Building a 0 - 3 GHz Spectrum Analyser


16 Bit ADC Notes:

Modification of ADC switches
Scotty has recently designated two bits in unused latch 4 to control the video filters. That provides values of 0 to 3 for a wide, mid, and narrow filter, plus one extra that I call XNarrow. We will implement that in the software once USB is up and running. You would then just need a one-of-four decoder and perhaps relay drivers like the MDC3105. It's probably a lot simpler to skip the relays and use a solid-state 1-of-4 multiplexer to select the capacitors. Then you can directly use the latch bits to control the filter selection. The operating frequency is so low and the isolation requirements so minimal that there are many possible solid-state solutions.
Sam W.

That switch sounds good, though its footprint looks like a pain to deal with. A lot of resistance on the ground end of the capacitor will partially defeat the filtering. But the resistance on the + side is very high--10 k for phase volts and 2.7 k for mag volts--so I'm not sure that 80 ohms is high enough to cause a problem.

I'm planning to use the ADG704, which has 2 ohms. Technically, those 2 ohms are on the + side of the capacitor, but the final filtered voltage will be taken from the + side of the switch resistance, so the resistance has the same effect as if it were on the - side of the capacitor.

For those who ordered Jurgen's accessory board, there is a video filter switch on that board, using the ADG704.

Actually, I was also hoping to add some series resistance to the capacitors, as protection against discharge through the switch. We'll see how that works out. My schematic actually shows 100 ohms, but I may have to lower it.

If you use tantalum capacitors for the narrow filter, be sure you use a low leakage capacitor. Otherwise, switching video filters could cause some small shift in phase or magnitude. I plan to have 4 filter capacitors, as follows:

Phase: None (i.e. use built-in), 0.1u, 3.3u and 10u, all ceramics. Mag: None (i.e. use built-in), 0.1u, 10u and 100u, the last being tantalum.

The 100u cap will be useful for extremely low level signals, especially during calibration. Note that I have smaller capacitors for phase. I have found that large phase capacitors create more problems than they solve, particularly by delaying the response to PDM inversion. Also, no matter what you do, phase is not very accurate when you are within 15 dB or so of the noise floor, so you don't have the need for as much filtering.

Sam W.
Hi Jeff and All,
 Do not be too concerned with long leads from the front panel to the SLIM-ADC. The video capacitors mounted inside the module act like noise filters anyway. If you wish, you can use shielded wire for the connections.
Hi William,
 I forgot to answer the question, "what voltage should the input to the AtoD read when nothing is connected to it?"

In reality, it should almost "float". Its input resistance is extremely high. The input shunt capacitor will help keep it at some indeterminate voltage level between 0 and 5 volts.

Touching it with even a high impedance scope probe will "pull" it to 0 volts.

As a test for a working Magnitude AtoD:
Disconnect the Log Det from the AtoD
Let the AtoD input float
Run in the SA Mode
Put your left hand finger on +5 volts and touch the AtoD input with your right hand finger. This should bring the input voltage up. Remove your hands. The Mag trace should either remain at a very high signal level or begin to creep downward. It should not just "snap" to the low level.

Then do the reverse:
Left hand ground
Right hand finger to input of AtoD
The Mag trace should be very low. Probably at -120 dB if the MSA is not yet calibrated.
Remove hands
Trace should remain low or creep upward, but no "snap".

Hi Harold,
 The three video bandwidths are somewhat arbitrary.  Wide video bandwidth for highest speed sweeping, medium for general speed, narrow video for very slow sweeping to get the most accurate magnitude and phase data.  I can see no problem for both to share the same selector switch.
 Mount the switch(s) on the outside of the module.  Either directly on the bottom of the module, or on the front panel and run short wires.
  If no switches are installed, both mag and phase will be in "wide" video bandwidth and will work fine for testing.
  The capacitors called out can be changed to the builders' preferences.  After some MSA use, the operator can adjust them accordingly.

Other Notes:

ADF4118 is a direct replacement for LMX2326 but requires the use of a 4.7K resistor R23