From: Jim Thompson on
On Tue, 23 Feb 2010 17:04:26 -0500, Phil Hobbs
<pcdhSpamMeSenseless(a)electrooptical.net> wrote:

>On 2/23/2010 4:13 PM, Chris wrote:
>> On Feb 23, 12:41 pm, Jim Thompson<To-Email-Use-The-Envelope-I...(a)My-
>> Web-Site.com> wrote:
>>> On Tue, 23 Feb 2010 13:31:14 -0700, Jim Thompson
>>>
>>>
>>>
>>> <To-Email-Use-The-Envelope-I...(a)My-Web-Site.com> wrote:
>>>> On Tue, 23 Feb 2010 12:14:02 -0800, Tim Wescott<t...(a)seemywebsite.now>
>>>> wrote:
>>>
>>>>> Jim Thompson wrote:
>>>>>> On Tue, 23 Feb 2010 11:06:08 -0800, Tim Wescott<t...(a)seemywebsite.now>
>>>>>> wrote:
>>>
>>>>>>> Chris wrote:
>>>>>>>> The data sheet just gives some ancient reference for finding the
>>>>>>>> values for R1, R2, and C1.
>>>
>>>>>>>> Is the RC network for the VCO is (R1, R2, C1)? Would I just use the
>>>>>>>> cutoff frequency for the R1 and C2 RC filter (i.e. fc=1/(2*Pi*R1*C1)).
>>>
>>>>>>>> Thanks,
>>>>>>>> Chris KQ6UP
>>>>>>> Find some other data sheet?
>>>
>>>>>>> Some of them are more informative than others. I don't know who still
>>>>>>> makes 'em, but I'd look to TI, ON Semiconductor, and Fairchild to start.
>>>
>>>>>>> The 74HC4046 is very similar (it has one more phase comparator, IIRC,
>>>>>>> and may delete some other function) -- if you can't find what you want
>>>>>>> in a 4046 data sheet a 74HC4046, or other 74xx4046 data sheet, may help.
>>>
>>>>>> The 4046 has as phase detectors a conventional edge-matching PFD plus
>>>>>> an XOR (posing as a four quadrant multiplier ;-)
>>>
>>>>>> ...Jim Thompson
>>>
>>>>> Either the 74HC4046 or the 74HC7046 adds an RS flip-flop phase detector,
>>>>> and (I think) deletes a voltage regulator or something.
>>>
>>>>> At any rate, one of the 'obscure' bits of the CD4046 is deleted, and a
>>>>> somewhat obscure phase detector is tacked on.
>>>
>>>> A PFD can be made from nine gates (ala Treadway/Thompson) or by using
>>>> a dual-D plus four 2-in-NAND's
>>>
>>>> ...Jim Thompson
>>>
>>> Just looked up a data sheet. The "extra" D-flop is theoretically a
>>> "lock detector"... known by the State of Thompson to be a fraud ;-)
>>>
>>> ...Jim Thompson
[snip sig]
>>
>> What would be a good way to indicate lock? I would like to have an LED
>> indicating a lock condition.
>>
>> Also, I was thinking of using a 555 to shut the PLL off when there is
>> no input signal present. This is needed for my application. I would
>> have the 555 hold for 1 period of the input frequency. If no signal
>> is detected in that time frame, I would have the INHIBIT go to logic 1
>> and shut off the loop.
>>
>> Thanks,
>> Chris
>
>With phase-frequency detectors it's pretty easy--the output goes to the
>rail when it's out of lock, so a window comparator works well.

What output is that?

>
>The 4046 also has the "phase pulses" output that (very roughly) tells
>you what the phase error is, so you can filter that and use a
>comparator. I don't think that works as well as the window comparator
>method.

Measuring something that is exactly zero width is a wee bit tricky.

>
>With an XOR gate or diode mixer PD, you need a second PD plus a 90
>degree phase shift. One excellent method is to use a divide-by-4
>Johnson counter (aka walking ring). Run the main PD off one output and
>the lock detector off the other.
>
>Cheers
>
>Phil Hobbs

What you mean is a "close-to-lock" detector is "easy" ?:-)

...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |

I love to cook with wine. Sometimes I even put it in the food.
From: Clifford Heath on
Marte Schwarz wrote:
> Right "very similar" expect the VCO. They need other values.
> CD4046 is much slower then 74HC4046. Earlier days I found only graphs in
> datasheets that leads me to useful values.

I've heard anecdotal evidence from several different designers saying
that the 74HC4046 doesn't work as a 4046 replacement in their designs,
suffering stability issues they found hard to diagnose.

Comments anyone?
From: Jim Thompson on
On Wed, 24 Feb 2010 09:18:47 +1100, Clifford Heath
<no(a)spam.please.net> wrote:

>Marte Schwarz wrote:
>> Right "very similar" expect the VCO. They need other values.
>> CD4046 is much slower then 74HC4046. Earlier days I found only graphs in
>> datasheets that leads me to useful values.
>
>I've heard anecdotal evidence from several different designers saying
>that the 74HC4046 doesn't work as a 4046 replacement in their designs,
>suffering stability issues they found hard to diagnose.
>
>Comments anyone?

I know they copied Treadway's and my designs, but don't know what they
bunged up :-(

...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |

I love to cook with wine. Sometimes I even put it in the food.
From: Ban on
Phil Hobbs wrote:
snip>
> With phase-frequency detectors it's pretty easy--the output goes to
> the rail when it's out of lock, so a window comparator works well.
>
> The 4046 also has the "phase pulses" output that (very roughly) tells
> you what the phase error is, so you can filter that and use a
> comparator. I don't think that works as well as the window
> comparator method.
>
> With an XOR gate or diode mixer PD, you need a second PD plus a 90
> degree phase shift. One excellent method is to use a divide-by-4
> Johnson counter (aka walking ring). Run the main PD off one output
> and the lock detector off the other.
>

Don't write about something you do not know well Phil, that makes "brutta
figura"

everything is wrong,
In lock is lets say within +/-30� Phaseangle. The phase comparator2 puts
out pulses according to how far apart are the edges of the input freqency
and the oszillator frequency. the control voltage of the VCO are these
pulses smoothed out by the loop filter. When it is in lock these pulses are
short, so you can drive a cap through a resistor and with a Schmitt you can
create a threshold switch to indicate this fact. It is only with an absent
input, that the oscillator drifts to the lowest frequency. And only if you
use phase comparator 2.
With PC1 the center frequency is put out with absent input signal. But "not
in lock" and absent input are two different conditions.

ciao Ban




From: Phil Hobbs on
On 2/23/2010 5:15 PM, Jim Thompson wrote:
> On Tue, 23 Feb 2010 17:04:26 -0500, Phil Hobbs
> <pcdhSpamMeSenseless(a)electrooptical.net> wrote:
>
>> On 2/23/2010 4:13 PM, Chris wrote:
>>> On Feb 23, 12:41 pm, Jim Thompson<To-Email-Use-The-Envelope-I...(a)My-
>>> Web-Site.com> wrote:
>>>> On Tue, 23 Feb 2010 13:31:14 -0700, Jim Thompson
>>>>
>>>>
>>>>
>>>> <To-Email-Use-The-Envelope-I...(a)My-Web-Site.com> wrote:
>>>>> On Tue, 23 Feb 2010 12:14:02 -0800, Tim Wescott<t...(a)seemywebsite.now>
>>>>> wrote:
>>>>
>>>>>> Jim Thompson wrote:
>>>>>>> On Tue, 23 Feb 2010 11:06:08 -0800, Tim Wescott<t...(a)seemywebsite.now>
>>>>>>> wrote:
>>>>
>>>>>>>> Chris wrote:
>>>>>>>>> The data sheet just gives some ancient reference for finding the
>>>>>>>>> values for R1, R2, and C1.
>>>>
>>>>>>>>> Is the RC network for the VCO is (R1, R2, C1)? Would I just use the
>>>>>>>>> cutoff frequency for the R1 and C2 RC filter (i.e. fc=1/(2*Pi*R1*C1)).
>>>>
>>>>>>>>> Thanks,
>>>>>>>>> Chris KQ6UP
>>>>>>>> Find some other data sheet?
>>>>
>>>>>>>> Some of them are more informative than others. I don't know who still
>>>>>>>> makes 'em, but I'd look to TI, ON Semiconductor, and Fairchild to start.
>>>>
>>>>>>>> The 74HC4046 is very similar (it has one more phase comparator, IIRC,
>>>>>>>> and may delete some other function) -- if you can't find what you want
>>>>>>>> in a 4046 data sheet a 74HC4046, or other 74xx4046 data sheet, may help.
>>>>
>>>>>>> The 4046 has as phase detectors a conventional edge-matching PFD plus
>>>>>>> an XOR (posing as a four quadrant multiplier ;-)
>>>>
>>>>>>> ...Jim Thompson
>>>>
>>>>>> Either the 74HC4046 or the 74HC7046 adds an RS flip-flop phase detector,
>>>>>> and (I think) deletes a voltage regulator or something.
>>>>
>>>>>> At any rate, one of the 'obscure' bits of the CD4046 is deleted, and a
>>>>>> somewhat obscure phase detector is tacked on.
>>>>
>>>>> A PFD can be made from nine gates (ala Treadway/Thompson) or by using
>>>>> a dual-D plus four 2-in-NAND's
>>>>
>>>>> ...Jim Thompson
>>>>
>>>> Just looked up a data sheet. The "extra" D-flop is theoretically a
>>>> "lock detector"... known by the State of Thompson to be a fraud ;-)
>>>>
>>>> ...Jim Thompson
> [snip sig]
>>>
>>> What would be a good way to indicate lock? I would like to have an LED
>>> indicating a lock condition.
>>>
>>> Also, I was thinking of using a 555 to shut the PLL off when there is
>>> no input signal present. This is needed for my application. I would
>>> have the 555 hold for 1 period of the input frequency. If no signal
>>> is detected in that time frame, I would have the INHIBIT go to logic 1
>>> and shut off the loop.
>>>
>>> Thanks,
>>> Chris
>>
>> With phase-frequency detectors it's pretty easy--the output goes to the
>> rail when it's out of lock, so a window comparator works well.
>
> What output is that?

The filtered PD output, e.g. at the VCO input. When there's a frequency
error, the 4046's output goes between tristate and one logic level, so
the integrated output rails and sits there. That's one of the good
things about it--I've often used it as a lock detector for PLLs whose
main phase detector was a diode bridge. (My fave is the Mini-Circuits
MPD-1.)

>
>>
>> The 4046 also has the "phase pulses" output that (very roughly) tells
>> you what the phase error is, so you can filter that and use a
>> comparator. I don't think that works as well as the window comparator
>> method.
>
> Measuring something that is exactly zero width is a wee bit tricky.

But you don't need to. You can decide how big a phase variance you're
willing to tolerate, and set the threshold accordingly. Thought is
required.

>
>>
>> With an XOR gate or diode mixer PD, you need a second PD plus a 90
>> degree phase shift. One excellent method is to use a divide-by-4
>> Johnson counter (aka walking ring). Run the main PD off one output and
>> the lock detector off the other.
>>
>> Cheers
>>
>> Phil Hobbs
>
> What you mean is a "close-to-lock" detector is "easy" ?:-)
>
> ...Jim Thompson

No. A PLL is in lock when the frequency error is zero, and the variance
of the phase error is bounded. You obviously haven't spent enough time
in low SNR situations. ;)


Cheers

Phil Hobbs


--
Dr Philip C D Hobbs
Principal
ElectroOptical Innovations
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058
hobbs at electrooptical dot net
http://electrooptical.net