that PLL again

I use Foxit Reader to view PDFs and CutePDF to make them.

Be careful if you install Foxit, to turn off the tag-along options.

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

Yikes, that's a bang-bang phase detector at 10 MHz!

It would presumably lock with the average PD output of VDD/2, and 3 ns of delay (1.5 with the DDR trick.) The PD output would be very noisy.

Rob was mumbling about a technique that makes sorta10 with better edge statistics, analogous to a 2nd order delta-sigma. That pushes up the spectrum of the noise from the PD, so it filters nicer.

The more serious issue is that this is a sufficiently fun problem that all sorts of people want to play with it, instead of doing the grunt work assigned to them.

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

Sort of, but without the huge-unknown-gain problem, and hence without the possible loop stability issues. You know the average value of K_phi

*a priori*.

The SNR would be the no worse than with the XOR--both the jitter signal and the desired error signal would be amplified the same amount. It would improve the DC stability by several times, though, for the same amount of hardware.

Not a bad idea at all. "Jitter shaping" has a nice marketing ring to it.

Except round here, where 75% of the people throw rocks instead. ;)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

I use PDF-Xchange Viewer (free): Since I've abandoned Acrobat, I've had no hangs, crashes, memory leaks, or other Acrobat "features". PDF-Xchange also has a built in OCR reader, that will make a scanned document readable.

--
Jeff Liebermann     jeffl@cruzio.com 
150 Felker St #D    http://www.LearnByDestroying.com 
Santa Cruz CA 95060 http://802.11junk.com 
Skype: JeffLiebermann     AE6KS    831-336-2558

[snip] [snip]

Ron Treadway, in the late '60's concocted basically a mixer with offset that did just that. That part may still exist in MECL/PECL, though I can't locate it. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| San Tan Valley, AZ 85142     Skype: skypeanalog  |             | 
| 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.

Everyone is always complaining about Acrobat... I've had it for years (but, indeed, did stop "upgrading" at v7.1.0)... but I've never experienced the crashes everyone complains of.

I also retained v4, which, after creating with v7, I "Reduce file size" to v4, because edits and mark-ups are so much easier in v4.

The only recent quirk is that I find v7 improperly "prints" many websites (screwing up visual presentation of links, etc.). I cure that by first printing to PostScript with MS Publisher Color Printer, the using Acrobat to convert to PDF. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| San Tan Valley, AZ 85142     Skype: skypeanalog  |             | 
| 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.

The pure bang-bang has no loop stability problem. It just hunts fast-early/slow-late. It's a thermostat. Running at 80 KHz, I'd have to wiggle the VCXO frequency by about 0.1 PPM, which is not unreasonable. Jitter smears out the PD transfer function into an s-curve, but doesn't seem to affect loop dynamics much.

(There is a loop stability problem from the added pole internal to all VCXOs, essentially a lowpass in the varicap driver circuit. The relevent VCXO spec is "modulation bandwidth", which is either not specified, or the universal "10 KHz minimum". 10K seriously mucks my loop. But there's a simple fix!)

Yes, it would be much better for DC stability than the XOR, or any

180-degree-range phase detector. I'll show it to Rob and maybe he can consider it. Or we might build it into the test board.

Nice hybrid! Nobody loses!

Well, they must be people-people. I'm an electronics-people.

Sounds like we have more fun than they do.

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

I'm sure we must be reinventing things, except that we want picosecond accurate *time* locking, and most PLL/RF jocks work in the frequency domain. Locking to and decoding high speed data streams is closer to what we want to do, but we want accuracy to 1/100000 UI, which data decoding doesn't need.

--

John Larkin         Highland Technology, Inc 
picosecond timing   laser drivers and controllers 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

Oh... conservation of energy, conservation of momentum, conservation of charge seems to be a good approximation.

A daily mail reader typically says, that theory is a load of bullocks, cos I did it. Unfortunatly,the daily mail reader does not understand statistics.

Kevin Aylward

- SuperSpice

"daily mail" == Daily Mail newspaper ?>:-}

That seems to be a scandal sheet that rivals Boston's Daily Record, which, when I was a student at MIT, we called the "Daily RinkyDink" ;-) ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| San Tan Valley, AZ 85142     Skype: skypeanalog  |             | 
| 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.

Rob has done a sim of the sorta10 generator in Python, and then FFT'd it, and got a forest of spurs spaced 160 KHz apart. The next trick would be to add the phase detector and the rest of the loop, and see what the ultimate jitter might be.

I think I can add a time-domain noise generator into my LT Spice sim of the bang-bang loop. I could add the jitter to the 10 MHz reference, with some noise spectrum shaping just for fun. That, and the loop filtering, would sort of linearize the bang-bang thing.

Days of simulation.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

Nah, my personal desktop runs Altium 3D view much faster than the one at work. ;-)

Oddly, 2D (schematic) is slightly slower.

Tim

--
Seven Transistor Labs 
Electrical Engineering Consultation 
Website: http://seventransistorlabs.com

Another example: turret tuner assemblies for TV sets had multiple coils, all tuned, on each of the dozen or so clickstops. Tuning was done with a motorized probe that deformed the coil, and took only seconds to 'trim' the LO and RF section, then advance the turret switch and do the next.

Modern laser-trimmed resistors (including on-chip) are ubiquitous, we just don't do it manually, and tend to forget that it IS being done. All that precision... is done with a mechanical hand on the trim adjustment.

I do not need 3D view much. Occasionally for a mechanical drawing and then even an old Foxconn G33 mobo with integrated graphics is fine.

--
Regards, Joerg 

http://www.analogconsultants.com/

that's strange because I have a XPS 8700 I7 system and it does have spare card slots in it along with a bay or two.

Maybe we aint got the same machine. My has the virtualization functions on the MB and it came with 12G upgradeable to 16G.

Jamie

Don't forget that Spice allows you to read a PWL signal from a file; this file could the output of Rob's Python script. This allows you to simulate the exact jitter waveform.

Regards, Allan

I don't know yet since mine won't be delivered much before New Years. As Forrest Gump put it, life is like a box of chocolates, you never know what you're gonna get.

I think they can be upped to 32GB but I figured 8GB is way enough for me. Maybe I even turn 2-3 GB into RAM disk for faster RAW file access in LTSpice. Or I'll get a SSD for it. The main thing was that many of the other XPS 8700 offers came with Windows 8 and that's not going to happen here. I wanted Win 7 64-bit Professional.

--
Regards, Joerg 

http://www.analogconsultants.com/

Google Earth won't work any more on my old HP computers. It demands a

3D graphics card. It does work on my newer Vaio laptop.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

. It

al

nd

ll

c n

r,

nd

ur

Probably not.

No. It's just a noise generator. You throw away the phase information by co nverting any phase error into +50nsec or -50nsec. Since - with fractional-n - the typical phase error is 3nsec, and systematic, you just add a lot mor e noise into the phase-feedback, which you have to filter out.

Just the level of error signal which you are having to average to zero. Run ning at 80kHz, rather than 10MHz, you have to wait a lot longer to collect enough signal to guarantee that the average is small enough not to wiggle y our VCO phase, and you've promoted what were originally sub-nanosecond phas e errors into either +50nsec or -50nsec. It's insanely dumb.

As discussed in Floyd M. Gardener's "Phaselock Techniques" ISBN 0-471-04294

-3 which was published in 1979. See Fig. 2.2

It would have a predictable pattern of finite offsets. Calling that "noise" misses the point that it averages to zero a lot faster - and more predicta bly - than random noise.

.

"Noise shaping" is better - the sigma-delta A/D converter business has spen t years lodging it in customer minds.

With an unhealthy dash of narcissism and NIH.

It depends how much you enjoy solving problems you have created for yoursel f. My preference was always to go for existing solutions - when they worked . There were plenty of real problems left to solve.

--
Bill Sloman, Sydney