Alfa TX power measurement results (and some other cards too)

Discussion in 'Wireless Networking Cards' started by Syys Valo, 25 Jan 2014.

  1. Syys Valo

    Syys Valo New Member

    Joined:
    23 Jan 2014
    Messages:
    9
    Likes Received:
    4
    Hi all,


    since I have access to a calibrated spectrum analyser with peak power metering, I thought that you might be interested in some real TX power measurements. Let's see how the manufacturer's promises stack up, shall we :)

    All measurements are made directly from the antenna port(s), channel 6, 802.11g 6Mbit unless otherwise noted. All adapters are set to FCC domain.

    Alfa AWUS036H (1W edition)
    Promised: 1000mW (30dBm)
    Measured power: 43,6mW (16,4dBm)
    Analysis: Very bad. My iPhone 5S has higher TX output! To be fair, this adapter was designed for 802.11b and has over 200mW output @802.11b 1Mbit C6.

    Alfa AWUS036NHR
    Promised: 500mW (27dBm) / 2000mW (33dBm) EIRP.
    Measured power: 363mW (25,6dBm)
    Analysis: OK, 72,5% of promised output. It's also nice that 802.11n has practically identical output (but naturally better range at the same bitrate). Beware of manufacturers advertising EIRP instead of true antenna port TX power. This is borderline fraud, since the antenna gain can be made arbitrary large. I could advertise a bog-standard Centrino card to have 5000000mW output (EIRP), and still be "right".

    Alfa AWUS036NH **COUNTERFEIT FROM EBAY**
    Promised: 2000mW (33dBm)
    Measured power: 45mW (16,5dBm)
    Analysis: "with genuine hologram" my ass. PCB designed by a 5-year old and of course horrible TX power to boot. Got my money back from Paypal - if you have bought a counterfeit I recommend you to do the same.

    Intel Centrino-N 6235
    Promised: n/a
    Measured power: 32mW (15dBm)
    Analysis: The amplifier will physically crap out at around 22dBm, so don't bother flashing the NVM unless you absolutely have no alternatives..

    Ubiquiti Networks SR71-E
    Promised: 600mW (27,8dBm)
    Measured power: 155mW (21,9dBm)
    Analysis: If I had paid for this with my own money, it would be on it's way back to the retailer by now. You receive 25,8% of the performance you paid for and as a nasty bonus, the Atheros AR9280 chipset it uses is actively hostile to tweaking. And by nasty I mean that if it detects tampering of the channel/TX power EEPROM settings, it silently limits TX power to 10dBm (10mW). So, the driver displays 27dBm but you'll get 10dBm out of the antenna port.. Thanks, Ubiquiti & Atheros!

    Realtek dual MAC reference card RTL8192DE
    Promised: n/a
    Measured power: 339mW (25,3dBm). 802.11n @ 2 chains: 661mW (28,2dBm)
    Analysis: Very nice dual-band 2T2R mini-PCIe card. Unlike the SR71-E, this fits to half-sized slots and has higher TX power to boot! This card is best sourced from high-end APs.

    Are there any other settings/adapters you'd like to see measured? I have access to many mini-PCIe cards but not that many USB models, since I have modded my Zenbook to have rpSMA ports (so no need to drag a USB adapter with me).

    Peace,


    Syysvalo
     
    • Like Like x 1
    • Winner Winner x 1
  2. GeekDeveloper

    GeekDeveloper Active Member

    Joined:
    21 Dec 2013
    Messages:
    31
    Likes Received:
    2
    Alfa AWUS036H (1W edition)
    Promised: 1000mW (30dBm)
    Measured power: 43,6mW (16,4dBm)
    Analysis: Very bad. My iPhone 5S has higher TX output! To be fair, this adapter was designed for 802.11b and has over 200mW output @802.11b 1Mbit C6.

    What? have you set the card to full power using iwconfig under linux?
     
    • Winner Winner x 1
  3. Syys Valo

    Syys Valo New Member

    Joined:
    23 Jan 2014
    Messages:
    9
    Likes Received:
    4
    Yes, all devices were tested at highest power available without resorting into EEPROM/NVM hacking. I re-tested the 036H with both FCC domain set and 0xFF (software set domain) with CRDA limit at 40dBm: results are within +-0,3dBm. Additional testing in Windows 7 x64, power control set to 100% gives identical results too.

    It is clear that this card doesn't handle OFDM well, at 54Mbps the TX power drops under 15dBm and there is jitter everywhere.
    The manufacturer simply inflated (=lied) about the TX power capabilities in order to get more sales, unfortunately this is still often considered as business as usual in Taiwan :(

    BTW this card was first marketed as a "500mW" card, then later when other "500mW" cards appeared on the market, this was magically transformed into a "1000mW" card - without drawing any additional current from the USB port. Wish I could run my 28dBm Realtek card with magic too, as with electricity it runs HOT after uploading >100GB.
     
    • Winner Winner x 1
  4. Kostja_V

    Kostja_V Well-Known Member
    VIP

    Joined:
    14 Jan 2013
    Messages:
    24
    Likes Received:
    9
    Watch this video on YouTube.

    Your results differs from, at least, this video...
    But its really interesting to get REAL TX power results of most popular adapters.
     
    • Winner Winner x 1
  5. kevsamiga

    kevsamiga Well-Known Member
    VIP

    Joined:
    15 Sep 2013
    Messages:
    406
    Likes Received:
    126
    These tests have been done before on the 036H, with totally different results to yours, with the drivers you can set iwpriv wlan0 highpower 1 with (r8187, ieee802.11 stack) :

    Google Translate

    This is a kinda interesting article, but it doesn't surprise me. No regulation means they can slap anything on the box and get away with it. You can't blame them, it's not just you that isn't allowed to use these higher powers, the sellers aren't even allowed to sell them with the potential to be rigged past the country of shipment power limits. And believe me I have tried to bypass reg domain limits. They have it all cooked up and worked out.

    The only way is a linear amplifier, which turns out is also illegal, but not to sell, just use.

    I already mentioned about the 036NHR in this post here :

    New | ALFA AWUS036NHR v2 | Page 2 | Xiaopan Forums

    Post #24, I said that the 036NHR has the strongest output in the test reports by the FCC.

    Pity it's a crap adapter for everything else...

    But in any case power isn't everything or the only relevant selling point, because if your in the ETSI domain like I am, nobody can transmit back to you anyway, unless everybody else goes illegal and amplifies their ends, so cranking up the power on only one side makes little, if any sense here.

    And well, if everyone gets the idea and turns up the power, you have one great big interference ridden :censored hole, where you can't get a signal at all above the noise.

    I think the 036NHA is capped @ 40mw too, because even tho it shows set at 100mw in the config utility, it won't display anything past 40mw.

    As for the 036NH, it would be good to find out the results against the NHR if you can get buy, borrow or steal a legit one.
     
    #5 kevsamiga, 25 Jan 2014
    Last edited: 25 Jan 2014
    • Winner Winner x 1
  6. Syys Valo

    Syys Valo New Member

    Joined:
    23 Jan 2014
    Messages:
    9
    Likes Received:
    4
    The person in the video is using a standing wave/integrating power meter meant for VHF/UHF antenna testing.
    That kind of meter is cheap and easily available, but unsuitable both for microwave frequencies and high-speed digital transmissions.
    I am using an Agilent E4445A spectrum analyser, which is about three orders of magnitude more expensive.

    For further proof, I measured the power consumption of the Alfa AWUS036H.
    When idle, the power consumption was 400mA @ 5V = 2W
    When transmitting at full power @ 802.11b 1Mbps (highest TX power the adapter achieves) the power consumption was 510mA @ 5V = 2,55W.
    There is no way to get 31dBm (1260mW) of RF energy out with an input of 550mW, no device in the universe has an efficiency over 100%!
    Or maybe you could actually create infinite amounts of energy just by chaining these Alfas together :p
    Getting ~200mW of RF energy out of 550mW electrical input is not that bad, sadly it's also the best the adapter can do.
     
    • Winner Winner x 1
  7. kevsamiga

    kevsamiga Well-Known Member
    VIP

    Joined:
    15 Sep 2013
    Messages:
    406
    Likes Received:
    126
    I don't doubt your testing my friend, it's not even my article. I was just wondering why the 036NHR was the only adapter to come with a Y USB cable, when the 036NH has the same specs listed on output anyway. :)

    Sure the 036H had a Y cable early on, but only for use on old laptops from the '90's where the USB ports didn't provide 500ma.

    I did my own testing of the NHR and H's power control drivers under Windows. For the most part they don't do anything, it's just a conjuring trick. I found out that the nominal value for the non power control drivers at least for the 036H case is 82% power control, which is what you have set if you don't use the power control versions.

    No there is not 30dbm of power or more than 500ma available from a single USB port.

    ....that is unless like me your using an external powered USB hub with only one of the ports connected where with a 2A mains adapter will happily power a single device up to 2000ma as long as nothing else is connected to the other ports on the hub. With a USB Y cable such as comes with the 036NHR then you have 1000ma current draw available for your plugged in device. It's simple maths. If a device can't draw more than 500ma, then it's not going to without blowing the poly fuses or limiting itself to some kind of low power mode to keep the current under the 500ma if your not using a Y cable or a powered hub. I don't know the actual answer to this because I aint rich enough to buy a spectral analyser, but it's worth a spin.

    Did you test under this extra powered configuration out of interest. Or were they all just plugged into a single USB port ? If you did, the testing methodology might be potentially flawed from the ground up. I don't know whether extra power supply matters or not, I didn't build the things.

    This will give an answer one way or another once any 500ma power restrictions of a single USB port are no longer a factor in these tests beyond any shadow of a doubt...

    And will also inform others if they are just pissing good money away on using a powered hub to plug in their weapon of choice, that makes 0% difference. :)

    Oh, and I just remembered, setting the TX power under Linux for the 036NH or any Ralink 3070 adapter doesn't actually do anything, sure it reports as 30dbm, but it's the drivers that don't effect any changes even though it reports the power change.

    So unlike the 036H with it's iwpriv highpower 1 setting, changing TX power on 036NH does diddly squat.

    I also found this snippet from 2010 which may already indicate the genuine 036NH's output:

    "Larry Butler Ampz Electronics says:

    The full test reports of both are online. The REAL output power of 036H is only 58mw, not 1000mw and 036NH is much better at 200mw, not 2000mw. Your laptop will be lucky to put out 20mw into its horrible PC board antenna buried behind the display so these units are a vast improvement from the embedded radio in any laptop/netbook.

    FCC regulations state that the EiRP (Equivalent isotropically radiated power: adapter + antenna - cabling loss) of ANY wifi adapter to keep the cable, phone and cellphone lobby happy is limited to 1 watt (1000mw) EiRP or 30 dBi. (See dBm to Watt Conversion Table.)

    The realtime test reports were way below that number which is why they were approved.

    There is no 2 watt wifi adapter in the USA. It would never pass FCC acceptance testing.

    Even a lie, my 036NH and 036H are fantastic units for the price and performance is very impressive.

    Those wanting the secret schematic will find it down-loadable as pdf on the 036NH at the FCC website above.

    The RF power amplifier is rated at +26dbm (about 400mw) and the whole thing barely gets warm in heavy use, which is very easy on the laptop batteries."

    On another note, there seems to be a significant difference between the 58mw output written in that article, and the 43.6mw you recorded in the order of 14.4mw bias. They did not specify which version of 036H was tested, nor did they say whether they tested a powered USB hub or Y cable, so the answer to this shall we say remains at least "interesting".

    But actually the FCC testing reports seem to indicate 70mw output.
     
    #7 kevsamiga, 25 Jan 2014
    Last edited: 26 Jan 2014
    • Winner Winner x 1
  8. Remington

    Remington Well-Known Member
    VIP

    Joined:
    17 Dec 2012
    Messages:
    168
    Likes Received:
    24
    I wondered the same thing when I got my NHR. Like I mentioned in another thread I'm using that Y USB cable with my current adapters with one end plugged in a powered hub so they get 1000ma. It would be interesting for someone could test all the Alfa's with a Y cable to see if there is any difference.
     
    • Winner Winner x 1
  9. Syys Valo

    Syys Valo New Member

    Joined:
    23 Jan 2014
    Messages:
    9
    Likes Received:
    4
    Very well said kevsamiga. Inflated/counterfeit cards, inflated/counterfeit antennas, I even see counterfeit cables every month.
    In addition, like you said having TX power reserves over others matters little unless you only need a one-way link or control all the endpoints.
    Beamforming capabilities, good STBC/SMPS and Greenfield support are seemingly never advertised outside professional circles.
    Personally, I'm mainly using higher power levels to simply keep the bitrate up for longer distances in my own WLAN and for a ~1,5km point-to-point link.

    I think I'll be able to borrow a 036NH next week, I'll check it out then. I might need to do an ETSI->FCC mod, do you remember the exact chipset used (RT3072L)?
     
    • Winner Winner x 1
  10. kevsamiga

    kevsamiga Well-Known Member
    VIP

    Joined:
    15 Sep 2013
    Messages:
    406
    Likes Received:
    126
    A 4 port powered USB hub with a mains adapter would be better, there are those models such as the 036NH, and 036NHR stating 2000mw over the 1000mw 036H, even if the claims as we suspect are highly unlikely to be correct, and leaves no doubts.
    --- Double Post Merged, 25 Jan 2014 ---
    036NH has 3070L, 3072 is the 2TR version that has beam forming.

    Also please try to get hold and test 036NHA, and Alfa 051 (gold) adapter for completeness.
    I wanna see how badly the 051 has a hissy fit using 802.11g OFDM output mode compared.

    And maybe a test of the 036NHR v2 to determine if the redesign is just a) cheaper to make or b) improved upon v1. (I'm assuming you tested only v1 model)

    The PCB size of the NHR v2 is half size of the v1's board and blue not green.

    Thanks.

    P.S.

    The TX power figures might come to the top of the ALFA oriental dung heap for the 036NHR, but the reception over the others is still utterly crap.:)
     
    #10 kevsamiga, 25 Jan 2014
    Last edited: 25 Jan 2014
    • Winner Winner x 1
  11. Syys Valo

    Syys Valo New Member

    Joined:
    23 Jan 2014
    Messages:
    9
    Likes Received:
    4
    Thanks for your research on the power control Windows drivers and the Ralink Linux drivers, very interesting (and shady!) stuff. Reminds me of a Windows audio player that had a big knob labelled "harmonic balance". Users would swear the music became much more detailed and easier to listen and even argued about the best settings for different genres on a forum. After comparing recorded files it was revealed that it did absolutely nothing.. it was then that I learned that the Internet was created for converting bits into rage ;)

    Anyway, back to the testing. I'm using a 5V, 20A bench power supply for the USB adapter tests so they're not current limited. I can do more power consumption measurements next week to check whether Y-cables are beneficial. I'd predict no, as otherwise the device would have to somehow measure the available current without tripping overcurrent protection in the host and dynamically change its power limits. Possible, yes, but too expensive in implement?
    I have an external HDD that requires a Y-cable, but it simply won't work at all without one (can't spin up).

    The difference between 44mW and 58mW is about 1dB, so it could be just variation and/or methological differences. FWIW the Agilent has better than +-0,2dB accuracy, so it can't explain the whole difference.
     
    • Winner Winner x 1
  12. kevsamiga

    kevsamiga Well-Known Member
    VIP

    Joined:
    15 Sep 2013
    Messages:
    406
    Likes Received:
    126
    Well there goes all the theory ppl about anyone needing a powered USB hub for the 036H or even a Y cable, or using one for any other ALFA adapter for that matter...cheers !

    Pure marketing hooey about anyone needing a powered hub, when in fact the adapters simply can't, or won't, or will never be able to chuck out the power levels necessary to require one.

    Not once, not ever. It would seem like everybody using special Y cables and hubs all this time on their ALFA adapters have been living in a dream world.

    I thought as much anyway, it's hardly an external HDD.

    I guess someone from the ALFA adapter factory got together and sat down with the people in the USB hub factory, along with the Y cable making factory, and said unto them:

    "hey, you know what, I've got a real good idea..." :cowboy:

    On a more serious note, whilst the power output figures for ALFA's may well be an accurate assesment, setting them aside for a moment, there are still huge variations across all the adapters, in the sense that even though the 036NHR has tested the most powerful here, there is still variation across all the adapters including connection stability, seeing networks others can't, being actually able to connect to networks others can't of which the NHR is not exempt due to higher power, and speed through the connection once connected, sensitivity factors etc. Linux suitability, and variations in driver quality etc. AP density, interference rejection. Then you have to deal with the fact there are no two radios or antenna's manufactured alike due to variations in components and design.

    So in the grand scheme of things, TX output power is not the be and end all of determining your effective range.

    The allowed power in Europe is only 100mw/20dbm anyway, but it really ends at 17dbm max because they take into account 3db of cable loss factored in the output already. So at 16.4dbm/16.5 your already doing quite well compared to the mini dongle adapters, because your never 100% efficiency of output anyway. Antenna gain will not influence output power neither.

    But people and impulse buyers still seem drawn to the flashy figures on the boxes from ALFA like pigeons. 2000mw ? Well the truth is out there...it's the Chinese ones though that still make me laugh that claim 6000mw and 48dbi.

    Still, even if the figures are over inflated it's no big deal, because for what they do for your connection compared to cheap and nasty dongle ones, or inbuilt laptop wifi, they still can't be beaten by pretty much any other junk out there that also sports ridiculous claims to ship their respective turds in a box.

    I think you will find your going to end being disappointed with all tests of various adapters, because nobody else either makes 1000 or 2000mw adapters anyway, and if they did they would be a) illegal b) you would probably microwave your testicles to a crisp and c) everybody would be buying them instead of falling for all this hyped up junk. But no, they shouldn't outright lie about it.

    Maybe high power relative to two paper cups, and a wet piece of string. In my old CB days before '86 when you got radiation sickness of all the RF there was a switch to knock the power down from 4W to 0.4w. I could still get pretty far even on 0.4W, but then the frequency was 27Mhz which penetrates easy. Not so lucky for low power at 2.4Ghz.
     
    #12 kevsamiga, 25 Jan 2014
    Last edited: 25 Jan 2014
    • Winner Winner x 1
  13. GeekDeveloper

    GeekDeveloper Active Member

    Joined:
    21 Dec 2013
    Messages:
    31
    Likes Received:
    2
    Ok, Now I am totally screwed after reading this article. I dont know why these Alfa cards are still considered good?
     
    • Winner Winner x 1
  14. kevsamiga

    kevsamiga Well-Known Member
    VIP

    Joined:
    15 Sep 2013
    Messages:
    406
    Likes Received:
    126

    Nah....... I'll say it again, power isn't everything. That's why I'm now mostly using an NHA
    these days (which is supposed to be the weakest adapter) and I put the 036H, NH, and NHR away for the time being.

    If you want more power it's easily solved with a linear amp or burner, but the problem is that while it amplifies the signal, it will also amplify the noise !

    Watch this video on YouTube.


    They are still orders of magnitude better than the reams of endless worthless junk out there on deal extreme and amazon.

    I even have a TP-Link WN722N with the same Atheros chip AR9271, but it isn't even close at picking up the same number of networks with the same quality as the 036NHA that has the exact same chip.

    It's kind of ironic that the most powerful tested adapter (036NHR) also happens to
    be the worst adapter ALFA has produced in most peoples minds.

    And anyway, you should be spending your money on rigging up decent antenna's because that's where the meat and potatoes really are.

    Considering just power alone, is only 1 element of the big wifi soup.

    And speed doesn't matter for wardriving, sensitivity, compatibility, and functionality does.
    --- Double Post Merged, 26 Jan 2014 ---
    I still want to believe......, but I have since yesterday discovered even more variance in your testing from FCC testing listed here on this WikiDev page appertaining to the 036H, so pardon my ignorance for going the extra mile over it :

    ALFA Network AWUS036H - WikiDevi

    "The FCC test report lists the max output power as 24.66 dBm (292.42mW) for 802.11b and 19.49 dBm (88.92mW) for 802.11g"

    That is too large a difference in comparison figures from your analysis from the FCC testing side by side to account for any margin of error of the Agilent spectral analyser.

    Infact it's practically double the milliwatts from the FCC analysis and testing, compared to your own tests which weighed in for the 036H at a paltry 43.6dbm in OFDM.

    Quite clearly, someone has their figures wrong and way off here, either that or broken equipment. The question is....who has their figures right ? You and your analyser, or the FCC ?

    Or did ALFA just infact ship a doctored version of the 036H off for FCC testing, or have they (instead) as another plausible conspiracy theory, completely rigged and downgraded the 036H V5's output compared to earlier model versions ?... :)

    It's feasible, but it doesn't make any sense to be going backwards in performance.

    No offence to you, your tests or any of your equipment, but there are too many conflicting reports out there floating around the tinternet regarding the venerable 036H's output.

    They're easily found for those looking hard enough. I'd just love to know the definitive (and correct) concrete answer as to the 036H's output is all.

    Since there is obviously more than one model of the 036H out there, and then we have infront of us completely different figures floating around from the FCC over the 036H on the one hand, and then other sources measuring half of what the FCC reports over the 036H to go along with them on the other.

    It still remains confusing as to which is correct, which (if any) model is underpowered, and still leaves no clear answer !
     
    #14 kevsamiga, 26 Jan 2014
    Last edited: 26 Jan 2014
    • Winner Winner x 1
  15. GeekDeveloper

    GeekDeveloper Active Member

    Joined:
    21 Dec 2013
    Messages:
    31
    Likes Received:
    2
    thanks mate that clears everything :)
     
  16. angel13

    angel13 Active Member

    Joined:
    9 Feb 2014
    Messages:
    16
    Likes Received:
    0
    • Winner Winner x 1

Share This Page

Loading...