A/D input levels

14mar08

Note: apr08. The a/d converters were modified so that their pkToPk range changed from 1.5 to .8volts. This data was taken with the old 1.5 pktopk setttings.

The the 12 bit A/D levels for the pdev spectrometer were measured by injecting a sine wave into digitizer I of the low band of box 110. The signal path was:

synth -> bpFilter -> scope -> 100MHz LpF -> digitizer input.
An Agilent 3 Ghz synth was set to 31 MHz generated the signal
A 10 MHz bw 30 MHz bp filter was used to remove any harmonics of the 31 MHz signal
the scope was set to 1MHz impedance (the tektronix 500 MHz input bandwidth scope). The peak to peak input voltage was read from here.
The 100 MHz low pass filter removes the lo from the mixer chassis. It is normally installed when we run.
pnet --sigstat was used to measure the rms input voltage in A/D counts.
For each setting data was also taken to verify that there were no A/D overflows. The unused digitizer was set to 0 so the spectra will come out symetric.

Sine wave measurement results
Scope PkToPk Millivolts	Scope pwr dbm	SynthPwr dbm	Signal Rms A/D cnts	Signal Pk A/D cnts	pshift	Notes
365	-4.8	-4	293	414	0x1ff5
560	-1.1	0	468	661	0x1ff5
975	3.8	5	835	1180	0x1ff5
1400	6.9	8	1175	1661	0x1ffe	0x1ff5,0x1ffa overflowed in the pfb
1560	7.8	9	1319	1865	0x1ffe
1650	8.3	9.5	1395	1972	0x1ffe
1690	8.5	9.7	1428	2019	0x1ffe

The scope Pwr dbm column was computed from the scope pk to pk voltage (assuming a 50 ohm termination in the line). Comparing this with the synth power gives a 1 db insertion loss for the 30MHz bp filter.

The plots show the A/D levels vs input signal (.ps) (.pdf):

Page 1 top: A/D levels vs rms input voltage.

The black line is the Peak counts, the blue line is the rms Counts.
The dash red line is a fit to the peak counts vs input rms voltage.
605 millivolts rms input gives a full scale (2047 counts output).
There are 3.41 A/D counts per millivolt rms input.

Pag 1 bottom: This plots the A/D counts (black peak, blue rms) vs the input power i n dbm (computed from the scope peak to peak voltage).
Page 2: The spectra for the sine waves for a/d Peak input=2019 and a/d peak input=414. The vertical scale is in dbs

Top full scale input sine wave . This used a pshift =0x1ffe
2nd : input sine wave with peak count=414. This used a pshift of 0x1ff5.
3rd. blowup about 31 MHz of the a/d=2019 cnt spectra,.
4th: blowup about 31 MHz of the a/d=414 cnt spectra.
The ratio of the sine waves: 5.92 matches the ratio of the pk a/d counts as long as you divide by 4. The extra divide comes from adding two more shifts in the butterfly stages (and this is a sine wave and not noise so it goes as 2^number of shifts).

Conclusions:

605 millivolt rms input (8.6 dbm) gives a full scale output:

2047 count digitizer output.
1448 rms digitizer counts.

There are 3.41 counts per millivolt rms input.
When inputting a full scale sine wave, any spurs are at least 70 db down.
The signal remained linear when reducing the full scale signal by 14db. This included a shift difference of 2 in pshift (the butterfly shifts).
The mixer chassis is currently not able to generate an 8.5 dbm output even with 0 attenuation set without compressing.

processing: x101/080313/injSine.pro

<- page up
home_~phil