Dynamic range of modified Delta44 A/D converter
(Nov 20 2001)
Comment added Sept 3 2002
This is an old page showing measurements with a tone generator on a Delta 44 that still has its input amplifier/filter in place. It is left here because it shows how such measurements can be done.

Two tone dynamic range

The two tone dynamic range is the ratio between the strongest signal that can be tolerated and the noise floor. The two tone dynamic range is usually limited by noise sidebands in transmitters and receivers. Look here for Some data on 1970 style transceivers

The two tone dynamic range is the same as the distance to the noise floor for the strongest signal that can be tolerated. The noise floor depends on the bandwidth, in the old days the noise floor was often measured in SSB bandwidth (2.5kHz or so) while modern measurements usually give noise floors in dB/Hz.

Typical two tone dynamic range at 25kHz frequency separation is 90 to 100dB for a good, 25 years old 144MHz transverter. This corresponds to 124 to 134dB/Hz. Modern equipment seems no better according to noise sideband measurements in QST.

Test equipment

The level an unwanted signal can have before causing a 3dB S/N loss on the desired signal is limited by A/D converter saturation when the receiver is a Delta44 A/D converter. There are some particular frequencies where the dynamic range is much lower compared to the rest of the spectrum and that is the overtone frequencies.

The linearity of the A/D converter is better expressed as the level of the overtones than as dynamic range numbers.

To be able to measure the level of the overtones created by the A/D converter it is necessary to have a signal with extremely low overtones. I have used a conventional audio oscillator, an OLTRONIX RCO-6 which gives a spectrum as shown in fig. 1.

Fig. 1. The audio generator without any filter.
2nd harmonic -53dB
3rd harmonic -66dB
4th harmonic -82dB

To test the band pass filter which has its center frequency near 10.5 kHz a square wave was used. The spectrum of the square wave without any filter is shown in fig. 2 and the same signal after passing the band pass filter is shown in fig. 3.

By combining the overtone content data of the audio generator from fig. 1 with the filter characteristics obtained from fig. 2 and fig. 3, the following overtone contents of the filtered test signal are obtained.

Fundamental 0dB
2nd harmonic -123dB
3rd harmonic -156dB
4th harmonic -172dB

The overtone levels obtained this way may be far too low, it is possible that overtones are generated in passive components for example.

Fig. 2. Spectrum of the square wave without filter.
level at 2nd harmonic -8dB
level at 3rd harmonic -12dB
level at 4th harmonic -15dB The levels are with respect to a fundamental at 10.5kHz.

The test signal obtained by passing the signal from the Oltronix RC generator through the filter is perhaps not quite as good as indicated by the measurements above. It is however good enough for the purpose. The harmonics produced by the Delta44 are much stronger than the harmonics contained in the test signal. This is verified by use of an iron free notch filter that attenuates the fundamental by 6dB while leaving the overtones unaffected. Applying the notch has the same effect on the overtones seen by the Delta44 as reducing the signal level by 6dB. This is definite proof that the overtones are generated within the Delta44.

Fig. 3. Spectrum of square wave after passing band pass filter. Taking the data of fig.2 into account, the following results are obtained for filter attenuation
fundamental 0dB
2nd harmonic -70dB
3rd harmonic -90dB
4th harmonic -90dB (?)

Overtone levels

The spectrum obtained by linrad for one channel of the Delta44 is shown in fig. 4 which shows noise floor and overtone content 1dB from saturation.

Fig. 4. Overtones and noise floor with a test signal 1dB from saturation.

The overtone levels depend on the fundamental as follows:
Fundamental       2nd harmonic     3rd harmonic      4th harmonic
dB below           dB below         dB below          dB below
saturation        fundamental      fundamental       fundamental

    1                118                86               107
    3                109                92               119
    5                110                96               116
   10                109               108                -
   15                115               110                -
The overtones have a slightly oscillatory behaviour, the table just gives an idea about typical values. Unlike earlier tests on an unmodified board, this test shows very low values for the second harmonic. It is unclear if the old measurements did not have a clean enough test signal or if the larger capacitors now present on the board affect the linearity.

The harmonic distorsion is low enough to make it difficult to make a frequency mixer that will match the performance of the Delta44.

The noise floor

The spectrum of fig. 4 indicates that the noise floor is 125dB below saturation. The spectrum is calculated with a bandwidth of 25Hz but it is presented on the screen with much lower resolution. Fig. 5 shows the same spectrum with expanded frequency scale.

Fig. 5. Zoomed in version of the same speectrum as in fig.4. The bandwidth is 25Hz and here the noise floor is 132 dB below saturation. The noise sidebands are most probably due to the audio generator.

The modified Delta44 has its noise floor at -146dB/Hz with respect to saturation. In order to obtain a good noise figure it will be necessary to add 20dB more noise from earlier receiver stages. The total receive system will then have a two tone dynamic range somewhere around -126dB/Hz or 92dB in SSB bandwidth. This will be adequate in most situations although it will be a good idea to provide an attenuator that can lower the RF signal should it become necessary. Contrary to a normal radio that looses sensitivity gradually when an undesired signal becomes too strong the digital receiver will have a catastrophic breakdown at the point where the A/D saturates.

When the Delta44 is used as a baseband receiver, receiving I and Q in two analog channels, the signal will increase by 6dB while the noise will increase by 3dB only. In case the antenna is a crossed yagi system in 45 degrees the strong signals will be split between the two channels adding another 3dB to the system dynamic range.

For details of the Delta 44 modifications check this link.