posts: 3763 Location: I'm the one on the plate.
Posted: Mon Oct 08, 2007 20:59 Post subject:
tparris wrote:
I hate to rain on the parade, but the test that WTS performed that shows the increadible TX powers is wrong.
I noticed that as well. I think he was primarily fishing for a free 20 dB attenuator. He kept saying that he needed an attenuator to choke down the input side to be within range of his meter, but right there in the video was a 10dB attenuator that he could have just as easily attached to the input and not the output, and would then have had 40 dB of range (assuming the meter could actually measure 1 watt as he purported it to be doing). Then he wanted to know the RX sensativity in watts. _________________ http://69.175.13.131:8015 Streaming Week-End Disco. Station Ripper V 1.1 will do.
I measured the output of a WHR-HP-G54 with an Agilent E4404B spectrum analyzer. I tried this with DD-WRT V23, sp2 and V24, RC3. There was no difference. Here are the results;
And for the power settings you asked me to measure;
28mW setting = 16.5dBm (44.7mW)
50mW setting = 19.5dBm (89.1mW)
80mW setting = 19.7dBm (93.3mW)
[these measured powers represent the RMS (root-mean-square) RF (radio frequency) power delivered to an ideal 50Ohm load during a transmit pulse. Channel 6 (2.437GHz) was used]
For the benefit of this thread....
You should notice that the actual TX output power levels off at 100mW. This is exactly 10 dB below the measurements that WTS posted. As I mentioned in my last post, WTS put his 10 dB attunuator on the wrong side of his meter, and skewed his measurments by 10dB... this the 30dBm that he saw was actually 20dBm=100mW.
I also opened up the bix to look at the cicruit board, and it is clear that there is an AGC (automatic gain control) sub-circuit that limits the output power to 100mW. There is nothing that can be done in software to get more than 100 mW out of these routers. I modification to the circuit which disables or bypasses the AGC could allow higher output powers. the componenet labeled 'IC26' is a directional coupler whic couples some of the forward power to the AGC (which is D1, IC2 and all of the associated capacitors and resistors). This measures the output power and feeds back to the BCM2050 radio chip (pins 14 and 32) and to the bias of the TX amp (at C25). I cut the trace between the coupler and the AGC, which removed any measure of actual TX power from the AGC, and the TX power was pegged at 20dBm regardless of how dd-wrt was set. This means that the AGC does work, and is limiting the TX power to 100mW. You MAY be able to get more TX power by completely bypassing or removing the coupler (IC26). I may test this in the future.
Also, to address some discusssions that I have seen on the WHR-HP-G54...
When I had one of these opened up, I measured the bias voltages to both the TX amplifier (labeled 'IC3') and the RX amplifier (which is not clearly labeled on the circuit, but is near the 'TR2' label). The bias of these amplifiers does NOT change when when bits 12 and 13 of 'boardflags' are toggled (ie when boradflags=0x1758 or 0x2758). Also, the reference voltage to the TX amplifier does NOT change. This means that setting 'boardflags' one way or the other does NOT affect the TX amplifier or the RX amplifier. It MAY affect some parameters of the BCM2050 radio chip, but I cannot know this without a datasheet for that chip... which seems impossible to get. Also, changing boardflags MAY change how dd-wrt works, but I have not looked at the source code to see how dd-wrt uses 'boardflag'. But, as far as the hardware external to the BCM2050 chip is concerned, boardflags[12,13] has NO effect.
Joined: 01 Aug 2006 Posts: 296 Location: Waldo Florida(trailer trash)
Posted: Tue Oct 09, 2007 19:09 Post subject:
tparris wrote:
Here yoiu go Bill;
I measured the output of a WHR-HP-G54 with an Agilent E4404B spectrum analyzer. I tried this with DD-WRT V23, sp2 and V24, RC3. There was no difference. Here are the results;
And for the power settings you asked me to measure;
28mW setting = 16.5dBm (44.7mW)
50mW setting = 19.5dBm (89.1mW)
80mW setting = 19.7dBm (93.3mW)
[these measured powers represent the RMS (root-mean-square) RF (radio frequency) power delivered to an ideal 50Ohm load during a transmit pulse. Channel 6 (2.437GHz) was used]
For the benefit of this thread....
You should notice that the actual TX output power levels off at 100mW. This is exactly 10 dB below the measurements that WTS posted. As I mentioned in my last post, WTS put his 10 dB attunuator on the wrong side of his meter, and skewed his measurments by 10dB... this the 30dBm that he saw was actually 20dBm=100mW.
I also opened up the bix to look at the cicruit board, and it is clear that there is an AGC (automatic gain control) sub-circuit that limits the output power to 100mW. There is nothing that can be done in software to get more than 100 mW out of these routers. I modification to the circuit which disables or bypasses the AGC could allow higher output powers. the componenet labeled 'IC26' is a directional coupler whic couples some of the forward power to the AGC (which is D1, IC2 and all of the associated capacitors and resistors). This measures the output power and feeds back to the BCM2050 radio chip (pins 14 and 32) and to the bias of the TX amp (at C25). I cut the trace between the coupler and the AGC, which removed any measure of actual TX power from the AGC, and the TX power was pegged at 20dBm regardless of how dd-wrt was set. This means that the AGC does work, and is limiting the TX power to 100mW. You MAY be able to get more TX power by completely bypassing or removing the coupler (IC26). I may test this in the future.
I hope this helps.
That is fantastic info thanks for the work you put in. I hope many people benafit from your info. Many Many Thanks, Bill _________________ DSL from Windstream
I feed a friend 400 meters through trees to Wi Fi station taped to his window from my bullet hp feeding 15db pringles type antenna
Joined: 06 Jun 2006 Posts: 3763 Location: I'm the one on the plate.
Posted: Tue Oct 09, 2007 19:42 Post subject:
tparris wrote:
these measured powers represent the RMS (root-mean-square) RF (radio frequency) power delivered to an ideal 50Ohm load
Can your meter take a peak reading instead of root mean square ? RMS means pretty much nothing unless you know the geometry of the waveform. Very likely it is not sine wave due to the somewhat complex modulation scheme. Even so, using the most conservative sine wave conversion you still have 141 mW of output. The FCC's own in house testing shows the unit with the factory firmware to peak at 316 mW, with scope shots and everything. The peak reading is what would commonly be known as the wattage output potential. You can find the links in another thread here labelled "The whole HP power issue" or something to that effect. After you get your specan set for the proper mode it would also be interesting to see the results of SP1 which handle the RF control system differently than the newer versions you tested. I use only SP1 on the HP radio just so I can twist it's nuts easier and get more power out. This just does not seem to be a good thread for parades. _________________ http://69.175.13.131:8015 Streaming Week-End Disco. Station Ripper V 1.1 will do.
Last edited by GeeTek on Tue Oct 09, 2007 20:40; edited 3 times in total
The WRT54Gv8 is roughly 1/10th the power of the WHR-HP-G54 (when running dd-wrt v24 RC3).
If you ever get a chance to measure the WHR-G125 please do as they seem to be selling plenty of them. I have five, 1 as an AP and 4 as client devices. It is a cheap way to use other channels as we are out of the country. Thanks again, Bill _________________ DSL from Windstream
I feed a friend 400 meters through trees to Wi Fi station taped to his window from my bullet hp feeding 15db pringles type antenna
Unfortunately I do not have access to an oscilloscope that can look at the raw 2.4GHz voltages... However, you do raise a good point. A spectrum analyzer cannot measure the 'peak power' of a single cycle of the carrier. I had not looked into the 802.11g satandards until today, but the modulation scheme of 802.11g will limit the RMS power we can get out of the WHR-HP-G54 to about 100mW.
802.11g uses Orthoginal Frequency Division Multiplexing (OFDM) to achieve the highest data rates (it reverts to complementary code keying at lower data rates). In OFDM, serveral (up to 52 for 802.11g) narrow band carriers are used to convey data (each carrier is then modulated using either amplitude modulation or phase shift keying (or both)). Since there are several carriers in the TX signal, the phases of the carriers will sometimes add consructively, and destructively, resulting in peak a voltage (or as you mentioned, peak power) that is much greater than the RMS voltage (power). This means that to prevent distortion of the TX waveform at time when these 'sub-carriers' are adding coherently, the TX RF section of the hardware must be able to handle the resulting larger voltages. This is why the TX amp in the WHR-HP-54G is rated for 30 dBm maximum output power, but is only being used for 20dBm max RMS power... to prevent saturation. This means that even if the AGC system can be bypassed to get more RMS power, doing so will greatly distort the TX waveform, and the resulting increase in power will only go to harmonics, and will not increase the TX power of the signal of interest.
802.11b, however, uses a single carrier with phase shift keying. Since there is only one carrier, there will not be any times when the peak power excedes 1.41 times the RMS power. So, if you stick to 802.11b speeds, the AGC can be bypassed, and the full power of the TX amp can be utilized.
Bill,
I probably wil not use the WHR-G125 in the near future. My application for dd-wrt is for long distance links (25+ km), so I need to be able to adjust the ACK (acknowledge) timing. This is not available in dd-wrt v24, which is necessary for the WHR-G125s.
Joined: 06 Jun 2006 Posts: 3763 Location: I'm the one on the plate.
Posted: Tue Oct 09, 2007 22:18 Post subject:
tparris wrote:
A spectrum analyzer cannot measure the 'peak power' of a single cycle of the carrier.
It can actually but that is not necessary. You have a lot to learn about your $25,000 machine. It also has a mode "H-11" I believe that is calibrated for wireless lan communications on 2.4 and 5.8 Ghz, which is specifically for wi-fi. Even if I am wrong on the mode number, the standard CRT read out will show the RF envelope in dBm which is the peak reading you are looking for. When you get the knobs set properly it should look a lot like what you see here...
https://fjallfoss.fcc.gov/prod/oet/forms/blobs/retrieve.cgi?attachment_id=569038&native_or_pdf=pdf
Look on page 29 and do the math on +25.2 dBm, which is the FCC's tested power output of this unit in "G" mode. It would behoove you and the rest of the forum to brush up on the operation of yer specan some more, my little homie. If you really do have access to one of these boxes and can actually run some accurate tests, there is a lot of joy to be had in checking these radios and various firmwares and output settings. So far you are as wet as a the wettest of the other "testers" that have ventured along and faded away. Give us a screen shot of the vertical deflection envelope. Seein iz beleavin....
Thanks for the vote of confidence GeeTek... believe it or not, I do know a little about what I am doing.
You will be happy to know that I can recreate the measurements in the FCC document pretty darn well. I do not have a camera with me tonight, but I will post some photos of the spectrum analyzer screen tomorrow, if that's what it takes. When I set up the measurments as the tester in the FCC doc, my results are very similar. Actually, for channel 6, I get 26.03dBm of band integrated power(over a 25MHz bandwidth centered at 2.437GHz, with a resBW of 300kHz, as in the FCC doc). I think this is the peak power that you are looking for [this is dd-wrt V23, sp2 set to 80mW Xmit power]. However... I would argue that the FCC measurements are a little bit off. The measurement in the FCC doc has the resolution bandwidth set to 300kHz which is comperable to the bandwith of the subcarriers in the 802.11g standard. This acts to make some of the power from the narrow band subcarriers 'leak' out to other frequencies (up to the 300kHz resolution bandwidth) near the sub-carrier freqs. If I decrease the resolution bandwidth to a reasonable value (say 2 or 10 kHz), then the integrated power is comparable to the powers that I posted previously. (in otherwords, to get a valid band power measurement, you need the resolution bandwidth of the scope to be much less than the beandwidth of your signal)
To give you details of how I measured the powers that I posted previously, I set the radio standard to 802.11g (similar to your suggestion), and used the burst power measurement function of the spectrum analyzer, which gives the min, max, and mean powers of a transmit pulse... I posted the mean values (which represent the RMS power during the pulse). I also used an averaging factor of 16 to get rid of some of the measurement noise. In theory, setting the radio standard on the spectrum analyzer to 802.11g sets all of the parameters to make a good measurment for that type of signal.
I really do think 100mW is the most that you can get out of one of these routers (without hardware modification, running dd-wrt)... even if FCC thinks otherwise.
... but I am open to suggestions if you want me to measure things a certain way.
Joined: 06 Jun 2006 Posts: 3763 Location: I'm the one on the plate.
Posted: Wed Oct 10, 2007 1:38 Post subject:
Sorry, I guess I was un-necessarily on the snyde side. There are a lot of different ways of looking at the RF envelope produced by an 802.11G radio. I am a general purpose radio tech, and not an expert on advanced modulation techniques. I beleive the "G" protocol employs a combination of both AM and FM modulation. The bandwidth and consequent amplitude of any given portion of the signal and sidebands is going to vary widely based on various parameters including air link rate and traffic levels. So as to not be arguing apples and oranges it is my understanding that these devices are generally rated by their peak envelope power, which would be measuring the center carrier frequency with minimal modulation. The maximum rf potential is what is of primary interest and is about as good of a measuring stick as any to use on comparing radios and wattage settings. The absolute wattage production (DC or RMS equivalent) can easily be calculated by multiplying the wattage going in from the wart by an approximate 81% device efficiency. This does not indicate the RF output potential since you have TX and RX cycles and some real complex modulation that is averaged into that figure.
tparris wrote:
..in otherwords, to get a valid band power measurement, you need the resolution bandwidth of the scope to be much less than the beandwidth of your signal...
I'm not so sure I follow your logic on that. You are simply limiting your scope of veiw (and measurement) to a thin slice of the spectrum which is only going to be what the modulation spreads over that portion of the band. I think you need to look at the entire signal at once, and measure center freq. to establish what the power output is. The top of the haystack is your maximum power output and is what everybody (including the FCC) is so interested in. _________________ http://69.175.13.131:8015 Streaming Week-End Disco. Station Ripper V 1.1 will do.
I could not agree more GeeTek. There are a number of ways to measre and define TX power. So... here are a few photos of actual TX spectra and some spectrum analyzer measurements. The DD-WRT community can interpret these data how they wish.
The second picture, http://gakona.gi.alaska.edu/parris/WHR-HP-G54/121_2126.JPG, shows a reproduction of the FCC test as shown on page 30 of the FCC document. DD-WRT was set to 80mW Xmit power. Here, the band power in the full displayed band of 25MHz band is 25.52dBm. This number is VERY similar to the FCC number (and judging from the FCC screen captures, this is likely how they got thier numbers. The spectrum, however, if a bit narrower (and 'dirtier') than in the FCC test, which makes me question if DD-WRT is limiting the data rate, or if buffalo changed how the radio works. I plan on getting some more WHR-HP-G54s next week, and I will test them with Buffalo's firmware.
The third picture, http://gakona.gi.alaska.edu/parris/WHR-HP-G54/121_2120.JPG shows the 'Channel Power' measurment that results when I set the spectrum analyzer to measure the 802.11g channel power (as GeeTek mentioned, the E4404B spectrum analyzer is set up to make lab quality measurments of a number of different standard radio signals, 802.11g being one of them). DD-WRT was set at 80mW. This shows a channel power of 20.01 dBm. This is how I got the numbers that I posted earlier.
Picture 4, http://gakona.gi.alaska.edu/parris/WHR-HP-G54/121_2122.JPG, shows the same measurement for 20mW setting.
I'll let the rest of the DD-WRT community debate how to use these data.
(and, when I get the new WHR-HP-G54s next week, I'll experiment with bypassing the AGC to see if we really can get more power out of these routers)
