C/N

[trungkstn]

Hi everybody.
Do you know how to measure carrier to noise (C/N) in CATV? If i use tv tuner, can i measure?
Thank a lot.

[RF_Developer]

This is one of those "if you have to ask the question, the answer must be no" type of things. I work in RF. Firmware and analogue stuff for power amplifiers in fact. I've got a lot of experience but I wouldn't dare to try and make a measuring device for that without asking the advice of some pretty serious RF guys.

You have to have a good RF/analogue front end to do decent measurements. The basis for the measurement is a spectrum analyser or a narrow band tunable receiver that can be swept across the band being measured, which is what a specrum analyser does. You then have to measure carrier power, which may not be as easy as it sounds, and sum up noise power across the measurement band. Because the noise is very low power and the carrier high you need a pretty good dynamic range. The C/N ratio is then just that: the ratio of the two measurements expressed in dB. If what you want is a quick and very dirty indicator for some simple application then it may be possible with simple kit. It would be simpler if you could control the source/unit under test to do the measurements one at a time, i.e. the noise without the signal, then the signal itself without having to worry about the noise, but as the noise includes many elements that are only there with the signal, e.g. phase noise and spurs, even harmonics on wide band measurements, this won't give the right answer.

All good fun :-)

RF Developer

[trungkstn]

My boss give me CXE101 and said that "measure C/N in CATV using microtroller PIC or dsPIC".
I answered:"CXE101 is only distribution amplifer. I think if you want to measure C/N in CATV then you should use method of spectrum analyzer: sweep or FFT. But I don't know how microcontroller can work with broad band."
Then, he said:"you must create a device that can measure C/N".
So sad.
I asked some friends to answer this problem. A friend said that i can use tivi tuner to calculate in some ways then i can recive C/N ratio.
I search internet and see http://www.mtmscientific.com/spectrum.html

Beside, i saw "Compact Spectrum Analyzer" by Neal Martini. You can search and down this article.

I really need some advice to solve this problem.

Thank a lot RF_Developer.

[temtronic]

OK..how much time and money is this project allowed? Honestly, it could take 4-5 man years at $70,000 per year to get 'close' to what the boss wants.
The slow scan spectrum analyzer site is a nice article, but HOW will you calibrate the device ? And, I'll bet no two have the same response curve to calibration is key!
You could 'google' for a week or two and get something already made,or close to the design spec, but it will chew up a lot of man hours. Been there, done that.
My background is in the Avionics field as well as Optical emission spectrometers, both areas of high noise,low signals.Nailing the signal/noise information is NOT a fun task.


RF means 'Real Fun'

[SherpaDoug]

1) This is not really related to using the CCS compiler on PICs. I think you might have better luck on a another BBS. You might try www.edaboard.com.

2) For anybody to give real help we need much more information about your application. How much accuracy do you need? Does it need to be calibrated, or is this just for peaking adjustments and short term relative readings are OK? What is your parts budget for the final product? What is the expected production volume, or is this for an in-house test fixture? What is your R&D time and materials budget? Is this actually for CATV, or is it just a signal similar to CATV?

The more you can tell us the more we can tell you.
_________________
The search for better is endless. Instead simply find very good and get the job done.

[trungkstn]

Thank a lot SherpaDoug, temtronic.
I'm really happy because of your replies. Because I'm worried that nobody help or answer my question.
1. @ temtronic
thank you $70,000 per year is too expensive with me. And only me on this project.
I like [RF means 'Real Fun'].
2. @ SherpaDoug:
[1]. Project's purpose is estimate the signal's quatity so that i think accuracy is not high.
[2]. budget is about $200
[3]. actually CATV
[4]. I'm used to using CCS complier on PICs. So that i hope you can help me.

P/s: really speaking, my English is not good. So that if you don't understand me, you should tell me about that. I will try to explain .
Again, thank a lot.

[bkamen]

The replies for this thread essentially ring true as to why when you look products from folks like Agilent and Tektronix (Spectrum analyzers) why the good ones are so darn expensive.

There's a reason...

Making a good one is hard.

Your boss should know this (shame on him if not).

His best bet is to look for used equipment on ebay or through some of the refurb'd equipment outlets -- BECAUSE IT WOULD PROBABLY BE CHEAPER in the long run.

(shrug) but I don't know all the goals behind his request to you.. so...

-Ben
_________________
Dazed and confused? I don't think so. Just "plain lost" will do. :D

[languer]

This is not that unusual to CATV. I think most of their high-speed modems now do this. Of course, they have nice DSPs inside. From a signal quality figure of merit (SQP), you can do this relatively easy (in theory, practice is quite a bit more complicated - but that is your work, reason you get paid for this). Does not mean you get much accuracy though (but a ballpark number).

To start with, you make two assumptions: (1) you have known channel power - you are sure you have power on channel, (2) you can estimate what the power level is when the channel power is absent.

To that end, you measure the channel power (again, assuming single tone, very narrow frequency). You then filter out the single tone, and measure channel power again. In a DSP this is quite straight forward; you have some A2D process, filter and perform some complex arithmetic to extract on-channel power. You add a filter (digitally) such that the single tone is filtered out and you're left with only the on-channel power without the single tone. Perform comparison and you get your figure of merit. Analog-wise this is done in different ways; you can mix-down the channel into a very low IF and then feed it through two paths, one with a BPF to the carrier, and another unfiltered to get the whole channel (i.e. carrier plus noise), you do the math stuff afterwards. You can look a two sections of the band, one which is known to have the signal, one which is known to be clear of the signal (each of these bands will require their very own BPFs).

None of this is new, it has been around for years. A simple uC is incapable of doing what you want by itself, it requires significant circuitry to do this. At this point I will second what everybody else has said here; what you're looking for is not available. Can it be made? Yes. Is it ready made for you? Unlikely.

However, if I was doing this as a HAM type thing. I would estimate before hand what the NF of the amp was , I would have a temperature sensor to estimate the ambient noise, I would have an ambient noise fudge figure (which is determined empirically). This would determine the No. I would then use a power detector (band limited to the band of interest) and I would measure its power. This would determine the C. And there you have your C/No. There could even be a niftier way, but it would require a bit more components (i.e. extra amplifier, detector, and filters - and more math).

Have fun and good luck.