While I’m not the first to post on the topic of USB3 hubs causing interference in 2.4 GHz, I thought I would add my experience to the mix. Here are a few quick reads for your interest:
I first noticed something strange in Ekahau Site Survey, using a 7-port USB 3.0 hub from Orico. My signal-to-noise ratio was negative (noise higher than signal)! What the heck, eh? The ever-helpful WLAN pro community on Twitter was quick to respond to my query, informing me that USB 3.0 hubs are not to be trusted for surveying. I followed up recently with some testing.
I was at the family cabin, where we don’t have our own internet connection. The neighbors do, however, with their router set to channel 6, and I can pick it up in our kitchen. I also set my phone’s hotspot to channel 1. My test setup consisted of:
- Two networks in 2.4 GHz, each on different channels
- A mid-2015 MacBook Pro Retina also running Windows 7 in Parallels
- A new Dell XPS 13 running Windows 10
- Ekahau Site Survey 8.5.1 w/ Ekahau NIC-300 (Proxim 8494)
- Metageek Chanalyzer w/ WiSpy DBx
- WiFi Explorer
- Acryilic WiFi
- Orico 7-port USB 3.0 hub mentioned above
- Aluratek USB 2.0 4-port hub
With both laptops side-by-side on the kitchen table, the networks looked like this in WiFi Explorer on the MBP.
Here is the view in Acrylic WiFi on the XPS 13.
Next, I took measurements in ESS on the MacBook with the NIC-300 connected with no hub, with the USB 2.0 hub, and finally with the USB 3.0 hub. Because the NIC-300 is plugged into the hub, the placement of the hub (velcro’d to back of display, sitting on desk, etc) is inconsequential. Here are the results in that order.
So, when attached to the MacBook Pro, the USB 3.0 hub was causing a roughly 20 dB increase in the noise floor! It also made the RSSI appear around 10 dB lower than with the USB 2.0 hub or no hub. I did the same tests using the XPS 13. Here are those results.
Using the USB 3.0 hub and Dell XPS 13, there was no effect. I thought I saw the noise bounce up for a split second, a few times, it didn’t last. However, that was enough for me to write off the USB 3.0 hub for surveying.
The next part of my test involved using Chanalyzer to look at the change in the noise floor depending on the position of the USB 3.0 hub, using both the MBP and XPS 13. I stuck the DBx on the velcro on the back of the displays and plugged it into a port on the laptop. I then took measurements with the hub sitting on the desk in front of the laptop, and again with the hub stuck to the velcro as one would do whilst surveying. I was focused strictly on the noise floor for these measurements. The following images show those setups.
The next image shows the 2.4 GHz spectrum without the USB 3.0 hub plugged in. The noise floor is in the neighborhood of -100 dBm, and you can see faint outlines of the signatures from the router on channel 6 and my phone on channel 1.
In the next image, look at the noise floor with the USB 3.0 hub plugged in but sitting on the desk in front of the laptop. The noise floor is slightly higher. I also remembered to turn off my phone hotspot and most of the other devices in the area.
Now, see what happens to the noise floor with the USB 3.0 hub stuck on the velcro, right beside the DBx, where it would likely be for a survey.
Here’s another look from the RTFM in ESS. The noise floor in the spectrum side on the left doesn’t look as high as it shows up in the scanning side on the right.
The noise floor has come up 10 dB. Now, a noise floor of -90 dBm doesn’t seem that bad for a production environment, but when you pair these results with the noise levels reported by ESS, there is definitely a problem. I’d love to repeat these tests using a Cisco AP with Clean Air, to see if the increased resolution picked up something else going on in the spectrum view.
The next two images show the spectrum with the hub and DBx plugged into the XPS 13. The first image is with the hub on the desk. The second is with the hub on the velcro, beside the DBx.
As with the results from the ESS measurements from the XPS 13, there is no perceivable change to the noise floor. Based on the Intel white paper, my guess is that the XPS has betting shielding or grounding on the USB 3.0 connectors, than the MacBook does. Now, I don’t really survey with my MBP due to the weight difference, but I still would rather avoid any chance of having my data messed with on the XPS.
So, I have purchased the 4-port USB 2.0 hub mentioned earlier. I definitely cheaped out on this hub, and it hasn’t impressed me with it’s build quality. I also overestimated the length of the USB tail, hence the additional right-angle extension cable. These images show what it looks like on both laptops.
It’s not very elegant, so I’m on the lookout for a good 7 port USB 2.0 hub. When I find one, I’ll be ordering a set of Hub Holsters to replace the velcro. If anyone is using something they like, please let me know in the comments.
In the last few weeks I’ve heard of some people say they have no issues with their USB 3.0 hub, and others say they definitely won’t survey with USB 3.0. If you take anything away from this post, let it be that you always test your equipment before using it in a production environment.
Thanks for reading, and please comment!