really slow PLL

It's certainly possible. However, within whatever tiny loop bandwidth you wound up with, the lockers would still have

20 log(40e6) = 152 dB

higher phase noise than the lockee.

It would be interesting to do the math to see whether it's possible to generate a concensus lock for the group: if you get everybody close enough, just sum their sine wave outputs and lock each one of them to that, with some bit of AC coupling or something so that they don't all wander together off to the edge of the tuning range.

Maybe have one doing the locking with a phase shifter and the others with VCOs, or something like that.

Definitely a partly-baked idea, but surely one could do better than 152 dB!

Cheers

Phil Hobbs

GPS has that problem too.

Each box is basically a multichannel power supply, but channels can be programmed to do stuff in timed sequences. I want different box outputs to time align within, say, one millisecond longterm once programs are kicked off together. So, many microseconds of equivalent RMS phase noise is OK as long as we stay time aligned longterm.

If a follower is told to start locking, it could timestamp the first incoming 1 PPS with a giant counter clocked by its local 40 MHz VCO. If a later 1 PPS edge appears to arrive too soon, we could speed up our VCXO by, say, 1 PPM, and vice versa. So longterm it walks into alignment with the 1 PPS and eventually dithers a microsecond per second. Noise on the coax gets fixed over time too.

That's better than just measuring the 1 Hz period once a second, tweaking the clock, and then throwing away that measurement. I want a time lock, not a frequency lock.

Absolutely. The scary 152 dB number doesn't mean that doing something like that is automatically a bad idea.

Being an old RF and ultrastable laser guy, though, it does make my ears perk up. ;)

Cheers

Phil Hobbs

It sounds like you're looking for a protocol like DMX if what you want is to trigger sequences of events across boxes to within a millisecond, I don't understand what this lock-the-40 MHz across boxes is about.

I like thermostats, single-bit-feedback control loops.

We have a couple of boxes that do fan control based on interior temperature. Once a second, if it's above the setpoint, ratchet fan speed up some fixed amount, 1% maybe. If it's cooler than the setpoint, step fan speed down. There's no acoustic drama and it's perfectly stable.

It dithers around the setpoint but nobody notices.

This is immune to classic control theory so the concept annoys some people but it works great.

Each box runs a bunch of timed processes. If they are triggered to start together, I don't want their timings to drift apart. Locking their clocks works.

My loop will also lock my boxes to a 1 PPS GPS source, so we can synchronize within a bigger system.

It's easy to explain 1 PPS pulses on BNC connectors.

On a sunny day (Wed, 20 Jul 2022 16:20:57 -0700) it happened John Larkin <jlarkin@highland_atwork_technology.com> wrote in snipped-for-privacy@4ax.com:

So why not use a RGB 24 bit 0xffffff master clock board that drives all slave modules those then need no local oscillator. For event 'sync' sent a DC step every now and then on the same coax and filter it out at the slave sides, to make the slaves jump to action.

Am 21.07.22 um 01:20 schrieb John Larkin:

I have a backburner project of locking 16 MTI-260 oscillators slooowy to another one, and when they are in sync, combine them with an array of Wilkinsons. That should have a nice effect on phase noise by averaging over 16. The CPLD has enough resources to implement that as a delay locked loop with 1 pps, but low hanging fruit first.

No. There is no 1PPS pulse from the sat nor the need for exactly 10 MHz. The sats transmit a pseudo noise sequence that is aligned to the second of their local clock source. The GPS receiver knows the polynomial and runs a local copy of the polynomial. It knows by cross correlation if the local pseudo noise is the same as that of the sat and therefore knows the start of the second. Usually that won't be the case. Then the receiver delays its own polynomial by omitting a clock to the shift register that generates it and tries again. Sooner or later it will fit.

One needs at least reception of 4 sats to get a set of (x, y, z, t) that fits together, usually more. Once one has a lock to one sat, it is possible to get the data of the current constellation of all sats. That helps to speed up the lock to the others by providing a guess of a better starting point for the search.

There is still a long way from synced polynomials to a good fix, for example removal of the Faraday effect by comparing the flight time difference for sats on different frequency bands.

95% of a GPS receiver is software. The 1PPS out of a typical GPS receiver used to be only a port bit of the CPU, synchronous to the second only in the very long run.

If the receiver delivers a nice 10 MHz, the CPU has better information to adjust it.

Gerhard

If you can tolerate 'a few microseconds' on a 40 MHz signal, that's not a phase-lock problem, it's a frequency-lock problem. Why not just run an up/down counter to generate a correction voltage for each non-leading VCO?

If you have an ethernet interface to each unit then Precision Time Protocol should do exactly what you want.

You really need to define longterm before the problem becomes well posed. Do you mean hours, days, weeks or months of runtime?

Have a free running counter on each of the followers and use the value of that after 1s, 10s, 100s to determine the correct tweak to apply locally. Tweaks of 1ppm at a time is rather crude you should be able to determine the right amount to tweak it by better than that. (especially over longer timebases)

Then you probably want to measure the cumulative error over many seconds. Each unit can work out how long it can free run without exceeding tolerance once it has the rough and ready count from the first second. After that you have a good idea of how many seconds you can free run for without having any ambiguities from residual drift.

This is an ancient trick from physics which avoids the smartest students from having to laboriously count every pendulum swing when determining g to maximum possible precision in a given time. It used to be (and probably still is a favourite exam practical). Components needed are very cheap and the whole thing is a good test of experimental technique.

It's actually a time lock problem. If a follower box starts up and sees its first 1 PPS input, it can thereafter declare 1 PPS internal events, based on its local VCO, and then do successive early/late comparisons to the external pulses. And trim its VCXO accordingly.

Where does the 10 MHz come from?

Each box has a 40.0something MHz oscillator. Some simple logic triggers on the 1 second GPS pulse and allows 40 million of these clocks through to the rest of the circuit, then waits for the next pulse, rinse and repeat.

Of course, this will only work in circumstances where it's ok to stop for a few cycles every second.

I wouldn't expect my VCXO to be more than 10 PPM off at the start of the lock request. So I can walk it to within 1 PPM, namely 1 microsecond error, in at most 10 seconds using 1 PPM jogs. If the osc were 50 PPM off, it would take 50 seconds to catch up to the external pulse.

Yes, I don't want to measure period once a second. I want to compare time alignment forever after receiving the first 1 pps pulse.

It's actually simple: first received pulse, start a mod 40 million counter. Every time it rolls over, do an early/late compare to the 1 PPS input, and jog the 40 MHz VCXO 1 PPM in the right direction.

The compare is a dflop, d is the msb of the counter, clock is the 1 PPS input. Occasional metastability is fine; it indicates success.

It doesn't even need to be a 40 million counter. Something a fraction of that will do. 10,000 maybe.

Maybe the counter can just free-run, never get initialized. Gotta do the math on that after I wake up.

Am 21.07.22 um 13:19 schrieb snipped-for-privacy@highlandsniptechnology.com:

Choise of implementer. One local clock generator is needed. This clock determines short term stabiity and phase noise.

My Lucent KS24361 uses 5 MHz MTI-260 double ovens; for redundancy/holdover it has a 2nd unit with another crystal oven without a receiver.

The redundancy units were really hard to sell without the receiver; that's why I have 20 of these MTI-260, got a good price. :-)

They were new old stock built by HP/Agilent for Lucent as replacement parts. They have never been on a telecom tower in China like most of those one gets on ebay.

I have expanded the Lucent to 10 MHZ and with a distribution amplifier:

< >

cheers, Gerhard

A real old time control guy like Tim Wescott would probably be a fan too--the great virtue of a bang-bang controller is that (as you say) it's highly resistant to variations in the _plant_.

Your furnace doesn't go nuts when you have a Christmas party, even though all those people generate a lot of heat, and there's lots of opening and closing of doors and ovens.

Cheers

Phil Hobbs

It's not as efficient as 'dry labbing'. ;)

The tradeoffs are sort of different when you have a $3 CPLD watching things rather than a student.

Cheers

Phil Hobbs (who didn't work that cheap even back when he was a student)

Yeah I don't quite get it, either. My rack of synthesizers can each play one voice of the Maple Leaf Rag via MIDI and they all stay synced together really well, at least over a timespan of several minutes...superficially at least it sounds like he wants a sequencer.

Using the nuts & bolts system clock for synchronization of "user tasks" also makes me uncomfortable, if the device behavior only need to align to the millisecond why not trigger them using some simple network protocol and let their hardware figure out how long a millisecond is independently. Do the timings of these boxes need to be tighter than the Maple Leaf Rag?