You are here

which is the top speed for a cable modem?


Hi group

That's my question. We had one cable modem connect it to a cmts and the top speed was 2.5 mbps for download and ~1.2 on upstream for a config file with 4mbps. A person told me that cable modems only can achieved 2 mbps. It's true?

Info for the test:
For downstream, 64qam, 6mhz channel width, Symbol Rate 5.056941 Msps
For upstream, Channel Width 3.200000 MHz, QPSK, Symbol Rate 2.560000 Msps
Docsis 1.0

Thanks in advanced

the 4 meg file, is that for upstream or downstream?
as for how fast the modem can go, depending on service class and qos, I've had 30 meg down and 21 meg up on a BSR2k docsis 2.0.
in your case, 64 qam down is around 27 meg shared. 3.2 wide at qpsk is around 3 meg shared.
change downstream to 256 qam and upstream to 16 qam. qos of 7. add a max burst over 1038 and you should double your speed


Hi group

Yes, the 4meg file is for upstream and downstream.

I will make what you told me.

Thank you so much!

Are you making any progress, cabo81? I sure am curious what the 2Mbps wall is all about and why some of us seem up against it and others don't in an interoperable protocol environment! ;-)


Hi cmcaldas

The cable modem doesn't get the top speed. This my config file

FileVersion = Version 6.0
03 (Net Access Control) = 1
18 (Maximum Number of CPE) = 6

24 (Upstream Service Flow Encodings)
S01 Service Flow Reference = 1
S06 QoS Parameter Set Type = 7
S07 Traffic Priority = 7
S08 Max Sustained Traffic Rate = 4096000
S09 Max Traffic Burst = 1038000

25 (Downstream Service Flow Encodings)
S01 Service Flow Reference = 5
S06 QoS Parameter Set Type = 7
S07 Traffic Priority = 7
S08 Max Sustained Traffic Rate = 4096000
S09 Max Traffic Burst = 1038000
29 (Privacy Enable) = 0

39 (DOCSIS 2.0 Enable) = 0

On the cmts I have the following

Cable 0/0 Downstream 0 is up
Frequency 561.000000 MHz, Channel Width 6 MHz, 256-QAM, Symbol Rate 5.360537 Msps
FEC ITU-T J.83 Annex B, R/S Interleave I=32, J=4
Downstream Channel ID: 1
Spectrum Group: grupo0, grupo1, grupo2, grupo3

Cable 0/0 Upstream 0 is up
Frequency 33.000000 MHz, Channel Width 3.200000 MHz, 16QAM, Symbol Rate 2.560000 Msps
Spectrum Group: grupo0
SNR 0 dBmv
Nominal Input Power Level 80 dBmV
Ranging Backoff (Start 2, End 8)
Ranging Insection Interval (200 ms)
Tx Backoff Start 2, Tx Backoff End 8
Modulation Profile Group 1
Concatenation is enabled
part_id=0x3137, rev_id=0x3
Range Load Reg Size=0x58
Request Load Reg Size=0xE
Minislot Size in number of Timebase Ticks is = 4
Minislot Size in Symbols = 64
Bandwidth Requests = 0x0
Piggyback Requests = 0x1764
Invalid BW Requests = 0x0
Minislots Requested = 0x0
Minislots Granted = 0x0
Minislots Size in Bytes = 32, Map Advance (Dynamic): 3129 usecs
UCD Count = 0
Slots 9821, NoUWCollNoEnergy 0, FECCorHCS 6, HCS 6
Req 176533659, ReqColl 0, ReqNoise 6, ReqNoEnergy 176529816
ReqData 0, ReqDataColl 0, ReqDataNoise 0, ReqDataNoEnergy 0
Rng 1786, RngColl 12582913, RngNoise 0
FECBlks 37589, UnCorFECBlks 15, CorFECBlks 16

And this is the service-flow-id and other info

CMTS1#show cable modem 000e.5c09.de1e svc-flow-id
Service flow id Interface Flow Direction Flow Max Rate
1 cable 0/0 Upstream 4096000
2 cable 0/0 Downstream 4096000
CMTS1#show cable modem 000e.5c09.de1e mac
MAC Address MAC Prim Ver Frag Concat PHS Priv DS US
State SID Saids Sids
000e.5c09.de1e online 1 DOC1.1 yes yes yes BPI+ 15 4

The signals in the cable modem are:
SNR: 41 dB
Power Level: 6 dB

Power level: 54 dBmV

But the speed is still low for the config file. The cable modem is a motorola sb5100, the only one connected to the cmts. Is there an error in my config file?

Thanks in advanced

how are you testing the speed?

I'm testing with a web site called Also I'm testing with downloads from differents pages, like ISO's from debian and Mandriva and calculating the BW.

None of this gave me more than ~2.0Mbps on downstream.

Thanks for your interest.

Problem with testing from a web site to see your speed is the link to the web that you have and the link the end has. to see what you have for a modem speed you need to test within your own network. if it's good there, then you know there a network or link issue. otherwise you can't rule anything out..
setup an ftp server behind the cmts and do download and upload testing with a pc connected behind the modem.

Q-check is cool as Carl said. Speakeasy is a very erratic testing site even for real world results. is the most consistent I've experienced.
I've compared tests between and another speedtest on a diagnostic server on my network with very similar results for a modem (SB5120 / DOCSIS 1.0) config'd for 10 down and 5 up. Get 9Mbps and change on a fairly consistent basis down, but can't crack 2Mbps up either.
64QAM downstream / 16QAM upstream 3.2MHz channel width -- the only active modem on the network with plenty of headroom both on the local network and to the Internet. My old LANcity QPSK modems blow the doors of the SB5120 on the upstream.


Blame the protocol lol the old Motorola cdlp was doing 2 Mbps with just a 600K wide return..
as for cracking the 2 Mbps, increase your burst rate ;-). 16qam @ 3.2Mhz wide is 10 Mbps, but should get around 7


I can only see what I can see -- which is better upstream performance out of my legacy modems -- on a production network, no less -- than a DOCSIS modem in a very controlled environment. I'd increase my burst rate if I had one set, but my understanding of the spec is that 0 = no limit.
So here's a question (or two); Am I too close to my diplexer roll-off skirt at an upstream center freq of 39MHz? At 6MHz channel width, I'd say I'd be asking for trouble, but no way at 3.2MHz. And is a specified max burst better than no limit? Ok, three questions. Am I better off in mixed mode upstream, or would that just be more of a workaround for more congested traffic conditions?


Hi group

I've been testing speed. I posted recently very happy that I achieved 4M on DS and 1M on US. And for a time it was good... (animatrix)

But now, when I wanted to make more tests, I made a config file with 20M on DS and 10M on US. And the DS was achieved. But US, only got ~1.3 Mbps. I don't know why this happen. Is the only cable modem plugged to the cmts. What I've been reading from you is about the protocol and congested traffic, but what happens in my case?

I paste a copy of my config file for your consideration.

FileVersion = Version 6.0
03 (Net Access Control) = 1
18 (Maximum Number of CPE) = 6

24 (Upstream Service Flow Encodings)
S01 Service Flow Reference = 1
S06 QoS Parameter Set Type = 7
S07 Traffic Priority = 7
S08 Max Sustained Traffic Rate = 10000000
S09 Max Traffic Burst = 1768
S10 Min Rsvd Traffic Rate = 6120
S11 Assumed Min Rsvd Rate Pkt Size = 64
S13 Timeout Admitted QoS Params = 200
S15 Service Flow Scheduling Type = 2
S16 Request/Transmission Policy = 00 00 00 00

25 (Downstream Service Flow Encodings)
S01 Service Flow Reference = 2
S06 QoS Parameter Set Type = 7
S08 Max Sustained Traffic Rate = 20000000
S09 Max Traffic Burst = 1768
S12 Timeout Active QoS Params = 200
S13 Timeout Admitted QoS Params = 200
S14 Maximum Downstream Latency = 2000
29 (Privacy Enable) = 0

As you can see, I've set the max burst. And I've made the test without max busrt too. On dowstream is fine. On upstream, I don't get the 10M.

I hope you can help with this.

Thanks in advanced.

I run an ftp server (on the dhcp server) and download / upload from the pc connected behind the modem. Ixia has a cool program called Q-check. there's also Jperf that's good. all are free and easy to use


Hi group

Sorry for the late post. We did the test. We create a new config file for 4M ds and 1M Us. We put a pc behind the cable modem and a laptop at cmts' end.
We used Jperf as speed tester. And yes, we got the speed in us and ds. Now we probed the docsis side is working well and the web test does not work accurately.

Now I think that a cable modem doesn't give me the real speed by the number of users on the downstream port wich is ~23mps shared with everybody. Now I have to calculate, how many BW do I have to give to users to make more efficient my ds port.

Thanks for the support.