[NLRS] RF Frequency Counters

[Dr. Gerald N. Johnson]

On Sat, 2006-04-22 at 08:37 -0500, John P. Toscano wrote:
> 
> Hello out there.  I'm looking for an opinion (or several) from the group 
> on RF frequency counters.
> 
> What I have now is a real inexpensive Radio Shack model that tops out at 
> about 1300 MHz (practically DC for some of you folks).  I am interested 
> in something that will work at least up through the 10 GHz ham band, 
> with good accuracy.
> 
> I'm looking at a used HP 5342A microwave frequency counter with Option 
> 001 (high-stability timebase) and Option 002 (amplitude measurement) 
> installed.  It has two inputs, Input 1 rated at 10 Hz to 520 MHz, and 
> Input 2 rated at 500 MHz to 18 GHz.
> 
> If anyone has experience with this model (good or bad), or has better 
> suggestions, I'd like to hear them.
> 
> Also, I'm wondering about accuracy, i.e. what is reasonable resolution 
> at 10368.1 MHz?  The specs on the Option 001 say:
>        Aging rage: < 5e-10 per day after 24-hour warmup
>       Temperature: < 7e-09 over the range of 0C to 50C
>        Short term: < 1e-11 for 1 second average time
>    Line variation:   1e-10 for 10% change from nominal
>           Warm-Up: < 5e-09 of final value 20 min after turn on, at 25C
> 
> The specs say that ACCURACY is +/- 1 count, +/- time base error.  The 
> time base (internal or external) is 10 MHz.  I can't believe that they 
> mean what (I think) this says, i.e. if the time base is accurate to 
> 1e-10 (for example), times 1e6 Hz, that the time base is accurate to 
> 1e-4 Hz and therefore the display at 10 GHz is also accurate to 1e-4 Hz, 
> plus or minus 1 display digit.  I would think, that at best, if the time 
> base was accurate to 1e-10 (for example), that 1e-10 x 1.3681e11 would 
> be 1.3681e1 HZ or 13 Hz, plus or minus 1 Hz on the display, or +/- 14 
> Hz.  The problem is that they don't seem to state the ACCURACY of the 
> timebase anywhere, only its stability.
> 
> The meter has 11 digits, and can DISPLAY 10 GHz to a resolution of 1 Hz, 
> but I wonder what the true ACCURACY is.  The specs above only talk about 
> stability, but who cares if it is stable to 1e-11 if it is 10 MHz off 
> frequency?  Obviously, my need to measure 10 GHz to 1 Hz accuracy is 
> limited, but I want to do a whole lot better than being 10's of KHz off 
> frequency like I am now, based on the IF rig readout.  (Yeah, I do know 
> how to subtract, but the apparent offset is so odd that it can be a 
> challenge to do in my head under pressure, and I'd rather not have to!)
> 
> Supposedly, the instrument I'm looking at has been calibrated against a 
> GPS-locked OCXO.  It also has an external timebase input, and I have 
> (but have not yet put together) a GPS-disciplined OCXO that could 
> potentially be attached.  Is it likely to be worth the effort?
> 
> Thanks in advance for your collective wisdom.
> 
> John (W0JT)

The key to long term accuracy is the drift rate per day. 5e-10 per day.
That says that it could have moved as much as 1830e-10 or 0.183e-6 in a
year since it was last set on frequency. Almost 2 KHz at 10368 MHz. All
those other errors are short time effects. The long term drift is
probably smaller when its not turned, but most consistent when its run
constantly.

As for quality, the only poor quality instruments HP has ever made were
oscilloscopes before the 1740. The 1740 and Tek 465 were competitive and
some like the HP1740 better than the Tek scope. Which was a landmark new
scope design for HP and their scopes probably have been keeping good.

Their counters and frequency standards have been good setting the state
of the are. 5e-10 per day is extremely good for a crystal oscillator.

If you want to set it closer you can use a frequency comparator to
compare for a week or so with WWVB at 60 Hz though I use a scope to get
faster results, my limits seem to be in the range of 1e-08, but that may
be the short instability of my Manson standard. Using the frequency
comparator is the classic way to set the frequency. Today you could use
1PPS pulses from a GPS receiver to set an external standard to at least
an order of magnitude smaller error than the daily drift of that
counter's standard. And then set the counter's standard to match.

What I do starts with a TRF receiver tuned to 60 KHz. Mine was made by
Instrument Corp of Florida and has been modified to tune up that high. I
put that signal on the vertical input of my Tek 475 scope. I trigger the
sweep with my local standard's 100 KHz output. What I see is a waveform
with three phases, just like the three phases of a three phase power
system when I sweep at 30 KHz sweep rate. Then after I give the scope
trigger some time to stabilize, I raise the sweep rate and the scope
gain to isolate one of the crossings of that three sweep waveform and I
watch the drift rate of that crossing. I ignore noise bursts and the
time when WWVB is on the lower power part of each second. The best I can
do is 100 nanoseconds per cm on the screen staying on the screen but
wandering back and forth in that 1 microsecond over a 15 minute time
period. Sometimes I get a bit better. That's maybe 5e-10 short term.
Which is probably as good my standard, though I wonder about the
stability of the scope trigger too. Achieving 5e-10 in a 15 minute
measurement off the air is unconventional and sometimes I wonder if some
of the few nanoseconds of variation I see might be the wire vertical at
WWVB moving in the wind... More likely its a short term VLF propagation
variation that's not been well documented.

The GPS stabilized OCXO is a good reference and could be a bit (two
orders or magnitude) better than the pretty damn good internal
oscillator. That's a better set of internal oscillator specs than you
could buy 30 or 40 years ago for an internal crystal oscillator without
any bounds on what you were willing to spend.

Accuracy is only as good as its been set. Then that's modified by the
long term drift.

Its a good counter.

-- 
73, Jerry, K0CQ,
All content copyright Dr. Gerald N. Johnson, electrical engineer