Prev: Passive Earthquake Proofing w/ Dry Stone
Next: Best Light for reading?
From: John Fields on 13 May 2010 09:50
On Wed, 12 May 2010 22:07:32 -0500, "krw(a)att.bizzzzzzzzzzzz"
<krw(a)att.bizzzzzzzzzzzz> wrote:

>On Wed, 12 May 2010 19:09:42 -0500, John Fields
><jfields(a)austininstruments.com> wrote:
>
>>On Tue, 11 May 2010 20:39:37 -0700 (PDT), Daku <dakupoto(a)gmail.com>
>>wrote:
>>
>>>Dear Sir,
>>>I do not intend to use a ring oscillator based VCO.
>>>However, as you have said below, literature search
>>>shows a lot of people using the ring oscillator based
>>>voltage controlled oscillator in phase locked loop
>>>designs, which is sounds strange to me, since in a
>>>phase locked loop, the VCO oscillation frequency
>>>must be very sensitive to the input voltage level.
>>>Maybe the ring oscillator is designed to oscillate
>>>at the central frequency, but frequency variation is
>>>difficult and tricky.
>>
>>---
>>Please bottom post, or inline post when it's necessary for clarity.
>>Thank you. :-)
>>
>>As others have noted, your original post seems to indicate that you
>>don't have a really good grasp on what's required to change the
>>frequency of an oscillator as a function of an applied voltage.
>>
>>For an LC tank, where:
>>
>> 1
>> f = -------------,
>> 2pi sqrt LC
>>
>>then either L or C must be varied as a function of voltage in order to
>>make f change.
>>
>>In the old days it was done with a saturable reactor, where the DC
>>voltage on the control winding and the attendant current through it
>>changed the reluctance of the core and, thererfore, the inductance
>>described by the secondary.
>>
>>Today, afaik, it's done with varactors.
>>
>>Comment?
>
>It can be done with ring oscillators, too. Gate delay isn't a strong function
>of Vcc, though so it's not done often. DLLs generally use a mux to select the
>number of gates in the ring.

---
Actually, the change in delay is quite pronounced, as shown by:

news:t6piu55sqenrovms7k5fuhrl8m7666thng(a)4ax.com

from TI's 1989 GMOS logic data book.

Also, for an HC00, tpd at 25C is 15ns max for Vcc = 6V, 18ns for 4.5V,
and 90ns for 2V.

From: John Fields on 13 May 2010 10:19
On Wed, 12 May 2010 20:07:26 -0700 (PDT), Daku <dakupoto(a)gmail.com>
wrote:

>On May 13, 5:09 am, John Fields <jfie...(a)austininstruments.com> wrote:
>>
>> ---
>> Please bottom post, or inline post when it's necessary for clarity.
>> Thank you. :-)
>>
>> As others have noted, your original post seems to indicate that you
>> don't have a really good grasp on what's required to change the
>> frequency of an oscillator as a function of an applied voltage.
>>
>> For an LC tank, where:
>>
>> 1
>> f = -------------,
>> 2pi sqrt LC
>>
>> then either L or C must be varied as a function of voltage in order to
>> make f change.
>>
>> In the old days it was done with a saturable reactor, where the DC
>> voltage on the control winding and the attendant current through it
>> changed the reluctance of the core and, thererfore, the inductance
>> described by the secondary.
>>
>> Today, afaik, it's done with varactors.
>>
>> Comment?
>
>Dear Sir,
>I am fully aware of the LC-tank circuit, and that a varactor needs to
>be added in the tank to get the voltage control. However, a simple
>literature search on voltage controlled oscillators indicates that a large number
>of designs are based on the CMOS ring oscillator.

---
After doing a cursory search for "ring counter VCO", such seems notv
to be the case.

Can you cite the links you found, please?
---

>As one of other posters
>noted before (and I fully agree with him) that it is difficult to
>dynamically control the oscillation frequency of the CMOS ring oscillator - this
>frequency is based on the characteristics of the MOSFETs in the inverters of
>the ring oscillator. Given that, how are these ring oscillator based
>designs being labelled "voltage controlled" since the voltage control merely
>switches the oscillations on and off, NOT control the frequency as in
>a LC tank.

---
That's not true, since if Vcc is varied, as shown in:

news:t6piu55sqenrovms7k5fuhrl8m7666thng(a)4ax.com

Then a VCO can be readily fashioned from HCMOS logic gates by
exploiting the fact that the gate delay varies over quite a wide range
(about 6:1) with Vcc changing from 2 to 6V.

From: krw on 13 May 2010 18:30
On Thu, 13 May 2010 08:50:37 -0500, John Fields
<jfields(a)austininstruments.com> wrote:

>On Wed, 12 May 2010 22:07:32 -0500, "krw(a)att.bizzzzzzzzzzzz"
><krw(a)att.bizzzzzzzzzzzz> wrote:
>
>>On Wed, 12 May 2010 19:09:42 -0500, John Fields
>><jfields(a)austininstruments.com> wrote:
>>
>>>On Tue, 11 May 2010 20:39:37 -0700 (PDT), Daku <dakupoto(a)gmail.com>
>>>wrote:
>>>
>>>>Dear Sir,
>>>>I do not intend to use a ring oscillator based VCO.
>>>>However, as you have said below, literature search
>>>>shows a lot of people using the ring oscillator based
>>>>voltage controlled oscillator in phase locked loop
>>>>designs, which is sounds strange to me, since in a
>>>>phase locked loop, the VCO oscillation frequency
>>>>must be very sensitive to the input voltage level.
>>>>Maybe the ring oscillator is designed to oscillate
>>>>at the central frequency, but frequency variation is
>>>>difficult and tricky.
>>>
>>>---
>>>Please bottom post, or inline post when it's necessary for clarity.
>>>Thank you. :-)
>>>
>>>As others have noted, your original post seems to indicate that you
>>>don't have a really good grasp on what's required to change the
>>>frequency of an oscillator as a function of an applied voltage.
>>>
>>>For an LC tank, where:
>>>
>>> 1
>>> f = -------------,
>>> 2pi sqrt LC
>>>
>>>then either L or C must be varied as a function of voltage in order to
>>>make f change.
>>>
>>>In the old days it was done with a saturable reactor, where the DC
>>>voltage on the control winding and the attendant current through it
>>>changed the reluctance of the core and, thererfore, the inductance
>>>described by the secondary.
>>>
>>>Today, afaik, it's done with varactors.
>>>
>>>Comment?
>>
>>It can be done with ring oscillators, too. Gate delay isn't a strong function
>>of Vcc, though so it's not done often. DLLs generally use a mux to select the
>>number of gates in the ring.
>
>---
>Actually, the change in delay is quite pronounced, as shown by:
>
>news:t6piu55sqenrovms7k5fuhrl8m7666thng(a)4ax.com
>
>from TI's 1989 GMOS logic data book.
>
>Also, for an HC00, tpd at 25C is 15ns max for Vcc = 6V, 18ns for 4.5V,
>and 90ns for 2V.

+/- 10% isn't very good for the capture ratio. The normal (lot-to-lot)
variation is likely much larger, making it more or less useless for a VCO.
From: George Herold on 14 May 2010 09:01
On May 13, 9:50 am, John Fields <jfie...(a)austininstruments.com> wrote:
> On Wed, 12 May 2010 22:07:32 -0500, "k...(a)att.bizzzzzzzzzzzz"
>
>
>
>
>
> <k...(a)att.bizzzzzzzzzzzz> wrote:
> >On Wed, 12 May 2010 19:09:42 -0500, John Fields
> ><jfie...(a)austininstruments.com> wrote:
>
> >>On Tue, 11 May 2010 20:39:37 -0700 (PDT), Daku <dakup...(a)gmail.com>
> >>wrote:
>
> >>>Dear Sir,
> >>>I do not intend to use a ring oscillator based VCO.
> >>>However, as you have said below, literature search
> >>>shows a lot of people using the ring oscillator based
> >>>voltage controlled oscillator in phase locked loop
> >>>designs, which is sounds strange to me, since in a
> >>>phase locked loop, the VCO oscillation frequency
> >>>must be very sensitive to the input voltage level.
> >>>Maybe the ring oscillator is designed to oscillate
> >>>at the central frequency, but frequency variation is
> >>>difficult and tricky.
>
> >>---
> >>Please bottom post, or inline post when it's necessary for clarity.
> >>Thank you. :-)
>
> >>As others have noted, your original post seems to indicate that you
> >>don't have a really good grasp on what's required to change the
> >>frequency of an oscillator as a function of an applied voltage.
>
> >>For an LC tank, where:
>
> >>               1
> >>     f = -------------,
> >>          2pi sqrt LC
>
> >>then either L or C must be varied as a function of voltage in order to
> >>make f change.
>
> >>In the old days it was done with a saturable reactor, where the DC
> >>voltage on the control winding and the attendant current through it
> >>changed the reluctance of the core and, thererfore, the inductance
> >>described by the secondary.
>
> >>Today, afaik, it's done with varactors.
>
> >>Comment?
>
> >It can be done with ring oscillators, too.  Gate delay isn't a strong function
> >of Vcc, though so it's not done often.  DLLs generally use a mux to select the
> >number of gates in the ring.
>
> ---
> Actually, the change in delay is quite pronounced, as shown by:
>
> news:t6piu55sqenrovms7k5fuhrl8m7666thng(a)4ax.com
>
> from TI's 1989 GMOS logic data book.
>
> Also, for an HC00, tpd at 25C is 15ns max for Vcc = 6V, 18ns for 4.5V,
> and 90ns for 2V.- Hide quoted text -
>
> - Show quoted text -

Hi John, You posted that link before, but I can't seem to make it
work. Is this on TI's web site?

It sounds like this would make an 'OK' VCO. And one could add some
outer control loop that would switch in (or out) two more inverters if
the control voltage got near the end of it's range. (As someone
already suggested.)


George H.

From: John Fields on 14 May 2010 09:41
On Fri, 14 May 2010 06:01:14 -0700 (PDT), George Herold
<ggherold(a)gmail.com> wrote:

>On May 13, 9:50�am, John Fields <jfie...(a)austininstruments.com> wrote:

>> news:t6piu55sqenrovms7k5fuhrl8m7666thng(a)4ax.com
>>
>> from TI's 1989 GMOS logic data book.
>>
>> Also, for an HC00, tpd at 25C is 15ns max for Vcc = 6V, 18ns for 4.5V,
>> and 90ns for 2V.- Hide quoted text -
>>
>> - Show quoted text -
>
>Hi John, You posted that link before, but I can't seem to make it
>work. Is this on TI's web site?

---
No, it's on USENET at alt.binaries.schematics.electronic, which Google
groups can't access, and Verizon shut down access to the alt hierarchy
in July of last year, I think. AFAIK they may not even offer any
access to USENET any more.

If you want to get USENET directly and not through a web interface
like Google groups, then you need to get a USENET ISP and access
USENET with a newsreader.
---

>It sounds like this would make an 'OK' VCO. And one could add some
>outer control loop that would switch in (or out) two more inverters if
>the control voltage got near the end of it's range. (As someone
>already suggested.)

---
I'm not a big fan of that circuit, but just posted it to prove a
point. ;)

BTW, I'll email you a copy of the PDF...

First  |  Prev  |  Next  |  Last
Pages: 1 2 3 4 5
Prev: Passive Earthquake Proofing w/ Dry Stone
Next: Best Light for reading?