time-nuts@lists.febo.com

Discussion of precise time and frequency measurement

View all threads

Fixing PN degradation via ADEV measurement

KT
Karen Tadevosyan
Sat, Jun 18, 2022 8:49 AM

Hello group,

I am developing a hamradio ground satellite station for the first geostationary satellite QO-100 (uplink 2.4 GHz / downlink 10 GHz).

Narrowband digital communication requires high frequency stability so I use a 10 MHz OCXO with an absolute Allan deviation (ADEV) of about 5 mHz as a single reference oscillator.

To form a reference signal of 25 MHz for the 9750 MHz synthesizer in LNB I use a PLL (to multiply 10 MHz OCXO frequency by 2.5 times).

When measuring the PN of a shaped reference signal at 25 MHz, measured with SA R&S FSQ8 achieved an instrument limit of -130 dBc/Hz @ 10 kHz. According to my calculations, the 25 MHz output PN should be in the range of -150 dBc/Hz. However, when operating via satellite, the nature of the received signal is more "noisy and smeared" with a lower S/N ratio (compared to operation from another low-noise generator at 25 MHz without PLL) which, in my opinion, indicates a worse PN value than expected -150 dBc/Hz.

At the same time I noticed such a feature - when measuring the absolute ADEV value of 25 MHz output signal the device does not show the calculated value of 5 MHz x 2.5 = 12.5 mHz but a value of approximately 60 - 70 mHz.

Additional tests at 10 GHz through the radio end-to-end confirmed a real decrease in S/N ratio of about 10-15 dB and I found the source of the problem.

Question: it turns out that the Pendulum CNT-91 frequency meter-analyzer (which I measure the Allan deviation) allow to record significantly lower PN values ​​than the R&S FSQ8?

That is, with PN borderline values ​​​​between -130 and - 150 dBc / Hz (in our particular case), the Allan deviation measurement allow to accurately fix the PN degradation where a not the worst SA R&S FSQ8 can not help.

I would like to hear the opinion of experts in this matter.

Karen, ra3apw

Hello group, I am developing a hamradio ground satellite station for the first geostationary satellite QO-100 (uplink 2.4 GHz / downlink 10 GHz). Narrowband digital communication requires high frequency stability so I use a 10 MHz OCXO with an absolute Allan deviation (ADEV) of about 5 mHz as a single reference oscillator. To form a reference signal of 25 MHz for the 9750 MHz synthesizer in LNB I use a PLL (to multiply 10 MHz OCXO frequency by 2.5 times). When measuring the PN of a shaped reference signal at 25 MHz, measured with SA R&S FSQ8 achieved an instrument limit of -130 dBc/Hz @ 10 kHz. According to my calculations, the 25 MHz output PN should be in the range of -150 dBc/Hz. However, when operating via satellite, the nature of the received signal is more "noisy and smeared" with a lower S/N ratio (compared to operation from another low-noise generator at 25 MHz without PLL) which, in my opinion, indicates a worse PN value than expected -150 dBc/Hz. At the same time I noticed such a feature - when measuring the absolute ADEV value of 25 MHz output signal the device does not show the calculated value of 5 MHz x 2.5 = 12.5 mHz but a value of approximately 60 - 70 mHz. Additional tests at 10 GHz through the radio end-to-end confirmed a real decrease in S/N ratio of about 10-15 dB and I found the source of the problem. Question: it turns out that the Pendulum CNT-91 frequency meter-analyzer (which I measure the Allan deviation) allow to record significantly lower PN values ​​than the R&S FSQ8? That is, with PN borderline values ​​​​between -130 and - 150 dBc / Hz (in our particular case), the Allan deviation measurement allow to accurately fix the PN degradation where a not the worst SA R&S FSQ8 can not help. I would like to hear the opinion of experts in this matter. Karen, ra3apw
BK
Bob kb8tq
Sun, Jun 19, 2022 2:41 AM

Hi

The “typical” approach to doing this is to multiply up to something around
100 MHz with a relatively narrowband PLL. You can get ( or build ) VCXO’s
in this range with phase noise (at 10KHz offset) in the 160’s to (maybe) 170’s.
This is much better performance than a 5 or 10 MHz reference can deliver
when multiplied to that frequency.

The “cross over” between multiplied reference noise and a fairly good VCXO
should typically be in the 50 to 200 Hz range. Since it is highly dependent on
the exact parts chosen, there is no “one size fits all” answer.

Bob

On Jun 18, 2022, at 12:49 AM, Karen Tadevosyan via time-nuts time-nuts@lists.febo.com wrote:

Hello group,

I am developing a hamradio ground satellite station for the first geostationary satellite QO-100 (uplink 2.4 GHz / downlink 10 GHz).

Narrowband digital communication requires high frequency stability so I use a 10 MHz OCXO with an absolute Allan deviation (ADEV) of about 5 mHz as a single reference oscillator.

To form a reference signal of 25 MHz for the 9750 MHz synthesizer in LNB I use a PLL (to multiply 10 MHz OCXO frequency by 2.5 times).

When measuring the PN of a shaped reference signal at 25 MHz, measured with SA R&S FSQ8 achieved an instrument limit of -130 dBc/Hz @ 10 kHz. According to my calculations, the 25 MHz output PN should be in the range of -150 dBc/Hz. However, when operating via satellite, the nature of the received signal is more "noisy and smeared" with a lower S/N ratio (compared to operation from another low-noise generator at 25 MHz without PLL) which, in my opinion, indicates a worse PN value than expected -150 dBc/Hz.

At the same time I noticed such a feature - when measuring the absolute ADEV value of 25 MHz output signal the device does not show the calculated value of 5 MHz x 2.5 = 12.5 mHz but a value of approximately 60 - 70 mHz.

Additional tests at 10 GHz through the radio end-to-end confirmed a real decrease in S/N ratio of about 10-15 dB and I found the source of the problem.

Question: it turns out that the Pendulum CNT-91 frequency meter-analyzer (which I measure the Allan deviation) allow to record significantly lower PN values ​​than the R&S FSQ8?

That is, with PN borderline values ​​​​between -130 and - 150 dBc / Hz (in our particular case), the Allan deviation measurement allow to accurately fix the PN degradation where a not the worst SA R&S FSQ8 can not help.

I would like to hear the opinion of experts in this matter.

Karen, ra3apw


time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com

Hi The “typical” approach to doing this is to multiply up to something around 100 MHz with a relatively narrowband PLL. You can get ( or build ) VCXO’s in this range with phase noise (at 10KHz offset) in the 160’s to (maybe) 170’s. This is much better performance than a 5 or 10 MHz reference can deliver when multiplied to that frequency. The “cross over” between multiplied reference noise and a fairly good VCXO should typically be in the 50 to 200 Hz range. Since it is highly dependent on the exact parts chosen, there is no “one size fits all” answer. Bob > On Jun 18, 2022, at 12:49 AM, Karen Tadevosyan via time-nuts <time-nuts@lists.febo.com> wrote: > > Hello group, > > > > I am developing a hamradio ground satellite station for the first geostationary satellite QO-100 (uplink 2.4 GHz / downlink 10 GHz). > > > > Narrowband digital communication requires high frequency stability so I use a 10 MHz OCXO with an absolute Allan deviation (ADEV) of about 5 mHz as a single reference oscillator. > > > > To form a reference signal of 25 MHz for the 9750 MHz synthesizer in LNB I use a PLL (to multiply 10 MHz OCXO frequency by 2.5 times). > > > > When measuring the PN of a shaped reference signal at 25 MHz, measured with SA R&S FSQ8 achieved an instrument limit of -130 dBc/Hz @ 10 kHz. According to my calculations, the 25 MHz output PN should be in the range of -150 dBc/Hz. However, when operating via satellite, the nature of the received signal is more "noisy and smeared" with a lower S/N ratio (compared to operation from another low-noise generator at 25 MHz without PLL) which, in my opinion, indicates a worse PN value than expected -150 dBc/Hz. > > > > At the same time I noticed such a feature - when measuring the absolute ADEV value of 25 MHz output signal the device does not show the calculated value of 5 MHz x 2.5 = 12.5 mHz but a value of approximately 60 - 70 mHz. > > > > Additional tests at 10 GHz through the radio end-to-end confirmed a real decrease in S/N ratio of about 10-15 dB and I found the source of the problem. > > > > Question: it turns out that the Pendulum CNT-91 frequency meter-analyzer (which I measure the Allan deviation) allow to record significantly lower PN values ​​than the R&S FSQ8? > > > > That is, with PN borderline values ​​​​between -130 and - 150 dBc / Hz (in our particular case), the Allan deviation measurement allow to accurately fix the PN degradation where a not the worst SA R&S FSQ8 can not help. > > > > I would like to hear the opinion of experts in this matter. > > > > Karen, ra3apw > > > > _______________________________________________ > time-nuts mailing list -- time-nuts@lists.febo.com > To unsubscribe send an email to time-nuts-leave@lists.febo.com
KT
Karen Tadevosyan
Sun, Jun 19, 2022 6:42 AM

Hi Bob,

Thank you for the advice. An issue with PN of  25 MHz reference source has been decided - I found the PN source, changed schematic and now all works well.

My question is about an opportunity to fix PN degradation by Pendulum CNT-91 frequency meter-analyzer (which I measure the Allan deviation) which allow to record significantly lower PN values than R&S FSQ8.

Karen, ra3apw

Hi Bob, Thank you for the advice. An issue with PN of 25 MHz reference source has been decided - I found the PN source, changed schematic and now all works well. My question is about an opportunity to fix PN degradation by Pendulum CNT-91 frequency meter-analyzer (which I measure the Allan deviation) which allow to record significantly lower PN values than R&S FSQ8. Karen, ra3apw
BK
Bob kb8tq
Sun, Jun 19, 2022 3:22 PM

Hi

The CNT-91 has a very high noise floor if you are trying to “see”
phase noise. It also does not give you much in the way of frequency
offset information. A single mixer setup running into a sound card
is a much better approach if you are looking for another opinion
vs the FSQ8. I have used the CNT 90 and CNT 91’s a lot. I have no
experience with the FSQ8.

Bob

On Jun 18, 2022, at 10:42 PM, Karen Tadevosyan ra3apw@mail.ru wrote:

Hi Bob,

Thank you for the advice. An issue with PN of  25 MHz reference source has been decided - I found the PN source, changed schematic and now all works well.

My question is about an opportunity to fix PN degradation by Pendulum CNT-91 frequency meter-analyzer (which I measure the Allan deviation) which allow to record significantly lower PN values than R&S FSQ8.

Karen, ra3apw

Hi The CNT-91 has a very high noise floor if you are trying to “see” phase noise. It also does not give you much in the way of frequency offset information. A single mixer setup running into a sound card is a *much* better approach if you are looking for another opinion vs the FSQ8. I have used the CNT 90 and CNT 91’s a lot. I have no experience with the FSQ8. Bob > On Jun 18, 2022, at 10:42 PM, Karen Tadevosyan <ra3apw@mail.ru> wrote: > > Hi Bob, > > Thank you for the advice. An issue with PN of 25 MHz reference source has been decided - I found the PN source, changed schematic and now all works well. > > My question is about an opportunity to fix PN degradation by Pendulum CNT-91 frequency meter-analyzer (which I measure the Allan deviation) which allow to record significantly lower PN values than R&S FSQ8. > > Karen, ra3apw >
KT
Karen Tadevosyan
Sun, Jun 19, 2022 5:58 PM

Hi

The CNT-91 has a very high noise floor if you are trying to “see” phase noise.

May be but with this Pendulum CNT-91 I can fix PN degradation in the range -130 ... -150 dBc/Hz where R&S FSQ8 can do nothing...

FYI: FSQ8 is not the worst SA: PN floor -130 dBc/Hz @10 kHz on 1 GHz.

Karen, ra3apw

Hi > The CNT-91 has a very high noise floor if you are trying to “see” phase noise. May be but with this Pendulum CNT-91 I can fix PN degradation in the range -130 ... -150 dBc/Hz where R&S FSQ8 can do nothing... FYI: FSQ8 is not the worst SA: PN floor -130 dBc/Hz @10 kHz on 1 GHz. Karen, ra3apw
BK
Bob kb8tq
Sun, Jun 19, 2022 6:36 PM

Hi

If the objective is “phase noise -130 dbc at a 10 KHz offset” and the signal is
at 25 MHz, the resultant jitter ( in a 10 KHz bandwidth ) is likely in the < 500 fs
range. The CNT-91 is only a 50 ps resolution device.

Since it’s a counter, there is no practical way to vary the bandwidth. You get
whatever the internal processing gives you. Even then, at 25 MHz a broadband
-130 dbc/Hz phase noise ( 10 MHz bandwidth) is still below 10 ps. Still below
the counter’s floor.

One of many calculators that let you play with jitter from phase noise:

https://rf-tools.com/jitter/ https://rf-tools.com/jitter/

It’s just the first one Mr Google turned up. There may be much better ones
out there.

Bob

On Jun 19, 2022, at 9:58 AM, Karen Tadevosyan ra3apw@mail.ru wrote:

Hi

The CNT-91 has a very high noise floor if you are trying to “see” phase noise.

May be but with this Pendulum CNT-91 I can fix PN degradation in the range -130 ... -150 dBc/Hz where R&S FSQ8 can do nothing...

FYI: FSQ8 is not the worst SA: PN floor -130 dBc/Hz @10 kHz on 1 GHz.

Karen, ra3apw

Hi If the objective is “phase noise -130 dbc at a 10 KHz offset” and the signal is at 25 MHz, the resultant jitter ( in a 10 KHz bandwidth ) is likely in the < 500 fs range. The CNT-91 is only a 50 ps resolution device. Since it’s a counter, there is no practical way to vary the bandwidth. You get whatever the internal processing gives you. Even then, at 25 MHz a broadband -130 dbc/Hz phase noise ( 10 MHz bandwidth) is still below 10 ps. Still below the counter’s floor. One of *many* calculators that let you play with jitter from phase noise: https://rf-tools.com/jitter/ <https://rf-tools.com/jitter/> It’s just the first one Mr Google turned up. There may be much better ones out there. Bob > On Jun 19, 2022, at 9:58 AM, Karen Tadevosyan <ra3apw@mail.ru> wrote: > > Hi > >> The CNT-91 has a very high noise floor if you are trying to “see” phase noise. > > May be but with this Pendulum CNT-91 I can fix PN degradation in the range -130 ... -150 dBc/Hz where R&S FSQ8 can do nothing... > > FYI: FSQ8 is not the worst SA: PN floor -130 dBc/Hz @10 kHz on 1 GHz. > > Karen, ra3apw > > >