Thanks for your help Mrclean0325. When I get back into work on Monday, ill look into what you suggested. Im not entirely sure what it means to "add pads", could you explain that a little more, thanks.
Just a bit more information about the testbed:
- The CMTS is model Cisco uBR7111, and all the cable modems are Cisco uBR900 series.
- From the CMTS we have one upstream and one downstream port, connected to each is a 6dB attenuator (ie. 6dB for US, and 6dB for DS)
- These go into a "Masthead Diplexer", so i presume its what we should be using. Then the single cable comes out of that towards the rest of our splitters and CM's.
We originally started with the attentuators at 24 for DS and 36 for US, this was when we were having locking problems. There is a diagram (maybe Cisco reccomendation) that states that we should use a 30dB for US and 40dB for DS, but we are still unsure as to what the correct values should be. We kept reducing the values until we could get up to 5 cable modems locking on at the same time. Previously with the higher attenuator values, they would not come up at the same time, some would, but others would never lock on to the frequency and would try to adjust TX power constantly (cant remember exact values).
Just an update, on Friday we discovered that the upstream 'channel-width' setting on the CMTS was for some reason set to just 200KHz, we increased this to the maximum available of 3.2MHz and have got some performance increases. DS is at about 6.5Mbit/s and US is about 1.8Mbit/s. As far as i know, we are using the DOCSIS 1.0 specification, and we do use a DOCSIS configuration file sent to the CM's with speed limits, these limits were well above what we actually observe on bandwidth tests, set to 30000000 US and 10000000 DS.
We have found the option in the CMTS to increase the transmit power, but havent adjusted it yet.
Thanks,
Shaun.
Main Topics
Browse All Topics
by: Mrclean0325Posted on 2006-02-04 at 10:11:11ID: 15873037
There are a couple of things that I see right off the bat:
1. You need a config file that will allow you whatever upstream and downstream speeds you need; i.e. if you want faster upstream speeds you need to change your config file.
2. Your upstream SNR if way too low for the set-up described. At 26, it may not work depending on what version of DOCSIS your CMTS and config file are set-up for. For the set-up you describe, the upstream SNR should be around 37 - 40. This will cause problems locking up modem and may cause upstream speed issues if your config file is set for higher than you are getting.
3. Your CMTS transmit power is way to low. I realise that it is a test bed, but the transmitterr in the CMTS gets noisier the lower the power. They like to be running between 80 - 90% threshold.
My suggestion is to do this:
I am assuming that you are using a diplex filter and not a splitter. If you are using a splitter to seperate the upstream and downstream at the CMTS - don't do it, get a diplex filter. (they look lke a 2 way splitter, but one leg will pass 0 - 42 MHz and the other leg will pass 55 MHz - whatever your top frequency would be). One leg of the diplex filter will go to your downstream transmit ( the 55MHz and up leg) and the other will go to one of your upstream ports. Set your CMTS for around 58 - 60 output level and verify with a meter (most CMTS will do up to around 65 - 68). You must make sure that you pad the higher frequency leg ( the 55Mhz and up) so that you get around 0 dB at the modem on your downstream channel. If you use the 'worst case' or the modem at the end of your splitter array will work best. You can do this easily by looking at the modem test page (in Motorola 192.168.100.1) or using a meter that will read your QAM levels and add pads until you get around 0 dB. This will take care of your downstream. The modems will go from -15 to +15 for the downstream; if trying DOCSIS 1.1, it calls for -10 - +10 - 0 dB is optimum and will work for what ever. You can achieve this in a testbed, in real life it not so easy.
To adjust your upstream, you will need to pad the 0 - 42Mhz leg of your diplex filter until your modem reads 48 - 50 TX (this will make it work for DOCSIS 1.0, 1.1, with 2 it will need to be 48 or less). The transmitters in the modem are also noisy when transmitting at lower levels.
Using a 'modem meter' will be the easiest to adjust everything and you should also have a way to look at the QAM constellation to make sure that there are no issues with the downstream. I use a Hukk CM-1000 for this, but there are others on the market that will do the same job. Once you have all of the above reading properly, then you can test your throughput. If the reading are the same, then you will definitiely need to change your config file for your modem. Each CMTS wll have it's own set of unique features that you can play with; but all will have to be physically setup close to what is described above. If this is a CMTS to replace a current one, make sure to get VERY familiar with it and don't try to implement all the advanced features at first - you will seriously regret it. Once everythng is working with a 'base' config, then you can slowly implement more advanced features. We are currently working to get all the modems and plant ready for DOCSIS 1.1/2 - not an easy task as our system is old and will not be rebuilt to do this. Hope this helps!