[time-nuts] Thunderbolt Harmonics
time at radio.sent.com
Fri Jan 20 16:36:08 EST 2017
I'm trying to be gentle, Rhys. :) I work with these issues every day at
work. Here are a few more comments. I assume you have the preamplifier
in the spectrum analyzer turned off.
The term "X harmonic" (such as 2nd or 3rd harmonic) means a
multiplication of the fundamental signal by the given factor. So the
term "1st harmonic" isn't used -- that's the fundamental. The 2nd
harmonic is 2X the fundamental, and the 3rd harmonic is 3X the
fundamental. So in your examples you should have said "2nd and 3rd
15 dB attenuation: 2nd harmonic is (-49.13 - +11.40)= -60.53 dBc
20 dB attenuation: 2nd harmonic is (-48.84 - +11.40)= -60.24 dBc
25 dB attenuation: 2nd harmonic is (-48.32 - +11.28)= -59.60 dBc
In nearly all cases it's silly to compare RF powers to 0.01 dB
resolution. The uncertainty of the signal powers being measured,
cable/connector loss, and instrumentation errors is in nearly all cases
larger than 0.1 dB. Your spectrum analyzer doesn't have separate
amplitude log linearly error specifications, but the total amplitude
error with 20 dB of attenuation is specified as +/- 0.7 dB. So the 2nd
harmonic values are not significantly changing as you change the
attenuation, so the source you are measuring probably has about -60 dBc
2nd harmonic output.
The 3rd harmonic results are going to cause me to wave my hands and make
uncomfortable assumptions. The 20 dB 3rd harmonic level seems to be an
outlier, but there is a possibility that a small amount of instrument
distortion is out of phase with the source signal so that they partially
null. RF measurements ARE magic in some cases. <LOL>
The use of the external 20 dB attenuator means that the spectrum
analyzer return loss is isolated from the signal source. What does that
mean? Any RF signal traveling down a cable is slightly reflected by
cable defects, connectors, filters, mixers, and imperfect attenuators or
terminators. The reflected signal is called "return loss" and in some
cases "VSWR" or just "SWR". If you had a perfect 50 ohm termination
(load) at the end of a perfect 50 ohm cable, all of the power sent into
the cable would be absorbed by the load and the return loss would be
infinite. The phase of the reflected signal at the source output
connector depends on the round-trip electrical length of the cable and
the nature of the reflection. The reflection from a short is 180 degrees
different from an open, and other types of load can produce different
reflected phases. By the time the reflection gets back to the source
connector, the phase of the reflected signal can cause the impedance to
appear to be nearly anything (greater or less than 50 ohms and probably
capacitive or inductive). If you change the source frequency there is a
different phase round-trip delay due to the wavelength changing, so in
general the RMS voltage at the source will have some ripple vs
frequency. If you place that 20 dB attenuator directly on the source
output connector, the return loss that the source "sees" is nearly
completely controlled by the quality of the attenuator. Even if the
cable had an open or short at the end, the signal passes both ways
though the attenuator so the return loss must be >40 dB (assuming a very
high quality attenuator). This is the same as saying that the VSWR
(Voltage Standing Wave Ratio) is close to 1. A 40 dB return loss
corresponds to a VSWR of 1.02. If an RF filter doesn't see a low VSWR
load, it may not produce the desired filtering behavior.
Bill Byrom N5BB
On Thu, Jan 19, 2017, at 10:48 PM, Rhys D wrote:
> Thanks for the detailed post Bill,
> I'm learning a lot here!
> So the spectrum analyser is indeed a "trap for young players"
> As you guessed, it is a Siglent SSA3000X series analyzer.
> I just looked at the same signal again with varied attenuations
> dialed in
> on the instrument (I am using an external 20dB attenuator from
> as well)
> Here is what I saw:
> Attenuation - Fundamental - 1st Harmonic - 2nd Harmonic
> 15 dB - 11.40 dB - 49.13 dB - 51.12 dB
> 20 dB - 11.40 dB - 48.84 dB - 56.48 dB
> 25 dB - 11.28 dB - 48.32 dB - 49.15 dB
> I guess these numbers mean I can't really trust what I can see on the
> instrument screen?
> By the way, I should just you know that I am not trying to solve a
> timing problem here, I'm more using it as learning opportunity
> and making
> sure that my setup is the best it can be.
> Thanks again for the input.
> On 20 January 2017 at 12:26, Bill Byrom <time at radio.sent.com> wrote:
>> You can't trust such low harmonic spurious measurements from a
>> analyzer unless you know how the spurs change with input level. The
>> second harmonic spur created in an amplifier or mixer inside the
>> spectrum analyzer input will typically increase by 2 dB for
>> every 1 dB
>> of input level increase. Anytime you see a frequency converting RF
>> component (such as the mixer in the input of a spectrum
>> analyzer), it is
>> nonlinear and will generate harmonics and intermodulation
>> products. All
>> you need to do is to keep the input level low enough so that the
>> distortion products generated in the analyzer are below the
>> signals you
>> are measuring. The best and easiest technique is to increase
>> the input
>> attenuation by the smallest step available (such as 5 dB or 10
>> dB) and
>> checking how the spurious components change.
>> ** If the harmonic or other spurious signal is coming from an
>> source, it should not change as the input attenuation changes.
>> ** If the harmonic or other spurious signal is generated inside the
>> analyzer, it should change relative to the fundamental signal as the
>> input attenuation changes.
>> ** I'm talking about the harmonics or other spurious signals
>> relative to
>> the fundamental frequency being displayed. If you remove the input
>> signal and still see the spur, it's a residual spur created
>> inside the
>> analyzer unrelated to the input signal.
>> If you graph fundamental signal displayed amplitude vs changing input
>> level, you will typically see the following for spurious signals
>> by most mixers or amplifiers:
>> (1) Fundamental signal = slope of 1
>> (2) Second harmonic signal = slope of 2
>> (3) Third order intermodulation (sum or different frequencies
>> caused by
>> mixing of two signals) = slope of 3
>> For more background, see:
>> If that is a SiglentSSA3000X series analyzer, here are the spurious
>> specifications from the datasheet:
>> ** Second harmonic distortion: -65 dBc (above 50 MHz input with
>> preamplifier off)
>> Note that the second harmonic distortion is only specified at 50 MHz
>> input and above and at a -30 dBm input power level with the
>> off. For comparison, here are the specifications of a Tektronix
>> portable spectrum analyzer. Disclosure: I work for Tektronix.
>> ** Second harmonic distortion: - 75 dBc (above 40 MHz input,
>> preamplifier OFF)
>> ** Second harmonic distortion: - 60 dBc (above 40 MHz input,
>> preamplifier ON)
>> I'm sure that the reason for a lower limit on the second harmonic
>> specification is that the results are worse at lower frequencies. So
>> it's quite possible that the harmonics you see are mainly coming from
>> the spectrum analyzer input mixer or preamplifier. As I suggested
>> earlier, try lowering the input level by 5 or 10 dB and see if the
>> harmonics go down linearly.
>> Bill Byrom N5BB
>> On Tue, Jan 17, 2017, at 08:40 PM, Rhys D wrote:
>>> Hi all,
>>> Before I start, let me say I'm rather a newbie at this sort of
>>> stuff so
>>> please be gentle.
>>> I was looking at the output of my Trimble Thunderbolt GPSDO and
>>> was rather
>>> surprised to see really "loud" harmonics in there. ~ 60dB down
>>> from the
>>> 10Mhz signal.
>>> Can anyone here shed some light on what I am seeing here?
>>> Surely this isn't what it is supposed to look like? Should I be
>>> trying to
>>> filter these before going to my distribution amplifier?
>>> Thanks for any light you can shed.
>>> time-nuts mailing list -- time-nuts at febo.com
>>> To unsubscribe, go to
>>> and follow the instructions there.
>> time-nuts mailing list -- time-nuts at febo.com
>> To unsubscribe, go to https://www.febo.com/cgi-bin/
>> and follow the instructions there.
> time-nuts mailing list -- time-nuts at febo.com
> To unsubscribe, go to
> and follow the instructions there.
More information about the time-nuts