EasyDABv2 measurements results

Log in to post comments

3 months are gone after release of EasyDABv2 hardware. It's time to publish some measurements and enchancements for this hardware. Here is short way of enchancing output signal.

The list of hardware, which has been used for measurements is:

R&S ETL Digital TV analyzer.
R&S FSU Spectrum analyzer.
QS1R - Direct Sampling Receiver with 16bit ADC.
RTL-SDR stick with patched gr-dab modules.

Here is list of fixes, that has been applied to the hardware/firmware update, released on the board's page:

Phase noise decreased by tuning PLL loop filter resistor and capacitors values. Previously PLL bandwidth was 50KHz, for now it has been increased to 1.2 MHz (it can be increased up to 2MHz, but crystal spures amplitude will be higher in that case).Here is spectrum images before and after values changed:
Gain control module has been removed, because of constellation map was not stable (amplitude jumping too highly between minimal and maximal values), but after this change, it's became much more "dot like", instead of "line like":

Also, after this change, ETL shows much better Modulation error rate:

Unfortunaly, this change decreases output power from 180mW down to 40mW. But this is ok, since distance, on which receiver can tune to the station decreases only about 25%. So signal is less stronger but pretty much stable from amplitude deviation view.
Sample rate of the I/Q data samples that is passed to AD9957 QDUC increased from 2048 kSamp/sec up to 8192 kSamp/sec. Also additional interpolation FIR filter added into firmware, it's target role - is to increase samplerate and filter unneeded images, that presented at +/- 4096 KHz:

Now 20 MHz spectrum span looks prettier than before:

Output spectrum - still have some spures and dead frequencies, which have to be filtered before transmitting sighal to the air or passed to the amplifier.

Here is some constants for calculations:

Crystal freq:
Fq = 24.576MHz

DAC frequency:

Fdac = 589.824 MHz

Samplerate:

Fs = 8.192 MHz

Output signal:

Fo = 194.064 MHz (it's relative level - is: -85dB )

The known dead frequencies or so called "spures" list (values are relative to main signal, band pass filter replaced to low pass one):

5*Fq = 122.880 MHz (5th harmonic of crystal): -113 dB

7*Fq = 172.032 MHz (7th harmonic of crystal): -120 dB

9*Fq = 221.184 MHz (9th harmonic of crystal): -123 dB

Fdac/4 = 147.456 MHz (DAC freq. / 4): -96 dB (the most powerful unwanted signal)

For specified "Fo" frequency the images would be:

Fo - 4*Fq = 194.064 - 24.576*4 = 95.760 MHz: -138 dB (mixed frequency image, can be filtered by band-pass)

Fo - 3*Fq = 194.064 - 24.576*3 = 120.336 MHz: -120 dB (mixed frequency image, can be filtered by band-pass)

Fo - 8*Fs = 194.064 - 8.192*8 = 128.528 MHz: -122 dB (mixed frequency image, can be filtered by band-pass)

Fo - 2*Fq = 194.064 - 24.576*2 = 144.912 MHz: -135 dB (mixed frequency image, can be filtered by band-pass)

Fo + 2*Fq = 194.064 + 24.576*2 = 243.216 MHz: -134 dB (mixed frequency image, filtered by band-pass or low-pass filter)

Fo + 3*Fq = 194.064 + 24.576*3 = 267.792 MHz: -120 dB (mixed frequency image, filtered by band-pass or low-pass filter)

Fo + 4*Fq = 194.064 + 24.576*4 = 292.368 MHz: -137 dB (mixed frequency image, filtered by band-pass or low-pass filter)

Fo*2 = 194.064 * 2 = 388.128 MHz: -100 dB (2nd harmonic of the main signal)

Fdac - Fo = 589.824 - 194.064 = 395.760 MHz: -106 dB (1st nyquist frequency image, can be filtered by band-pass or low-pass filter)

More spures that is not filtered, because it's close to the main signal:

186.924 MHz: -124 dB (mixed frequency image) it's nature is not known exatcly yet.

More measurements photos:

no_gain_2048 - R&S measurements result of 2048 kSamp/sec firmware, where PLL loop filter has been enchanced and gain module removed from firmware.
RS_FSU - This directory contains R&S FSU measurement results, that show enchancement of changing samplerate 2048->8192.