Some of you may know I did a survey of 900MHz recently, along with the real-world modem tests.  The spectrum analyzer inside the modem showed a very dire picture of the band, with the noise floor averaging about -60dBm.  Today, in preparation for 1.2GHz antenna installs, I ran some coax to the roof and was able to see the 900MHz vertical there directly on a proper spectrum analyzer.  The spectrum story is quite different!  Here's a view of a typical spectrum from 902-928MHz:

floor

The reference line at the top is -20dBm and each vertical graticule is 10dB.  The noise floor looks to be about 7 graticules below -20dBm, or -90dBm.  There is about 2dB loss in the 45ft feedline, so let's call the floor -88dBm.

There are however some sparse transmissions in this band.  I wish I could find a way to measure their duty cycle or plot their average power.  Sadly the best I can do is to record their maximum power using the peak-hold mode of the spectrum analyzer.  I captured about a 10 minute sample of air time, and here are the results:

peaks

The bright dot is the peak marker, and it reads -42dBm, which is really -40dBm at the antenna after you account for the coax loss.  The peaks are all suspiciously uniform in power, so this may be a single transmitter just jumping around the spectrum with each burst.

I can't explain why the Mikrotik 9HPn reports such a vastly different spectrum.  It may be mixing the 860-900MHz cell band into the 900MHz band?  I need to perform some further tests to figure this out.  There are definitely strong signals in the 860-900MHz range (not shown in these photos).

--Bart