At 14.15 25/09/2006, you wrote:
It is obvious that most involved with this group are aiming at
constructing their softrock equipment from the kits available. There
are, however, a group of us, including me, who would like to take
advantage of the great ideas presented, but still carry on with
own design including the component selection, board layout etc. Not
too familiar with the current developments in the op amp technology
and finding the softrock schematics with strange types (LM 741 I
hihi),I would like to know their criteria for selection. For the rx
part, low noise is one factor for sure, also 5 V supply voltage is
compatible with the rest of the board. What about GBW-product,
rate etc. , are these any critical since the opamps are in the audio
Before buying the parts proposed I want to search my (not so small)
junk box but as far as the opamps are concerned I would need some
advice on what to look for .
73 de Heikki (OH2LZI)
Hello Heikki and group,
I am one of those considering the SoftRock a nice test bench for new
ideas and for the optimization of SDR frontends.
Possibly my considerations will be well known to professional designers;
but in general we (OMs interested in the technical side of things) have a
lot to learn.
I spent the past weekend trying to characterize an M-Audio Audiophile 192
alone and connected to my SoftRock V6. Here are some results.
The A-192 declares a flat frequency response up to 80kHz while sampling
at 192kHz, 24 bits.
The measured response is flat up to over 90dB. The antialiasing filter
starts cutting here, so not to see aliasing signals in band I suggest
that it is useable up to 86kHz.
Unfortunately, the noise floor is not flat. From 0 to 3kHz it decreases
approx like 1/f, then it's flat up to 25kHz. From there up it increases
by approx 5dB per 10kHz, rising by 30dB at 85kHz. I have no idea of the
reason for this noise increment; the two channels are different, the R
being approx 3dB less noisy than L. Anyone may explain this noise
Returning to dynamic range, the saturation level is exactly 4Vrms as
declared. I used an unbalanced input, connecting the "-" input
Considering the good zone of the noise floor (from 2 up to 25kHz) the
dynamic range is in the order of 120dB in 2.4kHz. That's better than the
declared 113dB in "A" weighting filter, that has 1.74kHz
equivalent noise bandwidth (if I remember well).
The input sensitivity corresponds to a -94dBm (50Ohm) noise floor. So we
have to set our front-end output dynamic range from -94 to +26dBm.
By instance, if we want a -120dBm noise floor at the antenna, the
frontend should have 26dB voltage gain. I am deliberately ignoring the
noise figure issue here.
Now, my modified SoftRock has approx 20dB of voltage gain, so it is
quite "centered" in the range. The measured noise floor is
-113dBm in 2.4kHz, and is dominated by the front-end noise; by powering
down the SoftRock the noise decreases by 2dB. The saturation happens at
-5dBm, so the blocking dynamic range is 108dB.
I leave to you all considerations about the post-mixer opamps. In my
opinion they should be powered at least from +/- 6V or +/-9V rails
to use all the available headroom on 24 bit audio boards, and be able of
26-30dB gain in a 100kHz band and 120dB of dynamic range. Not trivial...
:-) how deep is your junk box?
A picture of the noise floor of the Audiophile 192 (at 96khz sampling
rate here) is available from
73 - Marco IK1ODO