How We Test Wireless Products - Revision 9

[thiggins]

Our V9 wireless test process introduces new test clients and new ways of using them.

Read on SmallNetBuilder

 

[RogerSC]

This brings up a question that I've been wondering about...now that the FCC does allow the same power in the lower segment of 5GHz. channels as the upper segment of 5GHz. channels, do newer routers like the Netgear R7800 take advantage of that? Can older routers take advantage of this via new firmware, or only newer routers? Not sure if the firmware that you can flash has control over radio power or not?

Curious about this, since you mentioned it *smile*.

Great article on your testing methodology, by the way. Appreciate your sharing this.

Thanks!

 

[RMerlin]

now that the FCC does allow the same power in the lower segment of 5GHz. channels as the upper segment of 5GHz. channels, do newer routers like the Netgear R7800 take advantage of that? Can older routers take advantage of this via new firmware, or only newer routers? Not sure if the firmware that you can flash has control over radio power or not?

I don't know how programmable Qualcomm's chipsets are, but in the case of a Broadcom-based router, the router manufacturer should be able to reconfigure any available channel and output power levels by compiling a new CLM file. This contains a table with all that power configuration data, and can be either embedded in the wireless driver at compile time, or loaded externally at boot time.

 

[TheLostSwede]

This makes a lot of sense, as there are so few consumer devices that are capable of more than 2x2.

 

[thiggins]

Any changes to transmit power levels require FCC recertification. I know of no older routers that have gone through the process. Please correct me if I am wrong.

New routers have been shipping with the new transmit power levels since the rules went into effect.

 

[dogzdogz]

I'm wondering why still use Chariot for 2x2 throughput testing since you already have IxVeriwave. Shouldn't keep the test bed as similar as possible ?

 

[thiggins]

I'm wondering why still use Chariot for 2x2 throughput testing since you already have IxVeriwave. Shouldn't keep the test bed as similar as possible ?

Tbe Veriwave is a very powerful system, but has three limitations when it comes to the type of testing we do. Its input range is limited to a 10dB range from around -25 to -35dBm. So it can't be used with external programmable attenuators for rate vs. range testing.

The second limitation is the WaveBlade does not have a mode where it automatically adjusts MCS rate in response to changes in RSSI.

The last limitation is that it can't be band-steered, i.e. you must set the Waveblade to a specific band and channel. This last point doesn't affect throughput vs. attenuation testing, but prevents the Veriwave from being used for testing "Smart Connect" and other automatic band-steering / load balancing technologies.

We are working with Veriwave to overcome these limitations. But for now, using a real STA was the best solution.

 

[sfx2000]

Any changes to transmit power levels require FCC recertification. I know of no older routers that have gone through the process. Please correct me if I am wrong.

New routers have been shipping with the new transmit power levels since the rules went into effect.

They would have to go back thru FCC testing again - most vendors likely won't due to cost concerns (very expensive)

It's beyond the scope of a "permissive change" which is a paperwork job... (i.e. we changed part X for part XrevB due to supply chain - form/function is same" type of thing).

 

[sfx2000]

I'm wondering why still use Chariot for 2x2 throughput testing since you already have IxVeriwave. Shouldn't keep the test bed as similar as possible ?

There is some overlap, but they do test different areas of the device.

 

[sfx2000]

Saw the D-Link review - nice shot of the chamber with the new radio antennas..

Quick question - the tether - is that ethernet and power?

The turntable - scripted/automated?

Anyways, really nice setup...

 

[sfx2000]

It would be interesting to see power consumption at the wall socket while the device under test is being stressed hard by both the wireless and router tests.

I've got a feeling some vendors will do a better job than others, even across the same vendor with different chipsets.

 

[thiggins]

The "tether" is not new... check the chamber shots in past reviews. Yes, it's for power and Ethernet. The cables are run through expando sleeving to prevent getting tangled with router antennas.

Turntable is also not new, but the platter is. Bigger than the previous one and with markings. Yes it it motor driven. This is described in the test process description.

 

[sfx2000]

I've seen the tether before in old lab shots, and assumed that it was power/network connections, but the danger is making the assumption

 

[sfx2000]

Tim -

For V10 of the Wireless Test Plan - the WaveTest 20 can do load/capacity testing, and that's part of the script package - now that we're getting ever more connected - not just Computers/Tablets/Phones/STB's, but also all the InternetOfThings (smarthome, quantified self, connected car, wearables) we're going to start seeing not 10 or 20 clients on the AP, but dozens of them... and these IoT thingies, many of them pop up, attached, transmit, and then shutdown at random..

It would be a slight modification perhaps of the MU-MIMO test plan, but scaled up a bit... this kind of relates to the discussion thread on the EA9500 - but the load/capacity aspects of a device under test, and the results would be great feedback to the vendors, and good info for potential customers.

It the objective testing that SNB does, along with strict editorial honesty, that many come here to SmallNetBuilder - keep up the great work - much appreciated.

(now that my nose is suitable brown )

 

[thiggins]

Load / capacity / stress testing is in the works. But need to work through some challenges with Veriwave and decide on the approach.

 

[sfx2000]

Load / capacity / stress testing is in the works. But need to work through some challenges with Veriwave and decide on the approach.

It's hard work - both for defining the test requirements, and also building the automation scripts behind it - been there myself helping out our performance test team.

 

[JClu]

Our V9 wireless test process introduces new test clients and new ways of using them.

Read on SmallNetBuilder

Hi
From How we test wireless R9, you mention you reconfiguring the test bed to use two channels of the octoScope quadAtten as fixed attenuators. Did you mean you setup those two channels of attenuator as 0dB and use it's default attenuators as fixed atteenuator?
Did you know the initial total path lost from station to DUT in 2.4G and 5G. I would like to calculate the max theory distance by your report.

Thanks!

 

[thiggins]

Two of the attenuators are set to 10dB for 2.4 GHz. The other two are adjusted to increase path loss.

I do not have a path loss estimation for either band.

 

[tiger_liem]

When will the result out?

 

[CHand]

Hi,
Do you think if the RAM and HDD of this Leveno PC will affect the test performance? I would like to buy the same one. Could you provide some information about this?( how many RAM/ HDD or SSD?)