[hpsdr] QSD gain and noise.

[Ahti Aintila]

On 24/09/2007, Phil Harman <phil at pharman.org> wrote:

> Are there any figures for the 180 degree QSD over the range 160m to 6m
> i.e. are we sure that changing to 180 degree switching really fixes the
> NF problem? If not then the H mode seems a better option.
>
Phil,
Based on quick experiments that remained undocumented more than a year
back, I only can say that the frequency response falls down sooner
with the 90 degree circuit than with the 180 degree version. Yesterday
I made some simulation assuming 4 ns shorter switch conduction times
than the theoretical 90 or 180 degree time. The simulation was made
using the IK1ODO circuit with the diplexer components matched to 200
ohm impedance level. The amplifier voltage gain was adjusted close to
0 dB referring to the 50 ohm input of the 1:2 input transformer.

The gain loss between 10 MHz and 40 MHz signals were 0.3 dB with 180
degree and 3.8 dB with 90 degree. To the noise calculations of the
TINA-TI simulator I don't trust in this switched signal case, but 90
degree circuits give lower noise if the sampling capacitors see high
impedance. In my case the matched low impedance partially degrades the
benefit of the 90 degree sampling. If I remember correct my early
measurements, at lower frequencies I got about 2 dB lower noise with
90 degree sampling using IK1ODO circuit at 200 ohm level. I cannot
emphasize too much the importance of the elimination of the switching
spikes as done by Marco, IK1ODO.

More experiments will be needed with better tools than I have. How it
looks to me, with the practical circuits and components of today, the
180 degree and H mode mixers may be better selection on 20 MHz and
beyond and on lower frequencies the atmospheric noise sets the limit
of the MDS anyhow.

73, Ahti OH2RZ

 1190787773.0