[hpsdr] SDR comparison

[Phil Harman]

Hi Pete,

There  is a modification that can be made to improve the performance of  the 
SDR1000 +  Janus + Ozy  combination.

Let me give a little background so you can see why this might be a good 
idea.

The stock SDR1000 was designed to be used with many different sound cards. 
In the early days this was the Santa Cruz or most likely what the user 
already had to hand or was fitted to their PC  mother board.

The noise figure of the various sound cards varied considerably, let's take 
a value of 30dB as being typical.  As a simple rule-of-thumb if we want to 
remove the effect of the sound card from the overall system noise figure 
then we need 10dB more gain before it  than the noise figure of the sound 
card itself.  So with a 30dB NF sound card we need  a gain of  30 + 10 = 
40dB before the card.

Any more gain than this does not reduce the overall system NF ( as measured 
at the antenna socket) it justs presents a larger signal level to the sound 
card which means it will overload sooner.

Early 3 board SDR1000s had the ability to switch the gain of the post QSD 
amplifier to a high setting (can't recall what this was) but it was soon 
found that this amount of gain was excessive.

One  of the reasons for the high noise figure of the typical sound card is 
that it has to accept audio industry standard signal levels - 0dBu or 
higher. To do this the designer will often add an attenuator (or switchable 
attenauator in the case of the Delta 44) at the input of the sound card. 
Let's say that the ADC by its self has a 10dB noise figure.  If we add a 
20dB attenuator in front of it to enable it to handle the signal levels we 
require then the overall card  noise figure will be  10 + 20 = 30dB.

In the case of Janus we are not building a consumer specification sound 
card, we want the lowest noise figure we can get together with a high 
dynamic range.

The noise figure of Janus appears to be in the order of 8dB. This will allow 
a QSD to be connected directly to  its (high impedance) balanced inputs 
without any intermediate amplifier - I hope someone will try this really 
soon - we added a small header on the Janus board so a QSD board could be 
plugged in without the need for additional connecting leads.

The low noise figure  means that the Janus ADC will overload if we put 
consumer level signals from an SDR1000 into it.  There are two ways to 
overcome this problem

1. The preferred method - reduce the gain of the post QSD amplifier (INA163) 
on the SDR1000 TRX board
2. Increase the large signal handling, but degrade the noise figure, of the 
Janus board by placing an attenutor in front of it.

I am unable to test option 1 since I have made so many circuit  changes to 
my SDR1000 it would not be feasable to revert to the original design. I know 
there are others in the group that have the necessary skills and test 
equipment to determine what the gain should be reduced to and can let us 
know the new component values.

With regards to option 2, we fitted a simple 15dB attenuator to the input of 
Janus that can be inserted/removed using a jumper. This is to enable those 
that do not want to modify their SDR1000 to still make use of the boards. As 
standard the Janus boards will be shipped with the attenuator in circuit.

Even with the attenuator in circuit the MDS of a standard SDR1000 is 
improved by about 4dB (from memory  - perhaps Bob N4HY can let us have the 
figure he measured) when using Janus + Ozy in comparison to a D44.

There is one down side in the use of the attenuator in that the frequency 
response of the system droops at the high end due to the R/C combination of 
the  attenuator and the C of the input RF filters.  This only effects the 
bandscope at the edges when using 192kHz sampling.  There are ways round 
this but I'll post them later if folks think it an issue.

A long reply to a simple question but I hope that those experimenters in the 
group can use the above to help us optimise the SDR1000 + Janus + Ozy 
combination.

We are aware that boards are about to be shipped and folks will want a 
version of PowerSDR etc that can be used with the boards. The necessary code 
and how best to access it is being worked on at the moment.

73's Phil...VK6APH














 1177739442.0