New front end for use of the Red Pitaya in SDR applications

The Red Pitaya SDR board is based on the Xilinx Zync SOC and has 14 bit external A/D converters. However, for SDR usage on the HF bands from 0.1-30 MHz (and for that matter up to 50 MHz) the Red Pitaya is a bit “deaf” in the stock configuration. I have made a broadband amplifier that has a fairly high gain and very good IIP3 properties. Below I have posed some pictures of the prototype amplifier.

20160424_153046-1This is the prototype amplifier. I inserted a ferrite ring on the input lead to roll off the VHF / UHF sensitivity to reduce problems with nearby broadcasters etc. There is a also a PI network attenuator on the ouput and I have inserted a couple of beads in that as well to roll of the outpu response when frequency increases. The other components in the lower part is a input pi attenuator I used when I did some VNA frequency response measurements. This as well as the RCA plus is not used (RCA plugs are surprisingly good for low level RF signal routing in the HF bands and nice to use in the lab).  I used a more professional attenuator with a large attenuation range and flat response to determine the proper attenuation level after the preamp into the Red Pitaya. Reducing gain after the first amplifier has very little effect on the noise figure. Reducing it before the first amplifier directly adds to the noise figure. I added some protection diodes over the input to reduce the risk of strong RF signals or static voltage build up damaging the input. Below I am measuring the response of the attenuator with the DG8SAQ VNA. It was flat from 0-1,3 GHz.






Wideband HF RF choke design for QRO operation. V2.

I have now continued the investigation into a possible design for a wideband HF RF choke for QRO applications. The criteria is that the R part of the R+jX load the choke present to the common mode current on a coax should be so high that there is low risk of overheating the core even for QRO operation. It turns out that two main loops that gives sufficient resistive (and reactive load) that covers the lower frequency range combined with another loop with smaller diameter and fewer cores that takes care of the upper frequency range will give quite good results. This is similar to what GM3SEK has observed and described in his publication. Vector network analyzer measurements confirms this.

Above the three choke sections can be observed
DSC_5174 DSC_5173 DSC_5167 DSC_5186 DSC_5175 DSC_5169
Above, the measuring setup can be seen. On the right there is a multicore choke that was  tried. It gives a big resistive and inductive peak low in the frequency range. Far more than required. The cost will be high due to the many cores and is not justified. Therefore a 3 + 3 + 2 design was found more optimal.

Above the resistive (blue) load to the common mode current, the inductive load (red) to the common mode current and the Q (green) for the choke can be studied. The choke presents around 1000 ohms resistive to HF current in the frequency range of approx 2,5-25 Mhz. Another material could probably have been added to prevent the drop above 25 Mhz. 10 Meter would be a bit marginal for QRO operation with a lot of common mode current.

Testing of a HF current transformer with a vector network analyzer

I wanted to check the load that a ferrite core with a secondary winding presents to the common mode current carrying conductor (the outside of a coax) when set up as a current transformer. The coax runs thru center, the secondary winding could be one or several turns loaded by a R+jø load.  (+jø load but the windings of the secondary will give some +jx component).

I did the following three S11 measurements with the VNA (M1, M2, M3):

M1) No secondary winding is present but the single turn is running thru the core

M2) A shorted two turn secondary and an open two turn secondary (winding is pulsed in and out off to be able to better detect a difference). The time domain is captured by the slow scan and sampling rate of the VNA.

M3) Only the primary winding is attached to the calibrated S11 measuring plane. (to measure the self inductance of the single turn in itself. This is an air-core measurements.)

Above is a similar test setup. The core is an “unknown” Amidon toroid core. The red conductor is the simulated coax common mode current path (a one turn loop). The green conductor is the secondary winding. The load could made by paralleling several resistors in series with the secondary (the blob on the left side, right picture). Note that the measurements below were done with a one turn loop and a short – no resistor. The measurement device is a VNA from DG8SAQ calibrated by O S L  references in the S11 measuring plane (the SMA in the end of the coax from the TX port).  
The left picture shows measurement 1) The mid picture shows measurement 2) the right picture shows measurement 3)

Edited: What is interesting to see is that the image to the left shows that this core has some resistive loss as can be seen on the blue trace. The Q is quite low. When there is a secondary winding present, the loss is shorted out but the inductancechanges since the inductance of the secondary is reflected into the S11 measurement plane. On the right image it can be seen that there is no resistive loss and a linear inductive reactance caused by the air core inductor (no appreciable drop off or frequency dependent effects, in the measuring frequency range)

What this tells me is that this core setup probably is not too well suited as a high current measuring setup for frequencies above 160m because it will affect the measuring circuit too much. Not in terms of the R element but because of the +jX element. The R is low to the RF current passing thru the core (in the common mode), but the +jX element will present a reactance to the RF current and thereby giving you lower current than should be expected without the core.  My analysis says that another core type should be selected or that a frequency compensation technique should be used. Alternatively that a lower turns ratio should be tried. However looking at the plot to the left, it can be seen that below 30 Mc*s^-1 the +jX component is too high. Perhaps this core has a too large permeability and that a lower permeability core should be used. The primary turn with secondary loading should give a low reactance on the primary. The voltage given over Rt would then be lower but the gain of the detector could be adjusted. (I may be wrong). Please comment if you have comments or suggestions.

Wideband HF RF choke design version 1

DSC_5151 DSC_5143 DSC_5155  9ferritter_2t_plus1f_2t

I here describe a wideband HF RF choke design in pictues. The choke covers most of HF from 160m to 10m and will give a resistive part of the impedance R ( R+jX) that should be high enough to not cause overheating of the ferrites. It will also give a significant reacive load, but the resistive load cannot be cancelled by capacitive effects easily. The 9 + 1 core design above will give the resistance shown on the blue trace in the image to the right above. The other traces (red, green, pink) shows  5, 4, 3 turn designs that was first tried. Note that the fewer # of cores in the choke, the higher up in frequency the resistance peak occurs. Note the single external series core with two turns. That single core takes care of extending the resistance up in frequency above the lower peak (the peak that goes out of the graticule to the upper left). The material is the Fair-Rite 2643167851 material.

VNA 0,5-600MHz measurements of Q, R and X. Fair-Rite 2643167851 high loss material


Today I did some further measurements on the same Fair-Rite material as the other post (Fair-Rite 2643167851). I did the measurements with a calibrated open, short, 50 ohm load S11 measurement. The green trace is the resistive part of the impedance, the red trace is the reactive part of the impedance and the blue trace is the unloaded Q (ratio stored energy in the magnetic field in the core and windings to lost energy in the effective series resitance). As expected, the Q is one when the green and red traces cross each other. This material is not useable as a regular coil for energy storage (high Q, filters and such) over approx 15 Mhz. Below 15 MHz the Q can be quite high, however. This core is probably very good as a dampening material for RFI applications as GM3SEK has indicated. For High power operation when coax common mode current causes problems I think this core may be good. The reason is that the resistive component (real component) of the impedance is dominant. This means that even if a capacitive reactance cancels the inductive reactance, the resistive part of the impedance is still always present. This can be seen from the green trace above. Center of the plot is approx 300Mc/sek. It also rises with QRG up around 450 Mc/sec. At 2 meters and 70 centimeters wavelength it looks like even one turn on a coax (in a low impedance point of the coax!) this choke will do some good. At HF, with more turns it is possible to achieve 1000 ohms over a fairly large range.

VNA measurements on Fair-Rite 2643167851 ferrite material

After reading the nice publication by GM3SEK ( about the Fair Rite ferrite matrial that gives a high resistive component when used for RF choking applications I wanted to do some measurements myself vith my VNA on that material used as a RF choke.


Above is a three choke setup


The blue trace above is the real Z ( R ) plotted from 0-30 MHz. The scale is 480 ohms/div. As you can see the material gives a resistance over a fairly large BW well over 1000 ohms peaking around 5K ohms resistive at the low frequency range. However over approx 18 Mhz, the resistive component is not that great. For QRO operation on 15 meter and 10 meter some further measures should be taken if RF currents are high.


Here is a three stage RF-choke setup. First the three element RF-choke, then a one element RF-choke and a two element RF-choke with the material stacked on top.


Note that the scale here is 620 ohms/div. The measurements show that the three in line setup didnt change the resistive part of the choke a lot. This may indicate that for QRGs over 18-20 MHz, a material with more loss in that frequency range could be found. I am later going to experiment with different winding diameters and cable diameters to see if that will change the resistive elements in the upper parts of the HF spectrum.