LowFER Receiver Using a _Software_ IF__B_.pdf

A LowFER Receiver Using a "Software" IF

http://www.qsl.net/k0lr/SW-RX/sw-rx.htm

A LowFER Receiver Using a "Software" IF

By Lyle Koehler, KØLR

Introduction

Downconverter with FET input stage

12 volt version with varactor tuning

Dual-Watch operation

Crystal calculator spreadsheet

Notes on using SDRadio

More fun with Spectrum Lab

Multipath propagation observations

Introduction -- This article describes a "software" receiver that uses a simple, low cost hardware downconverter in conjunction with sound-card software

such as DL4YHF's "Spectrum Lab" or I2PHD's "SDRadio", which act as a tunable DSP-IF system. Spectrum Lab is a powerful piece of software that can be

used for receiving a variety of signal modes, including very slow CW (QRSS) that must be read from a "waterfall" type of spectrogram display rather than

being copied by ear. SDRadio does not provide a built-in capability for QRSS, but is very convenient and easy to use for single sideband, CW, AM and ECSS

(Exalted Carrier Selectable Sideband) reception. You can download Spectrum Lab from DL4YHF's Web site , and a beta version of SDRadio is available at

I2PHD's software defined radio site. Because of its ease of use, SDRadio is a good choice for the initial checkout of the downconverter and your sound card

system. Most of the references to sound card software in this article are to Spectrum Lab, but notes on the operation of SDRadio are included at the end of

the article.

Note: Several references are made to other articles I have written, which can be found at my home page .

Most LowFER operation today utilizes weak-signal modes that require a computer to decode, or at least display, the signals. A few years ago, it required a

trained ear and a receiver with very narrow filters to dig weak LowFER signals out of the noise. Now all of that work is done by the computer's sound card

and digital signal processing (DSP) software. And that expensive box of receiving hardware sitting on your bench does little more than act as a frequency

converter to translate signals in the LF (30 to 300 kHz) range down to a frequency that can be handled by the sound card in a computer. Typically the

maximum sampling rate for a computer sound card is 48 kHz (although most applications will only show available rates up to 44.1 kHz), and the Nyquist limit

says that the sound card can handle input signals up to half of that frequency, or 24 kHz. The popular "Argo" spectrogram software uses a lower sampling

rate, and will accept incoming signals up to about 2.7 kHz.

1 z 27

2011-01-23 15:07

A LowFER Receiver Using a "Software" IF

http://www.qsl.net/k0lr/SW-RX/sw-rx.htm

Since the receiver is actually only serving as a down-converter, theoretically it can be replaced by a mixer and a local oscillator. In ham radio circles, that's

known as a "direct conversion" receiver, and the direct conversion circuit is very popular for simple homebrew receivers. Some of them perform amazingly

well for CW and SSB reception on the HF bands. There are a few disadvantages to direct conversion. One problem is that the receiver responds to signals on

both sides of the local oscillator frequency, so you have to contend with twice the noise and interference that you'd get with a "single signal" receiver. There

are ways to get rid of the image by using two mixers and phasing techniques, but now we're getting out of the "simple" receiver category. Another

disadvantage is that the incoming RF signals may be in the microvolt level, and it requires volts or millivolts to drive headphones or a speaker. If you try to do

all of the necessary amplification at audio frequencies, it takes really careful design to avoid hum pickup and oscillations. The so-called superheterodyne

circuit avoids these problems by converting the signal to an intermediate frequency (IF), where much of the filtering and amplification takes place. Modern

high-performance receivers typically use two or more conversion processes and intermediate frequencies. The first IF is chosen to be high enough so that the

image frequency can be rejected easily by the front-end filters in the receiver. The final IF is low enough to make it easy to achieve the desired narrow

bandwidth, or selectivity. Within the past few years, a number of receivers and ham transceivers have started using "IF DSP", where the final IF frequency is

in the range of perhaps 30 to 40 kHz, and the filtering and demodulation functions are performed by digital signal processing software.

The LowFER receiver described in this article is a highly simplified version of an IF DSP receiver. At LF, it doesn't take a lot of front-end selectivity or a

very high IF frequency to achieve usable (not necessarily good) image rejection. Suppose we choose an IF of 10 kHz, That puts the image frequency at 20

kHz away from the desired signal. If the response of the receiver's front end filter is down 10 or 20 dB and if there are no strong signals at the image

frequency, the receiver might perform almost as well as a much more expensive commercial unit. In order to receive today's narrowband LowFER modes, we

still need extremely good frequency stability and accuracy, and the ability to tune the receiver to the desired signal frequency. But if the tuning range isn't too

wide, it can be done by tuning the IF rather than the local oscillator. For example, we can cover the 180 to 190 kHz range, where most US LowFERs operate,

with a fixed local oscillator frequency and an IF that can be tuned over a 10 kHz range.

To put together a respectable LowFER receiver, I wanted a DSP IF that would tune from about 10 to 20 kHz, with a fairly narrow audio filter and a

"comfortable" BFO pitch. Of course, the computer doesn't care what the pitch of the audio output signal is, but I like to listen to the few remaining CW

beacons. I also want to hear what the receiver is picking up, even though the actual LowFER signal I'm trying to receive may be too slow to copy by ear, or

buried 20 dB below the noise. Simultaneously, so that I don't have to pipe the audio output to another computer, I want to be able to display a spectrogram of

the output signal, do automatic screen captures, save WAV files of the output, etc. That's a pretty big wish list. However, the "Spectrum Lab" software by

DL4YHF can do all of those things. In fact it will do a lot more, once you learn how to use all of the features.

Designing the actual receiver hardware was easy. All I needed was a down-converter, so I borrowed most of the circuit from the LF up-converter described

elsewhere on my web page in the "solderless homebrew projects" article. To cover the 180 to 190 kHz range, with a tunable IF of 10 to 20 kHz, the local

oscillator frequency needs to be either 170 or 200 kHz. 200 kHz is a bad choice, because that would place the image frequency in the range where there are

lots of strong non-directional beacons (NDBs). Crystals for the 170 kHz range are a bit hard to come by, but an HF crystal and a divider IC offer an easy and

inexpensive solution. One of the common microprocessor crystal frequencies is 11.0592 MHz, which comes out to 172.8 kHz when divided by 64. I decided

that this was close enough. The circuit of the down-converter, shown below, consists of an NE602 mixer (NE612 or SE612 parts are equivalent) and a

74HC4060 oscillator/divider operating in the divide-by-64 mode. Dan's Small Parts may have most of the parts required for this circuit. I didn't see the

11.0592 MHz crystal in his catalog, but he may have a 5.5296 crystal which only requires changing the output pin of the 74HC4060 from pin 4 to pin 5. To

receive signals at 137 kHz, it is necessary to select a crystal so that the local oscillator is in the vicinity of 150 kHz. A "low side" local oscillator would not be

2 z 27

2011-01-23 15:07

A LowFER Receiver Using a "Software" IF

http://www.qsl.net/k0lr/SW-RX/sw-rx.htm

a good idea because the image would be down in the really serious Loran splatter. Examples of acceptable crystals for 137 kHz reception are 10 MHz

(divided by 64) or 4.9152 MHz (divided by 32). The exact frequency of the local oscillator is not critical, as long as it is stable, because the tuning is

accomplished in the DSP IF. However, you do need a way to determine the crystal's frequency, and to calibrate the sound card's sampling rate if you are

going to try for very slow modes like QRSS30. When looking for needles in a haystack, it really helps to know which haystack! Possible methods for

calibration are discussed later in this article.

The receiver, like its up-converter predecessor, was built on a solderless protoboard. In the future I may revise the circuit to use varactor tuning with an

MVAM109 or NTE618 diode in place of the air variable capacitor, and with a FET stage in front of the mixer to improve the gain and selectivity. But this is

the circuit that was used for the "WE" and "WEB" reception examples described below. The receiver drew 8 milliamps when operated from my 6-volt

regulated supply. It could also be run on 4 AA cells, or from a higher voltage supply with a dropping resistor to keep the voltage within the limits for the

integrated circuits. For example; 9 volts with a 390 ohm resistor or 13.8 volts with a 1k resistor.

3 z 27

2011-01-23 15:07

A LowFER Receiver Using a "Software" IF

http://www.qsl.net/k0lr/SW-RX/sw-rx.htm

Assembled LF Downconverter

During the tests, the Spectrum Lab software ran on a 900 MHz Pentium III computer with 512 MB of RAM and a Creative Labs SoundBlaster Live sound

card. Even though the DL4YHF software is well documented, I would have a hard time starting from scratch and getting it to do everything I wanted for this

application. Fortunately the software comes with some pre-programmed examples that require only minor tweaking to get the desired result. I started with the

"SAQ receiver" settings, which can be used to receive the 17.2 kHz transmissions from the historic longwave transmitter at Grimeton, Sweden when it is fired

up once or twice a year. The default BFO pitch is 650 Hz, although I prefer something down around 450 Hz for receiving weak CW signals. With the local

oscillator in the down-converter at 172.8 kHz, and an incoming signal at 185.3 kHz, the IF is at 12.5 kHz. Going into the Spectrum Lab's "components"

window, I set the IF DSP local oscillator to 11850 Hz, which produced the required 650 Hz offset, and connected the receiver input to my pre-amplified loop.

After peaking the tuning capacitor (there is a delay between the signal input and the audio output, so you have to tune very slowly), there was a good strong

4 z 27

2011-01-23 15:07

A LowFER Receiver Using a "Software" IF

http://www.qsl.net/k0lr/SW-RX/sw-rx.htm

signal from LowFER "WE", who is about 70 miles from me. Later I changed the audio filter center frequency to 450 Hz and set the DSP local oscillator to

12050 Hz, which is where things were set to record this WAV file of WE's CW identifier. The original file as recorded by Spectrum Lab sounded cleaner, but

I cropped and compressed the audio to reduce the file size.

WE-SW_RX.wav

It took quite a bit of tweaking to set up the spectrogram feature of Spectrum Lab to receive the 1-watt, 189.950 kHz QRSS30 signal from Bill Bower's

"WEB" beacon in Texas. When Bill was in Oklahoma, I could usually hear his OK beacon every night, but now it's a little more of a challenge. I will try to

provide more details if I get to the point where I understand better what I'm doing, but the basic FFT settings were: Decimate by 16; FFT length 65536; 2000

millisecond scroll interval. The IF local oscillator was set for 16700 Hz (I'll leave the arithmetic up to the student), and I used a signal generator to make sure

the converter's input circuit was peaked to 189.950 kHz. I couldn't find a setting that made WEB visible prior to going to bed, but usually the band opens up a

couple of hours before sunrise. So I set up for screen captures every 20 minutes and hoped for the best. In the morning, the screen captures looked pretty bad

-- too much sensitivity, so they were full of psychedelic "snow". However, one of the many impressive features of Spectrum Lab is the scrolling buffer, which

lets you go back through the last couple of hours of data. That, plus the fact that when you adjust contrast, brightness, etc., it affects the past screen as well as

the future! By scrolling back through the data and adjusting the contrast and brightness settings, I was able to produce some readable and very colorful

renditions of the WEB identifier, but finally settled on the conservative color scheme shown below.

5 z 27

2011-01-23 15:07

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: