A/D Converter

Completed analog digital converter

The analog digital converter is responsible for digitizing the final analog signals in the analyzer.  It has 2 entirely separate converters.  This is because the analyzer can be extended to include vector network analyzer functions, which includes amplitude and phase measurements.  One converter is intended for amplitude (or magnitude) and one for phase.  I don’t have the phase detector module yet, but I may someday, so I populated both sides.  In the included image, I haven’t installed the voltage regulator yet, it’s there now, I just don’t have a photo of it.  Also, on the top of the module there are 2 sets of 3 holes for the pins of the “video filter”.  (The term video filter is mostly historical.  It’s from back in the days when spectrum analyzers were made with CRT’s like analog oscilloscopes.  If you wanted to have a smoother response, giving the signal time to settle, you’re use a more aggressive filter.)  Below those header pins, you can see there are a collection of capacitors that implement the filter function.  Building it was fairly straight-forward, there weren’t any change orders, or any other complications.  Once I installed it into the frame, I made a wiring harness for its connection to the control board.  Sam Wetterlin suggests testing the ADC right after the control board as it doesn’t depend on any other modules.

Test setup

A simple way to generate test voltages is to use a potentiometer with the wiper attached to the Log Detector’s input, 5 volts on one end and ground on the other.  Make sure that the potentiometer is high enough resistance that it doesn’t burn up from being connected like this.

As you can see in the video, the line in the graph is moving all around.  This corresponds quite nicely with my movements of the shaft of the potentiometer.  I was quite satisfied with these results.  I tried both video filter switch settings, though it didn’t appear to change the display.

Log detector installed with ADC

Because I was so satisfied from the results of the ADC testing, I decided to hook up the next module in the chain — the log detector.  The detector takes the final RF signal and converts it to a DC voltage that increases with the log of the RF voltage.  When I did this, I must have introduced a gremlin into the system.  Now, the ADC seems to “stick” its input to 5 volts.  Not only does it show 5 volts on the computer display, but 5 volts is also present on the wire.  I tried several things to get the system working again.  I tried to eliminate the log detector as the culprit by capping the input with a 50 ohm terminator, I tries cutting power to the log detector, and I tried different filter switch settings.  Nothing reliably helped.

Annoying problem

This plot is a pretty good example of what I’m talking about.  On the left, everything seems to be working fine, the noise floor seems low there may be some legitimate spikes from WiFi and such.  Then, in the middle, it gets a little whacky and locks up to a high voltage.

5 volts at ADC input

This is a photo of my oscilloscope display when the ADC is stuck.  The mean voltage is near 5 volts, and there were many little ticks of noise.

Because the symptoms didn’t change with the log detector off, I began trying to hunt-down the cause of the problem assuming the log detector probably wasn’t the cause.  First, I realized that the center pin of the filter switch header is wired to the ADC input, so I had an easy connection point for test equipment, which was nice.  With that known, I thought I’d determine how much current was leaking onto the input line.  To do this, I connected a ammeter from the input to ground.  The current reading was highly variable, but was usually about 300-1000 uA.  With this little amount of current, it was somewhat easy to over-power it with a potentiometer.  This leads me to believe that it’s possible that this problem existed before I attached the log detector, I just didn’t notice.  As a final point, the entire ADC module consumes approx. 26mA of current.

At this point, I have no idea what the problem is.  I’m going to check with the collective knowledge of the mailing list.  Hopefully they have a solution.

Update 5/29/2010 9PM:

Well, some time has elapsed, and many suggestions have been given.  I’ll list them, and explain how I tried to investigate.

  1. Sam: Try the Phase converter, and see if there are similar results
  2. Sam: It’s possible the ESD protection diode is shot from ESD or negligence.
  3. Sam: Current draw seems high, 10x too high
  4. Scotty: There might be a short between pins 2 and 3 of the ADC

So, with those suggestions in hand I began testing.  First, did a lot of measuring.  I measured the resistance from the ADC input to VCC, in attempt to address suggestion (2) and (4) as the IN+ pin and VCC pin are 3 and 2, respectively.  I used a standard ohmmeter and tested once with the positive side on the ADC input and once with the positive side on VCC.  I observed 17k ohm each direction.  This is not strictly a good thing.  In an ideal world, I’d expect that the resistance with positive voltage on the input to be less than the other way.  This is because the protection diode is intended to dump excess voltage on the input into VCC.  It’s not desirable for voltage to leak from VCC to the input, however.  This result could very well be the cause of the problem.  After taking these measurements, I confirmed that current draw is 26mA, which is still 10x what it should be, following suggestion (3).  I should say that when I started stuff up the supply sounded unhappy, like a subtle high-pitch whine.  I disconnected everything, and started by only powering the oscillator, then the log detector, then the ADC and the whine went away.  It’s possible there was a short or miswired connector.

Output of the phase ADC

Following Sam’s advice (1), I displayed the output of the phase measuring converter.  On the left side of the plot, the ADC is allowed to measure what is essentially on open circuit.  On the left, I’ve attached the probe of my oscilloscope.  I’m not sure that to make of this plot, so it is what it is…

At this point, I cleaned the board with alcohol and a toothbrush.  This dissolves solder residue, then I used a cloth to wick away the alcohol and dissolved residue.  I hooked the analyzer back up and did a measurement with the ADC alone.

Baseline Magnitude ADC output

Here is a recent plot from the magnitude ADC.  As you can see, there is a very clean output with a few occasional spikes.  I’m still not sure what causes the spikes, and if they’re a problem.

Going for broke

Feeling hopeful, I figured I’d plug the log detector back in.  Also, you can see the little rubber-duckie hooked up to the input.

 

ADC output with log detector

With the log detector powered up, and the rubber-duckie there is a fair amount of noise, though it’s still fairly low-level for the most part.  One thing that concerned me a little is that when I keyed my radio there was no change on the plot while I did notice a change in the log detector’s output using the oscilloscope.

Anyway, that’s a lot of new information.  Hopefully this is all really good news, but I fear I still have problems.  Basically, I’m debating when I should order new ADC ICs, not if.

Update 5/2/2010 8:46 PM:

After much testing, I’ve been able to determine that the primary problem with the ADC was a failed latch on the control board.  Everything appears to be functioning at this point.  Please see the control board page for details.

  1. No comments yet.
(will not be published)

Please complete this capcha. I get almost 1000 spam comments a day! * Time limit is exhausted. Please reload CAPTCHA.