PT
Pluess, Tobias
Fri, Feb 24, 2023 9:00 AM
Dear all,
for a long time I had this project on my todo list. I recently started with
building my own phase noise measurement equipment.
I have a batch of brand new UCT 108663 Oscillators (new old stock
possibly?) and had the possibility to measure them a while ago on a
Keysight E5052B analyser. I now want to make my own PLL + LNA to measure
the oscillators on my own spectrum analyser.
For this, I built the circuit shown in the image. A SRA-1 mixer is used as
phase detector. Since I want to avoid a negative voltage rail, I connected
the IF centre tap to a reference voltage, such that the IF output signal
swings around VCC/2 and not around ground.
The amplifier and integrator are configured in the same way. So far, the
PLL works very nicely. By changing the resistor R3 in the feedback path of
the first amplifier, I can easily adjust the loop bandwidth. Oscillators
stay nicely in quadrature. OK, so far this works.
Now comes the interesting part. The LNA. I would like to use AD797 or
similar (ADA4897 maybe?). I wonder whether it is, from a low noise
viewpoint, possible to use the schematic as shown? I have decoupled VCC/2
with several different capacitors and in theory, this configuration of the
amplifier should work. But will it in practice yield an acceptable result?
I would like to avoid a negative voltage rail if possible, because the
tuning voltage output will need to be positive anyways, and I would like to
avoid any switching converters.
thanks for any hints,
best
Tobias
[image: image.png]
Dear all,
for a long time I had this project on my todo list. I recently started with
building my own phase noise measurement equipment.
I have a batch of brand new UCT 108663 Oscillators (new old stock
possibly?) and had the possibility to measure them a while ago on a
Keysight E5052B analyser. I now want to make my own PLL + LNA to measure
the oscillators on my own spectrum analyser.
For this, I built the circuit shown in the image. A SRA-1 mixer is used as
phase detector. Since I want to avoid a negative voltage rail, I connected
the IF centre tap to a reference voltage, such that the IF output signal
swings around VCC/2 and not around ground.
The amplifier and integrator are configured in the same way. So far, the
PLL works very nicely. By changing the resistor R3 in the feedback path of
the first amplifier, I can easily adjust the loop bandwidth. Oscillators
stay nicely in quadrature. OK, so far this works.
Now comes the interesting part. The LNA. I would like to use AD797 or
similar (ADA4897 maybe?). I wonder whether it is, from a low noise
viewpoint, possible to use the schematic as shown? I have decoupled VCC/2
with several different capacitors and in theory, this configuration of the
amplifier should work. But will it in practice yield an acceptable result?
I would like to avoid a negative voltage rail if possible, because the
tuning voltage output will need to be positive anyways, and I would like to
avoid any switching converters.
thanks for any hints,
best
Tobias
[image: image.png]
EK
Erik Kaashoek
Fri, Feb 24, 2023 11:25 AM
Hi Tobias,
I can not add wisdom on the low noise opamp but I've build a very
similar PNA, including the VCC/2 shift of the mixer output, and would
like to suggest some additions to the schematic I found useful.
1: Make the LNA gain switchable. Unity gain for determining the "gain"
of the PNA, +30 and +60 dB for real measurements.
2: Add one gain stage driven by the output of U1 that drives two leds,
one for when above VCC/2, one for when below VCC/2. These will show if
the loop is not locked
3: Inject an zero to VCC offset voltage from a potmeter through a
resistor into the min input of U1 to allow pushing the loop into lock.
Without using a negative rail I could measure down to -155 dBm/Hz,
sufficient for my needs, so I stopped improving.
Erik.
On 24-2-2023 10:00, Pluess, Tobias via time-nuts wrote:
Dear all,
for a long time I had this project on my todo list. I recently started with
building my own phase noise measurement equipment.
I have a batch of brand new UCT 108663 Oscillators (new old stock
possibly?) and had the possibility to measure them a while ago on a
Keysight E5052B analyser. I now want to make my own PLL + LNA to measure
the oscillators on my own spectrum analyser.
For this, I built the circuit shown in the image. A SRA-1 mixer is used as
phase detector. Since I want to avoid a negative voltage rail, I connected
the IF centre tap to a reference voltage, such that the IF output signal
swings around VCC/2 and not around ground.
The amplifier and integrator are configured in the same way. So far, the
PLL works very nicely. By changing the resistor R3 in the feedback path of
the first amplifier, I can easily adjust the loop bandwidth. Oscillators
stay nicely in quadrature. OK, so far this works.
Now comes the interesting part. The LNA. I would like to use AD797 or
similar (ADA4897 maybe?). I wonder whether it is, from a low noise
viewpoint, possible to use the schematic as shown? I have decoupled VCC/2
with several different capacitors and in theory, this configuration of the
amplifier should work. But will it in practice yield an acceptable result?
I would like to avoid a negative voltage rail if possible, because the
tuning voltage output will need to be positive anyways, and I would like to
avoid any switching converters.
thanks for any hints,
best
Tobias
[image: image.png]
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com
Hi Tobias,
I can not add wisdom on the low noise opamp but I've build a very
similar PNA, including the VCC/2 shift of the mixer output, and would
like to suggest some additions to the schematic I found useful.
1: Make the LNA gain switchable. Unity gain for determining the "gain"
of the PNA, +30 and +60 dB for real measurements.
2: Add one gain stage driven by the output of U1 that drives two leds,
one for when above VCC/2, one for when below VCC/2. These will show if
the loop is not locked
3: Inject an zero to VCC offset voltage from a potmeter through a
resistor into the min input of U1 to allow pushing the loop into lock.
Without using a negative rail I could measure down to -155 dBm/Hz,
sufficient for my needs, so I stopped improving.
Erik.
On 24-2-2023 10:00, Pluess, Tobias via time-nuts wrote:
> Dear all,
>
> for a long time I had this project on my todo list. I recently started with
> building my own phase noise measurement equipment.
> I have a batch of brand new UCT 108663 Oscillators (new old stock
> possibly?) and had the possibility to measure them a while ago on a
> Keysight E5052B analyser. I now want to make my own PLL + LNA to measure
> the oscillators on my own spectrum analyser.
> For this, I built the circuit shown in the image. A SRA-1 mixer is used as
> phase detector. Since I want to avoid a negative voltage rail, I connected
> the IF centre tap to a reference voltage, such that the IF output signal
> swings around VCC/2 and not around ground.
> The amplifier and integrator are configured in the same way. So far, the
> PLL works very nicely. By changing the resistor R3 in the feedback path of
> the first amplifier, I can easily adjust the loop bandwidth. Oscillators
> stay nicely in quadrature. OK, so far this works.
> Now comes the interesting part. The LNA. I would like to use AD797 or
> similar (ADA4897 maybe?). I wonder whether it is, from a low noise
> viewpoint, possible to use the schematic as shown? I have decoupled VCC/2
> with several different capacitors and in theory, this configuration of the
> amplifier should work. But will it in practice yield an acceptable result?
>
> I would like to avoid a negative voltage rail if possible, because the
> tuning voltage output will need to be positive anyways, and I would like to
> avoid any switching converters.
>
> thanks for any hints,
> best
> Tobias
>
>
> [image: image.png]
>
>
> _______________________________________________
> time-nuts mailing list -- time-nuts@lists.febo.com
> To unsubscribe send an email to time-nuts-leave@lists.febo.com
BC
Bob Camp
Fri, Feb 24, 2023 3:49 PM
Hi
It would be a good idea to add a bit of L/C based lowpass filtering between
the mixer output and the op-amp input. It does not have to be anything exotic.
Having RF get to the op-amp can create unpredictable results.
Since you are looking for noise at the < 1nV / sqrt(Hz) level, decoupling the
bias supply will be a challenge. Getting this done at low offsets ( 10Hz …. 1Hz …)
likely will be very difficult.
The classic way to check noise floor on this sort of setup is to use a hybrid
network to split one oscillator output into two quadrature signals. They then
go into the two inputs of the mixer.
If you find your noise floor isn’t quite low enough, any of the mixers in the RPD-1
family will improve things a bit.
When you go to calibrate the setup with a beat note, You will need a way to
reduce the gain on your AD797. ( = you don’t want it saturating …).
Bob
On Feb 24, 2023, at 4:00 AM, Pluess, Tobias via time-nuts time-nuts@lists.febo.com wrote:
Dear all,
for a long time I had this project on my todo list. I recently started with
building my own phase noise measurement equipment.
I have a batch of brand new UCT 108663 Oscillators (new old stock
possibly?) and had the possibility to measure them a while ago on a
Keysight E5052B analyser. I now want to make my own PLL + LNA to measure
the oscillators on my own spectrum analyser.
For this, I built the circuit shown in the image. A SRA-1 mixer is used as
phase detector. Since I want to avoid a negative voltage rail, I connected
the IF centre tap to a reference voltage, such that the IF output signal
swings around VCC/2 and not around ground.
The amplifier and integrator are configured in the same way. So far, the
PLL works very nicely. By changing the resistor R3 in the feedback path of
the first amplifier, I can easily adjust the loop bandwidth. Oscillators
stay nicely in quadrature. OK, so far this works.
Now comes the interesting part. The LNA. I would like to use AD797 or
similar (ADA4897 maybe?). I wonder whether it is, from a low noise
viewpoint, possible to use the schematic as shown? I have decoupled VCC/2
with several different capacitors and in theory, this configuration of the
amplifier should work. But will it in practice yield an acceptable result?
I would like to avoid a negative voltage rail if possible, because the
tuning voltage output will need to be positive anyways, and I would like to
avoid any switching converters.
thanks for any hints,
best
Tobias
[image: image.png]
<image.png>_______________________________________________
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com
Hi
It would be a good idea to add a bit of L/C based lowpass filtering between
the mixer output and the op-amp input. It does not have to be anything exotic.
Having RF get to the op-amp can create unpredictable results.
Since you are looking for noise at the < 1nV / sqrt(Hz) level, decoupling the
bias supply will be a challenge. Getting this done at low offsets ( 10Hz …. 1Hz …)
likely will be very difficult.
The classic way to check noise floor on this sort of setup is to use a hybrid
network to split one oscillator output into two quadrature signals. They then
go into the two inputs of the mixer.
If you find your noise floor isn’t quite low enough, any of the mixers in the RPD-1
family will improve things a bit.
When you go to calibrate the setup with a beat note, You will need a way to
reduce the gain on your AD797. ( = you don’t want it saturating …).
Bob
> On Feb 24, 2023, at 4:00 AM, Pluess, Tobias via time-nuts <time-nuts@lists.febo.com> wrote:
>
> Dear all,
>
> for a long time I had this project on my todo list. I recently started with
> building my own phase noise measurement equipment.
> I have a batch of brand new UCT 108663 Oscillators (new old stock
> possibly?) and had the possibility to measure them a while ago on a
> Keysight E5052B analyser. I now want to make my own PLL + LNA to measure
> the oscillators on my own spectrum analyser.
> For this, I built the circuit shown in the image. A SRA-1 mixer is used as
> phase detector. Since I want to avoid a negative voltage rail, I connected
> the IF centre tap to a reference voltage, such that the IF output signal
> swings around VCC/2 and not around ground.
> The amplifier and integrator are configured in the same way. So far, the
> PLL works very nicely. By changing the resistor R3 in the feedback path of
> the first amplifier, I can easily adjust the loop bandwidth. Oscillators
> stay nicely in quadrature. OK, so far this works.
> Now comes the interesting part. The LNA. I would like to use AD797 or
> similar (ADA4897 maybe?). I wonder whether it is, from a low noise
> viewpoint, possible to use the schematic as shown? I have decoupled VCC/2
> with several different capacitors and in theory, this configuration of the
> amplifier should work. But will it in practice yield an acceptable result?
>
> I would like to avoid a negative voltage rail if possible, because the
> tuning voltage output will need to be positive anyways, and I would like to
> avoid any switching converters.
>
> thanks for any hints,
> best
> Tobias
>
>
> [image: image.png]
> <image.png>_______________________________________________
> time-nuts mailing list -- time-nuts@lists.febo.com
> To unsubscribe send an email to time-nuts-leave@lists.febo.com
JM
Jim Muehlberg
Fri, Feb 24, 2023 3:57 PM
Tobias,
Maybe you have seen Charles Wenzel's "teclib.com" site. Lots of fun
stuff there. Including a design for a LNA for precisely what you want:
http://techlib.com/files/lowamp.pdf
I built this about 15 years ago. There are probably better LN op amps
out there now.
Take a look at this guy: (Marco Reps)
https://youtu.be/XpbDMo8an5w
He's a "voltnut", but that's OK. We're all nuts.
Oh, and certainly, you need to read Linear application note #124 by Jim
Williams.
Good Luck!
Jim
On 2023-02-24 4:00 AM, Pluess, Tobias via time-nuts wrote:
Dear all,
for a long time I had this project on my todo list. I recently started with
building my own phase noise measurement equipment.
I have a batch of brand new UCT 108663 Oscillators (new old stock
possibly?) and had the possibility to measure them a while ago on a
Keysight E5052B analyser. I now want to make my own PLL + LNA to measure
the oscillators on my own spectrum analyser.
For this, I built the circuit shown in the image. A SRA-1 mixer is used as
phase detector. Since I want to avoid a negative voltage rail, I connected
the IF centre tap to a reference voltage, such that the IF output signal
swings around VCC/2 and not around ground.
The amplifier and integrator are configured in the same way. So far, the
PLL works very nicely. By changing the resistor R3 in the feedback path of
the first amplifier, I can easily adjust the loop bandwidth. Oscillators
stay nicely in quadrature. OK, so far this works.
Now comes the interesting part. The LNA. I would like to use AD797 or
similar (ADA4897 maybe?). I wonder whether it is, from a low noise
viewpoint, possible to use the schematic as shown? I have decoupled VCC/2
with several different capacitors and in theory, this configuration of the
amplifier should work. But will it in practice yield an acceptable result?
I would like to avoid a negative voltage rail if possible, because the
tuning voltage output will need to be positive anyways, and I would like to
avoid any switching converters.
thanks for any hints,
best
Tobias
[image: image.png]
time-nuts mailing list --time-nuts@lists.febo.com
To unsubscribe send an email totime-nuts-leave@lists.febo.com
--
Jim Muehlberg
Senior Engineer
National Radio Astronomy Observatory
ngVLA Local Oscillator Lead
1180 Boxwood Estates Rd B-111
Charlottesville, VA 22903-4602
P 434.296.0270
C 434.422.2017
Tobias,
Maybe you have seen Charles Wenzel's "teclib.com" site. Lots of fun
stuff there. Including a design for a LNA for precisely what you want:
http://techlib.com/files/lowamp.pdf
I built this about 15 years ago. There are probably better LN op amps
out there now.
Take a look at this guy: (Marco Reps)
https://youtu.be/XpbDMo8an5w
He's a "voltnut", but that's OK. We're all nuts.
Oh, and certainly, you need to read Linear application note #124 by Jim
Williams.
Good Luck!
Jim
On 2023-02-24 4:00 AM, Pluess, Tobias via time-nuts wrote:
> Dear all,
>
> for a long time I had this project on my todo list. I recently started with
> building my own phase noise measurement equipment.
> I have a batch of brand new UCT 108663 Oscillators (new old stock
> possibly?) and had the possibility to measure them a while ago on a
> Keysight E5052B analyser. I now want to make my own PLL + LNA to measure
> the oscillators on my own spectrum analyser.
> For this, I built the circuit shown in the image. A SRA-1 mixer is used as
> phase detector. Since I want to avoid a negative voltage rail, I connected
> the IF centre tap to a reference voltage, such that the IF output signal
> swings around VCC/2 and not around ground.
> The amplifier and integrator are configured in the same way. So far, the
> PLL works very nicely. By changing the resistor R3 in the feedback path of
> the first amplifier, I can easily adjust the loop bandwidth. Oscillators
> stay nicely in quadrature. OK, so far this works.
> Now comes the interesting part. The LNA. I would like to use AD797 or
> similar (ADA4897 maybe?). I wonder whether it is, from a low noise
> viewpoint, possible to use the schematic as shown? I have decoupled VCC/2
> with several different capacitors and in theory, this configuration of the
> amplifier should work. But will it in practice yield an acceptable result?
>
> I would like to avoid a negative voltage rail if possible, because the
> tuning voltage output will need to be positive anyways, and I would like to
> avoid any switching converters.
>
> thanks for any hints,
> best
> Tobias
>
>
> [image: image.png]
>
>
> _______________________________________________
> time-nuts mailing list --time-nuts@lists.febo.com
> To unsubscribe send an email totime-nuts-leave@lists.febo.com
--
Jim Muehlberg
Senior Engineer
National Radio Astronomy Observatory
ngVLA Local Oscillator Lead
1180 Boxwood Estates Rd B-111
Charlottesville, VA 22903-4602
P 434.296.0270
C 434.422.2017
LJ
Lux, Jim
Fri, Feb 24, 2023 7:39 PM
On 2/24/23 7:57 AM, Jim Muehlberg via time-nuts wrote:
Tobias,
Maybe you have seen Charles Wenzel's "teclib.com" site. Lots of fun
stuff there. Including a design for a LNA for precisely what you want:
http://techlib.com/files/lowamp.pdf
I built this about 15 years ago. There are probably better LN op amps
out there now.
Take a look at this guy: (Marco Reps)
https://youtu.be/XpbDMo8an5w
He's a "voltnut", but that's OK. We're all nuts.
Oh, and certainly, you need to read Linear application note #124 by
Jim Williams.
Good Luck!
Jim
There's a variety of very low noise current mode amplifiers, but they
are challenging to keep stable, because they have enormous GBW products
(1 GHz). And designing with current feedback is tricky - you need a
certain Rin and Rfb, but they add Johnson noise.
example ADA4927 1.4nV/sqrt(Hz), 14 pA/sqrt(Hz) at G=28, 100kHz -
estimating off the plot in the data sheet, the 1/f knee is around 10-20kHz
There's also the ADA 4817A 4nV/sqrt(Hz) 2.5 fA/sqrt(Hz) - this one
looks like it's 1/f is a bit lower. It's also challenging in low gain
applications.
On 2/24/23 7:57 AM, Jim Muehlberg via time-nuts wrote:
> Tobias,
>
> Maybe you have seen Charles Wenzel's "teclib.com" site. Lots of fun
> stuff there. Including a design for a LNA for precisely what you want:
>
> http://techlib.com/files/lowamp.pdf
>
> I built this about 15 years ago. There are probably better LN op amps
> out there now.
>
> Take a look at this guy: (Marco Reps)
>
> https://youtu.be/XpbDMo8an5w
>
> He's a "voltnut", but that's OK. We're all nuts.
>
> Oh, and certainly, you need to read Linear application note #124 by
> Jim Williams.
>
> Good Luck!
>
> Jim
There's a variety of very low noise current mode amplifiers, but they
are challenging to keep stable, because they have enormous GBW products
(1 GHz). And designing with current feedback is tricky - you need a
certain Rin and Rfb, but they add Johnson noise.
example ADA4927 1.4nV/sqrt(Hz), 14 pA/sqrt(Hz) at G=28, 100kHz -
estimating off the plot in the data sheet, the 1/f knee is around 10-20kHz
There's also the ADA 4817A 4nV/sqrt(Hz) 2.5 fA/sqrt(Hz) - this one
looks like it's 1/f is a bit lower. It's also challenging in low gain
applications.
BC
Bob Camp
Sat, Feb 25, 2023 12:13 AM
Hi
For a phase noise test set with the amp located at the mixer
output, you need good noise at a 1,10,100Hz sort of test point.
The current mode stuff is great for things like distribution amps
or for buffers driving the mixer. Indeed, a buffer into the mixer
is a pretty good idea.
Bob
On Feb 24, 2023, at 2:39 PM, Lux, Jim via time-nuts time-nuts@lists.febo.com wrote:
On 2/24/23 7:57 AM, Jim Muehlberg via time-nuts wrote:
Tobias,
Maybe you have seen Charles Wenzel's "teclib.com" site. Lots of fun stuff there. Including a design for a LNA for precisely what you want:
http://techlib.com/files/lowamp.pdf
I built this about 15 years ago. There are probably better LN op amps out there now.
Take a look at this guy: (Marco Reps)
https://youtu.be/XpbDMo8an5w
He's a "voltnut", but that's OK. We're all nuts.
Oh, and certainly, you need to read Linear application note #124 by Jim Williams.
Good Luck!
Jim
There's a variety of very low noise current mode amplifiers, but they are challenging to keep stable, because they have enormous GBW products (1 GHz). And designing with current feedback is tricky - you need a certain Rin and Rfb, but they add Johnson noise.
example ADA4927 1.4nV/sqrt(Hz), 14 pA/sqrt(Hz) at G=28, 100kHz - estimating off the plot in the data sheet, the 1/f knee is around 10-20kHz
There's also the ADA 4817A 4nV/sqrt(Hz) 2.5 fA/sqrt(Hz) - this one looks like it's 1/f is a bit lower. It's also challenging in low gain applications.
time-nuts mailing list -- time-nuts@lists.febo.com
To unsubscribe send an email to time-nuts-leave@lists.febo.com
Hi
For a phase noise test set with the amp located at the mixer
output, you need good noise at a 1,10,100Hz sort of test point.
The current mode stuff is great for things like distribution amps
or for buffers driving the mixer. Indeed, a buffer into the mixer
is a pretty good idea.
Bob
> On Feb 24, 2023, at 2:39 PM, Lux, Jim via time-nuts <time-nuts@lists.febo.com> wrote:
>
> On 2/24/23 7:57 AM, Jim Muehlberg via time-nuts wrote:
>> Tobias,
>>
>> Maybe you have seen Charles Wenzel's "teclib.com" site. Lots of fun stuff there. Including a design for a LNA for precisely what you want:
>>
>> http://techlib.com/files/lowamp.pdf
>>
>> I built this about 15 years ago. There are probably better LN op amps out there now.
>>
>> Take a look at this guy: (Marco Reps)
>>
>> https://youtu.be/XpbDMo8an5w
>>
>> He's a "voltnut", but that's OK. We're all nuts.
>>
>> Oh, and certainly, you need to read Linear application note #124 by Jim Williams.
>>
>> Good Luck!
>>
>> Jim
>
>
> There's a variety of very low noise current mode amplifiers, but they are challenging to keep stable, because they have enormous GBW products (1 GHz). And designing with current feedback is tricky - you need a certain Rin and Rfb, but they add Johnson noise.
>
> example ADA4927 1.4nV/sqrt(Hz), 14 pA/sqrt(Hz) at G=28, 100kHz - estimating off the plot in the data sheet, the 1/f knee is around 10-20kHz
>
> There's also the ADA 4817A 4nV/sqrt(Hz) 2.5 fA/sqrt(Hz) - this one looks like it's 1/f is a bit lower. It's also challenging in low gain applications.
>
> _______________________________________________
> time-nuts mailing list -- time-nuts@lists.febo.com
> To unsubscribe send an email to time-nuts-leave@lists.febo.com
