Winter 2.4GHz Propagation

Every year when the first snow cover and temps below 32 degrees F occurs, I begin seeing a sharp increase in associations on my

802.11b access point (actually dhcp leases on the downstream router). The AP is tower mounted, with a switched duplexing amplifier and a 12dbi colinear whip at 30 foot elevation. The leaves have long since fallen so they are not a factor. I have been of the conviction that the improved propagation conditions are due to better noise figures in the receiver and preamp from the cold temps but I wonder how much the snow cover may be a factor.

Anyone else have this experience?

Michael

Reply to
msg
Loading thread data ...

Hi, Temperature inversion does wonders for long range reach.

Reply to
Tony Hwang

An increase in the number of DHCP leases can be caused by many things, none of which have anything to do with temperature. For example, one user that is playing with spoofing MAC addresses can eat up DHCP pool. Transient users, such as those driving by your AP in cars, with their PDA's, cell phones, laptops, game machines, iPod, etc, set to automagically associate with any AP, will eat up addresses. Address pool consumption is a good indication that something is happening, but it's not necessarily a change in sensitivity.

There is no snow in this part of California, but I do have a bad guess if it really is a sensitivity change. The tower mounted bi-directional amplifier is the most likely culprit as it's about the only thing that could affect receiver sensitivity. My guess(tm) is that the bias on the front end GaFET or MMIC is drifting. Stabilizing the gain of these beasts over temperature is difficult. Looking at the data sheet of a typical MMIC amp:

See Fig 5 for the NF (noise figure) versus Temperature. At 2.4GHz, it's about 4.0dB what I would guess would be 0C and 4.5dB at 85C. That doesn't look like much of a change, but it's enough to be noticeable when measuring sensitivity. At the same time, the gain drops from

18dB at roughly 0C to 16dB at 85C, which is the opposite of what you're seeing. Oh well. I checked some other common RF MMIC's and found similar changes. Unlikely, but still possible.

Another dubious possibility is the antenna. It's made from copper sheet, maybe some brass, and pieces of coax cable inside. Some collinears have phasing coils. Some are just plated on a PCB (printed circuit board). A maker and model number would have been nice. I'm especially interested in a "whip" antenna at 2.4GHz. Perhaps you mean fiberglass radome? Anyway, the copper and brass move around with temperature. I would think that the antenna would have been optimized for room temperature or slightly above, but perhaps you ended up with one that is better at low temp.

Reply to
Jeff Liebermann

The change is dramatic; all throughout the warmer seasons I see perhaps one or two stranger leases per quarter (90 days); when the temp first dips to 20 degrees F (and there is a snow cover -- this seems to be essential as 20 degree days prior to the snow show no change) I see twenty or more in a single day. Last night with temps dipping to near zero degrees F, there were a dozen more leases added.

Too bad I didn't photograph the board before installing it so that now I could know what parts it contained. These PDFs describe this amp (I am using the 2400cx):

formatting link
formatting link

formatting link

When is a vertical a whip? This one flexes a little ;)

Too bad there is no real temp data for this antenna; I also wonder about moisture freezing and changing characteristics.

Reply to
msg

"The requested object does not exist on this server."

With only 12dBi of gain, I don't think the internal parts are that critical. Thermal expansion isn't going to move things around that much.

It's possible that the receive amp is unstable and oscillating. Maybe lowering the gain with the lower temperature returned it to sanity. Difficult to tell from here with near zero hardware info. However, if it is for real, then the receive amp is the most likely culprit.

Reply to
Jeff Liebermann

It's been a bad week, I'm lucky to be typing complete sentences; the urls are fixed here:

Yeah, I might try to borrow a spectrum analyzer and do some tests in a temp chamber sometime; it is just such a dramatic change that your probably right that something funky is happening in there.

Reply to
msg

Do the leases correspond to actual device connections? As in, can you ping the devices once they've obtained an address?

Are the MAC addresses legit? That can tell you what vendor device is making the request. Start a log collecting the MAC addresses and see what trends appear.

Finally, consider if there are cold weather activities that might have larger numbers of attendees that usual. Any schools nearby? Lots of people with wifi-enabled phones and laptops might be a likely culprit. A log over time would help reveal that.

-Bill Kearney

Reply to
Bill Kearney

The photo shows a Maxim MAX850 GaAs FET power supply.

There is a discrete GaAS FET amp in there somewher. My guess(tm) is that it's for TX, not RX. There's also an LT1632 op amp, which might be part of the Tx AGC power control. There are a bunch of discrete RF devices on the board, but no easy way to identify them. Offhand, my guess(tm) is that the RX chain (two devices) is not temperature stabilized and probably drifts. However, without a better schematic or block diagram, there's no way to be sure. Got an FCC ID number?

Hmmm. I found an HMC174 GaAS FET T/R switch on the lower right corner.

The SMD database doesn't show anything for an "H5" marking, so I can't guess what that does. Kinda looks like a voltage regulator. Anyway, I don't see any fancy DC stabilization for the two RF RX xsistors, which may explain the drift.

Well, if there's a gain change, the spectrum analyzer will certainly show something. If the circuit is oscillating or unstable, you might see something. However, if it's just a change in NF, the SA won't show anything.

Reply to
Jeff Liebermann

Many have client hostname set and I can reliably ping those that renew their leases regularly. Several are Sony game consoles (determined from the UID) and there is a mix of iPhones, laptops and what appear to be desktops too.

Since the temps have hit -10F, machines known to be at a winter carnival area on the waterfront called 'Bentlyville' have been given leases, about two miles away.

See above. The dhcp leases database shows the trends and so far the only correlation that I've been able to perceive is inverse to temperature (and perhaps proportional to snow cover).

Bentlyville has much less traffic in the extreme cold and was even shut down yesterday, yet the number of new leases stays stable or even increases in the colder weather.

Sure, but the lease activity is inversely proportional to the temperature and in our arctic climate this seems counterintuitive.

Michael

Reply to
msg

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.