Switched reactances in VCO
 

In the book "Introduction to Radio Frequency Design" of W7ZOI and in
the section 7.9 "Voltage Controlled Oscillators" I read about how to
switch in and out a capacitor or an inductor in a VCO, using a diode
switch in order to alter the frequency of the oscillator.

Unfortunately this information is theoritical to me and I couldn't
make my oscillator change its frequency.

Does any one have a practical example of a Switched reactance VCO ?

Thanks
Thanasis Lazos

Thanasis wrote:

In the book "Introduction to Radio Frequency Design" of W7ZOI and in
the section 7.9 "Voltage Controlled Oscillators" I read about how to
switch in and out a capacitor or an inductor in a VCO, using a diode
switch in order to alter the frequency of the oscillator.

Unfortunately this information is theoritical to me and I couldn't
make my oscillator change its frequency.

Does any one have a practical example of a Switched reactance VCO ?

Thanks
Thanasis Lazos

I worked on the Microdyne synthesizers using this method. It used strip
line inductors, with a heavy ground buss along the edge of the board.
The diodes were either forward or reverse biased to turn them on or off.
you can't just let them float. Use a capacitor at least 100 times the
highest value of the varicap to ground on the switching end of the
diode, and as short as possible connections for the both ends of the
diode. Connect a resistor to the diode and capacitor, and feed it a
positive or negative voltage for forward or reverse bias. They have to
be turned on hard, or they will rectify the RF and cause problems, but
you can't exceed the current ratings, or you will damage the diode. we
used + or - 12 volts for switching, and the diodes were used to switch
the range at about 10% of the operating frequency. Its simple when you
see it in front of you, but it is difficult to explain with text. This
design was used in their 1200, 1400, 700, and 2800 series receivers, so
you might find someone with a manual on one of these to see how they are
built.
--


Michael A. Terrell
Central Florida

Thanasis wrote:

In the book "Introduction to Radio Frequency Design" of W7ZOI and in
the section 7.9 "Voltage Controlled Oscillators" I read about how to
switch in and out a capacitor or an inductor in a VCO, using a diode
switch in order to alter the frequency of the oscillator.

Unfortunately this information is theoritical to me and I couldn't
make my oscillator change its frequency.

Does any one have a practical example of a Switched reactance VCO ?

Thanks
Thanasis Lazos

I worked on the Microdyne synthesizers using this method. It used strip
line inductors, with a heavy ground buss along the edge of the board.
The diodes were either forward or reverse biased to turn them on or off.
you can't just let them float. Use a capacitor at least 100 times the
highest value of the varicap to ground on the switching end of the
diode, and as short as possible connections for the both ends of the
diode. Connect a resistor to the diode and capacitor, and feed it a
positive or negative voltage for forward or reverse bias. They have to
be turned on hard, or they will rectify the RF and cause problems, but
you can't exceed the current ratings, or you will damage the diode. we
used + or - 12 volts for switching, and the diodes were used to switch
the range at about 10% of the operating frequency. Its simple when you
see it in front of you, but it is difficult to explain with text. This
design was used in their 1200, 1400, 700, and 2800 series receivers, so
you might find someone with a manual on one of these to see how they are
built.
--


Michael A. Terrell
Central Florida

Thanasis wrote:
Does any one have a practical example of a Switched reactance VCO ?

The GLB 2M synthesizer of the 70's used this method.
A trimmer cap was switched in by diode to lower 24MHz VCO
to 22+ MHz on receive. 1N4148 and 5 volts. simple.
73 W7ZFB

Thanasis wrote:
Does any one have a practical example of a Switched reactance VCO ?

The GLB 2M synthesizer of the 70's used this method.
A trimmer cap was switched in by diode to lower 24MHz VCO
to 22+ MHz on receive. 1N4148 and 5 volts. simple.
73 W7ZFB

"Michael A. Terrell" wrote in message ...

They have to be turned on hard, or they will rectify the RF
and cause problems ...

I was feeding the diode with 1mA. I couldn't imagine that should feed it
with ~20mA in order to be stable my oscillator.

Thanks for your help
Thanasis Lazos

"Michael A. Terrell" wrote in message ...

They have to be turned on hard, or they will rectify the RF
and cause problems ...

I was feeding the diode with 1mA. I couldn't imagine that should feed it
with ~20mA in order to be stable my oscillator.

Thanks for your help
Thanasis Lazos

Thanasis wrote:

"Michael A. Terrell" wrote in message ...

They have to be turned on hard, or they will rectify the RF
and cause problems ...

I was feeding the diode with 1mA. I couldn't imagine that should feed it
with ~20mA in order to be stable my oscillator.

Thanks for your help
Thanasis Lazos

We had problems with phase noise at lower currents.
--


Michael A. Terrell
Central Florida

Thanasis wrote:

"Michael A. Terrell" wrote in message ...

They have to be turned on hard, or they will rectify the RF
and cause problems ...

I was feeding the diode with 1mA. I couldn't imagine that should feed it
with ~20mA in order to be stable my oscillator.

Thanks for your help
Thanasis Lazos

We had problems with phase noise at lower currents.
--


Michael A. Terrell
Central Florida

"Michael A. Terrell" wrote:

Thanasis wrote:

In the book "Introduction to Radio Frequency Design" of W7ZOI and in
the section 7.9 "Voltage Controlled Oscillators" I read about how to
switch in and out a capacitor or an inductor in a VCO, using a diode
switch in order to alter the frequency of the oscillator.

Unfortunately this information is theoritical to me and I couldn't
make my oscillator change its frequency.

Does any one have a practical example of a Switched reactance VCO ?

Thanks
Thanasis Lazos

I worked on the Microdyne synthesizers using this method. It used strip
line inductors, with a heavy ground buss along the edge of the board.
The diodes were either forward or reverse biased to turn them on or off.
you can't just let them float. Use a capacitor at least 100 times the
highest value of the varicap to ground on the switching end of the
diode, and as short as possible connections for the both ends of the
diode. Connect a resistor to the diode and capacitor, and feed it a
positive or negative voltage for forward or reverse bias. They have to
be turned on hard, or they will rectify the RF and cause problems, but
you can't exceed the current ratings, or you will damage the diode. we
used + or - 12 volts for switching, and the diodes were used to switch
the range at about 10% of the operating frequency. Its simple when you
see it in front of you, but it is difficult to explain with text. This
design was used in their 1200, 1400, 700, and 2800 series receivers, so
you might find someone with a manual on one of these to see how they are
built.
--

Michael A. Terrell
Central Florida

On the Microdynes - did you use inductors on the DC lines
to the diodes to keep the RF off the DC ?

"Michael A. Terrell" wrote:

Thanasis wrote:

In the book "Introduction to Radio Frequency Design" of W7ZOI and in
the section 7.9 "Voltage Controlled Oscillators" I read about how to
switch in and out a capacitor or an inductor in a VCO, using a diode
switch in order to alter the frequency of the oscillator.

Unfortunately this information is theoritical to me and I couldn't
make my oscillator change its frequency.

Does any one have a practical example of a Switched reactance VCO ?

Thanks
Thanasis Lazos

I worked on the Microdyne synthesizers using this method. It used strip
line inductors, with a heavy ground buss along the edge of the board.
The diodes were either forward or reverse biased to turn them on or off.
you can't just let them float. Use a capacitor at least 100 times the
highest value of the varicap to ground on the switching end of the
diode, and as short as possible connections for the both ends of the
diode. Connect a resistor to the diode and capacitor, and feed it a
positive or negative voltage for forward or reverse bias. They have to
be turned on hard, or they will rectify the RF and cause problems, but
you can't exceed the current ratings, or you will damage the diode. we
used + or - 12 volts for switching, and the diodes were used to switch
the range at about 10% of the operating frequency. Its simple when you
see it in front of you, but it is difficult to explain with text. This
design was used in their 1200, 1400, 700, and 2800 series receivers, so
you might find someone with a manual on one of these to see how they are
built.
--

Michael A. Terrell
Central Florida

On the Microdynes - did you use inductors on the DC lines
to the diodes to keep the RF off the DC ?

In article ,
(Thanasis) writes:

"Michael A. Terrell" wrote in message
...

They have to be turned on hard, or they will rectify the RF
and cause problems ...

I was feeding the diode with 1mA. I couldn't imagine that should feed it
with ~20mA in order to be stable my oscillator.

Mike Terrell had a very good suggestion on switching diodes'
current. Lots of folks have found out that higher current levels
are absolutely needed. [me for one...:-) ]

I might suggest an alternative: The CMOS switch ICs such as Dan
Tayloe uses in his excellent Mixer for DC receivers. That CMOS
structure is designed for higher signal levels with relative isolation
of internal driving circuit to the switch portion.

I haven't tried that yet but am planning on that for next year.

Len Anderson
retired (from regular hours) electronic engineer person

In article ,
(Thanasis) writes:

"Michael A. Terrell" wrote in message
...

They have to be turned on hard, or they will rectify the RF
and cause problems ...

I was feeding the diode with 1mA. I couldn't imagine that should feed it
with ~20mA in order to be stable my oscillator.

Mike Terrell had a very good suggestion on switching diodes'
current. Lots of folks have found out that higher current levels
are absolutely needed. [me for one...:-) ]

I might suggest an alternative: The CMOS switch ICs such as Dan
Tayloe uses in his excellent Mixer for DC receivers. That CMOS
structure is designed for higher signal levels with relative isolation
of internal driving circuit to the switch portion.

I haven't tried that yet but am planning on that for next year.

Len Anderson
retired (from regular hours) electronic engineer person

On Sat, 08 Nov 2003 06:08:28 GMT, wrote:


Thanks. I'll have to experiment with resistors vs inductors.
I guess I always assumed you *had* to use inductors to keep
the RF off the supply line.

There is only one problem that I can think about when using only
resistors to isolate the diode control voltage is the thermal noise
voltage generated by any resistor. Especially in VCO control voltage
lines, the thermal noise voltage generated by the resistor will add up
to the control voltage, changing the capacitance and hence generate
phase noise in the oscillator.

However, a reverse biased switching diode at 12 V should have a quite
low absolute capacitance and thus, any switching voltage noise would
have a very minimal effect on the capacitance.

Paul OH3LWR

On Sat, 08 Nov 2003 06:08:28 GMT, wrote:


Thanks. I'll have to experiment with resistors vs inductors.
I guess I always assumed you *had* to use inductors to keep
the RF off the supply line.

There is only one problem that I can think about when using only
resistors to isolate the diode control voltage is the thermal noise
voltage generated by any resistor. Especially in VCO control voltage
lines, the thermal noise voltage generated by the resistor will add up
to the control voltage, changing the capacitance and hence generate
phase noise in the oscillator.

However, a reverse biased switching diode at 12 V should have a quite
low absolute capacitance and thus, any switching voltage noise would
have a very minimal effect on the capacitance.

Paul OH3LWR

Paul Keinanen wrote:

On Sat, 08 Nov 2003 06:08:28 GMT, wrote:


Thanks. I'll have to experiment with resistors vs inductors.
I guess I always assumed you *had* to use inductors to keep
the RF off the supply line.

There is only one problem that I can think about when using only
resistors to isolate the diode control voltage is the thermal noise
voltage generated by any resistor. Especially in VCO control voltage
lines, the thermal noise voltage generated by the resistor will add up
to the control voltage, changing the capacitance and hence generate
phase noise in the oscillator.

However, a reverse biased switching diode at 12 V should have a quite
low absolute capacitance and thus, any switching voltage noise would
have a very minimal effect on the capacitance.

Paul OH3LWR


You also have a capacitor to ground at both ends of the resistor.
--


Michael A. Terrell
Central Florida

Paul Keinanen wrote:

On Sat, 08 Nov 2003 06:08:28 GMT, wrote:


Thanks. I'll have to experiment with resistors vs inductors.
I guess I always assumed you *had* to use inductors to keep
the RF off the supply line.

There is only one problem that I can think about when using only
resistors to isolate the diode control voltage is the thermal noise
voltage generated by any resistor. Especially in VCO control voltage
lines, the thermal noise voltage generated by the resistor will add up
to the control voltage, changing the capacitance and hence generate
phase noise in the oscillator.

However, a reverse biased switching diode at 12 V should have a quite
low absolute capacitance and thus, any switching voltage noise would
have a very minimal effect on the capacitance.

Paul OH3LWR


You also have a capacitor to ground at both ends of the resistor.
--


Michael A. Terrell
Central Florida

In article , Paul Keinanen
writes:

On Sat, 08 Nov 2003 06:08:28 GMT, wrote:


Thanks. I'll have to experiment with resistors vs inductors.
I guess I always assumed you *had* to use inductors to keep
the RF off the supply line.

There is only one problem that I can think about when using only
resistors to isolate the diode control voltage is the thermal noise
voltage generated by any resistor. Especially in VCO control voltage
lines, the thermal noise voltage generated by the resistor will add up
to the control voltage, changing the capacitance and hence generate
phase noise in the oscillator.

That should not normally be a problem. The basic formula for
RMS noise voltage is SQRT(4 k T Bw R) with k = 1.38 x 10^-23,
T is temperature in Kelvin, Bw is bandwidth of noise, R is Ohms
of the resistor.

Assuming a VCO control loop bandwidth of 5 KHz, temperature on
the warm side at 320 K, and a resistor value of 10 KOhms, the
RMS noise voltage would be about 0.94 microVolts.

If the VCO control voltage range is 4 VDC and the tuning range is
approximately linear over 4 MHz, that 0.94 uV will produce a phase
noise of 0.94 Hz RMS. Not a great deal...:-)

It's more than likely that stray voltage garbage in the circuit from
other sources (such as inadequately bypassed supply rails) would
be a potential problem. [pun intended]*

Len Anderson
retired (from regular hours) electronic engineer person













* when puns are outlawed, only outlaws will have puns...

In article , Paul Keinanen
writes:

On Sat, 08 Nov 2003 06:08:28 GMT, wrote:


Thanks. I'll have to experiment with resistors vs inductors.
I guess I always assumed you *had* to use inductors to keep
the RF off the supply line.

There is only one problem that I can think about when using only
resistors to isolate the diode control voltage is the thermal noise
voltage generated by any resistor. Especially in VCO control voltage
lines, the thermal noise voltage generated by the resistor will add up
to the control voltage, changing the capacitance and hence generate
phase noise in the oscillator.

That should not normally be a problem. The basic formula for
RMS noise voltage is SQRT(4 k T Bw R) with k = 1.38 x 10^-23,
T is temperature in Kelvin, Bw is bandwidth of noise, R is Ohms
of the resistor.

Assuming a VCO control loop bandwidth of 5 KHz, temperature on
the warm side at 320 K, and a resistor value of 10 KOhms, the
RMS noise voltage would be about 0.94 microVolts.

If the VCO control voltage range is 4 VDC and the tuning range is
approximately linear over 4 MHz, that 0.94 uV will produce a phase
noise of 0.94 Hz RMS. Not a great deal...:-)

It's more than likely that stray voltage garbage in the circuit from
other sources (such as inadequately bypassed supply rails) would
be a potential problem. [pun intended]*

Len Anderson
retired (from regular hours) electronic engineer person













* when puns are outlawed, only outlaws will have puns...

On 08 Nov 2003 23:35:19 GMT, (Avery Fineman)
wrote:


Assuming a VCO control loop bandwidth of 5 KHz, temperature on
the warm side at 320 K, and a resistor value of 10 KOhms, the
RMS noise voltage would be about 0.94 microVolts.

Typically the DC end of the resistor is heavily bypassed to ground,
thus the resistor is more or less in parallel with the resonant
circuit reactances. In many cases a 10 kohm resistor would heavily
load the Q of the resonant circuit and thus, a larger resistor (say 1
Mohm) might have to be used. This would increase the noise voltage ten
times.

Looking first at a free running VCO with the control signal coming
through a large resistor. The noise is generated at an infinite
bandwidth, but the varactor (and other stray) capacitances (often in
the order of 10 pF) in the resonant circuit will form a first order RC
low pass filter with the -3 dB point at 100 kHz .. 1 MHz. Above this
frequency the noise power drops by -3dB/octave, thus it contributes
the same amount of noise power as below the crossover point.

Thus, for a free running VCO, the noise voltage could be up to 10
times larger than previously assumed at 5 kHz bandwidth. With a larger
resistor, the boise voltage would be in an order of 100 uV RMS.

If the VCO is part of a PLL, the only effect is to the noise
frequencies _inside_ the loop bandwidth. The noise outside the loop
bandwidth is _not_ attenuated. The loop bandwidth can be quite narrow
in the order of 100 Hz in some channellised systems especially if a
fixed prescaler is used. The loop bandwidth could be well below 100 Hz
in VHF/UHF FM gears if the audio (and CTCSS) modulation is injected
directly into the PLL control voltage line. Thus, the PLL cleanup
operation affects only a minuscule part of the noise bandwidth.

If the VCO control voltage range is 4 VDC and the tuning range is
approximately linear over 4 MHz, that 0.94 uV will produce a phase
noise of 0.94 Hz RMS. Not a great deal...:-)

At least in many VHF/UHF systems, the VCO sensitivity could be 10
MHz/V and with a noise voltage of 100 uV, the phase noise would be 1
kHz. This would severely limit the SNR obtainable when using FM in
addition to the increased reverse mixing. These kinds of problems are
common with VHF/UHF FM gears.

However, some badly designed simple up converting general coverage HF
receivers could suffer from quite high phase noise, if a single VCO is
used without switching in reactances. In such receivers, any CW signal
would sound as it is coming through the aurora :-).

Clearly, this discussion shows that picking the right initial
assumptions, quite different conclusions can be made.

Paul OH3LWR

On 08 Nov 2003 23:35:19 GMT, (Avery Fineman)
wrote:


Assuming a VCO control loop bandwidth of 5 KHz, temperature on
the warm side at 320 K, and a resistor value of 10 KOhms, the
RMS noise voltage would be about 0.94 microVolts.

Typically the DC end of the resistor is heavily bypassed to ground,
thus the resistor is more or less in parallel with the resonant
circuit reactances. In many cases a 10 kohm resistor would heavily
load the Q of the resonant circuit and thus, a larger resistor (say 1
Mohm) might have to be used. This would increase the noise voltage ten
times.

Looking first at a free running VCO with the control signal coming
through a large resistor. The noise is generated at an infinite
bandwidth, but the varactor (and other stray) capacitances (often in
the order of 10 pF) in the resonant circuit will form a first order RC
low pass filter with the -3 dB point at 100 kHz .. 1 MHz. Above this
frequency the noise power drops by -3dB/octave, thus it contributes
the same amount of noise power as below the crossover point.

Thus, for a free running VCO, the noise voltage could be up to 10
times larger than previously assumed at 5 kHz bandwidth. With a larger
resistor, the boise voltage would be in an order of 100 uV RMS.

If the VCO is part of a PLL, the only effect is to the noise
frequencies _inside_ the loop bandwidth. The noise outside the loop
bandwidth is _not_ attenuated. The loop bandwidth can be quite narrow
in the order of 100 Hz in some channellised systems especially if a
fixed prescaler is used. The loop bandwidth could be well below 100 Hz
in VHF/UHF FM gears if the audio (and CTCSS) modulation is injected
directly into the PLL control voltage line. Thus, the PLL cleanup
operation affects only a minuscule part of the noise bandwidth.

If the VCO control voltage range is 4 VDC and the tuning range is
approximately linear over 4 MHz, that 0.94 uV will produce a phase
noise of 0.94 Hz RMS. Not a great deal...:-)

At least in many VHF/UHF systems, the VCO sensitivity could be 10
MHz/V and with a noise voltage of 100 uV, the phase noise would be 1
kHz. This would severely limit the SNR obtainable when using FM in
addition to the increased reverse mixing. These kinds of problems are
common with VHF/UHF FM gears.

However, some badly designed simple up converting general coverage HF
receivers could suffer from quite high phase noise, if a single VCO is
used without switching in reactances. In such receivers, any CW signal
would sound as it is coming through the aurora :-).

Clearly, this discussion shows that picking the right initial
assumptions, quite different conclusions can be made.

Paul OH3LWR