A Broadband and ADSL forum. BroadbanterBanter

Welcome to BroadbanterBanter.

You are currently viewing as a guest which gives you limited access to view most discussions and other FREE features. By joining our free community you will have access to post topics, communicate privately with other members (PM), respond to polls, upload your own photos and access many other special features. Registration is fast, simple and absolutely free so please, join our community today.

Go Back   Home » BroadbanterBanter forum » Newsgroup Discussions » uk.telecom.broadband (UK broadband)
Site Map Home Register Authors List Search Today's Posts Mark Forums Read Web Partners

uk.telecom.broadband (UK broadband) (uk.telecom.broadband) Discussion of broadband services, technology and equipment as provided in the UK. Discussions of specific services based on ADSL, cable modems or other broadband technology are also on-topic. Advertising is not allowed.

What factors affect "speedtest" speed, when it is less than router's sync speed?



 
 
Thread Tools Display Modes
  #1  
Old February 28th 19, 02:05 PM posted to uk.telecom.broadband
NY
external usenet poster
 
Posts: 492
Default What factors affect "speedtest" speed, when it is less than router's sync speed?

Our router used to report a sync speed of around 1.5-2 Mbps down and 0.4 up.
Speedtest achieved speeds of somewhere between 0.8 and 1.8 Mbps, depending
on time of day (contention?).

Then our line experienced a total loss of phone (no dialling tone etc) which
has now been fixed and was due to an underground cable fault, according to
the BT OR engineer who phoned to say "fault has been fixed".

During the phone outage, broadband continued with sync of 0.6 D and 0.4 U,
with fairly similar Speedtest results.

Now that the line has been fixed, the router syncs at 1.5 D / 0.4 U once
more, but Speedtest is only getting 0.6 D / 0.4 U.


I know that when there has been a line fault, it takes a while for the
modems in the exchange and the router to "train" themselves to find out the
best speed that the line can support. But does that affect sync speed or
real-world Speedtest result? What factors affect Speedtest result when it is
significantly less than sync speed at a time of day when normally Speedtest
would be better? How long does it typically take for training to complete
and the sync/Speedtest speed to level out?


This is for Ookla Speedtest app on Android and Windows.

  #2  
Old February 28th 19, 06:04 PM posted to uk.telecom.broadband
Woody
external usenet poster
 
Posts: 722
Default What factors affect "speedtest" speed, when it is less thanrouter's sync speed?

On Thu 28/02/2019 14:05, NY wrote:
Our router used to report a sync speed of around 1.5-2 Mbps down and 0.4
up. Speedtest achieved speeds of somewhere between 0.8 and 1.8 Mbps,
depending on time of day (contention?).

Then our line experienced a total loss of phone (no dialling tone etc)
which has now been fixed and was due to an underground cable fault,
according to the BT OR engineer who phoned to say "fault has been fixed".

During the phone outage, broadband continued with sync of 0.6 D and 0.4
U, with fairly similar Speedtest results.

Now that the line has been fixed, the router syncs at 1.5 D / 0.4 U once
more, but Speedtest is only getting 0.6 D / 0.4 U.


I know that when there has been a line fault, it takes a while for the
modems in the exchange and the router to "train" themselves to find out
the best speed that the line can support. But does that affect sync
speed or real-world Speedtest result? What factors affect Speedtest
result when it is significantly less than sync speed at a time of day
when normally Speedtest would be better? How long does it typically take
for training to complete and the sync/Speedtest speed to level out?


This is for Ookla Speedtest app on Android and Windows.


I replied to a similar problem the other day - might have been on a
caravan forum.

For a telephone to work there has to be d.c. continuity in the line so
that current to power the phone and ringing volts can get down the cable
to the instrument. However B/B used radio frequency (RF) signals.

If your phone stops working due to a tiny break in the cable the RF
signals will often jump that gap so your B/B will keep working albeit
degraded.

BT are sods for knicking linking pairs within the exchange so again your
phone would stop working (silent, no hiss or sidetone) but the broadband
would still work as that is added just before the line leaves the exchange.

--
Woody

harrogate three at ntlworld dot com
  #3  
Old February 28th 19, 07:00 PM posted to uk.telecom.broadband
Martin Brown[_2_]
external usenet poster
 
Posts: 174
Default What factors affect "speedtest" speed, when it is less thanrouter's sync speed?

On 28/02/2019 14:05, NY wrote:
Our router used to report a sync speed of around 1.5-2 Mbps down and 0.4
up. Speedtest achieved speeds of somewhere between 0.8 and 1.8 Mbps,
depending on time of day (contention?).

Then our line experienced a total loss of phone (no dialling tone etc)
which has now been fixed and was due to an underground cable fault,
according to the BT OR engineer who phoned to say "fault has been fixed".

During the phone outage, broadband continued with sync of 0.6 D and 0.4
U, with fairly similar Speedtest results.

Now that the line has been fixed, the router syncs at 1.5 D / 0.4 U once
more, but Speedtest is only getting 0.6 D / 0.4 U.


Dropped packets would be one possible cause. Force a sync at around mid
day to get the fastest sync speed (and then speed test also at night)
and vice versa - sync a couple of hours after sunset to get the slowest
possible sync speed and see if that makes the link more reliable.

Unless you have lots of neighbours with fibre or are somewhere very
remote I'd be surprised if contention was an issue.

Routerstats light is invaluable for finding intermittent line faults if
your router is supported and provides the right diagnostics.

I know that when there has been a line fault, it takes a while for the
modems in the exchange and the router to "train" themselves to find out
the best speed that the line can support. But does that affect sync
speed or real-world Speedtest result? What factors affect Speedtest
result when it is significantly less than sync speed at a time of day
when normally Speedtest would be better? How long does it typically take
for training to complete and the sync/Speedtest speed to level out?


For small changes in SNR it can take a couple of days before the effects
of a repair take full effect if your router has been on the spiral of
doom. It can be hastened if the ISP does a reset for you. Even then it
may bounce up and down for a few days to find the sweet spot.

--
Regards,
Martin Brown
  #4  
Old February 28th 19, 07:12 PM posted to uk.telecom.broadband
Java Jive
external usenet poster
 
Posts: 463
Default What factors affect "speedtest" speed, when it is less thanrouter's sync speed?

Lost NY's OP so replying to Woody ...

On 28/02/2019 18:04, Woody wrote:

On Thu 28/02/2019 14:05, NY wrote:

This is for Ookla Speedtest app on Android and Windows.


As it happens, trying to get my 4G USB stick flying on my hacked BTHH5a,
and now assessing it's performance and stability (after four days I'd
say pretty good on both counts), I've been running an awful lot of speed
tests recently. Here are my findings:

BT Wholesale
http://speedtest.btwholesale.com/

Though much lambasted by others, until recently generally my preferred
test because the site has no advertising corrupting the results, but
currently hanging doing the upload test, so have been forced to look
elsewhere.

My Broadband Speed Speed Checker
http://www.mybroadbandspeed.co.uk/speedtester/

Quite good, has some advertising, but not excessive, so it's probably
not affecting the result much, if at all, as shown by it agreeing well
with the next choice.

Which? - Broadband Checker
https://broadbandtest.which.co.uk/

Good. No adverts. Agrees well with above.

www.netmeter.co.uk - Broadband speed test
https://www.netmeter.co.uk/

Good. No adverts while testing. Currently my preferred. Today gives
slightly better figures than above two, but that could be normal
fluctuation.

Broadband Speed Checker
http://www.broadbandspeedchecker.co.uk/

Crap. Crowded out with advertising which continues loading while
testing, thus rendering the result unreliable at best, and useless over
a slow connection.

Speedtest.net by Ookla - The Global Broadband Speed Test
http://www.speedtest.net/

Almost as crap. Again has advertising which continues loading while
testing, thus rendering the result unreliable at best, and useless over
a slow connection.
  #5  
Old February 28th 19, 08:10 PM posted to uk.telecom.broadband
Graham J[_3_]
external usenet poster
 
Posts: 54
Default What factors affect "speedtest" speed, when it is less thanrouter's sync speed?

NY wrote:
Our router used to report a sync speed of around 1.5-2 Mbps down and 0.4
up. Speedtest achieved speeds of somewhere between 0.8 and 1.8 Mbps,
depending on time of day (contention?).

Then our line experienced a total loss of phone (no dialling tone etc)
which has now been fixed and was due to an underground cable fault,
according to the BT OR engineer who phoned to say "fault has been fixed".

During the phone outage, broadband continued with sync of 0.6 D and 0.4
U, with fairly similar Speedtest results.

Now that the line has been fixed, the router syncs at 1.5 D / 0.4 U once
more, but Speedtest is only getting 0.6 D / 0.4 U.


I know that when there has been a line fault, it takes a while for the
modems in the exchange and the router to "train" themselves to find out
the best speed that the line can support. But does that affect sync
speed or real-world Speedtest result? What factors affect Speedtest
result when it is significantly less than sync speed at a time of day
when normally Speedtest would be better? How long does it typically take
for training to complete and the sync/Speedtest speed to level out?


This is for Ookla Speedtest app on Android and Windows.


It is good practice to make all Speedtest measurements on a device
connected by wire to the router, to avoid any confusion introduced by
WiFi. Also, to have only one device connected (by whatever means) to
the router - so you are not losing channel capacity to other services.

Did you make a note of the SNR margin, loop attenuation, and error
counts before the fault?

Ordinarily the downstream SNR margin is about 6dB. If it rises above
that it indicates the line is noisy, and the router adjusts its settings
so that it makes less use of the frequencies that are affected by the
noise. Consequently the sync speed drops. A rule of thumb is that for
every 3dB increase in SNR margin you expect to see about 500kbits/sec
drop in sync speed. You might expect a small variation day to night.

For upstream the speed is capped by the service you have, so for ADSL it
is either 448kbits/sec, or double that, or triple that. The router
generally responds to this cap by adjusting to a higher SNR margin - so
for upstream the SNR margin is not a very useful parameter.

When the line is first commissioned with ADSL the Dynamic Line
Management may set the DS SNR margin at 18dB with a consequent low sync
speed. Over time the DLM will reduce the SNR margin and increase the
speed, aiming for optimum performance - this "training" can easily take
a week. Sometimes it never happens, in which case you have to talk to
your ISP to get the issue resolved. Some ISPs won't understand.

The loop attenuation depends on your line length, and is different
between downstream and upstream. It should not vary with time; if it
does it suggests a faulty line.

The Speedtest result can show up to 88% of the sync speed; no higher,
because ADSL has a protocol overhead which uses some of the available
channel capacity. It can show less - usually because of congestion
(depends on time of day, and on your ISP); or because of noise which
causes packet re-transmission; or because of a "profile" limitation.

During your phone fault, the broadband signal probably jumped the break
which prevented the phone from working - as others here have explained.

After the "repair" there may be noise which causes packet
re-transmission, or the repair may have swapped your line so a different
DSLAM device in the exchange.

Further, while the fault was present, the DLM may have adjusted your
"profile" to the lower speed that the line was then capable of; and it
may take time to re-adjust upwards. The "profile" is separate to the
training that takes place between your router and DSLAM - it adjusts to
ensure that data packets are not delivered to the DSLAM faster than the
channel can deal with them - it minimises the need for buffering and
flow control. This "profile" can also get "stuck", so again, a call to
your ISP should resolve it. A good ISP will have a history of the
profile changes so will understand the issue. And Zen Internet or A&A
can show you these parameters via a management portal.


--
Graham J
  #6  
Old March 1st 19, 09:23 AM posted to uk.telecom.broadband
NY
external usenet poster
 
Posts: 492
Default What factors affect "speedtest" speed, when it is less than router's sync speed?

"Graham J" wrote in message
...
This is for Ookla Speedtest app on Android and Windows.


It is good practice to make all Speedtest measurements on a device
connected by wire to the router, to avoid any confusion introduced by
WiFi. Also, to have only one device connected (by whatever means) to the
router - so you are not losing channel capacity to other services.

Did you make a note of the SNR margin, loop attenuation, and error counts
before the fault?


Unfortunately no, but from memory, the values are very similar to what I
remember: attenuation about 31 U and 59-62 D, SNR around 6 dB in both
directions.

What we have now is the router syncing at a speed (1500-2000 kb/sec) which
is *always* greatly in excess of the results that Speedtest gives.

Previously the Speedtest results varied throughout the day, being worst in
the evening (presumably due to contention), over the range 1900 at best, 500
at worst (kb/sec down).

Ordinarily the downstream SNR margin is about 6dB. If it rises above that
it indicates the line is noisy, and the router adjusts its settings so
that it makes less use of the frequencies that are affected by the noise.
Consequently the sync speed drops. A rule of thumb is that for every 3dB
increase in SNR margin you expect to see about 500kbits/sec drop in sync
speed. You might expect a small variation day to night.

For upstream the speed is capped by the service you have, so for ADSL it
is either 448kbits/sec, or double that, or triple that. The router
generally responds to this cap by adjusting to a higher SNR margin - so
for upstream the SNR margin is not a very useful parameter.


Yes, I was aware about the upstream speed being a multiple of 448 kb/sec.



When the line is first commissioned with ADSL the Dynamic Line Management
may set the DS SNR margin at 18dB with a consequent low sync speed. Over
time the DLM will reduce the SNR margin and increase the speed, aiming for
optimum performance - this "training" can easily take a week. Sometimes
it never happens, in which case you have to talk to your ISP to get the
issue resolved. Some ISPs won't understand.

The loop attenuation depends on your line length, and is different between
downstream and upstream. It should not vary with time; if it does it
suggests a faulty line.

The Speedtest result can show up to 88% of the sync speed; no higher,
because ADSL has a protocol overhead which uses some of the available
channel capacity. It can show less - usually because of congestion
(depends on time of day, and on your ISP); or because of noise which
causes packet re-transmission; or because of a "profile" limitation.


I hadn't realised there was as much overhead as that. I'm sure I've seen
speeds that are closer to the sync speed.

I'm wondering whether the sync speed before the fault was greater than the
2000 kb/sec that the router is currently reporting.

The stats that the router (Technicolour TG582n - Plusnet firmware) is
reporting a

Uptime: 0 days, 17:11:30
DSL Type: ITU-T G.992.1
Bandwidth (Up/Down) [kbps/kbps]: 448 / 1.952
Data Transferred (Sent/Received) [GB/GB]: 1,32 / 1,18
Output Power (Up/Down) [dBm]: 12,6 / 16,6
Line Attenuation (Up/Down) [dB]: 31,5 / 59,0
SN Margin (Up/Down) [dB]: 16,0 / 3,4
System Vendor ID (Local/Remote): TMMB / ----
Chipset Vendor ID (Local/Remote): BDCM / IFTN
Loss of Framing (Local/Remote): 9 / -
Loss of Signal (Local/Remote): 1 / -
Loss of Power (Local/Remote): 0 / -
Loss of Link (Remote): -
Error Seconds (Local/Remote): 1.087 / -
FEC Errors (Up/Down): 44 / 233.592
CRC Errors (Up/Down): 25 / 3.820
HEC Errors (Up/Down): -5 / 3.820 / 6.626

(I think the router uses a "." instead of a "," as a thousands separator,
and a "," as a decimal point.)

Those Forward Error Correction (FEC) and other error numbers don't look
good.

During your phone fault, the broadband signal probably jumped the break
which prevented the phone from working - as others here have explained.


The line sounds clean to me now, but my wife has noticed that if she calls
me from her mobile, there is a loud random percussive noise (like a lot of
manual typewriters) which was there before but has got worse. I didn't
mention it when I reported the fault to BT because I was hopeful that the
work to get the line working (repairing the break in the wire) would also
clear that fault.



Anyway, I've reopened the fault with BT: line noise and dramatically slower
internet. They are coming on Monday to investigate, and this time to make a
house visit as opposed to working elsewhere. I didn't see a BT van in the
village (though I may have missed it) so I presume I'm still on the original
line pair, and they just tidied up connections to that at a cabinet or the
exchange.

  #7  
Old March 1st 19, 11:03 AM posted to uk.telecom.broadband
Graham J[_3_]
external usenet poster
 
Posts: 54
Default What factors affect "speedtest" speed, when it is less thanrouter's sync speed?

NY wrote:

[snip]

Did you make a note of the SNR margin, loop attenuation, and error
counts before the fault?


Unfortunately no, but from memory, the values are very similar to what I
remember: attenuation about 31 U and 59-62 D, SNR around 6 dB in both
directions.

What we have now is the router syncing at a speed (1500-2000 kb/sec)
which is *always* greatly in excess of the results that Speedtest gives.


How long ago was the line fault repaired?

Previously the Speedtest results varied throughout the day, being worst
in the evening (presumably due to contention), over the range 1900 at
best, 500 at worst (kb/sec down).


[snip]

The Speedtest result can show up to 88% of the sync speed; no higher,
because ADSL has a protocol overhead which uses some of the available
channel capacity.* It can show less - usually because of congestion
(depends on time of day, and on your ISP); or because of noise which
causes packet re-transmission; or because of a "profile" limitation.


I hadn't realised there was as much overhead as that. I'm sure I've seen
speeds that are closer to the sync speed.


True for FTTC.

See https://aa.net.uk/kb-broadband-how-atm.html

Simpler explanation at:

https://blog.ipspace.net/2009/03/adsl-overhead.html

Different percentage in:
http://www.dslreports.com/forum/r28413343-Modem-sync-speed-VS-speed-test

Also

https://community.plus.net/t5/Fibre-Broadband/Why-is-it-88-2-re-IP-speed-vs-sync-speed/td-p/1233279

I'm wondering whether the sync speed before the fault was greater than
the 2000 kb/sec that the router is currently reporting.


Unlikely.

The stats that the router (Technicolour TG582n - Plusnet firmware) is
reporting a


[snip]

SN Margin (Up/Down) [dB]:*** 16,0 / 3,4


Looks very low at 3.4dB ...

[snip]

Those Forward Error Correction (FEC) and other error numbers don't look
good.


What SNR margin and error counts do you see immediately after rebooting
the router?

Do you have another router you can try?

The line sounds clean to me now, but my wife has noticed that if she
calls me from her mobile, there is a loud random percussive noise (like
a lot of manual typewriters) which was there before but has got worse. I
didn't mention it when I reported the fault to BT because I was hopeful
that the work to get the line working (repairing the break in the wire)
would also clear that fault.


Get somebody to call you from a known good landline. If the noise is
there report it as a voice fault - don't mention broadband - it only
confuses them.

Anyway, I've reopened the fault with BT: line noise and dramatically
slower internet. They are coming on Monday to investigate, and this time
to make a house visit as opposed to working elsewhere. I didn't see a BT
van in the village (though I may have missed it) so I presume I'm still
on the original line pair, and they just tidied up connections to that
at a cabinet or the exchange.


From which I presume your ISP is BT. It's unlikely that their helpdesk
will understand the concept of the BRAS Profile. But you can learn
about it he

https://kitz.co.uk/adsl/IPprofile.htm

Tell us what happens ...


--
Graham J
  #8  
Old March 1st 19, 11:47 AM posted to uk.telecom.broadband
NY
external usenet poster
 
Posts: 492
Default What factors affect "speedtest" speed, when it is less than router's sync speed?

"Graham J" wrote in message
...
NY wrote:

[snip]

Did you make a note of the SNR margin, loop attenuation, and error
counts before the fault?


Unfortunately no, but from memory, the values are very similar to what I
remember: attenuation about 31 U and 59-62 D, SNR around 6 dB in both
directions.

What we have now is the router syncing at a speed (1500-2000 kb/sec)
which is *always* greatly in excess of the results that Speedtest gives.


How long ago was the line fault repaired?


Yesterday about 3 PM.

I am aware that it often takes time for the *sync speed* to return to
normal, as the modem at the exchange and the one in the router need time to
negotiate the best speed that the line will support.

In this case, the sync speed is significantly higher than the Speedtest
speed (1500 versus 600 kbps), which is different to the normal case of sync
speed being slower than it used to be.

I wish I'd been keeping a note of router stats before the fault - but then
everything was working OK, so there was no need to - except for interest and
as a defence against future problems.

It may well be that there are problems with the line which were exacerbated
when the gross error of the discontinuity which caused the loss of dialling
tone / voice circuit were rectified.


Anyway, I've reopened the fault with BT: line noise and dramatically
slower internet. They are coming on Monday to investigate, and this time
to make a house visit as opposed to working elsewhere. I didn't see a BT
van in the village (though I may have missed it) so I presume I'm still
on the original line pair, and they just tidied up connections to that at
a cabinet or the exchange.


From which I presume your ISP is BT. It's unlikely that their helpdesk
will understand the concept of the BRAS Profile. But you can learn about
it he


No, the ISP is Plusnet. I referred to BT because they are the people to
report a voice fault to, since the voice service is with BT. The broadband
changes are incidental. I have reported them to Plusnet for them to check
what their tests find, in case they wanted to open a fault of their own with
BT OR. However their tests didn't find any faults, so they are sending me a
new, slightly more modern router, which they think will help - maybe some
routers' modems work better than others with borderline lines. I was hoping
they'd say "yes, this line isn't as good as it was" from any stats that they
may keep, and get BT OR to investigate.

So all I can do when the engineer comes on Monday is to talk about
intermittent line noise which only seems to affect incoming calls from my
wife's mobile (and don't affect calls she makes to other mobiles) and which
are usually only heard at her end and not mine, though yesterday even I
could hear a loud rustling when she phoned.

And when I talk to the engineer mention that ADSL is affected as well and
that whatever is causing the voice noise is also, as you'd expect, affecting
ADSL.

Speedtest results which were around 1500, dropping to 600 after a broken
line fault and its repair, suggests an incomplete repair.

  #9  
Old March 1st 19, 11:59 AM posted to uk.telecom.broadband
Andy Burns[_5_]
external usenet poster
 
Posts: 388
Default What factors affect "speedtest" speed, when it is less thanrouter's sync speed?

NY wrote:

Graham J wrote:

The Speedtest result can show up to 88% of the sync speed;


I hadn't realised there was as much overhead as that.


Some (most? all?) ISPs set the IP Profile to a fixed percentage of the
sync rate, in Plusnet's case that is 88.2% for ADSL, it's higher for
FTTC connections (presumably due to lower overhead of 1508 byte ethernet
packets rather than 53 byte ATM cells)

my router reports

Data Rate: 79.556 Mb/s, 19.999 Mb/s

BTW tester reports

IP Profile for your line is - 73.38 Mbps
Upstream Rate IP profile on your line is - 20 Mbps

So 92.2%, though I understand this can vary depending on your line's
retransmission rate.
  #10  
Old March 1st 19, 12:04 PM posted to uk.telecom.broadband
Andy Burns[_5_]
external usenet poster
 
Posts: 388
Default What factors affect "speedtest" speed, when it is less thanrouter's sync speed?

NY wrote:

Speedtest results which were around 1500, dropping to 600 after a broken
line fault and its repair, suggests an incomplete repair.


To me it suggests the IP profile has not (yet) been set based on the
repaired sync rate, PN support should be able to give it a nudge, what
does the "further diagnostics" page from

http://www.speedtest.btwholesale.com

report? (warning it may look like it's doing nothing, just leave it
until you get a result)
 




Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
Newly upgraded from ADSL to FTTC/VDSL - fast sync but very slow speedtest.net speed NY uk.telecom.broadband (UK broadband) 4 August 26th 18 11:53 PM
Less than 12mo contact available? uk.telecom.broadband (UK broadband) 5 April 28th 06 11:03 AM
less than 12month contracts? uk.telecom.broadband (UK broadband) 8 August 16th 05 09:54 PM
[UK-Bug] News ... Broadband for less than a tenner ? Andy M Jenkins uk.telecom.broadband (UK broadband) 1 May 31st 04 06:51 PM


All times are GMT +1. The time now is 09:49 AM.


Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2019, Jelsoft Enterprises Ltd.Content Relevant URLs by vBSEO 2.4.0
Copyright 2004-2019 BroadbanterBanter.
The comments are property of their posters.