Return to KLUBNL.PL main page

rsgb_lf_group
[Top] [All Lists]

LF: Re: Notch for DCF39/Spectrograms/Luxembourg effect

To: [email protected]
Subject: LF: Re: Notch for DCF39/Spectrograms/Luxembourg effect
From: "James Moritz" <[email protected]>
Date: Sat, 18 Jan 2003 12:42:22 +0000
In-reply-to: <[email protected]>
Reply-to: [email protected]
Sender: <[email protected]>
At 12:09 17/01/2003 -0500, you wrote:

<< This is the wrong way to achieve high impedance (and good IMD),
 it is no need to keep the "source current low" {etc} >>

I do not see anything in the Web page that suggests keeping the source
current low.  It only says "reasonable amount," which I took to mean "be sure
to use enough" to keep IMD low.

Dear LF Group,

The output impedance of a follower is roughly the inverse of it's transconductance, ie. 1/gm. A J310 JFET has a gm of about 10 - 20 mS, so the output impedance is roughly 50 - 100 ohms. To get low distortion, and gain close to unity with such a follower requires that the load it is driving should be much greater than the output impedance of the follower. So a J310 source follower would be fine if driving several hundred ohms or greater, but not for a 50 or 75 ohms system.

To get around this, one could use a special FET like the CP640 device used in the AMRAD active antenna circuit, with much higher gm. A small power MOSFET would give a low output impedance, but rather high input capacitance. Bipolar transistors have much higher gm (roughly Ic/25mV), but would have relatively low input impedance, in the kilohm range, and be relatively noisy with a high source impedance. Various high speed op-amps and IC buffers can be used, but they tend to be quite noisy.

A cheap and easy alternative is to use a compound follower, with JFET and bipolar followers cascaded. I tried the one shown in the attachment - Zin is basically 500k in parallel with the FET capacitance, the gain with 50ohm load is about 0.95. The noise at 136kHz with the input shorted was about 0.05uV in 400Hz BW. About 1V RMS can be applied to the input before there is much distortion whilst driving 50ohms, so it ought to be OK with most antennas, and is probably better than most receivers. The FET was a J310, but other devices with Idss > about 8mA should be OK. A lower gm device with lower Idss could be used if the 100R resistor were increased in value, but would be noisier. The output BJT used was a 2N3053, but can be a wide range of small power transistors with Hfe > 50 when biased to 30mA or so and fT in the 10s of MHz - eg. BD131, BFY50, ZTX650 etc. Zout is a few ohms - If driving a capacitive load, a resistor in series with the output of 10 - 20 ohms will prevent oscillation. Extra decoupling will be required if the 12V supply is noisy.

Using spectrogram software to monitor the band for signals works very well - it allows you to see CW signals that are inaudible, which you can then try to "zoom in" using a narrow filter. Although you can't copy the morse from the screen display, you can detect the presence of signals that are too weak to hear even in the narrowest filter, whilst monitoring the whole band segment at once. Some of the spectrogram programs have DSP audio filters built in, too.

The broadcast station noises M0WYE is hearing may well be "Luxembourg effect", where cross-modulation of the DCF39 carrier by the BC signals occurs in the ionosphere, and is often quite strong at my QTH. Usually I can here the audio of at least two stations at once.

Hope this is helpful,

Cheers, Jim Moritz
73 de M0BMU

JPEG image

<Prev in Thread] Current Thread [Next in Thread>