Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[roseway]

A question has been raised concerning the Y axis units in QLN graphs. Every reference I've seen gives these units as dBm/Hz (i.e. power divided by frequency) but that does seem a little strange, and it's been suggested that the units should really be dBm. Can anyone throw any light on this?

[renluop]

Probabl of minimal use, if at all, but by putting dBm/Hz in to Google I found

http://www.electronicspoint.com/threads/dbm-dbm-hz-what-is-the-relationship.72492/

[JGO]

Yes White Noise Density does puzzle people !

Fundamental thermal noise due to electron motion is equal to kTB where
k = Boltzman's constant
T = Absolute temperature
B = Bandwidth in Hz

so the noise per Hz is - 174dBm/Hz (or - 204 dBw) at room temperature so in a channel B Hz wide is + B dB more .
For QLN graphs the conversion factor from channel noise to noise density is the equivalent rectangular noise bandwidth of the channel width, about 36 dB more for 4 kHz.

Roseway, please confirm the 4 kHz !

REIN is not white noise so strictly the density is the equivalent white noise giving the same average  per channel.   
 
 

[Bald_Eagle1]

JDSU use dBm/Hz for their graphs, as I do for mine (see attached extract from one of their documents).

[roseway]

Many thanks for the comments.

Roseway, please confirm the 4 kHz !

In xDSL the sub-channel (tone) bandwidth is 4.3125 kHz (if that's what you meant).

So these values are actually White Noise Density (noise power per Hz of bandwidth), and dBm/Hz is the correct unit.

[burakkucat]

Probabl of minimal use, if at all, but by putting dBm/Hz in to Google I found

http://www.electronicspoint.com/threads/dbm-dbm-hz-what-is-the-relationship.72492/

I thought I had just about understood what was being discussed until a contributor got on to comparing the weight of a pig relative to an elephant per fly fart when my brain gave way . . .     

[renluop]

I doubt the fly fart would be noticed, were it standing beside the elephant. But the noise would likely zilch everyone's SNR, no to say his system.

[JGO]

Roseway, please confirm the 4 kHz !

In xDSL the sub-channel (tone) bandwidth is 4.3125 kHz (if that's what you meant).

So these values are actually White Noise Density (noise power per Hz of bandwidth), and dBm/Hz is the correct unit.

In all systems the bandwidth is shaped because an ideal rectangular bandwidth is impossible, (it would cause infinite delay for one thing !)  so I would expect the noise bandwidth to be slightly less than 4.3125.
If you are looking at comparative results with the same system it doesn't matter if you quote noise as per Hz or per channel but beware of comparing with someone else as there may be 30 odd dB because of units. Then "We have a problem "  !!!

[roseway]

I see (I think). In this case we're talking mainly about DSL monitoring programs, and the common factor is the Broadcom CLI. Assuming that Broadcom are consistent in the way they define the reported QLN values, then I suppose that it's best that we all use the same terms. But it's possible that the JDSU uses a different definition, so we can't directly compare JDSU QLN graphs with those in in our monitoring programs.

I seem to recall someone posting a link to a JDSU manual. Does anyone have this manual, and does it define QLN as they use it?

[JGO]

Assuming that Broadcom are consistent in the way they define the reported QLN values, then I suppose that it's best that we all use the same terms. But it's possible that the JDSU uses a different definition, so we can't directly compare JDSU QLN graphs with those in in our monitoring programs.

Don't know JDSU but there is always some silly **!"& who decides to do it differently ! As there is some possible error as REIN noise doesn't have the same peak to mean power ratio as white noise, there is always a potential error without assistance !

[boost]

Dunno if this helps...

PSD (dBm/Hz) Power spectral density. The portion of
the power for each frequency component
of the signal. (Infineon SIMs only)