Efficient Use of Spectrum: The Importance of a Safe Gap Between Channels

[LLT71]

If we have 300kHz bandwidth and every channel uses 100kHz BW, then theoretically we can have total of 3 channels that won't overlap. Why do we need to assign some of the bandwidth for "safe gap" between channels if they "don’t" overlap?

 

[NascentOxygen]

The filter you construct to confine each channel to within a 100kHz BW is always going to depart from the "ideal".

 

[LLT71]

The filter you construct to confine each channel to within a 100kHz BW is always going to depart from the "ideal".

I'm sorry I'm beginner at this can you make it a bit simpler? :D

 

[scottdave]

The shape of a response from a band pass filter (it only let's a certain band of frequencies to pass through) will look like a curve. Think like an upside down parabola. This is not the exact shape, but just trying to give you some idea. Typically the cutoff frequency will be about where the filter is cutting the power down to half of in the peak. So there is some wider frequency range where some signal still gets through.

 

[Baluncore]

If we have 300kHz bandwidth and every channel uses 100kHz BW, then theoretically we can have total of 3 channels that won't overlap. Why do we need to assign some of the bandwidth for "safe gap" between channels if they "don’t" overlap?

Does the 100kHz BW specification already include a guard band?
Are the 100kHz BW channels analogue or digital data?
The channel separation needed will depend on how the information is modulated onto one or more carriers.
The receive demodulators may be able to operate in each others sidebands, so long as the carriers remain separate. For example, if AM modulation is used, the two channels could be transmitted as independent sidebands on one carrier. The separation can then be done by IQ mixing, without BP-filters or a guard band.

 

[tech99]

The filter you construct to confine each channel to within a 100kHz BW is always going to depart from the "ideal".

Further, for digital transmission, to preserve the pulse shape we need filters that roll off gently.

 

[skeptic2]

Not only are the receive channel filters similar to concave downwards parabolas, the transmitted signal tends to look like a trapezoid with the narrow side upwards. Two 100 kHz wide signals, 100kHz apart will always overlap their "skirts" to some extent. The important thing is to make the skirts so steep that at the point where they overlap, the sidebands are so far down it doesn't matter. With an FM signal the steepness of the skirts is largely dependent on the deviation index, the frequency deviation divided by the highest modulating frequency.

Perhaps with more information we could help you more. What country are you in? What frequency band are you using? What type of information are you transmitting (analog, digital, type of modulation and what is your highest frequency or data rate?)

 

[sophiecentaur]

We can design an build 'brick wall' filters for medium and low level signals and that allows us to use very narrow guard bands, for instance for cables, carrying multiple signals but you can't make the filters for modulators and high power transmitters as good as that. If you try to fill the two channel gap then someone nearby who is on the edge of their service area will risk getting Adjacent Channel Interference. It won't happen to every one with fringe reception but the system has to cover all eventualities that arise from having a two dimensional area with time varying propagation conditions that needs usable coverage all over. If it were up to the advertisers, they wouldn't care about the small percentage of people on the fringes but the Authorities look after the interests of minorities (in this case).

Compare broadcast reception philosophy with mobile phone service and fast broadband. No one sticks up for people in rural situations with no broadband because laying on fibre is costly and will only be done if there are enough customers.

 

[Bart McCoy]

Without diving into filter theory too deeply... you said that each channel uses 100 kHz bandwidth. By that, I assume you mean that something has filtered each channel with a 100 kHz bandwidth filter. Filtering is not ideal, and the meaning of "bandwidth" is important. Ideally, INSIDE your 100 kHz window, the filter keeps your signal and OUTSIDE your bandwidth, it attenuates/kills unwanted signals completely. What really happens is that INSIDE your 100 kHz window, the filter keeps most of your signal. OUTSIDE your bandwidth, it attenuates unwanted signals *somewhat*. At the edges of your 100 kHz window, it may only attenuate unwanted voltages by 0.707 (-3 dB) or 1/2 (-6 dB), unless you have a kick butt filter. As you move further outside your window, the filter performs better. But what if you have garbage signals at the edge of your 100 kHz window-- can you live with attenuating those by 71% or 50%? Depending on your signal level in your 100 kHz window, the noise levels just outside your 100 kHz window, etc, you may or may not care. If you give yourself a little extra bandwidth for margin-- not allowing neighboring channels to be side-by-side, the filter will do a MUCH better job of cleaning up unwanted noise/signals (noise = other channels).

 

[Baluncore]

Why do we need to assign some of the bandwidth for "safe gap" between channels if they "don’t" overlap?

We do not have to waste spectrum on a "safe gap".
Where is the reference that says we must have a "safe gap".

 

[Tom.G]

We do not have to waste spectrum on a "safe gap".

It is to avoid adjacent channel interference. See: https://www.fcc.gov/wireless/bureau-divisions/mobility-division/700-mhz-guard-bands

Where is the reference that says we must have a "safe gap".

See page 4: https://transition.fcc.gov/realaudio/mt120301.pdf

Or generally: https://www.google.com/search?q=guard+band+site:fcc.gov

 

[Baluncore]

It is to avoid adjacent channel interference.

Those 700MHz guard bands are utilised, they are not empty. “The 700 MHz Guard Bands licensees must adhere to specific technical and operational measures designed to minimize interference to public safety licensees …”

See page 4: https://transition.fcc.gov/realaudio/mt120301.pdf

FreqDivDuplex only needs that guard band because it is a co-located two way system. If there was no separation, the duplexers could not prevent the transmitters desensitising the receivers. That guard band can be used at other sites that are not part of the same network. Notice that there is no guard band between the channels in the upstream block, or between the channels in the downstream block. Adjacent channels do not require separation if they are all TX or all RX.
The co-location of TX and RX channels on UHF usually get a 10 MHz separation. That local site guard band spectrum is utilised by systems at other sites.

It comes down to definitions; If a channel requires a 100kHz band then that is what it needs and gets. That service should not radiate significantly outside that 100kHz band. Radiation outside the band is usually required to be more than 70dB below the peak in-band signal.

The OP did not specify the 100kHz channels were mixed TX/RX.

 

[sophiecentaur]

We do not have to waste spectrum on a "safe gap".
Where is the reference that says we must have a "safe gap".

The ITU have recommendations for the efficient use of spectrum in a way that helps to maximise the number of users. I am referring to over-air spectrum and not cable systems here, of course. Adjacent channel interference is a problem with practical implementations so it is a lot better to occupy every third channel from any given site and leave the two adjacent channels for use at sites with service areas outside your local area. That way, users the fringe of your area are unlikely to be swamped by transmissions from the adjacent area because they are only near the fringe of the other service area. It reduces the problem of adjacent channels interference by using spatial diversity.
This link has a picture of how a Cellular network for mobile phones can maximise the coverage and allow frequency re-use. The 'safe gap' is used in conjunction with a topographical gap. (I did look for a similar diagram for UHF TV transmitter networks but failed)