PLL input filter vs. loop filter

[Ken Smith]

The problem with a filter in front of the PLL is that you can't "remove"
missing pulses. A low-pass filter would work fine if you were seeing
extra pulses...

That isn't true. A filter can fill in a cycle or two. It can remove the
side bands that represent the rapid AM modulation to zero. Another way to
think of it is a tuned circuit with a high Q can make a continuous output
if it only gets driven on every other cycle.

 

[Jim Thompson]

finally a good answer by someone who knows. Take a venerable NE567 or
XR215A, put a bit of amplification and maybe 200Hz highpass in front of it,
so the signal is in the 20 to 100mV rms range. Set the BW to 1% and after
200cycles the output goes low and signalles your PLL locked. You can tap
into the squarewave output to get a stable reference without missing pulses.

I was doing it with TACAN receivers in the mid '60's.

...Jim Thompson

 

[Ken Smith]

Can't I just notch the 18Hz?

Yes, a fairly simple notch filter can remove 18Hz and not effect signals
above 200Hz enough to matter. You can combine it with a high pass filter.

Consider this circuit:

FB
!
--- 0.2u
--- FB
85K ! 85K !\ !
-+-/\/\/--+--/\/\/---+--+-----! >--+-- Out
! ! ! !/
--!!----+---!!----- \ Gain=4
0.1u ! 0.1u /
/ \ 10K
\ /
/43K !
\ GND
!
FB


This circuit only has one more resistor ( 10K ) than the simple notch case
but it gives a -20dB cut to the low frequency end.

 

[Tom Becker]

No and [] I told you where I saw it used ...

You did? I missed that and I think it's not in this thread. Where?


Tom

 

[Tom Becker]

If you are making [a] multiplying PLL and have a lowpass before the comparitor...

Terminology, please. Is a multiplying PLL a clock-multiplying PLL? Or
do you mean a multiplying phase detector, mentioned earlier? The
fed-back sawtooth does which?


Tom

 

[RST Engineering \(jw\)]

With 6J6s, 12BA6s, and 12AT6s?

{;-)

Jim

 

[Jim Thompson]

With 6J6s, 12BA6s, and 12AT6s?

{;-)

Jim

In the mid '60's we were well into the integrated circuit era
(Motorola SPD).

Now in the mid '50's I was building tube stuff.

...Jim Thompson

 

[RST Engineering \(jw\)]

You must have a couple of years on me, Jim, because in the mid '50s I was
still in sixth grade. It was 8th grade before I started building tube
stuff.

And, as I vaguely recall from an early sophomore college job in the
early-mid 50s, we were still gaga over the 709 opamp and Fairchild's
3-device array of digital RTL stuff.

Jim

 

[Jim Thompson]

You must have a couple of years on me, Jim, because in the mid '50s I was
still in sixth grade.

I just turned 66.

It was 8th grade before I started building tube
stuff.

In 1955 I was in 9th grade. (Last year of "Junior High", now called
Middle School.)

My father owned a hardware store and a radio/TV repair shop, so I had
all kinds of goodies available. In 1956 my father became a Raytheon
wholesaler, so I then had CK722's and CK760's as well ;-)

And, as I vaguely recall from an early sophomore college job in the
early-mid 50s, we were still gaga over the 709 opamp and Fairchild's
3-device array of digital RTL stuff.

Jim

...Jim Thompson

 

[RST Engineering \(jw\)]

I just turned 66.

17 Nov '43

My father owned a hardware store and a radio/TV repair shop, so I had
all kinds of goodies available. In 1956 my father became a Raytheon
wholesaler, so I then had CK722's and CK760's as well ;-)

Lucky SOB. I remember I spent an entire month's earnings in '57 from a
paper route making a transistor headphone amp for a crystal set -- yes, with
galena and a catswhisker. The device was in fact a CK722.

Jim

 

[Tom Becker]

... Take a venerable NE567...

I dug through the junk box and came up with an LM565 and a zillion 567s.
I never really studied the 567's schematic before but there it is, a
balanced modulator. 565, too.

I'm off to play with old technology. Thanks, Ban.

Tom
ex da Saluzzo, multi anni fa.

 

[Ken Smith]

No and [] I told you where I saw it used ...

You did? I missed that and I think it's not in this thread. Where?

Sorry, I posted that before I really was awake.

It was supposed to read:

"No and if I told you where I saw it used ...
I'd have to kill you."

Its an old joke and in this case not really true but since I don't own the
design and don't have contact with the owner, It isn't likely I'll be able
to prvide a link.

 

[Ken Smith]

If you are making [a] multiplying PLL and have a lowpass before the

comparitor...

Terminology, please. Is a multiplying PLL a clock-multiplying PLL? Or
do you mean a multiplying phase detector, mentioned earlier? The
fed-back sawtooth does which?

Ok, I'll try to be clear, although this is a Monday:

Assume you have a VCO running at several times the input frequency. ie: a
"multiplying PLL".

It is a fairly easy matter to filter the input signal so that it has no
components above some fraction of the VCO's frequency. The filter really
should remove the 3rd harmonic of the expected frequency and above.

The VCO's output can be made into a (not super good) sawtooth with a
simple low pass filter.

If the input signal is connected to one input of a comparitor and the
other side of the comparitor is driven with a scaled version of the
sawtooth, you get a PWMed output from the comparitor. If the sawtooth's
amplitude is adjusted so that it is just a little bigger than the biggest
signal, the PWM doesn't saturate.

Now when this PWM train goes through the XOR, you get a bunch of
frequencies on the output of the XOR.

(a) The frequency content of the input multiplied by the fundamental of
the "feedback" signal of the PLL. This is the signal that you want for
making the PLL track the input.

(b) The frequency contents of the input signal multiplied by the
harmonics of the "feedback" signal. This is unwanted and the loop filter
on the PLL usually will remove it.

(c) The PWM frequency and its harmonics. This is another unwanted signal
that gets eaten up by the loop filter.

(d) The PWM frequency and its harmonics intermixed with the harmonics of
the "feedback" signal. Assuming that the feedback signal is 50-50 duty
cycle and that the VCO is running at 2^N times the feedback, all of the
energy in this also ends up at high frequencies. This also gets eaten by
the loop filter.

 

[Tom Becker]

Ok, I'll try to be clear...

Well done; thanks. I think I get it.

I believe you described a synchronous voltage to time converter, sort
of, i.e. a synchronous voltage to phase converter, ahead of the phase
comparator. Seems a natural. Yes, sort of?

Off to the breadboard. Thanks, all.


Tom

 

[Ban]

I dug through the junk box and came up with an LM565 and a zillion
567s. I never really studied the 567's schematic before but there it
is, a balanced modulator. 565, too.

I'm off to play with old technology. Thanks, Ban.

Tom
ex da Saluzzo, multi anni fa.

Tom, If you want something more sophisticated this is a freqency difference
discriminator, called
Quadricorrelator by Richman/Bellisio:

.-------.
| | .----. delta sin
.--o Mult |-----o LP |---------o--------------------.
| | | '----' | |
| '---o---' | |
| | loop3 | loop1 |
| | cos.-------. | .---o---.
| '--------| | .-o-. .----. | |
o--o | VCO o-----|Addo----| LP o----| Mult |
| .--------| | '---' '----' | |
| | sin'-------' '---o---'
| | loop2 |
| .---o---. .-------. |
| | | .----. | | |
'--o Mult |-----o LP |--------------o d/dt |-------'
| | '----' delta cos | |
'-------' '-------'
(created by AACircuit v1.28 beta 10/06/04 www.tech-chat.de)

Loop1 and 2 form a wideband frequency locked loop and loop3 is a
phase-locked-loop. When the signal comes into the lock range of the PLL the
FLL can be disabled to avoid interaction. The circuit allows fast response.

 

[Ken Smith]

I believe you described a synchronous voltage to time converter, sort
of, i.e. a synchronous voltage to phase converter, ahead of the phase
comparator. Seems a natural. Yes, sort of?

Its more a voltage to pulse width than to phase. remember the input to the
comparitor is nearly symmetrical in time.

 

[Tom Becker]

Its more a voltage to pulse width than to phase. remember the input to the
comparitor is nearly symmetrical in time.

Pulse width, yes, synchronized with the VCO. The trailing edge is
phase. No?


Tom

 

[Ken Smith]

Pulse width, yes, synchronized with the VCO. The trailing edge is
phase. No?

Yes but so is the rising edge and it shifts in the other direction:

 

[Tom Becker]

Yes, but so is the rising edge and it shifts in the other direction:

Do you mean sawtooth - or triangle wave?


Tom

 

[Ken Smith]

Do you mean sawtooth - or triangle wave?

Triangle, actually.


* *
* * * *
* * *