Tech Notes from the Field

Earlier this week, Tuesday to be exact, I noticed something not quite right with my diversity delay.  HD radio has added a little more complexity to our systems which requires delaying analog audio to match digital audio so when a radio switches between analog and HD audio there is no skipping or jumping.  This makes the transitions smoother in high multipath areas and only sounds like a quality change if the delay is set properly.  Knowing I had an issue I began  to look into it  when I received a call from our third-party frequency observer.  Our carrier frequency was off by -780 Hz.  Though within the legal limits, this was a large difference than the month prior.  What changed?

The technology for running IBOC, In Band On Channel, digital radio brought along some new challenges and some more things to monitor and look at.  In this case we have two things occurring:  Diversity delay drifting and a carrier off frequency.  With the new technology a source clock is required to maintain sync among all facets of transmission, and this is a 10 MHz reference.  All normal installations have a GPS signal fed to the Exporter which has a built-in receiver.  This produces the reference 10 MHz clock.  All normal installation uses this clock through the Exporter to sync their exciters, so the exciter is now synced to the same 10 MHz reference.  Most of the time this works.  In our case, on one station, it does not.  All four stations have Exporters located at the studios, and three of the four have no issues with the exciters syncing.  The third seemed to have some issues with this, so a 10 MHz GPS reference was installed at the transmitter site to clock the exciter directly.  This was my starting point.

It was safe to conclude that there was an issue, somewhere, with regard to the reference clocking since both diversity delay and carrier frequency were affected.  It was time to check things out at the site.  The ESE-110 GPS reference front panel display showed 2 green LEDs.  GPS lock and power.  Good sign.  The Flexstar exciter showed no errors.  Green LEDs and diagnostic screens showed good things.  I placed a spectrum analyzer on the 10 MHz output of the ESE.  I measured 10 MHz.  Actually it was slightly lower, like 10 Hz, but then again it is an older spectrum analyzer and probably needs calibration.  I used a TFT 844 to measure the frequency shift in conjunction with the spectrum analyzer.  Yup, it was there.  Since I wanted a third party measurement to confirm, I called him up.  I switched to the internal clock of the Flexstar.  Frequency changed and swung to +110 Hz.  Better.  The diversity delay settled down, too.  What was up with the 10 MHz input?

If the ESE output was within spec, and a call to ESE verified this along with the only indicators, the green LEDs, then what is up with the exciter?  Again, the exciter seemed to be good with the external reference according to the indicators on it along with the diagnostic screens.  I did not find the schematic at the site, another long story, so all I had was the operations manual and block diagrams.  I remeasured the 10 MHz signal.  I stretched out the bandwidth to look at the “hump.”   All the manual says is a max of 10 dBm, 0 dBm nominal, and it is used for carrier sync purposes.  They better modify that to say it is for ALL frequency references!  Mental note on  that.  I’ve been running this configuration for a year without issue, so what changed?  I stood there staring at the simplified block diagram of the Modulation Chain.  That PLL feeds the FPGA and D/A module.  Two different mixers.  Both having issues.  I read the short paragraph on the PLL.  The 10 MHz reference input is first amplified and put through an AGC loop for level stability.  I lacked stability.

My what if moment came at this point.  I have had zero, none, nada, problems for a year with my current configuration and a fairly sudden change occurred.  If they are amplifying the reference input, what if the amplifier has gone bad in some way.  I dug through the spectrum analyzer bag knowing I had a couple of in-line BNC pads.  I took out the -20 dBm pad, slapped it on the output of the ESE box, and reconnected it to the Flexstar exciter.  I gave it about 10 minutes, though I didn’t have to, and took a diversity delay measurement.  Yup, had to change it, so it was time to dial it in and see if it stays.  I got the delay down to a very respectable -0.0008 s.  At that value I should surely see a change if my solution did not work.  I gave it about 2 minutes and rechecked.  -0.0004 s.  I gave it 10 minutes more; -0.0005 s.  Before it was changing rapidly, now it seems to be locked in.  I gave a call to Harris.  While talking with support, I checked again, -0.0003 s.  OMG, this may be working.  I emailed my third party to take a measurement while I was busy talking with Harris.  No immediate response, but I was not too concerned.  The Harris tech was actually surprised at this solution, and after our talk concluded that a new PLL on site would be prudent as all indications now pointed to that as the culprit.

I packed up and blessed the site.  I returned to the station and took another reading.  The first in about an hour.  If it was drifting now, it would show it.  -0.0005 s.  I was beside myself at this point. What did I stumble upon?  It has never been this stable.  Before leaving for the day and my tax appointment, I took another reading:  -0.0009 s.  So far so good.  After my appointment, we were out eating dinner (yes, we owe the government a couple of bucks) I received an email from my contact.  He listed the times of measurements and the error, the last entry was “6:47 PM.   0 Hz error ** Wow!  What did you do?” My solution was holding!  I came into work this morning, and I checked my delay again:  -0.0004 s.

I now have a PLL module on order with Harris as we have no idea how long this one will last.  From what I can figure the amplifier section of the PLL has an issue.  The 10 MHz reference was getting distorted and creating  a new frequency that offset the carrier frequency.  In addition, the distorted signal was “outside” the specification for the unit to lock the diversity delay which was free-running allowing it to drift over a half second.  Talking with a colleague we both concluded they may want to install a test point or provide a software controlled pad on the external 10 MHz input.  For now, the problem is solved.  I am tempted to see how long this lasts or if there will be further deterioration causing drift once again.  Happy troubleshoot!

Cheers!