Tunable BPF for power line communication receiver.Help

[zzm_3392]

Hi,
The receiver works in the frequency from 10kHz to 500kHz. But because there exist many strong interferences in this frequency on power line, i want to divide it into several frequency band, such as 10-100kHz,100-200kHz,200-300kHz,300-400kHz,400-500kHz. So the receiver can work in another band when there exists interference.
The source impedance of the tunable BPF is 1k Ohms, and load impedance is 10k Ohms. I'm not sure Whether it can be realized by LC or other method. If ok, please give me some advices. Or if you have another better idea, please let me know.Thanks.
Best Regards.

 

[chuckey]

You have not told us the frequencies of these interfering signals. Are the power lines acting as longwave/mediumwave receiving aerial? You have not said what your required pass bandwidth is. I rather fancy upconverting the signal to some frequency out side of you immediate interest (1MHZ?), then filtering it, then down converting it. You will use the same oscillator and in effect are tuning the filtering agent down to your frequency. You must be careful of cross modulation.
Frank

- - - Updated - - -

Just had another thought, If you transmit balanced on adjacent conductors and receive using a wide band differential amplifier, the interference will be common mode and the set up will automatically give you 40 - 60 dB rejection of it.
Along time ago I worked for a wired TV company, the channels were available on balanced pair multicore cables. The balance on the lines should be better then 20 dB, but if somewhere in some terminal box, a bad connection occurred the cables then pick up AM radio. So I built a shield transistor radio with a very accurately balanced transformer at its input. To use, you stuck a piece of wire into one of the two pins of the input connector and tuned for a radio station. You then plugged the set into a consumers outlet, with a decent balance on the cable, the radio set did not pick up anything at all, if a cable had less then perfect balance, the AM station came in at good strength.
Frank

 

[zzm_3392]

Hi, Frank.
The center frequency of interesting signal (OFDM) is selectable in the 50-500kHz. The bandwidth is 90kHz. The strong interference does not come from the aeiral effect of the power lines. It's from the wired conductive interference when the electrical equipment and mechine are powerd on and off, even during their working. So it may cause my receiving terminals into large signal working area. At the same time, the weak interesting signal is decreased and even more cann't be processed by the receiver.
As a result, i want to design a tunable BPF at the front-end of the receiver and set the receiver to work far from the interfered frequency . So, this BPF can reduce the strong interference as much as possible. But i'm not sure whether it can be realized or not by LC.

Thanks.

 

[tony_lth]

You can use a switch to select different fixed band BPF.

 

[eraste]

Hi,
I think that at this low frequencies you can use localised elements (capacitor, inductor, varicap ...) to do your filter. Moreover at this freqencies the Q (quality factor) of inductance are better than at high frequencies.

 

[zzm_3392]

Hi tony_lth,
For this low frequency, can you recommand some low-cost SP4T ? Thanks.

- - - Updated - - -


- - - Updated - - -

Hi,
I think that at this low frequencies you can use localised elements (capacitor, inductor, varicap ...) to do your filter. Moreover at this freqencies the Q (quality factor) of inductance are better than at high frequencies.

Hi eraste,
Can you give more details about the localised elements, or show a draft, or recommand a software to design this tunable filter? Thanks.

 

[eraste]

In fact, localised elements (e.g L, C in cms ) are in contrast with distributed elements used in high frequencies (e.g capacitor or inductor made with microstrip line).
For software, you can use : ansoft designer, agilent ads (both commercial) or RFsim99 (freeware) to make simulation and calculate the components of your filter.

 

[biff44]

performance wise, the best thing to do would be a DSP based filter. That way you could adaptively change the filter bandwidths, add notch filters, etc, based on exactly what was interfering with the signal.

If you want to go cheap, I would probably go with analog (op amp based) bandpass filters, with FETs switching in/out elements to make the center frequency/bandwidth change.

 

[zzm_3392]

performance wise, the best thing to do would be a DSP based filter. That way you could adaptively change the filter bandwidths, add notch filters, etc, based on exactly what was interfering with the signal.

If you want to go cheap, I would probably go with analog (op amp based) bandpass filters, with FETs switching in/out elements to make the center frequency/bandwidth change.

Hi biff44,
Digital filter is not applicable, because the strong interference cause the A/D into saturation, and the interested signal with interference may be clipped. So i think the suitable way is to reduce the interference in analog domain before A/D.
For FETs switching in/out elements to make the center frequency/bandwidth change, do you mean like this in the following picture? The CTRL signal control the FET on/off, then change the total capacitance between.

 

[FvM]

Oc course you can switch individual filter elements with MOS transistors. Using CMOS analog switches is most likely preferable. I also assume that the suggestion by Tony involves less overall effort, but it depends on the filter order. I didn't yet hear details about the filter specification,

 

[ninju]

You aren't using any isolation transformers?

Or any other form of galvanic isolation?

 

[zzm_3392]

You aren't using any isolation transformers?

Or any other form of galvanic isolation?

Hi niuju,
The receving filter accesses into the power line through a transform. So this filter deals with the small signal. I want to design a tunable filter with numerical controlling. It can cover the whole 50-500kHz band by several segments. Can you give me some advices on it?

- - - Updated - - -

Oc course you can switch individual filter elements with MOS transistors. Using CMOS analog switches is most likely preferable. I also assume that the suggestion by Tony involves less overall effort, but it depends on the filter order. I didn't yet hear details about the filter specification,

I
At the current stage, i only want to assure whether the tunable filter covering 50-500kHz can be realized by LC or not. The insertion loss is much better less than 1dB, input and output impedance are 1k Ohm and 10k Ohm respectively. It is best to have small dimension. So it's the last choice to use several fixed BPF with switches. I most want to know whether there is another way. Can you give me some suggestion?

 

[ninju]

What i suggest is a LC filter. You want a constant BW of 90khz, with different center frequencies also, you want to match impedances. So, i think you should go in for a fixed L and variable C. The C can be an array of capacitors in parallel, which you can numerically switch depending on the center frequency.