Thanks to everyone who replied. I've now discovered the reason for the
disastrous noise on the Mini-Circuits VCO: Obviously I was using *extremely*
good RF PCB design techniques/layout, but these devices need even better
than the very best. They are abnormally sensitive - not so much to noise on
Vtune (20V tuning devices are less sensitive than the 5V wideband types) -
but to earthing - much more than normal - and much more than Mini-Circuits
own suggested PCB layout for the device suggests. And despite the fact that
the tiny metal can is mostly directly bonded earth pins. It's not feasible
how bad the results can be, even on a very good RF board! Apparently, the
quoted -75dBc phase noise figure is almost unattainable, except using
Mini-Circuits own metal block test jig.

I've spent a week questioning myself and my Analyzer, so I hope this might
help others to get acceptable results using this device. Use vias
everywhere, solder the can all round, really go over-the-top on everything,
in addition to all the normal RF grounding/decoupling practices, and then
you might just get barely acceptable results.

Al.
---

The problem is the tuning voltage, and how sensitive it is. This applies to
all wide-range VCO's.
I can't find the original message, but here's an example...

Assume VCO tunes 1GHz to 2 GHZ with 20 volts of tuning voltage. While the
tuning voltage
is not linear, simply to linear assumption. Then the sensitivity of the
tuning line is 1 GHz / 20 volts,
this is the same as 1 KHz / 20 microvolts. It's really hard to get noise
levels that low! Practically,
the entire VCO as well as the circuit that drives the tuning line and all
the componenets therein
have to be inside a well shielded enclosure. Further, extraordinary care has
to be taken in grounding.
This means figuring out all the noise sources, ground return paths, etc. You
cannot just hook everything
to a groundplane, that's nowhere near good enough. It's necessary to group
'vias' to the ground plane
based on which return currents they share, and then wire each ground
connection or group of
connections to the exact specific via needed. One cm of groundplane can have
millivolts of potential
difference across it.

-- Tom




"Alex Collins" wrote in message
...
Thanks to everyone who replied. I've now discovered the reason for the
disastrous noise on the Mini-Circuits VCO: Obviously I was using
*extremely*
good RF PCB design techniques/layout, but these devices need even better
than the very best. They are abnormally sensitive - not so much to noise
on
Vtune (20V tuning devices are less sensitive than the 5V wideband types) -
but to earthing - much more than normal - and much more than Mini-Circuits
own suggested PCB layout for the device suggests. And despite the fact
that
the tiny metal can is mostly directly bonded earth pins. It's not feasible
how bad the results can be, even on a very good RF board! Apparently, the
quoted -75dBc phase noise figure is almost unattainable, except using
Mini-Circuits own metal block test jig.

I've spent a week questioning myself and my Analyzer, so I hope this might
help others to get acceptable results using this device. Use vias
everywhere, solder the can all round, really go over-the-top on
everything,
in addition to all the normal RF grounding/decoupling practices, and then
you might just get barely acceptable results.

Al.
---

The problem is the tuning voltage, and how sensitive it is. This applies to
all wide-range VCO's.
I can't find the original message, but here's an example...

Assume VCO tunes 1GHz to 2 GHZ with 20 volts of tuning voltage. While the
tuning voltage
is not linear, simply to linear assumption. Then the sensitivity of the
tuning line is 1 GHz / 20 volts,
this is the same as 1 KHz / 20 microvolts. It's really hard to get noise
levels that low! Practically,
the entire VCO as well as the circuit that drives the tuning line and all
the componenets therein
have to be inside a well shielded enclosure. Further, extraordinary care has
to be taken in grounding.
This means figuring out all the noise sources, ground return paths, etc. You
cannot just hook everything
to a groundplane, that's nowhere near good enough. It's necessary to group
'vias' to the ground plane
based on which return currents they share, and then wire each ground
connection or group of
connections to the exact specific via needed. One cm of groundplane can have
millivolts of potential
difference across it.

-- Tom




"Alex Collins" wrote in message
...
Thanks to everyone who replied. I've now discovered the reason for the
disastrous noise on the Mini-Circuits VCO: Obviously I was using
*extremely*
good RF PCB design techniques/layout, but these devices need even better
than the very best. They are abnormally sensitive - not so much to noise
on
Vtune (20V tuning devices are less sensitive than the 5V wideband types) -
but to earthing - much more than normal - and much more than Mini-Circuits
own suggested PCB layout for the device suggests. And despite the fact
that
the tiny metal can is mostly directly bonded earth pins. It's not feasible
how bad the results can be, even on a very good RF board! Apparently, the
quoted -75dBc phase noise figure is almost unattainable, except using
Mini-Circuits own metal block test jig.

I've spent a week questioning myself and my Analyzer, so I hope this might
help others to get acceptable results using this device. Use vias
everywhere, solder the can all round, really go over-the-top on
everything,
in addition to all the normal RF grounding/decoupling practices, and then
you might just get barely acceptable results.

Al.
---

In article , says...
Thanks to everyone who replied. I've now discovered the reason for the
disastrous noise on the Mini-Circuits VCO: Obviously I was using *extremely*
good RF PCB design techniques/layout, but these devices need even better
than the very best. They are abnormally sensitive - not so much to noise on
Vtune (20V tuning devices are less sensitive than the 5V wideband types) -
but to earthing - much more than normal - and much more than Mini-Circuits
own suggested PCB layout for the device suggests. And despite the fact that
the tiny metal can is mostly directly bonded earth pins. It's not feasible
how bad the results can be, even on a very good RF board! Apparently, the
quoted -75dBc phase noise figure is almost unattainable, except using
Mini-Circuits own metal block test jig.

Did you read my post about phase-locking it at a low loop bandwidth
before trying to measure PN?

I've spent a week questioning myself and my Analyzer, so I hope this might
help others to get acceptable results using this device. Use vias
everywhere, solder the can all round, really go over-the-top on everything,
in addition to all the normal RF grounding/decoupling practices, and then
you might just get barely acceptable results.

I don't think measuring -75 dBc/Hz at 1 kHz requires superhuman effort.
It does require an understanding of the measurement process and the
nature of the device being characterized, though.

-- jm

------------------------------------------------------
http://www.qsl.net/ke5fx
Note: My E-mail address has been altered to avoid spam
------------------------------------------------------

In article , says...
Thanks to everyone who replied. I've now discovered the reason for the
disastrous noise on the Mini-Circuits VCO: Obviously I was using *extremely*
good RF PCB design techniques/layout, but these devices need even better
than the very best. They are abnormally sensitive - not so much to noise on
Vtune (20V tuning devices are less sensitive than the 5V wideband types) -
but to earthing - much more than normal - and much more than Mini-Circuits
own suggested PCB layout for the device suggests. And despite the fact that
the tiny metal can is mostly directly bonded earth pins. It's not feasible
how bad the results can be, even on a very good RF board! Apparently, the
quoted -75dBc phase noise figure is almost unattainable, except using
Mini-Circuits own metal block test jig.

Did you read my post about phase-locking it at a low loop bandwidth
before trying to measure PN?

I've spent a week questioning myself and my Analyzer, so I hope this might
help others to get acceptable results using this device. Use vias
everywhere, solder the can all round, really go over-the-top on everything,
in addition to all the normal RF grounding/decoupling practices, and then
you might just get barely acceptable results.

I don't think measuring -75 dBc/Hz at 1 kHz requires superhuman effort.
It does require an understanding of the measurement process and the
nature of the device being characterized, though.

-- jm

------------------------------------------------------
http://www.qsl.net/ke5fx
Note: My E-mail address has been altered to avoid spam
------------------------------------------------------

On Fri, 7 May 2004 07:55:14 +0100, "Alex Collins" wrote:

Thanks to everyone who replied. I've now discovered the reason for the
disastrous noise on the Mini-Circuits VCO: Obviously I was using *extremely*
good RF PCB design techniques/layout, but these devices need even better
than the very best. They are abnormally sensitive - not so much to noise on
Vtune (20V tuning devices are less sensitive than the 5V wideband types) -
but to earthing - much more than normal - and much more than Mini-Circuits
own suggested PCB layout for the device suggests. And despite the fact that
the tiny metal can is mostly directly bonded earth pins. It's not feasible
how bad the results can be, even on a very good RF board! Apparently, the
quoted -75dBc phase noise figure is almost unattainable, except using
Mini-Circuits own metal block test jig.

I've spent a week questioning myself and my Analyzer, so I hope this might
help others to get acceptable results using this device. Use vias
everywhere, solder the can all round, really go over-the-top on everything,
in addition to all the normal RF grounding/decoupling practices, and then
you might just get barely acceptable results.

Al.
---



Could be the analyzer is noisey...

Try measuring additive phase noise with two VCOs at the same time using a phase
shifter to calculate the detector constant. subtract 3dB from your measurement.







Remove "HeadFromButt", before replying by email.

On Fri, 7 May 2004 07:55:14 +0100, "Alex Collins" wrote:

Thanks to everyone who replied. I've now discovered the reason for the
disastrous noise on the Mini-Circuits VCO: Obviously I was using *extremely*
good RF PCB design techniques/layout, but these devices need even better
than the very best. They are abnormally sensitive - not so much to noise on
Vtune (20V tuning devices are less sensitive than the 5V wideband types) -
but to earthing - much more than normal - and much more than Mini-Circuits
own suggested PCB layout for the device suggests. And despite the fact that
the tiny metal can is mostly directly bonded earth pins. It's not feasible
how bad the results can be, even on a very good RF board! Apparently, the
quoted -75dBc phase noise figure is almost unattainable, except using
Mini-Circuits own metal block test jig.

I've spent a week questioning myself and my Analyzer, so I hope this might
help others to get acceptable results using this device. Use vias
everywhere, solder the can all round, really go over-the-top on everything,
in addition to all the normal RF grounding/decoupling practices, and then
you might just get barely acceptable results.

Al.
---



Could be the analyzer is noisey...

Try measuring additive phase noise with two VCOs at the same time using a phase
shifter to calculate the detector constant. subtract 3dB from your measurement.







Remove "HeadFromButt", before replying by email.

On Fri, 07 May 2004 22:02:42 GMT, eatitcold wrote:

On Fri, 7 May 2004 07:55:14 +0100, "Alex Collins"
wrote:

Thanks to everyone who replied. I've now discovered the reason for the
disastrous noise on the Mini-Circuits VCO: Obviously I was using *extremely*
good RF PCB design techniques/layout, but these devices need even better
than the very best. They are abnormally sensitive - not so much to noise on
Vtune (20V tuning devices are less sensitive than the 5V wideband types) -
but to earthing - much more than normal - and much more than Mini-Circuits
own suggested PCB layout for the device suggests. And despite the fact that
the tiny metal can is mostly directly bonded earth pins. It's not feasible
how bad the results can be, even on a very good RF board! Apparently, the
quoted -75dBc phase noise figure is almost unattainable, except using
Mini-Circuits own metal block test jig.

A place I worked at years ago had nightmare probs with VCO phase
noise.
It depended on which freqs we were using though ( Z-Comm), some were
ok, others..yeesh!
Always had to have them in a can with soldered tops & undersides,
shielding everywhere, & filters , filters, filters :-)


newbies...

Improving VCO Phase Noise Performance Through Enhanced Characterization


http://www.highfrequencyelectronics....FE0104_Vye.pdf




Remove "HeadFromButt", before replying by email.

On Fri, 07 May 2004 22:02:42 GMT, eatitcold wrote:

On Fri, 7 May 2004 07:55:14 +0100, "Alex Collins"
wrote:

Thanks to everyone who replied. I've now discovered the reason for the
disastrous noise on the Mini-Circuits VCO: Obviously I was using *extremely*
good RF PCB design techniques/layout, but these devices need even better
than the very best. They are abnormally sensitive - not so much to noise on
Vtune (20V tuning devices are less sensitive than the 5V wideband types) -
but to earthing - much more than normal - and much more than Mini-Circuits
own suggested PCB layout for the device suggests. And despite the fact that
the tiny metal can is mostly directly bonded earth pins. It's not feasible
how bad the results can be, even on a very good RF board! Apparently, the
quoted -75dBc phase noise figure is almost unattainable, except using
Mini-Circuits own metal block test jig.

A place I worked at years ago had nightmare probs with VCO phase
noise.
It depended on which freqs we were using though ( Z-Comm), some were
ok, others..yeesh!
Always had to have them in a can with soldered tops & undersides,
shielding everywhere, & filters , filters, filters :-)


newbies...

Improving VCO Phase Noise Performance Through Enhanced Characterization


http://www.highfrequencyelectronics....FE0104_Vye.pdf




Remove "HeadFromButt", before replying by email.