how-to block ads
|
Bicephale
join:2005-09-24
 ·TekSavvy Solutions..
| reply to Bicephale
Re: From the ground up!
%20.GIF "48401 bytes")
ST516v6 Long-Term Averaged Curves (2009-Sep, 12-18, 22-30) | %20.GIF "353985 bytes")
Groups (5) of individual daily records (DS SNRM is Green, Rx CRC/HEC are Blue, DS BitRate is Yellow | %20.GIF "92299 bytes")
ST516v6 Averaged RxCRC Groups (2009-Sep, 12-18, 22-30) | %20.GIF "152955 bytes")
MoDem Comparisons - SS4200 vs GBB2060 vs ST516v6 (2009-Sep, 12-18, 22-30) |  ST516v6 FW v···-18,.zip 753,970 bytes
(ST516v6 FW v6.1 - Analysis & MoDem Comparisons (2009-Sep, 12-18, 22-30) .ZIP) | | |
This late data harvest was done using the same Thomson
ST516v6 as before, my experimenting setup was identical
except for its FirmWare (which i've changed to v6.1.0.5).
MoDem comparisons weren't straightforward to make this
time: i could hardly expose any daily 24 h sinuosidal noise
floor from the long-term curves as i did before, except for
the noisiest averaged group-curve (Sep. 15, 17 & 27)!
It's not clear if this was due to my change of FirmWare or
some radical modification in the phone-line behaviour as a
consequence of some seasonal influence but my Thomson
now compares reasonably well with the other DSL units...
It's still noisier once the aberrations have been rejected,
in my opinion, but this represents a definite improvement!
One feature of my ST hasn't changed a bit after i flashed
it with another FirmWare, though:
%20.GIF)
ST516 - Speeds Degrade over time, Bicephale, 2009-Oct-4
The DS SNR Margin Bleeding Syndrome remains as obvious
as ever!
 | |
Bicephale
join:2005-09-24
·TekSavvy Solutions..
|
Oups! There's been a problem with the previous archived log-file...
 | |
Bicephale
join:2005-09-24
·TekSavvy Solutions..
| %20.GIF "46465 bytes")
ST516v6 - Averaged Analysis (2009-Oct, 4-8, 16-19) | %20.GIF "457119 bytes")
ST516v6 - Daily Records - RF Choked (2009-Oct, 2-19) | %20.GIF "35455 bytes")
ST516v6 - Events (2009-Oct, 7-8, 20) | %20.GIF "29647 bytes")
ST516v6 - Grouping Pre-Analysis (2009-Oct, 2-21) | ST516v6 FW v···20) .ZIP 425,275 bytes
SpreadSheet-ready .CSV Formatted Data | | |
Today's post reveals some significant changes/differences.
1st of all, the SNR Margin "bleeding syndrome" is much less
severe but this could be a result of the phone-line being a
lot more stable instead of the FirmWare's quality, go figure!
The marginal events revealing a 3+ dB SNR Margin boost of
at least twenty minutes each are still as puzzling as ever...
Analysis of "look-alike" daily records has been made: 1) by
grouping the records in respect to their slope (see the last
graph); 2) by keeping only those that were "in sync" (e. g.
with the expected number of samples)... Two main groups
emerged, each was 5 (Oct. 4-8) and 4 (Oct. 16-19) days
respectively. The top graph which compares them reveals
a tremendous improvement relatively to the records taken
in the previous months. The cause could be better phone-
line conditions, a more suitable FirmWare or a mix of both.
Take note how a manual reset took care of the SNR Margin
"bleeding syndrome"... | |
Bicephale
join:2005-09-24
·TekSavvy Solutions..
| %20.GIF "45832 bytes")
GNet BB0060B - Pre-Analysis (2009-Nov, 3-23) | %20.GIF "41587 bytes")
GNet BB0060B - Analysis (2009-Nov, 3-23) | GNet BB0060B···23) .ZIP 429,891 bytes
GNet BB0060B - .CSV SpreadSheet Analysis (2009-Nov, 3-23) | | |
This one has been waiting on the back-burner long enough, it
is about time to see what that long-term record has to say!!!
Grouping of "look alikes" was done simply by using the Rx CRC
daily rate. Seven groups emerged, group #4 formed a majority
while the others remained marginal (especially group #7 which
is composed of "Not-in-Sync" and/or "eventful" daily records).
| |
-
|
