One technique to improve the phase noise in wide-band VCO's was shown by
Ulrich Rhode. He uses several diodes in parallel in order to decrease the
RF
current in each diode, reportedly lowering losses and improving noise. I
never tried it myself.

If the diodes are the limiting factor, it can help.
The second diode seemed to do the most good,
the third, a little more, but any more didn't help
(in my designs).

The way the varactor is made will affect the
noise performance. Ask the manufacturers
which process yields their best phase-noise
performance. Most of them are very helpful.
As I recall, planar, epitaxial construction,
with thermal compression bonded leads
gave my best results. Hyperabrupt diodes
gave the worst. But things change rapidly
these days, so ask.

73, John - K6QQ

On Fri, 9 Jan 2004 16:31:56 -0800, "John Moriarity"
wrote:

One technique to improve the phase noise in wide-band VCO's was shown by
Ulrich Rhode. He uses several diodes in parallel in order to decrease the RF
current in each diode, reportedly lowering losses and improving noise. I
never tried it myself.

Or is it simply that when using multiple varactors in parallel, the
amount of capacitance needed from each varactor is reduced, hence a
higher tuning voltage must be used with the same inductance ? With a
high tuning voltage, the capacitance/voltage ratio is smaller, hence
the capacitance difference would be smaller on the separate RF half
cycles. This would reduce the phase noise.

If the diodes are the limiting factor, it can help.
The second diode seemed to do the most good,
the third, a little more, but any more didn't help
(in my designs).

Your observations would support my theory, since adding further
varactors would only increase the tuning voltage slightly with the
same inductance.

Paul OH3LWR

On Fri, 9 Jan 2004 16:31:56 -0800, "John Moriarity"
wrote:

One technique to improve the phase noise in wide-band VCO's was shown by
Ulrich Rhode. He uses several diodes in parallel in order to decrease the RF
current in each diode, reportedly lowering losses and improving noise. I
never tried it myself.

Or is it simply that when using multiple varactors in parallel, the
amount of capacitance needed from each varactor is reduced, hence a
higher tuning voltage must be used with the same inductance ? With a
high tuning voltage, the capacitance/voltage ratio is smaller, hence
the capacitance difference would be smaller on the separate RF half
cycles. This would reduce the phase noise.

If the diodes are the limiting factor, it can help.
The second diode seemed to do the most good,
the third, a little more, but any more didn't help
(in my designs).

Your observations would support my theory, since adding further
varactors would only increase the tuning voltage slightly with the
same inductance.

Paul OH3LWR

In article , "W3JDR"
writes:

Steve,
I think the main reason for back-to-back diodes is to prevent rectification
of the RF. Rectification can cause several bad things, including pushing DC
current back out the tuning voltage line, instability, and increased phase
noise. Having said that, I tried back-to-back diodes a couple times and I
don't ever recall ending up with it in the final design, so it must not have
added all that much value. On the negative side, it halves the capacitance.
One technique to improve the phase noise in wide-band VCO's was shown by
Ulrich Rhode. He uses several diodes in parallel in order to decrease the RF
current in each diode, reportedly lowering losses and improving noise. I
never tried it myself.

Variable-capacitance diodes exhibit that phenomenon only in
reverse-DC-bias connections. In forward-DC-bias connection they
behave generally like ordinary diodes, conducting very little current
until about 0.6 V across the junction.

A single variable-capacitance diode across an RF circuit will conduct
- and thus have an effect on the RF tuned circuit - when the
combination of DC tuning bias and RF voltages are above the
forward-conduction breakpoint.

The purpose of "back-to-back" connection is to keep the (now two)
variable-capacitance diodes always in reverse-conduction...the RF
voltage (peak-to-peak) is not supposed to exceed either the break-
down voltage of the diodes or cause either of them to be forward-
biased during any part of the RF cycle.

With no forward conduction, the variable capacitance diodes remain
just that - variable capacitances.

When forward conduction occurs, it adds more non-linearity to the
RF circuit and tends to decrease the action of the variable capacitance.

When used in low-level RF stages of a receiver input, the RF voltage
hardly ever exceeds 1 V peak-to-peak and thus the variable
capacitance diode never goes into forward conduction. Single diodes
can be used there without doubling-up.

Len Anderson
retired (from regular hours) electronic engineer person

In article , "W3JDR"
writes:

Steve,
I think the main reason for back-to-back diodes is to prevent rectification
of the RF. Rectification can cause several bad things, including pushing DC
current back out the tuning voltage line, instability, and increased phase
noise. Having said that, I tried back-to-back diodes a couple times and I
don't ever recall ending up with it in the final design, so it must not have
added all that much value. On the negative side, it halves the capacitance.
One technique to improve the phase noise in wide-band VCO's was shown by
Ulrich Rhode. He uses several diodes in parallel in order to decrease the RF
current in each diode, reportedly lowering losses and improving noise. I
never tried it myself.

Variable-capacitance diodes exhibit that phenomenon only in
reverse-DC-bias connections. In forward-DC-bias connection they
behave generally like ordinary diodes, conducting very little current
until about 0.6 V across the junction.

A single variable-capacitance diode across an RF circuit will conduct
- and thus have an effect on the RF tuned circuit - when the
combination of DC tuning bias and RF voltages are above the
forward-conduction breakpoint.

The purpose of "back-to-back" connection is to keep the (now two)
variable-capacitance diodes always in reverse-conduction...the RF
voltage (peak-to-peak) is not supposed to exceed either the break-
down voltage of the diodes or cause either of them to be forward-
biased during any part of the RF cycle.

With no forward conduction, the variable capacitance diodes remain
just that - variable capacitances.

When forward conduction occurs, it adds more non-linearity to the
RF circuit and tends to decrease the action of the variable capacitance.

When used in low-level RF stages of a receiver input, the RF voltage
hardly ever exceeds 1 V peak-to-peak and thus the variable
capacitance diode never goes into forward conduction. Single diodes
can be used there without doubling-up.

Len Anderson
retired (from regular hours) electronic engineer person

On Sat, 10 Jan 2004 12:50:46 +0200, Paul Keinanen
wrote:

On Fri, 9 Jan 2004 16:31:56 -0800, "John Moriarity"
wrote:

One technique to improve the phase noise in wide-band VCO's was shown by
Ulrich Rhode. He uses several diodes in parallel in order to decrease the RF
current in each diode, reportedly lowering losses and improving noise. I
never tried it myself.

Or is it simply that when using multiple varactors in parallel, the
amount of capacitance needed from each varactor is reduced, hence a
higher tuning voltage must be used with the same inductance ? With a
high tuning voltage, the capacitance/voltage ratio is smaller, hence
the capacitance difference would be smaller on the separate RF half
cycles. This would reduce the phase noise.

If the diodes are the limiting factor, it can help.
The second diode seemed to do the most good,
the third, a little more, but any more didn't help
(in my designs).

Your observations would support my theory, since adding further
varactors would only increase the tuning voltage slightly with the
same inductance.

Paul OH3LWR


suppose the anti-paralell configuration could improve it further like
http://home.online.no/~la8ak/images/1et27.gif , first got the idea
from high level PIN switch in Electronics design, but have not seen
the application elsewhere

JM
http://home.online.no/~la8ak/L2.htm
--
Amount of SPAM is so large that MailWasher must delete 99% of the incoming mails
Cannot check every email manually. Please use intelligent title for email.
Mails without titles or using just "hi" is deleted

On Sat, 10 Jan 2004 12:50:46 +0200, Paul Keinanen
wrote:

On Fri, 9 Jan 2004 16:31:56 -0800, "John Moriarity"
wrote:

One technique to improve the phase noise in wide-band VCO's was shown by
Ulrich Rhode. He uses several diodes in parallel in order to decrease the RF
current in each diode, reportedly lowering losses and improving noise. I
never tried it myself.

Or is it simply that when using multiple varactors in parallel, the
amount of capacitance needed from each varactor is reduced, hence a
higher tuning voltage must be used with the same inductance ? With a
high tuning voltage, the capacitance/voltage ratio is smaller, hence
the capacitance difference would be smaller on the separate RF half
cycles. This would reduce the phase noise.

If the diodes are the limiting factor, it can help.
The second diode seemed to do the most good,
the third, a little more, but any more didn't help
(in my designs).

Your observations would support my theory, since adding further
varactors would only increase the tuning voltage slightly with the
same inductance.

Paul OH3LWR


suppose the anti-paralell configuration could improve it further like
http://home.online.no/~la8ak/images/1et27.gif , first got the idea
from high level PIN switch in Electronics design, but have not seen
the application elsewhere

JM
http://home.online.no/~la8ak/L2.htm
--
Amount of SPAM is so large that MailWasher must delete 99% of the incoming mails
Cannot check every email manually. Please use intelligent title for email.
Mails without titles or using just "hi" is deleted

Ok, many of you have suggested rather than just simulation I should actually
build something. Joe (W3JDR) has kindly donated a circuit diagram which I
started off with.

Best way to get going quickly is dead bug style, soldered straight to the
copper side of a PCB. I have never used this style of construction as it
looked flimsy, but it looked like a good way to go just for prototyping.
Well, the circuit went together fast and the finished product is very
steady, so I am now a convert to this construction method! I may need a new
iron though, as the basic low power model I have struggled to heat up the
large area of board.

Firstly I built the circuit (almost) as directed, slight modifications were
made to suit parts I had. Fixed capacitors were used as this allowed
evaluation without worrying about the diodes performance. The oscillation
voltage across the tank was about 1 vp-p measured with an older 50MHz CRO
and a 1:1 probe. Oscillation seemed stable and I liked the result.

Question: I used the ARRL equation for a cylindrical inductor and seemed way
off (50%). Should I expect the copper sheet to affect the inductance, and
how.

I then replaced the fixed capacitors with BB909 (3-30pF) diodes. The result
wasn't so good. Oscillation voltage was less (lower Q tank maybe) and the
stablility seemed poor (again poor Q). I will attempt more measurements
tonight.

James.

Ok, many of you have suggested rather than just simulation I should actually
build something. Joe (W3JDR) has kindly donated a circuit diagram which I
started off with.

Best way to get going quickly is dead bug style, soldered straight to the
copper side of a PCB. I have never used this style of construction as it
looked flimsy, but it looked like a good way to go just for prototyping.
Well, the circuit went together fast and the finished product is very
steady, so I am now a convert to this construction method! I may need a new
iron though, as the basic low power model I have struggled to heat up the
large area of board.

Firstly I built the circuit (almost) as directed, slight modifications were
made to suit parts I had. Fixed capacitors were used as this allowed
evaluation without worrying about the diodes performance. The oscillation
voltage across the tank was about 1 vp-p measured with an older 50MHz CRO
and a 1:1 probe. Oscillation seemed stable and I liked the result.

Question: I used the ARRL equation for a cylindrical inductor and seemed way
off (50%). Should I expect the copper sheet to affect the inductance, and
how.

I then replaced the fixed capacitors with BB909 (3-30pF) diodes. The result
wasn't so good. Oscillation voltage was less (lower Q tank maybe) and the
stablility seemed poor (again poor Q). I will attempt more measurements
tonight.

James.