100M : Power levels went up a notch and speed down?

[ileikcaek]

Hi,

I just checked my graph and Last night at exactly 1am my TBB Ping graph went up a notch, though it doesn't look "that" bad.



(after 4am was me rebooting it to try and get another channel)

When I checked the modem network stats I saw the power levels have literally gone up by around 1.0dBmV across all channels, this happened at 1am on the dot looking at the TBB graph.

My speed went down to 95 on the channel I was on, I rebooted the superhub and I got stuck on another channel that is giving a whopping 5Mb maximum. (my pings are fine though).



Pretty sure those power levels are well within spec so what could be going on here? What could cause the power levels to raise by 1.0dBmv at 1am exactly?

Edit: This is odd, I just went and pulled the cord for a minute, I have the same "default" channel (22) but now I am getting 105 again, I will have to monitor this I guess.

[AbyssUnderground]

Power levels do fluctuate during the day so a 1dBmV change isn't surprising. It is a little out of spec though, the closer to 0dBmV the better if I remember correctly.

[qasdfdsaq]

Spec is -15 to +15 and optimal is -7 to +7 or +3 depending who you ask. They're borderline in the "not good" category.

[ileikcaek]

When I first got 100Mb the power levels were consistently around +3 but after a while the connection went unstable and there were tons of dropouts and the like I believe they replaced a faulty amplifier in the cabinet to fix that issue because the power levels were around +5 after that with different channel ID's. more recently they have started to creep up but I guess the cold weather could be having an affect?

Still odd these problems started at 1am last night which is evident from a spike of packet loss followed by an increase of base latency of 1 or 2ms at the time. it seems to me that this isn't by fault but somebody has changed something to cause this due to it happening at 1am exactly.

Guess I will have to watch this as I doubt it's something VM will care about considering my net still works albeit with speed drops from time to time.

[Sephiroth]

In winter, the rise in downstream power can be as high as +3dBmv. So if it got colder at 01:00 that would explain it.

[General Maximus]

i have noticed mine creep up by 1dBmv recently. Whatever happened though that if the cumulative power went above 33 (or whatever it was) the thing blows up?

[Sephiroth]

Yes - but remember the cumulative power is calculated on the 3dB doubling principle.

It's not a case of just adding up the dBmv listed on your 4 or 8 channels. The (logarithmic) dBmv scale is relative to 0dBmv and a +3 dBmv increase in power level doubles the power arriving at the SH.

Let's say that the power level on a single channel is 0 dBmv. Now let's double that to 2 channels so you aggregate power in dBmv terms is +3 dBmv. Make that 4 channels and it's +6 dBmv; 8 channels and it's 9 dBmv. Your SH currently sees a spread of 16 DS channels (each 8 MHz wide) in your set from which you can actually use 4 or 8 (or whatever VM set) according to the valid BPI+ key associated with your lease.

But all that stuff hits your SH and is sifted by the tuners. 16 = 2^4 so the aggregate downstream power is at least 4 x 3 dBmv = 12 dBmv (when the reported power is 0 dBmv).

You will bust the EuroDOCSIS 3 limit of 33 dBmv if the average power across the downstream channels exceeds 22 dBmv.That's calculated by adding 3 dBmv to the 22 dBmv for each doubling of the number of channels. So, for 16 channels, that's 22+3+3+3+3 = 34 dBmv. Clear as mud, eh?

[General Maximus]

i sort of get it but i cant do the backwards maths to understand why the ops modem hasnt blown up. If you are using 8 channels as you have given in the example above and 8 channels is 2^3 = 22+9 = 31dBmv (in the ops case) when the reported power is 0, what on earth is it when the reported power is 8dBmv per channel?

[Sephiroth]

It's 16 channels that hit the modem and you tune to 4, 6 8 or whatever. The whatever doesn't matter.

If all the power levels were equal (say 23 dBmv which is really a small amount of milliwatts), then add 3d Bmv if the number of channels doublws from 1 to 2; add 3 dBmv if the number of channels doubles to 4; add 3 dBmv if the number of channels doubles to 8; add 3 dBmv if the number of channels doubles to 16.

Then 23+3+3+3+3 = 35

It's a sort of reverse logic based on 3dB increase doubles the power, You can see that's goota be right becasue if everyone was 2.1 dBmv with 16 channels coming in, you'd bust the power limit. Clearer?

[General Maximus]

no, i still dont understand where the reported power levels come into it. Where does the 21 or 23 come into it which we start off with (is that from the cmts?) and as we assuming that we star off at 0 do we then add the reported power levels to the +3? Can you do a worked example based on an average of 8dBmv over 8 channels which is what I think the op has.

And putting my confusion about the reported power levels aside, are you saying that in order for VM to ever use a 16 channel modem that the power levels would have to be 0? (21 +(4x3))=33

[Sephiroth]

The three things to grasp (this was all explained to me carefully by Igni some time ago) are:

1/
Power doubles at a 3 dBmv increase in reported level

2/
The SH tuner has 16 channels from which it can tune up to 8

3/
The number of tuned channels is not relevant to the calculation.

If only one channel was coming in, then taking your 8 dBmv level, the aggregate power is 8 dBmv. If 2 channels are coming in, that doubles the power, adding 3 dBmv (not 8 dBmv); double that to 4 channels, add another 3 dBmv; and so on till you reach the 16 channels which is 2^4. So the aggregate power is 8+3+3+3=3 = 20 dBmv.

The only caveat to the above is that I'm assuming the average dBmv; individual channels vary according to amplfieir slope and frequency; hence the use of the term "average".

If the SH allowed you to tune 16 channels, it would be the same calculation.

I wish I could say Simples!

[qasdfdsaq]

Also remember cumulative power (in any sense that matters) is the amount of RF power coming in through the tuner. The tuner may not be 8 channels wide (i.e. exactly 40Mhz). Remember the VMNG300 had a 126Mhz tuner, even though it "could" only use 20Mhz of the bandwidth, any channel power in the remaining 106Mhz would still be going through the downstream ICs, so you would have to add up the power of anything in that remaining space.

The average doesn't matter, it's the total that counts. Although mathematically the average and total are directly linked anyway...

---------- Post added at 05:56 ---------- Previous post was at 04:48 ----------

Originally Posted by Sephiroth

Your SH currently sees a spread of 16 DS channels (each 8 MHz wide) in your set from which you can actually use 4 or 8 (or whatever VM set) according to the valid BPI+ key associated with your lease.

But all that stuff hits your SH and is sifted by the tuners. 16 = 2^4 so the aggregate downstream power is at least 4 x 3 dBmv = 12 dBmv (when the reported power is 0 dBmv).

Ah, I see you've already covered this to some extent. Though I'm not sure how you know the SH's tuner is 16 channels wide. I can't find any info on it myself...

[Sephiroth]

Igni told me, Qasi.

The average does matter for the sake of illustration. Peops look at their 4, 6 or 8 channels and say that's around the 4 dBmv mark; but then the hidden channels may be at a higher frequency which may well arrive at lower power.

[General Maximus]

thanks dude, I understand now. Although I was the only person to question it I am sure many of us have been enlightened by your words of wisdom the the numbers of reads has nearly doubled since we started doing some maths

[horseman]

Originally Posted by qasdfdsaq

... The tuner may not be 8 channels wide (i.e. exactly 40Mhz). Remember the VMNG300 had a 126Mhz tuner, even though it "could" only use 20Mhz of the bandwidth, ….. Though I'm not sure how you know the SH's tuner is 16 channels wide. I can't find any info on it myself...

Probably my oncoming dementure but this explanation seems a tad ambiguous / confusing and not easily related to CableLabs (Euro)DOCSIS specs?

1. The inference above could be interpreted that 8 channels is equivalent to 40MHz? Obviously 8 (euroDOCSIS 8MHz d/s) channels would be 64MHz capture window with 8 adjacent channel constraint and thus 40MHz corresponds to only 5 adjacent channels?. That (8 adj channel constraint) of course only applies to minimum Standard Receive Connection Profile and as I'm sure you know the CM initially transmits a minimum of both a Std RCP and a Vendor specific RCP to CMTS so perhaps you know for certain which one any specific CMTS is actually using?

2. Equally "VMNG300 has a 126MHz tuner but only 20MHz bandwidth" seems to make little sense to me as neither appear to relate to adjacent channel capture window? Clearly I've misunderstood the context, perhaps you'd be kind enough to clarify?

Originally Posted by Sephiroth

...Your SH currently sees a spread of 16 DS channels (each 8 MHz wide) in your set from which you can actually use 4 or 8 (or whatever VM set) according to the valid BPI+ key associated with your lease..

Again that implies a Vendor RCP as either a single Receiver Module would need a 16 x 8 = 128MHz capture window or possibly more than one Receiver module? eg:



I'm also struggling with relating the "BPI+" to the RCP exchange at the moment as well?

Overall I still find the original "power" explanation a tad incongruous as depending on the modulation profile on each channel I find it somewhat difficult to conceive where you could have an average channel power(and the 3dB multiplier) that exceeds the max input spec of 33dBmv in any scenario that either doesn't already exceed +/- 17dBmv limit for a single channel or consequently doesn't then have extreme equalisation issues between individual channels anyway?

If however you perversely added the individual channel powers and then say deducted the 3dB uplift for doubling the channels you could more easily fit the scenarios?