Possible PLL Lock Indicator?

Or maybe you're talking an unbounded VCO? That can cause harmonic locks, or no lock at all. ...Jim Thompson

It's the unlocked condition we're talking about, so the PD phase null is no t in view--the PD is putting out a mildly distorted sine wave.

It's the phase of the loop filter/amp transfer function (amplitude/phase vs beat frequency) that matters for pull-in and false lock.

Cheers

Phil Hobbs

Are you disobeying Bode ?:-) You're supposed to put your extra poles well-beyond where the loop gain passes thru 0dB.

Maybe I'm missing something? But I've built a LOT of PLL's that didn't have such a problem. ...Jim Thompson

The loop is unlocked, so the beat frequency is much larger than the loop bandwidth. That's why the asymptotic phase is relevant.

Sure, lots don't. It's not an issue with PFDs, of course, but XOR, multiplier, S/H, and D-flop PDs are vulnerable.

Cheers

Phil Hobbs

Aha! There's an equation somewhere in Gardner that says that loop bandwidth must be such-and-such to guarantee acquisition... otherwise you need "aided" acquisition.

All my analog loops have been multipliers of some sort. I don't use digital equivalents in analog situations.

I once tried that TI digital PLL abortion in a modem application, then ended up using it _only_ for transmit... that's where that S-curve discriminator for receive originated (~1980 :-)

I ran a test using that approach... out from Phoenix to Louisville (a friend's house) then back on a separate line... a really noisy connection (acoustic muff's ;-)... got an excellent BER.

As you narrow the loop BW, the acquisition time goes as 1/f_0**2. That gets old in a hurry. But the real limit is the pull-in signal becoming smaller than the worst-case offset voltage. That's why the Schmitt in my auto-sweep circuit has to provide 2-3x the maximum offset voltage in the op amp and PD combined.

I'm a fan of the Mini Circuits MPD-1, which has a volt or so of output and pretty low offset--1 mV or less. Less offset means less AM-PM conversion.

Cheers

Phil Hobbs

What offset voltage? Are you trying to cope with signals so small that an analog multiplier offset is troublesome? Loop gain, with an analog PD, is signal amplitude dependent.

I don't use off-the-shelf PD's >:-) ...Jim Thompson

We need to distinguish clearly between the in-lock and out-of-lock behaviour. (This has been a continual difficulty in this exchange.)

When the loop is in lock, there's some nice big voltage slope on the PD output--on the order of a volt per radian--so small offsets cause only small phase errors.

Those phase errors do result in small amounts of AM-PM conversion, which you may or may not care about, depending on how much AM noise there is and how tight the phase noise spec is.

When you're out of lock, the nice big PD slope goes away, leaving a beat note and a small amount of DC from the sum of the pull-in signal and the offsets. This is true irrespective of the input amplitude.

The DC pull-in signal is very small, so you care a good deal about offsets (if it weren't, pull-in wouldn't be so slow). With the self-sweeper, unless you short it out when the lock detect is true, it imposes a DC offset at least a few times the worst-case sum of the offsets. You pick it to be that size in order to get reliable locking in a reasonable time, without trashing the AM-PM performance too badly.

Cheers

Phil Hobbs

Gack! Looked it up. That's just a mixer. No wonder you fret offsets.

I have built on-chip PD's up to 5GHz. ...Jim Thompson

-- | James E.Thompson | mens | | Analog Innovations | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | STV, Queen Creek, AZ 85142 Skype: skypeanalog | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at

Spice is like a sports car... Performance only as good as the person behind the wheel.

It's a very strong mixer, with high output voltage (1V/rad), very low noise, and low offset (200 uV or so). In general it's a much better choice than Gilbert cells for making rugged analogue PLLs.

Cheers

Phil Hobbs

With what analog input signal level? ...Jim Thompson

Here's an analog phase detector (~1974) with coherent AGC...

to get both the loop gain constant _and_ get synchronous detection of the modulation. ...Jim Thompson

-- | James E.Thompson | mens | | Analog Innovations | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | STV, Queen Creek, AZ 85142 Skype: skypeanalog | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at

Spice is like a sports car... Performance only as good as the person behind the wheel.

Should have mentioned... this was BC (before CAD), no simulation, breadboarded, tested in a farmer's field next to Bowmar, then "chip'd" ;-) ...Jim Thompson

Nominal is about +7 dBm LO, slightly less on the RF.

Cheers

Phil Hobbs

That's 1.4159V P-P (assuming 50 Ohms :-( ...Jim Thompson

Yup, that's been the standard combination for about that long. You also need wideband AGC, or else when the loop unlocks the gain will max out and the resulting distortion and noise intermod will prevent the loop from locking.

Cheers

Phil Hobbs

Wouldn't it be useful to have an analog version with an output voltage proportional to the phase difference?

Possibly. Usually a lock indicator is used to connect critically timed devices _after_ lock is assured. ...Jim Thompson

What I'm thinking about is that lock is never perfect, so in your circuit there the NOT gates are required to have just the right speed to indicate the amount of phase error you're looking for. But if there was an analog output you could monitor it for exactly the max error you want.