Let's clarify some things here before this starts wandering all over the
1. I wasn't trying to splits tenths of a db in terms of impact. It was
intended to be a simple relative comparison with as close to real life
audio as I could get.
2. There are multiple ways to improve S/N ratio. One is the
potentially large pattern benefit (noise discrimination off the side or
rear) that even a slight improvement in forward gain might give.
Another is the use of narrow filters. Another is a rig with good DSP
noise reduction. Another is the use of binaural means to tap into your
gray matter filters. None of that was relevant to what I was trying to
show, which was merely the difference in intelligibility of different
S/N's ... using real life noise ...once the other factors had played
3. I used some nasty 80m band noise for the background specifically to
avoid "uniform" noise (white, gaussian, etc). The sputter you get on
the air is much harder to receive through because it takes very little
burst to obliterate or blur a few dits. Some simple math will tell you
where a db is relevant or not relative to white noise, but I tried to
shift the comparison to a more practical context. In my opinion, and I
invite the experts to correct me here if I am wrong, a db or two
difference against an unstable background can have a larger impact on
intelligibility than against white noise.
Jim Lux wrote:
> At 04:57 PM 5/29/2007, WA3GIN in Alex. City, VA wrote:
>> I think your efforts should be rewarded but the test is flawed.
>> Try using QRP and an attenuator. Get a distant station to record the
>> transmissions...one with and one without the 1db attenuator in the
> This too, isn't necessarily a fair test. It will changed the received
> SNR at the receiver IF the atmospheric noise dominates over the
> receiver noise, so in that sense it's fair as a test of "is a dB on Tx
> worth it".
> However, it doesn't address the effects of directivity on receive..
> increased gain on receive implies smaller beamwidth, which might
> reduce received noise. As long as the loss in the system isn't so
> much that receiver noise starts to be significant, you could make the
> tradeoff of less antenna efficiency (more loss) in exchange for more
> But it's that question of "is 1 dB worth it" that makes this all fun..
> just because it's somewhat subjective, and not all dBs are the same.
> Rarely are we dealing with nice symmetrical signalling in a additive
> white gaussian noise environment, where you could mathematically
> calculate the effect of a 1dB change.
> Jim, W6RMK
TowerTalk mailing list