Page 4 of 8

Re: HIGH PASS FILTER FOR OPTIMUM HF RECEPTION

Posted: Wed Jan 30, 2019 3:03 pm
by glovisol
PRACTICAL TEST OF THE CHEBYSHEV N=9-r=0.1- Zin=Zout=50 Ohm (2)

Here High Pass Filter performance 0 - 1900 KHz is shown. The proposed filter not only eliminates the strong local MW Broadcast signals, but also protects the RSP receiver's front end from strong or very strong signals below the MW Broadcast band, which is left unattenuated by the internal MW notch filter. Thus the High Pass filter guarantees complete elimination of spurious potentially caused by signals in the LW band, no matter if the receiver is in LIF or ZIF mode.

Re: HIGH PASS FILTER FOR OPTIMUM HF RECEPTION

Posted: Wed Jan 30, 2019 4:15 pm
by glovisol
CHEBYSHEV N=9-r=0.1dB-Zin=Zout=50Ohm HIGH PASS FILTER - PROTOTYPE #6 (FINAL) (1)

Apart from good performance, this filter is very easy to build, because hard to find toroid cores are not necessary for the coils. Since there is excellent correlation between Coil 32 software and actual measured turns, the filter can be readily reproduced, without testing the coils, from data uploaded below.

As coil formers I used pieces of plastic garden hose pipe with an external diameter of 17 mm. This is a common, transparent plastic garden hose with internal fibre reinforcement. I compared two coils on the Q meter, one on the garden hose and the other in air and the difference in Q and in self capacitance was negligible. Two small holes are drilled in the coil former using the heated tip of a small screwdriver.

Capacitors must be polystirene, mylar or mica (mica is best, but more expensive). Ceramic capacitors (unless they are NPO grade) cannot be used because of great capacitance shift with temperature.

The filter topology problem (e.g. how to simply accommodate HI Z long wire antennas and the RSP HI Z inputs) remains open, but I have some ideas to be tested next.

Re: HIGH PASS FILTER FOR OPTIMUM HF RECEPTION

Posted: Wed Jan 30, 2019 4:18 pm
by glovisol
CHEBYSHEV N=9-r=0.1dB-Zin=Zout=50Ohm HIGH PASS FILTER - PROTOTYPE #6 (FINAL) (2)

Re: HIGH PASS FILTER FOR OPTIMUM HF RECEPTION

Posted: Fri Feb 01, 2019 10:38 am
by glovisol
SOLVING THE FILTER TOPOLOGY PROBLEM FOR THE HI Z BALANCED INPUT

Chebyshev filter N=9, r=1.5 dB, ZIN/ZOUT=0.6 - PROTOTYPE #7

Filter prototype #6 previously presented works between 50 Ohm input/output impedances and is therefore able to accommodate 50 Ohm antenna lead-ins and 50 Ohm RSP inputs, but what about receiving systems based on long wire antennas and RSP receivers equipped with a HI Z balanced input?

Choosing a suitable prototype among the available Chebyshev low pass filter tabular data, transforming to High Pass and manipulating the ripple parameter, one can obtain a filter network with Zin = 600 Ohm, Zout = 1000 Ohm at the cost of an insertion loss (due to ripple reflections) between 3 and 6 db, but with inductance levels low enough to still allow the use of air core coils. In other words the HP filter also matches the long wire antenna to the Hi Z input and only a 1:1 balun will be necessary at the antenna lead-in.

Tests on such a filter prototype are not easy, as two baluns are required for testing in 50 Ohm systems: therefore insertion loss and frequency response of these devices tend to influence the measurements. All data necessary for building the filter will be presented in the next post. 50 Ohm test baluns are made with Amidon T94-26 toroids: to obtain the tightest coupling possible, the 50 Ohm winding is wound in between the turns of the high Z winding. Such baluns are only good for bench testing, but not for operation, as their noise rejection is very bad due to high input/output capacitance.

PC screens uploaded below show filter performance. Some unwanted signal feedtrhu can be noticed with the High Pass filter inserted: this is due to the fact that the test prototype has no shielding enclosure while working at high impedance. Ultimate filter performance will be presented once it will be built & shielded in its final form.

Re: HIGH PASS FILTER FOR OPTIMUM HF RECEPTION

Posted: Fri Feb 01, 2019 12:33 pm
by glovisol
FILTER TOPOLOGY FOR THE HI Z SDR BALANCED INPUT

Chebyshev filter N=9, r=1.5 dB, ZIN/ZOUT=0.6 - PROTOTYPE #7

Here below complete data for the HI Z Hi Pass filter and spectrum analyser response. It is difficult to determine if the ripple near cutoff (2 MHz) is inherent to the filter itself or caused by the balun response. In any case the selectivity slope is as anticipated. In normal use a 600 to 600 Ohm isolating transformer, with primary connected to the long wire antenna and secondary connected to A1/A2 should be used. Terminals B1 & B2 must be connected to terminals P & N of the RSP. The two 50 Ohm to Hi Z test baluns are the two yellow-white toroids.

A fully balanced version of this filter will be shown in the next post.

Re: HIGH PASS FILTER FOR OPTIMUM HF RECEPTION

Posted: Fri Feb 01, 2019 7:19 pm
by glovisol
ADDITIONAL INFORMATION ON MEASUREMENT BALUNS USED FOR HI Z HIGH PASS FILTER TESTING

Chebyshev filter N=9, r=1.5 dB, ZIN/ZOUT=0.6 - PROTOTYPE #7

The baluns used for filter testing, as described in the previous post, were made with iron powder (TYPE 26) material and were not entirely suitable, but this was the material I had on hand. Today I received a much more suitable (for wideband transformers) ferrite binocular cores and here below I explain the transformer design procedure in detail, giving measurement data which is otherwise almost impossible to find.

ERRATA CORRIGE: previous binocular core was by mistake mentioned as BN-43-202, while in fact it is BN-73-202, as amended below.
=====
The binocular core is model BN-73-202, where 73 is the ferrite material and 202 is the size: B(length)=14.35 mm / A(width)=13.3 mm / C(thickness)=7.5 mm. With material type 73, mu=2500.

LC meter and Q meter measurements gave the uniform result that N=7 turns produce L=580 uH and a wideband Q response of Q=3.5 @ 1 MHz. From this data:

Al = ((100^2)*L)/(N^2) = 118*10^3 This Al is now our parameter for wideband transformer calculations.
=======

For our wideband test baluns the 50 Ohm windings must have a reactance Xl=50*4=200 Ohm @ 1 MHz, hence:

L= (200*10^6)/(2*PI*10^6) = 32 uH
N = 100 * SQRT(32/118,000) = 1.6 turns, we shall use 2 turns.

Therefore:

For the 50-600 Ohm balun: N1 = 2*SQRT(600/50) = 7 turns
For the 1000-50 Ohm balun: N2 = 2*SQRT(1000/50) = 9 turns

Tests with the new baluns demonstrated that the filter insertion loss is negligible, apart from the passband ripple of 5dB, which is confirmed and is the price to pay for the other filter characteristics. Uploaded below the PC screen of filter operation with new impedance matching transformers. Effectiveness of the baluns made with ferrite material 73 is that attenuation of the test unit does not increase with frequency any more.

Re: HIGH PASS FILTER FOR OPTIMUM HF RECEPTION

Posted: Sat Feb 02, 2019 12:32 am
by Mike2459
I wound a transformer for the Hi-Z port on my RSP2 with a BN73-202 Binoc Core It rolls off a little above 20 Mhz, not sure if it's the transformer or the mini-whip. Anyway, the published data that I saw shows the BN43-202 with an Al of 2200 ±20%. uH=(Al*turns²)/1000

Re: HIGH PASS FILTER FOR OPTIMUM HF RECEPTION

Posted: Sat Feb 02, 2019 9:28 am
by glovisol
Hi Mike, welcome back! I am glad you are checking, as it is so easy to make mistakes.

There is little difference, really, between what you and I say if we take into consideration the fact that for powdered iron (low mu values) Al is normally given in uH/100 turns units, while for ferrites (high mu values) Al is normally given in mH/1000 turns units,

uH (microhenries) being 10^-6 Henries
and
mH (millihenries) being 10^-3 Henries

Therefore coil turns N can be calculated as follows:

N = 100 * SQRT(L[uH]/Al) [1]

or

N= 1000*SQRT(L[mH]/Al) [2]

If we now take [2] (the one you apparently use) we see that to get inductance L the correct formula is:

L = (N^2)*Al/(10^6) where L is expressed in mH and Al is expressed in mH/1000turns .

In order to compare with previous work done with powder iron cores, I preferred to express the Al value in uH/100 turns, coming to the value of 118,000 which is equivalent to the value of 11,800 when expressed in mH/1000 turns.

Now for the difference Al between rated value (Al=2,200) and value found by me (Al=12,000) the reason is I made a mistake in mentioning material 43, while in fact I was using material 73. In other words the binocular cores measured are: BN-73-202, with a rated Al = 12,000 +/-20 %. Additionally the rated tolerance of +/- 20% must be taken into account and furthermore:
Experience shows that this formula provides only a rough estimate of inductance. You will
need to actually measure the inductor on the network analyzer to determine its inductance
precisely. From: "Toroid Inductor construction"
And this is what I did, measuring the core both with an inductance meter (Monacor LCR-4000) and Q meter (HP 4342A) and obtaining identical results. To conclude it is good to know that the cited binocular core is capable of working as a wideband 1 - 20 Mhz impedance matching transformer with practically no losses and with a small number of turns.

Thanks again for your help, attention & assistance,

glovisol

Re: HIGH PASS FILTER FOR OPTIMUM HF RECEPTION

Posted: Sat Feb 02, 2019 2:06 pm
by glovisol
Chebyshev filter N=9, r=1.5 dB, ZIN/ZOUT=0.6 - PROTOTYPES #7 & 8

Uploaded below data for single ended filter prototype # 7 with ferrite measurement baluns in place of powder iron and prototype #8, balanced version, which should help in local noise suppression.

Re: HIGH PASS FILTER FOR OPTIMUM HF RECEPTION

Posted: Mon Feb 11, 2019 5:25 pm
by glovisol
FILTER CHEBYSHEV HIGH PASS N=9 - 50/50 Ohm - UNBALANCED

From prototype to finished filter

Today, after a long wait, the die cast boxes arrived and I could assemble the Chebyshev N=9 / 50-50 Ohm filter. Performance & completed unit uploaded below. I have resorted to rigid tubing for coil formers for better mechanical stability. Details in next post.