STARTSEITEN

>>>Links:

Formulas2

Engineering - Electronics - Content :

Lumped electronic RF-filters content:

 

Lumped electronic RF-filters

Rf-filter design starts in the same manner than usual filter design:

  • Clarifying specification values of bandwidth, band middle frequency, roll off, rejection, impedance.
  • Find out which kind of filter shall be used: Butherworth,Tschebischeff, Cauer, Wave, Zig Zag or Parametric filter.
  • To minimize the number of coils, a Parametric filter shout be the first choice .
  • Estimation of pole and zeros and solving the filter equations with a computer software or find the filter values in a catalogue.

Filter catalogues:

  • [1] ZWEREV, “ELECTRONIC FILTERS” ;
  • [2] SAAL, “THE CATALOG OF NORMALIZED LOW PASS FILTERS”;
  • [3] H: RÜHL H.H.NGUYEN, “PRAKTISCHER ENTWURF VON WELLENPARAMETERFILTERN”;
  • [4]ECKHARD GLEISSNER , “ZUM ENTWURF VON HOCHFREQUENZ BANDBASSFILTERN”>>, Dissertation TU München - Telefunken- Sonderdruck

Up to this design activity, the design of RF-filter differs not  from the usual filterdesign, but  the number of electrical components which must solve the characteristic equation, is very limited .This elements are:

                                  Fig.1 Circuit components to solve filterroots desingning RF-filters

The characteristic equation must be solved, using only this elements to nullify the roots of the equation [1]. At the same time ,expected parasitic values which my appear in a practical circuit, and the coil losses, must be watched. Using series resonators my lead to unacceptable parasitics. A computeranlysis of the circuit will at least show if it really works what has been designed . After that we can simplify the circuit using Norton’s equations watching the appearing component values.This maximum and minimum values in the frequency range from 70 MHz to 1 GHz are:

  • 12 nH < L < 500 - 3000 nH
  • 0.2 pF < C < 5 nF
  • Qmin = 20  , Qmax =140

 

Parametric filter dedects corona in molted transformers

The insulation of high voltage transformers usually is measured by watching the DC-leakage current. But in molted transformers the problem of unknown tiny air bubbles exist. This bubbles my create ionization in the captive air.The corona effect may exist in a transformer for years after the high voltage breakdown will happen.The ionization bubbles can be found, watching the MHz leakage current with a scope. But this will lead to lot of mismeasuremends due of added radio frequencys.To filter this radiowaves , a proper filter must used. The requirement of the filter is : Fo = 1 MHz, Zin = 1 kOhm, Zout = 50 Ohm, B =100 khz, Fig. 1 shows the measurement circuit, using a sensitive oscilloscope to watch the corona.

 

 

 

 

 

             Fig 1 Corona measurement equipment

The Filter is a transforming parametric circuit using lumped components rare then an OP-active filter which could be damaged from high voltage sparcs. Fig.2. Fig.3 shows S21 .

 

 

Fig.2 Corona measurement filter                                                                                                                                                                                                              
                                                  Fig.3 corona measurement filter S12

Retour:

 

2Pole parametric IF bandpass matches OP-AMP

The above simple 2 pole parametric bandpass can be modified to be a 10Mhz IF-Filter matched to the OP-AMP in and outputs.The input of the filter has 1 Ohm to fit to the OP-ouput, whereas the output is resistive matched with a 100 Ohm resistor to fit the Megaohm input of OP-AMPs.The series connection of two or three filters between OPs give an high selectivity If-Filter.Fig 1 shows the circuit and Fig.2 the selectivity of one single filter.

Fig.1 OP-AMP bandpass

           Fig.2 S21 of op bandpass

Return:

 

A simple parametric IF-Filter

We get a simple narrowband IF filter using:

The circuit is gained using Norton’s transformation and matches the output of RF-Amp to a transistor. Fig.1

  Fig. 1 IF-Filter 10 MHz

  • C1 = 200 pF; C2 = 100 pF; C3 = 400 pF; C4 =1 2 pF; C5 = 560 pF; C6 = 200 pF; C7 = 200 pF; C8 = 2 pF; L1 = 450 nH; L2 = 430 nH; L3 =4 000 nH; L4 = 1014 nH;
  • C1, C8 and L3 are problem less components and may be fixed SMD-components.
  • The parasitics at this frequencies are uncritical

Fig. 2 Simple IF-Filter

Retour:

Bandwidth switchable If-filter

To enhance the selectivity in a receiver, it is often help full to decrease the bandwidth with a switch. Fig.1 shows a IF-filter having two bandwidthes.The switches can be a small coaxial relais.Their little parasitic capacitors will not affect the function of the filter to much, because on their nodes are big capacitors of 400 pF:

Fig.1 Double bandwidth if-filter

The component values are:

  • C1 = 200 pF; C2 = 100 pF; C3 = 400 pF(+10 pF); C4 =15 pF (-7 pF); C5 = 555 pF (+5 pF); C6 = 300 pF; C7 = 2000 pF; C8 = 2 pF; L1 =440 nH; L2 = 430 nH; L3 = 4000 nH; L4 =1014 nH;

             Fig.2 S12 of the double bandwidth filter

Retour:

DVB-Covering Filter

Outside band signals my disturb the digital television DVB especially at the upper frequencies of 800 Mhz. To reject this unwanted signals, a DVB-Covering filter was developed.We get a practical circuit using a the following parametric filter:

  The circuit shows Fig 1 and is developed using a parametric filter software and Norton’s transformation.

 Bild 3 DVB-Covering Filter   !!! Z=50Ohm

  • C1 = 0,5 pF; C2 =10 pF; C3 = 2,2 pF; C4 = 1,5 pF; C5 = 2,4 pF; C6 = 0,5 pF;
  • C7 = 0,5 pF; C8 = 1 pF; Cpar.= 0,5 pF
  • L1 =15 nH; L2 = 20 nH; L3 = 160 nH; L4 = 18 nH;  L5 = 11,6 nH;
  • C -Values are CAD-values, in a practical circuit, the measured parasitic values must taken into account

                                      Fig. 2 S21of Pf341

  • Frequency band = 400 MHz to 800 Mhz

Retour:

1 GHz Mixer-output filter

A Mixer must work into some resonating device to gather the wanted mixing products. A small filter can do this job, but it must be properly matched. We use: and find the circuit of Fig.1 where the parasitics do not affect the function as Fig.2 shows.The low output coil in a practice circuit is only a small path on the PCB and is calculated with a Q of 20. The black component values are computed values.The PCB -capacitors must be subtracted.

Fig.1 Mixeroutputfilter-circuit

 

                                                               Fig.2 S21 of Mixer filter.   Matching values are :S11 and S22 = 12dB.

Retour:

A receiver-unversal-polefilter

This filter is an example of filters having certain frequency poles.This example shows the high increase on rejection outside the band. We use a wave parameter filter from [3] having two resonators at Fo and two poles at frequencies near the bandend. As we want high low insertion loss in the band, we add a third resonator at Fo. This filter is an universal filter to be changed for many purposes. For instance -narrow band -channel filter or high selectivity if filter. Impedance or frequency changing is easy. To change the bandwidth , the poles at band end must equally changed in percentage to broaden the band. This filter is easy to handle in a CAD-program, the three resonators are always at Fo and the poles can than be adjusted.This allows to build a bandwidth switchable filter. Fig.1 shows the circuit for Fo =100 Mhz and Fig.2 the transmission S21setting Q=100 . Lower impedance on the input or output can be realized as it is well known , using taping of the first coil.

                                                   Fig.1 Wpo2 -high selectivity universal filter

                                              Fig.2 wpo2 universal-high selectivity-filter

Return:

Noise measurement using a parametric if filter

To measure the noise produced in a communication system, the bandwidth of the noise measurment device has to be limited over the bandwidth of the system. The noise power must be measured after the filter and the generated noise in the source can be computed from the ENR value:

     Fig:1 Noise measurement in a communication channel

                          Fig.2 Formulas to compute the generated noise

The necessary data’s are: fo=140 Mhz; Bandwidth 1dB =+-6 Mhz, Bandwidth 3dB =+- 10 MHz ,  Bandwidth 20dB = +-15 Mhz. The used filter is a parametric filter having the grade of 11. The poles are :   4,1,3,  Grad =11 . A  PCB circuit using fixed air coils is used. To tune the filter, SMD-trimmers have been used .To manufacture a series product a lot of tolerance considerations still has to be made.The circuit up to now is shown in Fig. and the transmission S12 in Fig.4.

                                                     Fig.3 140Mhz noise measurementfilter

                                      Fig.4 S21; 413-Filter

                                          Fig.5 S12; 413-Filter

                                            Fig.6 S22; 413-Filter

Return:

QPSK-Data Filters for digital radio

Today , digital modulated communication channels are everywhere. Ignorant people may think its digital communication, using data jumps from 1 to zero, the same as digital logic . But this is not the case.We look how the communication bit’s look like and show how to design lumped RF-data filters for high bit rates. Look at two different digital bit’s, the hard jump and the soft cosines shape jump and its amplitude density. We get this density from a Fourier transformation . Fig.1.

                                                 Fig.1 Amplitude density of data bits

 Fig. 2 shows the graphical results of this power density .The hard data bit produces power  at multiples of the bitrate. To avoid data confusion due of this unwanted power, soft square cosines bit’s are used in digital data communication.

Fig.2 Power density of different bit’s..                                     Fig. 3 Filter specification of digital radio mod-demod filter

 

                        

 

 

 

 

Now we look to a classical specification of a filter to change the digital output bit’s of a QPSK-modulator into a cosines square shape. And look too, to a QPSK- demodulator specification to transform the data bit’s content , back to analog values. Fig.3. We see the shape of a very broadband low pass filter having a little sinx /x over swing before crossing the Nyquist frequency fbit. At this frequency the roll off function is working, which is a cosines square function having a variable steepness n. The demodulator filter is indeed a sinus function, written in cosines form without sin x/x increase. At Fig 4, the equations can be found:

Fig.4 Digital radio mod-demod -filter specification. formulas

 The Filters

Mod and demod filters are both low pass filters, I use the same simple basic filter , it’s a n-pole low pass from a catalogue[3] and select it to match approximately the specification requirements. Now how can we get the cos roll off and the sinx / x increase? We get the roll off, by shaping the poles resonator with a resistor only damping at a certain series resonator frequency .The sinx / x  increase is very easy to get. A little differentiating resistor into the infinite pole capacitor enhances the corner frequency over swing.This filter is much more easier to adjust than an full computed overswing filter without resistors, where all components my change the roll off too. Fig 5 shows one pole of the filter.  Fig.6 Shows the transmission of such a modulator filter, using 5 poles in series.At the 2 upper frequencys, the poles work without resistors..

Fig.5 Universal data filter pole

Fig.6 Example of communication modulator filter transmission

Return:

Minimum Coil Parametric 70 MHz If Filter

Parametric filters need less coils than every other filtertype. Here is a minimun coil broadband 70 Mhz Parametric Filter The circuit of Fig.1 is the result of  computerized solving of the H(s) and K(s) equations, of a parametric filter.The components are found in a computer  iteration process to nullify the equation H(s).The Circuit shows the typical C-Input of Parametric filters and has only 5 coils to get a grade of 11. Fig.2 presents the transmission .

Infinit poles = 3. Zero poles =4. Finite Poles = 2. Grad = 2*2+4+3=11.

                                       Fig.1 Circuit and values of Fil342

                        Fig.2 Fil.342 Transmission