Resonant circuit design

[bitrex]

In e.g. VCO design, assuming a less than ideal varactor/varicap, is
there a particular resonant circuit/tank topology or ratio of components
where given a base operating frequency with no control voltage applied,
the VCO frequency deviation either side is maximized with respect to
change in varicap capacitance?

 

[Jan Panteltje]

On a sunny day (Fri, 9 Aug 2019 12:39:39 -0400) it happened bitrex
<user@example.net> wrote in <glh3F.26424$YO5.23824@fx29.iad>:

In e.g. VCO design, assuming a less than ideal varactor/varicap, is
there a particular resonant circuit/tank topology or ratio of components
where given a base operating frequency with no control voltage applied,
the VCO frequency deviation either side is maximized with respect to
change in varicap capacitance?

If you want the varicap to have maximum effect it middle range value should
be the only capacitance around in the LC tuned circuit.
If there are other Cs in the circuit such as coil winding capacitance, trimmers,
feedback capacitors, Cce, Ccb, etc.
then the effect of the varicap is reduced.
You can use varicaps in parallel though.

 

[bitrex]

On 8/9/19 1:18 PM, George Herold wrote:

On Friday, August 9, 2019 at 12:39:44 PM UTC-4, bitrex wrote:
In e.g. VCO design, assuming a less than ideal varactor/varicap, is
there a particular resonant circuit/tank topology or ratio of components
where given a base operating frequency with no control voltage applied,
the VCO frequency deviation either side is maximized with respect to
change in varicap capacitance?

Hmm well besides just having the varicap as the only capacitance in circuit..
Freq. ~ 1/sqrt(L*C)... I'm not sure what else you are looking for.
(I've never made a varicap VCO.)
George H.

Hi George, thanks for getting back, to clarify I meant "tank" in a kind
of a general sense for resonant circuit topologies. You can have
different kind of tanks like series, parallel, ones with tapped
inductors or tapped capacitors and so forth.

I was curious if any particular topology of that type when used as the
oscillator frequency determining network was more amenable to maximizing
deviation with respect to the tuning capacitance value than some other one

 

[George Herold]

On Friday, August 9, 2019 at 12:39:44 PM UTC-4, bitrex wrote:

In e.g. VCO design, assuming a less than ideal varactor/varicap, is
there a particular resonant circuit/tank topology or ratio of components
where given a base operating frequency with no control voltage applied,
the VCO frequency deviation either side is maximized with respect to
change in varicap capacitance?

Hmm well besides just having the varicap as the only capacitance in circuit..
Freq. ~ 1/sqrt(L*C)... I'm not sure what else you are looking for.
(I've never made a varicap VCO.)
George H.

 

[John Larkin]

On Fri, 9 Aug 2019 12:39:39 -0400, bitrex <user@example.net> wrote:

In e.g. VCO design, assuming a less than ideal varactor/varicap, is
there a particular resonant circuit/tank topology or ratio of components
where given a base operating frequency with no control voltage applied,
the VCO frequency deviation either side is maximized with respect to
change in varicap capacitance?

You seem to be proposing to forward-bias the varicap?

A schematic sketch would help.

 

[Tim Williams]

"bitrex" <user@example.net> wrote in message
news:glh3F.26424$YO5.23824@fx29.iad...

In e.g. VCO design, assuming a less than ideal varactor/varicap, is there
a particular resonant circuit/tank topology or ratio of components
where given a base operating frequency with no control voltage applied,
the VCO frequency deviation either side is maximized with respect to
change in varicap capacitance?

Either side of what?

As others have suggested, it's not clear that you're asking for anything
besides "make the variable part be as much of the total as you can".

If you want more dF/dV, you can always run it in the GHz then mix it down.
Deviation of many decades, if you like.

You can probably come up with a tank where dF/dV is arbitrarily high in some
local range (but tending to constrain the average value over the full
control range, I would suspect).

You can certainly come up with one where the F(V) curve goes through
hysteresis loops (more commonly described as, jumping from one resonant mode
to another). The dF/dV would be, well, you might argue it's infinite at the
edges of such a loop, but "undefined" might be more accurate.

Tim

--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Design
Website: https://www.seventransistorlabs.com/

 

[George Herold]

On Friday, August 9, 2019 at 1:23:57 PM UTC-4, bitrex wrote:

On 8/9/19 1:18 PM, George Herold wrote:
On Friday, August 9, 2019 at 12:39:44 PM UTC-4, bitrex wrote:
In e.g. VCO design, assuming a less than ideal varactor/varicap, is
there a particular resonant circuit/tank topology or ratio of components
where given a base operating frequency with no control voltage applied,
the VCO frequency deviation either side is maximized with respect to
change in varicap capacitance?

Hmm well besides just having the varicap as the only capacitance in circuit..
Freq. ~ 1/sqrt(L*C)... I'm not sure what else you are looking for.
(I've never made a varicap VCO.)
George H.


Hi George, thanks for getting back, to clarify I meant "tank" in a kind
of a general sense for resonant circuit topologies. You can have
different kind of tanks like series, parallel, ones with tapped
inductors or tapped capacitors and so forth.

I was curious if any particular topology of that type when used as the
oscillator frequency determining network was more amenable to maximizing
deviation with respect to the tuning capacitance value than some other one

Some capacitive multiplier that worked at AC?
If it's opamp frequency, I wonder if one of those two opamp GIC's
(general impedance converter) could make the capacitance look bigger....
That doesn't really help though, cause the frequency range is
related to the proportional change in C.
As Jan said keep all other stray C small.

George H.

 

[Phil Hobbs]

On 8/9/19 3:19 PM, Tim Williams wrote:

"bitrex" <user@example.net> wrote in message
news:glh3F.26424$YO5.23824@fx29.iad...
In e.g. VCO design, assuming a less than ideal varactor/varicap, is
there a particular resonant circuit/tank topology or ratio of components
where given a base operating frequency with no control voltage
applied, the VCO frequency deviation either side is maximized with
respect to change in varicap capacitance?

Either side of what?

As others have suggested, it's not clear that you're asking for anything
besides "make the variable part be as much of the total as you can".

If you want more dF/dV, you can always run it in the GHz then mix it
down. Deviation of many decades, if you like.

You can probably come up with a tank where dF/dV is arbitrarily high in
some local range (but tending to constrain the average value over the
full control range, I would suspect).

You can certainly come up with one where the F(V) curve goes through
hysteresis loops (more commonly described as, jumping from one resonant
mode to another).  The dF/dV would be, well, you might argue it's
infinite at the edges of such a loop, but "undefined" might be more
accurate.

Tim

Or put variable feedback on the cold end. That way you could use a pot,
a fast op amp, and a fixed cap.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[bitrex]

On 8/9/19 4:47 PM, Phil Hobbs wrote:

On 8/9/19 3:19 PM, Tim Williams wrote:
"bitrex" <user@example.net> wrote in message
news:glh3F.26424$YO5.23824@fx29.iad...
In e.g. VCO design, assuming a less than ideal varactor/varicap, is
there a particular resonant circuit/tank topology or ratio of components
where given a base operating frequency with no control voltage
applied, the VCO frequency deviation either side is maximized with
respect to change in varicap capacitance?

Either side of what?

As others have suggested, it's not clear that you're asking for
anything besides "make the variable part be as much of the total as
you can".

If you want more dF/dV, you can always run it in the GHz then mix it
down. Deviation of many decades, if you like.

You can probably come up with a tank where dF/dV is arbitrarily high
in some local range (but tending to constrain the average value over
the full control range, I would suspect).

You can certainly come up with one where the F(V) curve goes through
hysteresis loops (more commonly described as, jumping from one
resonant mode to another).  The dF/dV would be, well, you might argue
it's infinite at the edges of such a loop, but "undefined" might be
more accurate.

Tim


Or put variable feedback on the cold end.  That way you could use a pot,
a fast op amp, and a fixed cap.

Cheers

Phil Hobbs

The problem is that more or less I have an inductor of a given size
that's pretty much fixed by physical requirements, and a variable
capacitor that's in the some-pF range that it is, and only wants to vary
so much in response to a signal.

But I don't want to run the tank at the frequency it wants to run at
given the constraints so I have to add some series or parallel fixed
elements to bring it down in frequency. But then naturally I don't get
enough dF/dV as I'd like because when the frequency comes down so does that.

I didn't know if there was a particular resonant circuit/oscillator
topology that would let me cheat there somehow (there are a lot of ways
to connect up RLCs in a feedback loop and make them oscillate and I
don't know the dF/dV for all of them, if one of the elements is
variable, off the top of my head) or if I should resort to
down-converting as TW suggested, or perhaps do it indirectly by running
the main oscillator at a high frequency and generating a CV for a VCO in
a PLL or somesuch.

 

[bitrex]

On 8/9/19 4:47 PM, Phil Hobbs wrote:

On 8/9/19 3:19 PM, Tim Williams wrote:
"bitrex" <user@example.net> wrote in message
news:glh3F.26424$YO5.23824@fx29.iad...
In e.g. VCO design, assuming a less than ideal varactor/varicap, is
there a particular resonant circuit/tank topology or ratio of components
where given a base operating frequency with no control voltage
applied, the VCO frequency deviation either side is maximized with
respect to change in varicap capacitance?

Either side of what?

As others have suggested, it's not clear that you're asking for
anything besides "make the variable part be as much of the total as
you can".

If you want more dF/dV, you can always run it in the GHz then mix it
down. Deviation of many decades, if you like.

You can probably come up with a tank where dF/dV is arbitrarily high
in some local range (but tending to constrain the average value over
the full control range, I would suspect).

You can certainly come up with one where the F(V) curve goes through
hysteresis loops (more commonly described as, jumping from one
resonant mode to another).  The dF/dV would be, well, you might argue
it's infinite at the edges of such a loop, but "undefined" might be
more accurate.

Tim


Or put variable feedback on the cold end.  That way you could use a pot,
a fast op amp, and a fixed cap.

Cheers

Phil Hobbs

it's slow enough freqs (MHz) that I can use some op amps.

 

[bitrex]

On 8/9/19 6:06 PM, bitrex wrote:

On 8/9/19 4:47 PM, Phil Hobbs wrote:
On 8/9/19 3:19 PM, Tim Williams wrote:
"bitrex" <user@example.net> wrote in message
news:glh3F.26424$YO5.23824@fx29.iad...
In e.g. VCO design, assuming a less than ideal varactor/varicap, is
there a particular resonant circuit/tank topology or ratio of
components
where given a base operating frequency with no control voltage
applied, the VCO frequency deviation either side is maximized with
respect to change in varicap capacitance?

Either side of what?

As others have suggested, it's not clear that you're asking for
anything besides "make the variable part be as much of the total as
you can".

If you want more dF/dV, you can always run it in the GHz then mix it
down. Deviation of many decades, if you like.

You can probably come up with a tank where dF/dV is arbitrarily high
in some local range (but tending to constrain the average value over
the full control range, I would suspect).

You can certainly come up with one where the F(V) curve goes through
hysteresis loops (more commonly described as, jumping from one
resonant mode to another).  The dF/dV would be, well, you might argue
it's infinite at the edges of such a loop, but "undefined" might be
more accurate.

Tim


Or put variable feedback on the cold end.  That way you could use a
pot, a fast op amp, and a fixed cap.

Cheers

Phil Hobbs


it's slow enough freqs (MHz) that I can use some op amps.

Oooh I see what you're saying. C multiplier. Oooooh that might work.

 

On Fri, 9 Aug 2019 12:39:39 -0400, bitrex <user@example.net> wrote:

In e.g. VCO design, assuming a less than ideal varactor/varicap, is
there a particular resonant circuit/tank topology or ratio of components
where given a base operating frequency with no control voltage applied,
the VCO frequency deviation either side is maximized with respect to
change in varicap capacitance?

I assume that you are asking how to maximize the tuning range.

On an LC circuit with a fixed L in order to get a specific frequency
range, the capacitance range must be squared. So to get 1:3 frequency
range, you need a 9:1 capacitance range. Minimizing any stray
capacitances helps extending the high frequency limit.

On the high capacitance side, when he control voltage is low, compare
this to the RF voltage present. The RF voltage will change the
capacitance during the RF cycle. In the worst case the capacitance
diode becomes forward biased during the RF-cycle, loading the
resonance. The phase noise performance may suffer badly.

One thing that may help is using a dual capacitance diode (common
cathode), In this case one diode is reverse biased and acts a
capacitor while the other diode is forward biased acting as a resistor
and during the opposite RF cycle the role of the diodes are reversed.
This distorts the RF waveform symmetrically.even at zero control
voltage.

At least with only a single capacitance diode there should be a low
limit for the control voltage compared to RF voltage, so that the
diode is always reverse biased.

The RF voltage can be an issue at least in receivers. Some high
quality receivers drop the front section impedance level to 25 or 12
ohms instead of the ordinary 50 ohms. This also drops the RF voltage
compared to lowest capacitance diode control voltage, reducing the
risk for spurious mixing products, mainly the second order.

In an oscillator, you may have to look at the RF voltage levels
compared to the capacitance control voltage.

Of course if you can also change the L e.g. by a variable DC magnetic
field on some suitable ferrite material will also help extending the
frequency range.

 

[Jan Panteltje]

On a sunny day (Fri, 9 Aug 2019 18:06:17 -0400) it happened bitrex
<user@example.net> wrote in <u7m3F.75406$3m1.70413@fx17.iad>:

>it's slow enough freqs (MHz) that I can use some op amps.

74HC4046?
Works from 3V upwards.
Always have some in stock here.

 

[Jan Panteltje]

On a sunny day (Sat, 10 Aug 2019 07:57:42 +0300) it happened
upsidedown@downunder.com wrote in
<sfhsketlphn7am5kealrrhop1u1tkh4hj5@4ax.com>:

On Fri, 9 Aug 2019 12:39:39 -0400, bitrex <user@example.net> wrote:

In e.g. VCO design, assuming a less than ideal varactor/varicap, is
there a particular resonant circuit/tank topology or ratio of components
where given a base operating frequency with no control voltage applied,
the VCO frequency deviation either side is maximized with respect to
change in varicap capacitance?

I assume that you are asking how to maximize the tuning range.

On an LC circuit with a fixed L in order to get a specific frequency
range, the capacitance range must be squared. So to get 1:3 frequency
range, you need a 9:1 capacitance range. Minimizing any stray
capacitances helps extending the high frequency limit.

On the high capacitance side, when he control voltage is low, compare
this to the RF voltage present. The RF voltage will change the
capacitance during the RF cycle. In the worst case the capacitance
diode becomes forward biased during the RF-cycle, loading the
resonance. The phase noise performance may suffer badly.

One thing that may help is using a dual capacitance diode (common
cathode), In this case one diode is reverse biased and acts a
capacitor while the other diode is forward biased acting as a resistor
and during the opposite RF cycle the role of the diodes are reversed.
This distorts the RF waveform symmetrically.even at zero control
voltage.

At least with only a single capacitance diode there should be a low
limit for the control voltage compared to RF voltage, so that the
diode is always reverse biased.

The RF voltage can be an issue at least in receivers. Some high
quality receivers drop the front section impedance level to 25 or 12
ohms instead of the ordinary 50 ohms. This also drops the RF voltage
compared to lowest capacitance diode control voltage, reducing the
risk for spurious mixing products, mainly the second order.

In an oscillator, you may have to look at the RF voltage levels
compared to the capacitance control voltage.

Of course if you can also change the L e.g. by a variable DC magnetic
field on some suitable ferrite material will also help extending the
frequency range.

If you do not have 2 varicaps I have uses transistors Ccb and Cbe
http://panteltje.com/pub/rubidium_locked_24MHz_circuit_diagram_IMG_6641.JPG
It is a PNP, the base is always positive so the bc and be diodes are revese biased.

I hAVe nOt AnALyZeD IT BuT IT wOrKS

 

[Phil Hobbs]

On 8/10/19 12:56 AM, Jan Panteltje wrote:

On a sunny day (Fri, 9 Aug 2019 18:06:17 -0400) it happened bitrex
user@example.net> wrote in <u7m3F.75406$3m1.70413@fx17.iad>:

it's slow enough freqs (MHz) that I can use some op amps.

74HC4046?
Works from 3V upwards.
Always have some in stock here.

Yeah, that part has about the crappiest oscillator in history.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[Phil Hobbs]

On 8/10/19 5:24 AM, Jan Panteltje wrote:

Phil Hobbs opiniated:
On 8/10/19 12:56 AM, Jan Panteltje wrote:
On a sunny day (Fri, 9 Aug 2019 18:06:17 -0400) it happened bitrex
user@example.net> wrote in <u7m3F.75406$3m1.70413@fx17.iad>:

it's slow enough freqs (MHz) that I can use some op amps.

74HC4046?
Works from 3V upwards.
Always have some in stock here.


Yeah, that part has about the crappiest oscillator in history.

Well depends on what you define as 'crappy'
It is great for audio modulation and demodulation (PLL)
and a zillion other things.
No idea what the OP wants to do, and if it MUST be a LC.
You better respect the CD4046 74HC4046 and 74HCT4046,
it is more popular than the 555 almost ? ;-)

The metal-gate 4046 is quite a nice part for what it is. Its oscillator
tunes more than 100:1 with pretty good linearity. Its two real warts
are the very poor predictability of the oscillation frequency and the
deadband in the phase-frequency detector PD2, helpfully positioned right
at the servo point. You can fix that by adding a resistor to ground to
pull the servo point off the deadband, but there's nothing you can do
about the poorly controlled oscillation frequency except make the
deviation range very large.

All the HC versions have _horrible_ oscillators: they're wildly
nonlinear (up to 5:1 changes in slope), and they just quit oscillating
for control voltages below a volt or so. (Some slightly better, some a
bit worse.) A couple of the glorified versions (7046 and 9046) claim
that their PDs don't have the dead band. However, the dead band is
easily worked around and their oscillators still stink on ice.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[Jan Panteltje]

Phil Hobbs opiniated:

On 8/10/19 12:56 AM, Jan Panteltje wrote:
On a sunny day (Fri, 9 Aug 2019 18:06:17 -0400) it happened bitrex
user@example.net> wrote in <u7m3F.75406$3m1.70413@fx17.iad>:

it's slow enough freqs (MHz) that I can use some op amps.

74HC4046?
Works from 3V upwards.
Always have some in stock here.


Yeah, that part has about the crappiest oscillator in history.

Well depends on what you define as 'crappy'
It is great for audio modulation and demodulation (PLL)
and a zillion other things.
No idea what the OP wants to do, and if it MUST be a LC.
You better respect the CD4046 74HC4046 and 74HCT4046,
it is more popular than the 555 almost ? ;-)

 

[Jan Panteltje]

On a sunny day (Sat, 10 Aug 2019 05:38:11 -0400) it happened Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote in
<qim3a5$3us$1@dont-email.me>:

On 8/10/19 5:24 AM, Jan Panteltje wrote:
Phil Hobbs opiniated:
On 8/10/19 12:56 AM, Jan Panteltje wrote:
On a sunny day (Fri, 9 Aug 2019 18:06:17 -0400) it happened bitrex
user@example.net> wrote in <u7m3F.75406$3m1.70413@fx17.iad>:

it's slow enough freqs (MHz) that I can use some op amps.

74HC4046?
Works from 3V upwards.
Always have some in stock here.


Yeah, that part has about the crappiest oscillator in history.

Well depends on what you define as 'crappy'
It is great for audio modulation and demodulation (PLL)
and a zillion other things.
No idea what the OP wants to do, and if it MUST be a LC.
You better respect the CD4046 74HC4046 and 74HCT4046,
it is more popular than the 555 almost ? ;-)


The metal-gate 4046 is quite a nice part for what it is. Its oscillator
tunes more than 100:1 with pretty good linearity. Its two real warts
are the very poor predictability of the oscillation frequency and the
deadband in the phase-frequency detector PD2, helpfully positioned right
at the servo point. You can fix that by adding a resistor to ground to
pull the servo point off the deadband, but there's nothing you can do
about the poorly controlled oscillation frequency except make the
deviation range very large.

All the HC versions have _horrible_ oscillators: they're wildly
nonlinear (up to 5:1 changes in slope), and they just quit oscillating
for control voltages below a volt or so. (Some slightly better, some a
bit worse.) A couple of the glorified versions (7046 and 9046) claim
that their PDs don't have the dead band. However, the dead band is
easily worked around and their oscillators still stink on ice.

Cheers

Phil Hobbs

snif snif, will have to throw all those 47HC4046 away...
Anyways raspberry can also make all sorts of frequencies.
Somebody even makes DVB signals with one, WITHOUT any external parts added...
https://hackaday.com/2016/04/03/dvb-s-from-a-raspberry-pi-with-no-extra-hardware/
https://github.com/F5OEO/rpidatv

 

[Bill Sloman]

On Saturday, August 10, 2019 at 6:31:55 PM UTC+10, Phil Hobbs wrote:

On 8/10/19 12:56 AM, Jan Panteltje wrote:
On a sunny day (Fri, 9 Aug 2019 18:06:17 -0400) it happened bitrex
user@example.net> wrote in <u7m3F.75406$3m1.70413@fx17.iad>:

it's slow enough freqs (MHz) that I can use some op amps.

74HC4046?
Works from 3V upwards.
Always have some in stock here.


Yeah, that part has about the crappiest oscillator in history.

The original 4046 oscillator worked over a large range of voltages, and the Philips 74HC and 74HCT didn't. The 74CT9046 did have an improved pahse detector.

--
Bill Sloman, Sydney

 

[Gerhard Hoffmann]

Am 09.08.19 um 18:39 schrieb bitrex:

In e.g. VCO design, assuming a less than ideal varactor/varicap, is
there a particular resonant circuit/tank topology or ratio of components
where given a base operating frequency with no control voltage applied,
the VCO frequency deviation either side is maximized with respect to
change in varicap capacitance?

It would help to know your VCO design.

If it is the usual VHF VCO, then almost all use a CC circuit:
collector at RF ground, maybe through a small resistor to get some
output energy, emitter high impedance, most often capacitively loaded
and some series resonance in the base. That is the textbook case of
an oscillating follower.

If you measure into the base, you see a capacitor in series with a
negative resistance. That un-damps anything inductive in the base
connection unless there is a large enough positive resistance to
cancel the negative one. If the circuit is meant as an amplifier
that is the job of the base/gate/grid stopper.

If you have a varicap in the base LC, its tuning range is reduced
by the series capacitance of the BE impedance. That may be quite
small. I once did try to pull down the range of a Z-Comm VCO. That
worked to some extend, but there was a frequency I could not get
below, even with large ceramic caps. That was the frequency of the
base resonator stripline and the BE capacitance.

There are 2 oscillator books by one Randall Rhea; he described
the working of the VCO quite well. It seems that the transistor type
is quite important for both the negative resistance and also for
the series BE capacitance. Rhea is the author of the Genesys system
that is now part of Agilent ADS, eh, Keysight.

On the positive side, the varicap inductance is absorbed by the
stripline.

There are newer varicaps that are optimized for large capacitance
ratio. Take a look at Infineon, for example.

At low tuning voltages, dC/dV is greatest. Unfortunately the Q
is worst there, also. Avoid having them in forward direction even
for a short part of the cycle. The usual trick with anti-series
diodes helps.

Rohde says that many small diodes in par. yield better phase noise
than a large one because of averaging. While I'm normally happy
to plow with the power of a thousand chicken, I'm not really convinced.
It cannot make a difference when I connect several subdiodes onchip
vs. onboard.

cheers, Gerhard