FE-5680A's suitability for use as a 10 MHz reference for microwave transverters
Hello,
As has been mentioned here before, the output of the "non-tunable"
FE-5680A's has been noted to have low-level spurs in it - no doubt due
to the way the various frequency loops are derived within, some using
DDS techniques. It is for this reason that when I packaged my FE-5680A
in its own, stand-alone enclosure I included a fairly narrow band (+/-6
kHz @ -6dBc) crystal-based bandpass filter in the output.
After more recent testing of two FE-5680A's using two different 10 GHz
microwave transverters, I've determined that this filtering just isn't
enough. At first, it was assumed/hoped that the racket that I was
hearing was coming from somewhere else - perhaps the switching
up-converter or some other interaction - or just something "odd" about
my homebrew 10 GHz transverter, but this is, unfortunately, not the case.
In testing with a DownEast Microwave 10 GHz transverter fitted with an
N5AC synthesizer, the CW notes sounded nice and clean when locked to the
Z3801 and there was only a trace of modulation that I'd not really
noticed before when I used the Efratom LPRO-101, but when the '5680A was
connected, the incidental PM was bad enough that it was difficult to
determine where, exactly, zero beat was! Since the synthesizer uses a
fairly high reference frequency internally there was little impediment
to the low-level phase modulation on the reference.
I compared this with my own homebrew 10 GHz transverter. This unit uses
an 18.4 MHz Butler VCXO that is multiplied to 110.4 MHz which is then
fed to a "brick" oscillator with the 110.4 MHz being compared to the 10
MHz reference using a harmonic mixer, locking to the 400 kHz residual.
Since this unit has a comparatively low loop bandwidth in the VCXO the
grunge was considerably reduced, but still objectionable, giving some
hope that a simple VCXO scheme might make the '5680A usable.
I still have yet to do a more-detailed analysis of the phase modulation
that is appearing on the 10 GHz signals, but I can clearly hear a low
frequency modulation source (perhaps the lock-in amplifier) plus a
myriad of other audio frequency components and their harmonics. Again,
with the LPRO-101 was very "clean" by comparison and I could just hear
some similar, very low-level noises in the background that I'd not
really noticed before.
As it is, the '5680A-based reference is unusable with the N5AC
synthesizer and its wide loop bandwidth and "almost" usable with my
homebrew transverter and its comparatively narrow loop bandwidth. I'm
now bent on making the '5680A usable as a microwave reference, but my
current plans are to build a simple 10 MHz Butler VCXO and then lock it
to the '5680A using a very "slow" loop filter: In that way, I'm hoping
that the phase noise will be largely that of the 10 MHz VCXO and its
cheap CPU-type crystal rather than the '5680A!
Clint
KA7OEI
Hi
The phase noise of the FE's is not great, even inside +/- 3 KHz. To do an
adequate job of cleaning them up, something like a < 1 Hz loop bandwidth PLL
would be needed.
Bob
-----Original Message-----
From: time-nuts-bounces@febo.com [mailto:time-nuts-bounces@febo.com] On
Behalf Of C. Turner
Sent: Tuesday, February 07, 2012 12:48 PM
To: time-nuts@febo.com
Subject: [time-nuts] FE-5680A's suitability for use as a 10 MHz reference
for microwave transverters
Hello,
As has been mentioned here before, the output of the "non-tunable"
FE-5680A's has been noted to have low-level spurs in it - no doubt due
to the way the various frequency loops are derived within, some using
DDS techniques. It is for this reason that when I packaged my FE-5680A
in its own, stand-alone enclosure I included a fairly narrow band (+/-6
kHz @ -6dBc) crystal-based bandpass filter in the output.
After more recent testing of two FE-5680A's using two different 10 GHz
microwave transverters, I've determined that this filtering just isn't
enough. At first, it was assumed/hoped that the racket that I was
hearing was coming from somewhere else - perhaps the switching
up-converter or some other interaction - or just something "odd" about
my homebrew 10 GHz transverter, but this is, unfortunately, not the case.
In testing with a DownEast Microwave 10 GHz transverter fitted with an
N5AC synthesizer, the CW notes sounded nice and clean when locked to the
Z3801 and there was only a trace of modulation that I'd not really
noticed before when I used the Efratom LPRO-101, but when the '5680A was
connected, the incidental PM was bad enough that it was difficult to
determine where, exactly, zero beat was! Since the synthesizer uses a
fairly high reference frequency internally there was little impediment
to the low-level phase modulation on the reference.
I compared this with my own homebrew 10 GHz transverter. This unit uses
an 18.4 MHz Butler VCXO that is multiplied to 110.4 MHz which is then
fed to a "brick" oscillator with the 110.4 MHz being compared to the 10
MHz reference using a harmonic mixer, locking to the 400 kHz residual.
Since this unit has a comparatively low loop bandwidth in the VCXO the
grunge was considerably reduced, but still objectionable, giving some
hope that a simple VCXO scheme might make the '5680A usable.
I still have yet to do a more-detailed analysis of the phase modulation
that is appearing on the 10 GHz signals, but I can clearly hear a low
frequency modulation source (perhaps the lock-in amplifier) plus a
myriad of other audio frequency components and their harmonics. Again,
with the LPRO-101 was very "clean" by comparison and I could just hear
some similar, very low-level noises in the background that I'd not
really noticed before.
As it is, the '5680A-based reference is unusable with the N5AC
synthesizer and its wide loop bandwidth and "almost" usable with my
homebrew transverter and its comparatively narrow loop bandwidth. I'm
now bent on making the '5680A usable as a microwave reference, but my
current plans are to build a simple 10 MHz Butler VCXO and then lock it
to the '5680A using a very "slow" loop filter: In that way, I'm hoping
that the phase noise will be largely that of the 10 MHz VCXO and its
cheap CPU-type crystal rather than the '5680A!
Clint
KA7OEI
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
I am just finishing my promised stability and phase noise measurements
on a batch of inexpensive Rb standards; I hope to publish the results
tomorrow evening.
In the meantime, I've looked at two of the FE-5680s and their phase
noise is significantly worse than either the Efratom FRS or the Datum
LPRO -- in particular, there is a forest of spurs all the way from 1 Hz
on out, most of them at around -80dBc or worse. By the time you
multiply that to 10 GHz, that's only about 20 dB below the carrier!
Apart from the spurs, the noise floor is significantly higher than the
other two types.
A clean-up oscillator would be an interesting add-on.
John
On 2/7/2012 12:48 PM, C. Turner wrote:
Hello,
As has been mentioned here before, the output of the "non-tunable"
FE-5680A's has been noted to have low-level spurs in it - no doubt due
to the way the various frequency loops are derived within, some using
DDS techniques. It is for this reason that when I packaged my FE-5680A
in its own, stand-alone enclosure I included a fairly narrow band (+/-6
kHz @ -6dBc) crystal-based bandpass filter in the output.
After more recent testing of two FE-5680A's using two different 10 GHz
microwave transverters, I've determined that this filtering just isn't
enough. At first, it was assumed/hoped that the racket that I was
hearing was coming from somewhere else - perhaps the switching
up-converter or some other interaction - or just something "odd" about
my homebrew 10 GHz transverter, but this is, unfortunately, not the case.
In testing with a DownEast Microwave 10 GHz transverter fitted with an
N5AC synthesizer, the CW notes sounded nice and clean when locked to the
Z3801 and there was only a trace of modulation that I'd not really
noticed before when I used the Efratom LPRO-101, but when the '5680A was
connected, the incidental PM was bad enough that it was difficult to
determine where, exactly, zero beat was! Since the synthesizer uses a
fairly high reference frequency internally there was little impediment
to the low-level phase modulation on the reference.
I compared this with my own homebrew 10 GHz transverter. This unit uses
an 18.4 MHz Butler VCXO that is multiplied to 110.4 MHz which is then
fed to a "brick" oscillator with the 110.4 MHz being compared to the 10
MHz reference using a harmonic mixer, locking to the 400 kHz residual.
Since this unit has a comparatively low loop bandwidth in the VCXO the
grunge was considerably reduced, but still objectionable, giving some
hope that a simple VCXO scheme might make the '5680A usable.
I still have yet to do a more-detailed analysis of the phase modulation
that is appearing on the 10 GHz signals, but I can clearly hear a low
frequency modulation source (perhaps the lock-in amplifier) plus a
myriad of other audio frequency components and their harmonics. Again,
with the LPRO-101 was very "clean" by comparison and I could just hear
some similar, very low-level noises in the background that I'd not
really noticed before.
As it is, the '5680A-based reference is unusable with the N5AC
synthesizer and its wide loop bandwidth and "almost" usable with my
homebrew transverter and its comparatively narrow loop bandwidth. I'm
now bent on making the '5680A usable as a microwave reference, but my
current plans are to build a simple 10 MHz Butler VCXO and then lock it
to the '5680A using a very "slow" loop filter: In that way, I'm hoping
that the phase noise will be largely that of the 10 MHz VCXO and its
cheap CPU-type crystal rather than the '5680A!
Clint
KA7OEI
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hi
Here is a little more on how much of a problem you have.
If you would like the spurs to be down 70 dbc at 10 GHz. They go up by 20 log N. in this case N is 1000. That gets you 60 db. Spurs at 10 MHz would have to be down at -130 dbc to make it at 10 GHz.
If you want noise over 10 KHz to be 60 db down, it goes by 10 log BW. That gets you to 1 Hz noise at -100. The same 60 db to 10 MHz then applies. You would need -160 dbc phase noise at 10 MHz to hit that.
Both of those would be hard to hit with any Rb. Fortunately you can use a multi step multiplication chain. Once you make that decision, noise and spurs on the Rb are not a big issue.
Bob
On Feb 7, 2012, at 1:52 PM, John Ackermann N8UR jra@febo.com wrote:
I am just finishing my promised stability and phase noise measurements on a batch of inexpensive Rb standards; I hope to publish the results tomorrow evening.
In the meantime, I've looked at two of the FE-5680s and their phase noise is significantly worse than either the Efratom FRS or the Datum LPRO -- in particular, there is a forest of spurs all the way from 1 Hz on out, most of them at around -80dBc or worse. By the time you multiply that to 10 GHz, that's only about 20 dB below the carrier! Apart from the spurs, the noise floor is significantly higher than the other two types.
A clean-up oscillator would be an interesting add-on.
John
On 2/7/2012 12:48 PM, C. Turner wrote:
Hello,
As has been mentioned here before, the output of the "non-tunable"
FE-5680A's has been noted to have low-level spurs in it - no doubt due
to the way the various frequency loops are derived within, some using
DDS techniques. It is for this reason that when I packaged my FE-5680A
in its own, stand-alone enclosure I included a fairly narrow band (+/-6
kHz @ -6dBc) crystal-based bandpass filter in the output.
After more recent testing of two FE-5680A's using two different 10 GHz
microwave transverters, I've determined that this filtering just isn't
enough. At first, it was assumed/hoped that the racket that I was
hearing was coming from somewhere else - perhaps the switching
up-converter or some other interaction - or just something "odd" about
my homebrew 10 GHz transverter, but this is, unfortunately, not the case.
In testing with a DownEast Microwave 10 GHz transverter fitted with an
N5AC synthesizer, the CW notes sounded nice and clean when locked to the
Z3801 and there was only a trace of modulation that I'd not really
noticed before when I used the Efratom LPRO-101, but when the '5680A was
connected, the incidental PM was bad enough that it was difficult to
determine where, exactly, zero beat was! Since the synthesizer uses a
fairly high reference frequency internally there was little impediment
to the low-level phase modulation on the reference.
I compared this with my own homebrew 10 GHz transverter. This unit uses
an 18.4 MHz Butler VCXO that is multiplied to 110.4 MHz which is then
fed to a "brick" oscillator with the 110.4 MHz being compared to the 10
MHz reference using a harmonic mixer, locking to the 400 kHz residual.
Since this unit has a comparatively low loop bandwidth in the VCXO the
grunge was considerably reduced, but still objectionable, giving some
hope that a simple VCXO scheme might make the '5680A usable.
I still have yet to do a more-detailed analysis of the phase modulation
that is appearing on the 10 GHz signals, but I can clearly hear a low
frequency modulation source (perhaps the lock-in amplifier) plus a
myriad of other audio frequency components and their harmonics. Again,
with the LPRO-101 was very "clean" by comparison and I could just hear
some similar, very low-level noises in the background that I'd not
really noticed before.
As it is, the '5680A-based reference is unusable with the N5AC
synthesizer and its wide loop bandwidth and "almost" usable with my
homebrew transverter and its comparatively narrow loop bandwidth. I'm
now bent on making the '5680A usable as a microwave reference, but my
current plans are to build a simple 10 MHz Butler VCXO and then lock it
to the '5680A using a very "slow" loop filter: In that way, I'm hoping
that the phase noise will be largely that of the 10 MHz VCXO and its
cheap CPU-type crystal rather than the '5680A!
Clint
KA7OEI
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
El 09/02/2012 01:40, Bob Camp escribió:
Hi
Here is a little more on how much of a problem you have.
If you would like the spurs to be down 70 dbc at 10 GHz. They go up by 20 log N. in this case N is 1000. That gets you 60 db. Spurs at 10 MHz would have to be down at -130 dbc to make it at 10 GHz.
As a side note (since I've just asked by), does the phase noise also go
up by 20 log N when multiplying?
Regards,
Javier
Hi
Indeed the phase noise goes up by 20 log N when multiplying. To be precise, only the phase spurs go up by 20 log N. If they are incoherent they still multiply, but you get a bit of an offset.
Bob
On Feb 9, 2012, at 2:03 AM, Javier Herrero wrote:
El 09/02/2012 01:40, Bob Camp escribió:
Hi
Here is a little more on how much of a problem you have.
If you would like the spurs to be down 70 dbc at 10 GHz. They go up by 20 log N. in this case N is 1000. That gets you 60 db. Spurs at 10 MHz would have to be down at -130 dbc to make it at 10 GHz.
As a side note (since I've just asked by), does the phase noise also go up by 20 log N when multiplying?
Regards,
Javier
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.