dslreports logo
site
 
    All Forums Hot Topics Gallery
spc

spacer




how-to block ads


Search Topic:
uniqs
622
share rss forum feed


xymox1
Premium
join:2008-05-20
Phoenix, AZ

High freq attentuator

I need to bring down my cable modem downstream levels. My upstream are fine.

So I need a 6db attenuator that does not effect below 54mhz just the 800Mhz band.

Ive searched all over the net and im having trouble finding where i can buy one.


bchandler02

join:2011-07-08
Oklahoma City, OK

what are your current signal levels?



xymox1
Premium
join:2008-05-20
Phoenix, AZ

+7 and yea they are within range, but I have some technical reasons I want to hit 0.

My up stream are 39.5 and I dont want to increase my output power.

I want a mfgr of a low pass attentuator with low insertion loss below 54mhz and 6db up at 850Mhz.


bchandler02

join:2011-07-08
Oklahoma City, OK
reply to xymox1

Not sure that I've ever seen such a thing.

I buy most of my stuff from summitsource.com, it may be worth calling them and asking?



bbeesley
VIP
join:2003-08-07
Richardson, TX
kudos:5
reply to xymox1

said by xymox1:

My up stream are 39.5 and I dont want to increase my output power.

Actually, you do want to increase your output. The higher you can get it without going over ~52db the better your signal to noise ratio will be

as long as you aren't exceeding the DOCSIS 55db spec and/or clipping the output, higher is better


odog
Cable Centric Vendor Biased
Premium,VIP
join:2001-08-05
Atlanta, GA
kudos:11
Reviews:
·Comcast
reply to xymox1

said by xymox1:

+7 and yea they are within range, but I have some technical reasons I want to hit 0.

My up stream are 39.5 and I dont want to increase my output power.

I want a mfgr of a low pass attentuator with low insertion loss below 54mhz and 6db up at 850Mhz.

Your technical reasons are what? FYI that type of filter is generally not made. They also don't normally knock the signal down just a few dB. Most low pass filters have 30-40dB rejection in the stopband, sometimes more. About the only way to get what you want is to dilpex the signal, then pad the "high" side. That 0dB input probably isn't worth the return penalties of adding another diplex.
                   HIGH 6dB pad----"out" two-way
signal---diplex                                             "in" two-way-- CM
                   LOW------------"out" two-way
 

said by bbeesley:

Actually, you do want to increase your output. The higher you can get it without going over ~52db the better your signal to noise ratio will be

as long as you aren't exceeding the DOCSIS 55db spec and/or clipping the output, higher is better

Higher is not better, and not really higher. Ultimately the signal is still at 0dBmV at the upstream receiver in the CMTS, and same levels as other modems in most of the plant before it gets there.

For example, you have a modem on the "through" leg of a 12dB direction coupler.(aka uneven splitter) A DC12 has three ports, an input, a 12dB down tap leg, and through leg with loss of 1dB. For demonstrative purposes a modem hooked directly to the input cable has a TX power of 30dB. Hooking your modem up on the through leg of the tap gives you a return power of ~31dB. I hook my modem up to the DC12 tap leg and my modems power is ~42dB. If you look at our modems power levels, they are ONLY different until we get to the DC12. After that we have identical power on the wire. If it happened any other way this whole DOCSIS thing wouldn't work.


bbeesley
VIP
join:2003-08-07
Richardson, TX
kudos:5

said by odog:

If you look at our modems power levels, they are ONLY different until we get to the DC12. After that we have identical power on the wire.

True, but I want to exit my house as high above the noise floor as possible without clipping as this is where the majority of return path impairments would be potentially generated on the plant


xymox1
Premium
join:2008-05-20
Phoenix, AZ

2 edits
reply to xymox1

My understanding of a CMTS is that the handshake that occurs with the modem looks at the output level seen by the CMTS and then adjusts the modem output power to a "0" db level compared with a reference level. This keeps all the modems at the same level at the CMTS. This is a critical function.

You did however make sense in another way. Using a attenuator for the upstream and forcing the modem to generate a higher level of output would bring its output higher and could increase S/N on the output of the modem. Assuming distortion and things all stay the same. The attenuator then knocks down the level as seen by the CMTS. Its the same level at the CMTS but the modem's output is much higher.

Now is that a good thing ? Without really sophisticated test equipment its impossible to know. As that power output increases lots of things happen. Hard to even guess what would happen. Its possible it might produce a better response.

There is one thing i have seen is that as the modem needs to produce more output power less channels bond. This makes sense as a wider bandwidth is harder to do at high power so it may fall back in bonded channels in order to provide a higher output.

Cisco and Motorola ( broadcom design ) are pretty smart guys/gals and the CMTS / Modem link is most likely very well thought out.

My overall understanding of how it all works is the lower the CMTS sets your modem output power during the handshake the better your connection must be.

I have a FRSA50-15 and a 6 which are low freq attenuators which work below 50Mhz and I will play with and see if having a higher output power helps. »www.arcomlabs.com/33ReturnStepAt···ors.html

Why do I want to hit 0db in downstream ? Its the RF stage 0 point. It would make sense its designed to work best at 0db. 6db is a lot in voltage terms. Hitting zero will also give me the maximum range to accommodate any transient issues giving me +- 10db. Actually I cant think of a reason not to hit 0db as long as the trade off in the attenuation is not costly in other areas as you mention using a diplexer. This is why im searching for a attentuator that only effects the 800-900Mhz band.

There are companies that make a high freq attenuator. The ACS-870/6 is a "cable equivalent" filter that would work for me but you cant buy them online anywhere. »www.arcomlabs.com/34LineConditioning.html

Arcom makes lots of filters »www.arcomlabs.com/31CableTelevis···ers.html



xymox1
Premium
join:2008-05-20
Phoenix, AZ

Another example of a cable simulator filter that you just cant buy online anywhere... »www.eaglecomtronics.com/pdfs/Equalizers.pdf



odog
Cable Centric Vendor Biased
Premium,VIP
join:2001-08-05
Atlanta, GA
kudos:11
Reviews:
·Comcast
reply to xymox1

I 'm still not a fan of adding anything to the mix. I try to minimize even the passive connections, and all potential reflection points. The less pieces in the puzzle, the less problem you tend to have.

An CS would probably be the best case for what you are looking to do. I don't agree with doing it, since I don't think you gain anything tangible. I also got stuck on notch filters when you said you wanted to only affect the 800MHz band. Same with using a step attenuator to bump the level up, I don't think you gain anything tangible. You're only really bumping the level up until it hits the step attenuator, then it is right back at the same level it was before that. Due to noise funneling the artificially increase level would only help if the ingress was coming from inside your house.



bbeesley
VIP
join:2003-08-07
Richardson, TX
kudos:5

said by odog:

I 'm still not a fan of adding anything to the mix. I try to minimize even the passive connections, and all potential reflection points. The less pieces in the puzzle, the less problem you tend to have.

This is one point on which we agree. The fewer connections and parts you can have in an RF network the better


odog
Cable Centric Vendor Biased
Premium,VIP
join:2001-08-05
Atlanta, GA
kudos:11
Reviews:
·Comcast

said by bbeesley:

said by odog:

I 'm still not a fan of adding anything to the mix. I try to minimize even the passive connections, and all potential reflection points. The less pieces in the puzzle, the less problem you tend to have.

This is one point on which we agree. The fewer connections and parts you can have in an RF network the better

Which is good, because we normally agree


DrDrew
That others may surf
Premium
join:2009-01-28
SoCal
kudos:15
reply to xymox1

said by xymox1:

Another example of a cable simulator filter that you just cant buy online anywhere... »www.eaglecomtronics.com/pdfs/Equalizers.pdf

If you really want to mess with levels on the high end of the spectrum affecting the low end as little as possible.... a 100' coil of RG6 has about 6 dB of loss at 800 Mhz and 1 dB of loss at 30 Mhz.

Drop that between the wall plate and modem in place of a 6' jumper.

In reality though, your signals are fine.
--
If it's important, back it up... twice. Even 99.999% availability isn't enough sometimes.


xymox1
Premium
join:2008-05-20
Phoenix, AZ

quote:
This is one point on which we agree. The fewer connections and parts you can have in an RF network the better
In this we all agree

I do still want to be able to hit 0db and play with some measurements in the RF section and look at some levels after the LNA.

I was thinking tho. Cable techs routinely use a splitter to drop levels when they really are too hot. If attenuation is really needed why not use one of these selective attenuators and just drop rcv levels ? Using a attenuator that also effects TX seems to make no sense.


DrDrew
That others may surf
Premium
join:2009-01-28
SoCal
kudos:15

said by xymox1:

Cable techs routinely use a splitter to drop levels when they really are too hot. If attenuation is really needed why not use one of these selective attenuators and just drop rcv levels ? Using a attenuator that also effects TX seems to make no sense.

Cost. Selective attenuators are more expensive and less commonly used then splitters. Plus it's more inventory to stock, track, and maintain.

In addition, it's very rare the forward levels are too high without the upstream levels being too low. In most cases, if they are, it's a distribution system balancing issue.
--
If it's important, back it up... twice. Even 99.999% availability isn't enough sometimes.