linrad support: Analog signal level and A/D board gain.
(Dec 12 1 2001)

Do not waste precious dynamic range by incorrect signal levels.

The audio board

Most A/D boards (every board I am familiar to) has computer controlled switches that can be set for different gain levels. If the A/D board sensitivity is set high, the amplifier that preceeds the A/D converter on the A/D board will add some noise. Different inputs may behave differently, normally the microphone input is much noisier than the line input.

Once you have sucessfully got linrad running, start with any selection of parameters that your system accepts and look at the noise floor on the main spectrum (blue dB scale) while running in "Weak CW mode". Remove all signal wires to the board and leave all inputs open. Set all volume controls to minimum and try to select different inputs (line, mic, CD ....) if the board allows you any choice. Note if there is any difference in noise level.

Select the best (lowest noise) input in case there is a difference. Then bring up the volume until you see the noise floor go up by 1dB. Then reduce the voltage gain by a factor of two (6dB) to make the noise floor degradation caused by the amplifier 0.25dB.

Once you have selected an input and the maximum volume that you can use with it, connect a signal source to the input and run linrad in the F=Hardware test mode. Increase the signal level until you see saturation on the screen. It is best to use a sine wave like source, but any signal will do. Fig. 1. shows the screen you will get.



Fig 1. Linrad in "Hardware test mode".

What you should look for is AD max and AD min and the numbers that follow. With a 24 bit A/D board you should read 8388607 and -8388608 as in fig.1. With a 16 bit board you should read 32767 and -32768.

In the not unlikely case you get lower readings, saturation does not happen in the A/D converter but somewhere else. It could be your signal source but it could also be the amplifier on the sound board.

If it is the signal source, just get another one, if it is not the signal source, increase the volume switch setting on the sound board by 6dB and loose 1dB instead of 0.25dB of the dynamic range. If saturation still happens in the sound board audio amplifier, try the other inputs if their noise floor is equally good or get a better sound board!

Once you have set the sound board for the maximum gain it can have without loss of dynamic range and made sure the D/A can be saturated, measure the signal level required at the A/D board input for the boart to just saturate. For the Delta44 board one needs about 11V p-p or 3.9V RMS.

The ideal analog hardware

The analog hardware has to be able to supply the required voltage with very low distorsion.

With ideal analog hardware, the noise floor does not raise at all when it is connected to the audio board if the antenna and first amplifier is disconnected

With ideal hardware the only contributions to the noise floor one can see on the screen should come from the antenna, the first amplifier and the audio board. Ideally the noise floor should increase by 20dB when the first amplifier and the antenna is connected. Then 99% of the noise comes from antenna/preamp and 1% from the A/D board and system noise figure is degraded by only 0.05dB due to the sound board.

In difficult situations it could be enough to allow the noise floor to raise by only 10dB when antenna/preamp is connected. Then 90% of the noise is from the front end and 10% from the A/D converter. The loss of system noise figure is 0.4dB due to the sound board so a 10 dB increase of dynamic range can be traded for a S/N loss of 0.35dB.

In really difficult cases the gain can be reduced by 20dB from ideal. Then 50% of the noise will come from the sound board and 50% from the antenna/preamp. The associated loss of NF will be 3dB but that is more or less ok in this situation. With signal levels high enough to require 20dB below optimum gain, the sideband noise from the troublesome signal will degrade sensitivity anyway - and probably by more than 3dB.

Using non-ideal analog hardware

If the analog hardware is a "normal" receiver, disable AGC or turn down the RF volume to make sure AGC does not act under normal operating. Set the volume controls for the noise floor to be placed 6dB above the sound board noise floor when antenna and the first amplifier is disconnected.

Random noise adds by power. If the sound board noise level is P, placing the noise floor with preamp and antenna disconnected 6dB higher means that the power level then is 4 * P which means that 25% of the noise comes from the sound board and 75% from the analog hardware. In this case, the analog hardware to which the preamp will be connected has 33% more noise compared to its normal use directly off the loudspeaker due to the contribution from the sound board. The output of the preamplifier will go to a receiver with a noise temperature that is degraded by 33% or 1.25dB. In case the noise floor does not rise by at least 16dB when antenna and preamp is connected you need more preamp gain (or lower losses) or a better second amplifier. In case the noise floor rises by more than 20dB your system is good and you should reduce the gain until the preamp does not add more than 20dB.

Check the dynamic range

Once the noise floors are correctly placed, check the performance of your system by listening to a strong signal from the antenna input. Tune the analog hardware to place the signal at the low side of the spectrum. It is essential that the frequency is low enough so three times the audio frequency is within the audio passband. Now, increase the signal gradually and watch the second harmonic. Very probably your analog hardware is not even nearly as good as the sound board so you may see the second and/or third harmonic grow above a reasonable level (-40dB below the fundamental) long before the system is saturated. Saturation may well be in the analog hardware and not in the soundboard which you should check with the "Hardware test mode" as described above.

For more details and some examples, check: Checking the dynamic range

In case you are using a normal SSB radio, the bandwidth is limited to about 2.5kHz so a very limited dynamic range is no problem. Strong signals have to be kept outside the 2.5kHz passband and the analog filters of the radio is responsible for rejecting them. In case the strongest signal the analog hardware can handle is far below sound board saturation, the sound board is not used to its full capacity but that is no problem at all. To process weak signals in noise 8 bit A/D converters would be perfectly ok! More bits are needed in order to process strong signals correctly.

If you are using a wideband analog system, regardless whether you have wide filters (real signals) or a complex baseband signal I and Q (direct conversion radio) the problem is the same. It is difficult to to make analog hardware that takes full advantage of the dynamic range of a good sound board. On the other hand It is very easy to build the analog harware for a wideband radio with linrad if dynamic range requirements are modest.

Assuming that -40dB is the maximum acceptable level of the strongest spur, finding what maximum level a signal may have before this limit is exceeded will tell if the hardware is acceptable. If the maximum level as defined this way is many dB below saturation of the sound board, better analog hardware is required.

With really good hardware it is possible to get extremely good performance, the dynamic range of a modified Delta44 can be preserved, which allows quite good performance, comparable to what conventional receivers/transceivers can give.