Maker Pro
Maker Pro

PLL Lock to an Offset Frequency

J

Jim Thompson

Jan 1, 1970
0
Jim, with all due respect, you need to think about what I wrote in my
original post. The sign of the loop gain is opposite for USB and LSB
for a given PD null, i.e. if you use an XOR or a diode mixer whose nulls
are at quadrature, if the loop wants to lock up at +pi/2 on USB, it'll
lock up at -pi/2 on LSB. With a PFD, one sideband will have the right
sign of loop gain to lock up at 0, where (as you point out) everything
is copacetic.

Aha! Now I see our miscommunication. You said, "If you use a
**frequency-phase detector** or other very asymmetric PD". An XOR is
NOT a PFD (phase-frequency detector, AKA edge-matcher), it's a PD
(phase detector, AKA MIXER). PFD's are used in synthesizers. I am
co-inventor of the first PFD, the Motorola MC4044.
The point I was making in my original post is that the other sideband
will have to try locking up at +- pi, where there's a ruddy great
cliff--its PD gain there is like Vdd /2*pi*(f_0*t_PD), i.e. something
like 500 times larger than the other null. Of course it's noisy and
possibly metastable there, but the point is that the loop gain is
_huge_, so no lock will occur there. Thus with a tiny bit of
acquisition aiding, e.g. 2 resistors and a cap in positive FB around the
loop amplifier, you can make a reliable lock to one sideband and not
the other, _without_needing_a_SSB_mixer_.

Cheers,

Phil Hobbs

Huh? The OP wanted to **generate** either an upper or lower sideband.
An XOR mixer isn't going to do that.

...Jim Thompson
 
P

Phil Hobbs

Jan 1, 1970
0
Jim said:
Aha! Now I see our miscommunication. You said, "If you use a
**frequency-phase detector** or other very asymmetric PD". An XOR is
NOT a PFD (phase-frequency detector, AKA edge-matcher), it's a PD
(phase detector, AKA MIXER). PFD's are used in synthesizers. I am
co-inventor of the first PFD, the Motorola MC4044.




Huh? The OP wanted to **generate** either an upper or lower sideband.
An XOR mixer isn't going to do that.

...Jim Thompson
Forget it. Other people understood.

Cheers,

Phil Hobbs
 
J

Jim Thompson

Jan 1, 1970
0
Forget it. Other people understood.

Cheers,

Phil Hobbs

OK, Maybe I'm dense this afternoon (not that unusual for me :).
Explain on me how *exactly* you're getting from 8MHz to 8.125MHz using
an XOR.

And the "Other people <who> understood" please join in. Laughing at
my density is allowed ;-)

...Jim Thompson
 
P

Phil Hobbs

Jan 1, 1970
0
Jim said:
[snip]
The point I was making in my original post is that the other sideband
will have to try locking up at +- pi, where there's a ruddy great
cliff--its PD gain there is like Vdd /2*pi*(f_0*t_PD), i.e. something
like 500 times larger than the other null. Of course it's noisy and
possibly metastable there, but the point is that the loop gain is
_huge_, so no lock will occur there. Thus with a tiny bit of
acquisition aiding, e.g. 2 resistors and a cap in positive FB around the
loop amplifier, you can make a reliable lock to one sideband and not
the other, _without_needing_a_SSB_mixer_.

Cheers,

Phil Hobbs


Huh? The OP wanted to **generate** either an upper or lower sideband.
An XOR mixer isn't going to do that.

...Jim Thompson

Forget it. Other people understood.

Cheers,

Phil Hobbs


OK, Maybe I'm dense this afternoon (not that unusual for me :).
Explain on me how *exactly* you're getting from 8MHz to 8.125MHz using
an XOR.

And the "Other people <who> understood" please join in. Laughing at
my density is allowed ;-)

...Jim Thompson

Perhaps we were starting from different points.

Instead of using a SSB mixer, use an offset PLL as suggested by the OP,
running at 1:1. This is an alternative to the 64x PLL with the 125-kHz
reference frequency, that avoids the phase noise penalty of frequency
multiplication (in this case, 20*log(64) or 36 dB). Something like this:

USB/LSB----------------------------------------------+
|
|
|
125 kHz ref--------------------------------+ |
| |
_V___ _V__
8 MHz carrier------\\---\ RC=700 ns | | | X |
)) >---RRRR---+--->| PFD |---| +-1|---------+
+-//---/ | |_____| |____| |
| R
| (xor) CCC R
| | +-CCC--RRR--+ R
| GND ___ | /| | |
| / \ | /-|----+---+
+-----------<----------| VCO |<--+---< |
| | \+|----+
\___/ \| |
GND

The nice feature of a correctly-designed offset PLL, using a PFD, is
that you can get rid of the other sideband to any accuracy you like with
a lowpass filter on the loop control voltage--using all jellybean parts
zero inductors, and zero tweaks. You can choose USB or LSB by switching
the sign of the loop gain, e.g. by swapping PFD outputs, negating them
with an extra XOR each, or using a +-unity gain amp. It would be almost
as cheap to generate all three signals and switch the outputs with a mux.

Thus instead of needing to use fancy things like SSB mixers, narrow
filters, or 500 MHz divide-by-Ns, or accepting the 36-dB phase noise
penalty of multiplication, you can do it all with a PFD, one or two op
amps, and a few gates, none of which will need any adjustment. This is
all a consequence of the nice phase properties of offset PLLs, as
described previously. Which works the best depends on the VCO design
and other constraints--e.g. an RC VCO is probably noisy enough that the
36-dB phase noise added to the reference signal is a drop in the bucket.


Cheers,

Phil Hobbs
 
J

Jim Thompson

Jan 1, 1970
0
On Fri, 14 Oct 2005 18:39:50 -0400, Phil Hobbs

[snip]
Perhaps we were starting from different points.

Instead of using a SSB mixer, use an offset PLL as suggested by the OP,
running at 1:1. This is an alternative to the 64x PLL with the 125-kHz
reference frequency, that avoids the phase noise penalty of frequency
multiplication (in this case, 20*log(64) or 36 dB). Something like this:

USB/LSB----------------------------------------------+
|
|
|
125 kHz ref--------------------------------+ |
| |
_ _V__ _ _V__
8 MHz carrier------\\---\ RC=700 ns | | | X |
)) >---RRRR---+--->| PFD |---| +-1|---------+
+-//---/ | |_____| |____| |
| R
| (xor) CCC R
| | +-CCC--RRR--+ R
| GND ___ | /| | |
| / \ | /-|----+---+
+-----------<----------| VCO |<--+---< |
| | \+|----+
\___/ \| |
GND

The nice feature of a correctly-designed offset PLL, using a PFD, is
that you can get rid of the other sideband to any accuracy you like with
a lowpass filter on the loop control voltage--using all jellybean parts
zero inductors, and zero tweaks. You can choose USB or LSB by switching
the sign of the loop gain, e.g. by swapping PFD outputs, negating them
with an extra XOR each, or using a +-unity gain amp. It would be almost
as cheap to generate all three signals and switch the outputs with a mux.

Thus instead of needing to use fancy things like SSB mixers, narrow
filters, or 500 MHz divide-by-Ns, or accepting the 36-dB phase noise
penalty of multiplication, you can do it all with a PFD, one or two op
amps, and a few gates, none of which will need any adjustment. This is
all a consequence of the nice phase properties of offset PLLs, as
described previously. Which works the best depends on the VCO design
and other constraints--e.g. an RC VCO is probably noisy enough that the
36-dB phase noise added to the reference signal is a drop in the bucket.


Cheers,

Phil Hobbs

OK. Now I follow you. Somehow I missed the "offset loop".

I like it!

I've seen similar things done using a D-Flop as a harmonic mixer.

...Jim Thompson
 
J

Jim Thompson

Jan 1, 1970
0
On Fri, 14 Oct 2005 18:39:50 -0400, Phil Hobbs

[snip]
Perhaps we were starting from different points.

Instead of using a SSB mixer, use an offset PLL as suggested by the OP,
running at 1:1. This is an alternative to the 64x PLL with the 125-kHz
reference frequency, that avoids the phase noise penalty of frequency
multiplication (in this case, 20*log(64) or 36 dB). Something like this:

USB/LSB----------------------------------------------+
|
|
|
125 kHz ref--------------------------------+ |
| |
_ _V__ _ _V__
8 MHz carrier------\\---\ RC=700 ns | | | X |
)) >---RRRR---+--->| PFD |---| +-1|---------+
+-//---/ | |_____| |____| |
| R
| (xor) CCC R
| | +-CCC--RRR--+ R
| GND ___ | /| | |
| / \ | /-|----+---+
+-----------<----------| VCO |<--+---< |
| | \+|----+
\___/ \| |
GND

The nice feature of a correctly-designed offset PLL, using a PFD, is
that you can get rid of the other sideband to any accuracy you like with
a lowpass filter on the loop control voltage--using all jellybean parts
zero inductors, and zero tweaks. You can choose USB or LSB by switching
the sign of the loop gain, e.g. by swapping PFD outputs, negating them
with an extra XOR each, or using a +-unity gain amp. It would be almost
as cheap to generate all three signals and switch the outputs with a mux.

Thus instead of needing to use fancy things like SSB mixers, narrow
filters, or 500 MHz divide-by-Ns, or accepting the 36-dB phase noise
penalty of multiplication, you can do it all with a PFD, one or two op
amps, and a few gates, none of which will need any adjustment. This is
all a consequence of the nice phase properties of offset PLLs, as
described previously. Which works the best depends on the VCO design
and other constraints--e.g. an RC VCO is probably noisy enough that the
36-dB phase noise added to the reference signal is a drop in the bucket.


Cheers,

Phil Hobbs

OK. Now I follow you. Somehow I missed the "offset loop".

I like it!

I've seen similar things done using a D-Flop as a harmonic mixer.

...Jim Thompson

Phil,

The more I study your scheme the more I like it! I have a 1056MHz ±
132MHz situation that would be a perfect application of your
configuration, because it's clearly superior for high ratios of
carrier to offset.

...Jim Thompson
 
Frank said:
Why would you want to do that? Divide the 8MHz by 640 and use that as
your new reference.

Dividing down real-world signals is fine, as long as they are squeaky
clean.
Phase-locked loops can act like flywheels, filtering out occasional
spikes and ringing artefacts, which can do nasty things to the counter
doing the divide-by-640.
We only know that the input frequency is 8MHz.

That's his example, not his problem - which he hasn't revealed.
That's as absolute
as it comes, and 1/64th of it is therefore an absolute 125KHz.

Again, we don't know what he is really trying to do.
____________
Bill Sloman, Nijmegen
 
R

Rich Grise

Jan 1, 1970
0
Hi
I want to design a PLL which must lock to a
frequncy which is a bit different from input frequency.
For example input Frequncy is 8MHz and PLL must synthisize
a 8.125.000Hz or 7.875.000Hz (8MHz+- 125KHz) output signal.
How can i do this?

Finals week again?

Good Luck!
Rich
 
R

Rich Grise, Plainclothes Hippie

Jan 1, 1970
0
[snip]
Customer: "Hey! Your equipment is a piece of xxxx and my foot hurts
really bad!"

This is worse when your customer is also your boss, which is why I'm now
a consultant...

I decided that in 1973 ;-)

...Jim Thompson

Whoa, dood! You _ARE_ old! In 1973 I was getting laid and smoking Thai
Stick. ;-P (not necessarily simultaneously, but usually. ;-P )

Cheers!
Rich
 
J

Jim Thompson

Jan 1, 1970
0
Finals week again?

Good Luck!
Rich

It was a fruitful post. From Phil Hobbs I gained an idea I really
should have known.

...Jim Thompson
 
J

Jim Thompson

Jan 1, 1970
0
[snip]
Customer: "Hey! Your equipment is a piece of xxxx and my foot hurts
really bad!"

This is worse when your customer is also your boss, which is why I'm now
a consultant...

I decided that in 1973 ;-)

...Jim Thompson

Whoa, dood! You _ARE_ old! In 1973 I was getting laid and smoking Thai
Stick. ;-P (not necessarily simultaneously, but usually. ;-P )

Cheers!
Rich

I was 33 and already had 4 kids.

...Jim Thompson
 
T

Terry Given

Jan 1, 1970
0
Jim said:
It was a fruitful post. From Phil Hobbs I gained an idea I really
should have known.

...Jim Thompson

and thats the great thing about SED.

Cheers
Terry
 
R

Rich Grise

Jan 1, 1970
0
and thats the great thing about SED.

OK, I can live with this.

Frankly, I'd say it's just _one_ of the great things about s.e.d,
not the least of which is, the people here are probably the smartest
on the planet, and they have a sense of humour!

Ever checked out comp.lang.c ? Boy! Talk about a bunch of prigs! ;-)

Thanks,
Rich
 
X

xray

Jan 1, 1970
0
I was 33 and already had 4 kids.

...Jim Thompson

In those days Catholicism was indistinguishable because effective birth
control was just catching on.

Dunno, just an observation.
 
J

Jim Thompson

Jan 1, 1970
0
In those days Catholicism was indistinguishable because effective birth
control was just catching on.

Dunno, just an observation.

Nonsense! The classic condom has been around forever, you could even
buy 'em in Massachusetts (under the counter), and diaphragms and birth
control pills were already in use.

We had two girls and then I decided I'd like two boys.

Besides the dirt bag liberal weenies were already in motion trying to
restrict birth rate. Remember Paul Erlich and his multiple
appearances on The Tonight Show, touting Zero Population Growth (ZPG)?

So I decided SOMEONE had to keep conservatism going ;-)

...Jim Thompson
 
P

Phil Hobbs

Jan 1, 1970
0
Jim said:
OK. Now I follow you. Somehow I missed the "offset loop".

I like it!

I've seen similar things done using a D-Flop as a harmonic mixer.

Yes, I've done that back in the palmy days--about 1982, in my first
engineering job (me with my new astronomy B. Sc.), I had to design and
prototype a pilot tone generator at 70 +- (10/22) MHz--i.e.
70.45454545... and 69.54545454... MHz, don't ask me why, I didn't design
the frequency plan. IIRC I used a 74LS293 as a divide-by-11 on the 10
MHz reference signal (jam reset on 12-count) followed by a divide-by-two
to get 50% duty cycle, and clocked both halves of a 74S74 with the
same 10 MHz. Those were the harmonic mixers for two offset 1:1 PLLs at
454.5... kHz, using Mini-Circuits RPD-1 diode bridge phase detectors,
rather than PFDs. I really liked RPD-1s, because you got about a volt
out of them, and they were really quiet.

The S74 output was very noisy due to metastability--the trailing edge
jitter was especially non-pretty, and once in awhile it missed a pulse.
I got round this by using a homemade VCXO on each side, and keeping
the loop BW small. This was okay for this board, since the difference
between the two pilot tones was going to be frequency-multiplied and
used as the frequency reference for a 14 GHz direct-broadcast satellite
system, and the PTG only had to be in the base station, not the
remotes. (I think this was the first commercial DBS ever--it went on
sale in early 1984.)

The spectrum analyzer trace of the signal from that S74 harmonic mixer
was not a pretty sight--it jiggled around like a water fountain. I
certainly wouldn't use that technique again--XORs work as harmonic
mixers too, though you don't get the same signal amplitude of course.

It may be that using more recent logic parts and choosing the locking
phase very carefully would keep the metastability under control better
than I was able to do. As an 10:1 harmonic mixer at 70 MHz, the edges
were just too close together for STTL.

Cheers,

Phil Hobbs
 
Top