Low Jitter 20MHz oscillator

my turn : Jim tom tom tom ... -D

I'd buy a packaged oscillator, and make sure it and the ADC have very quiet, well bypassed power supplies. It's best if the oscillator supply and the ADC logic supply are the same rail, so thresholds track. If your oscillator slews at a volt per ns, a millivolt of equivalent noise is a picosecond.

--

John Larkin                  Highland Technology Inc 
www.highlandtechnology.com   jlarkin at highlandtechnology dot com    

Precision electronic instrumentation 
Picosecond-resolution Digital Delay and Pulse generators 
Custom timing and laser controllers 
Photonics and fiberoptic TTL data links 
VME  analog, thermocouple, LVDT, synchro, tachometer 
Multichannel arbitrary waveform generators

You are absolutely right. A home-made oscillator may not even oscillate, and is unlikely to be on-frequency. An HC gate will have a prop delay tempco in the 10s of PS per degree C, so local temperature fluctuations convert to jitter. It will be very power supply sensitive, too. Crystal drive level is uncontrolled, usually too high. An HC will make slow edges, another jitter hazard.

A $2 oscillator can have a couple ps RMS cycle-cycle jitter, sub-ns risetime, and will usually arrive working to withing a few PPM of the advertised frequency.

--

John Larkin                  Highland Technology Inc 
www.highlandtechnology.com   jlarkin at highlandtechnology dot com    

Precision electronic instrumentation 
Picosecond-resolution Digital Delay and Pulse generators 
Custom timing and laser controllers 
Photonics and fiberoptic TTL data links 
VME  analog, thermocouple, LVDT, synchro, tachometer 
Multichannel arbitrary waveform generators

He's a redneck, and at the low end of that scale.

--

John Larkin Highland Technology Inc

jlarkin at highlandtechnology dot com

Precision electronic instrumentation Picosecond-resolution Digital Delay and Pulse generators Custom timing and laser controllers Photonics and fiberoptic TTL data links VME analog, thermocouple, LVDT, synchro, tachometer Multichannel arbitrary waveform generators

    ...Jim Thompson       |    mens     |     |     et      |  |       |

Hi Jim, What's the advantage of the U version for xtal drive?

Thanks, George H.

Silicon. ;-)

Standard HC is buffered (one on the front, one on the back, plus whatever logic lies inbetween, so an HC04 is actually three inverters), which doesn't bias well. HCU is unbuffered, so it makes linear amps just as well as the old CD4000s did.

Tim

--
Deep Friar: a very philosophical monk. 
Website: http://seventransistorlabs.com

Ok my first plan was to power the analog card with a 3V6 battery.

I realize that driving modern ADC clock input is not so trivial (like LTC2203)

Ok.

Thx, Habib.

Jim is a little bit condescendant with French guys like me but he must not as an analog innovatOR and as a good wines amatOR. I guess he spent too a long time in the cellar and returning here afterwards pissing on my shoes.

These old men are incorrigible!

Habib.

[snip]

The "U" stands for "unbuffered".

A standard 74HC04 (no "U") inverter is actually 3 inverters in series, to scale up the power gain.

The "U" version is just a single (2-transistor) inverter, so it's simply a transconductance device... in other words, an amplifier ;-)

Most crystal oscillators use some variation of the 'HCU04 configuration. In more critical applications, AGC is applied to... guess what?... the 'HCU04 configuration... to make it a completely linear amplifier, with good spectral characteristics.

My MC1648 (circa 1968) is a bipolar example of an AGC'd oscillator to get good spectral characteristics. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| Phoenix, Arizona  85048    Skype: Contacts Only  |             | 
| 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.

Jitter is a function of the phase noise profile.

Set your jitter spec., calculate your jitter from the phase noise

of your chosen oscillator.

Then you don't need to argue with anybody.

It's not difficult.

The LTC2203 has a single-ended clock input, which is OK if handled very carefully. It's better to use a clock and an ADC with differential PECL or LVDS levels; the edges are faster, and power supply and ground noise are much smaller hazards.

We did a box that uses an LTC2242-12, a 12-bit ADC running at 250 MHz. The clock is AC-coupled differential, which makes it pretty much immune to low frequency and common-mode noise.

Often, maybe usually, picosecond jitter is caused by *low* frequency effects: power supply ripple, ground loops, switcher noise, cmos analog noise, temperature fluctuations, clock 1/f phase noise.

--

John Larkin                  Highland Technology Inc 
www.highlandtechnology.com   jlarkin at highlandtechnology dot com    

Precision electronic instrumentation 
Picosecond-resolution Digital Delay and Pulse generators 
Custom timing and laser controllers 
Photonics and fiberoptic TTL data links 
VME  analog, thermocouple, LVDT, synchro, tachometer 
Multichannel arbitrary waveform generators

You condem others, but have to ask for help. That piss on your shoes is your own.

Read Thompson's posts. Unprovoked insults and slurs. Not to mention mostly wrong.

--

John Larkin                  Highland Technology Inc 
www.highlandtechnology.com   jlarkin at highlandtechnology dot com    

Precision electronic instrumentation 
Picosecond-resolution Digital Delay and Pulse generators 
Custom timing and laser controllers 
Photonics and fiberoptic TTL data links 
VME  analog, thermocouple, LVDT, synchro, tachometer 
Multichannel arbitrary waveform generators

clock

Yes surely the single ended input clock requires more attention by us (PCB stack up issues, Power supplies and signal integrity) thank you for pointing these topics and we have been some experience on that subjects.

Seems to be such a thing approching what we plan to do. Great piece of electronics design.

That is always the same guys, Spehro, you and a few others who are speaking about electronics designs down here.

Recently someone tells me to not question anything about electronics and just let some guys like Jim Thompson the great guru analog innovaTOR pissing me off.

What a pity !

Habib.

clock

frequency

Jim Thompson a Master Circuit Designer; no less authority than Jim Thompson told us so.

At 16 bits, clock jitter might be swamped by analog circuit noise.

--

John Larkin                  Highland Technology Inc 
www.highlandtechnology.com   jlarkin at highlandtechnology dot com    

Precision electronic instrumentation 
Picosecond-resolution Digital Delay and Pulse generators 
Custom timing and laser controllers 
Photonics and fiberoptic TTL data links 
VME  analog, thermocouple, LVDT, synchro, tachometer 
Multichannel arbitrary waveform generators

ted

      ...Jim Thompson     ...Jim Thompson       |    mens     |     |     et      |  |       |

ide quoted text -

Got it, thanks Jim (and Tim)

George H.

Aha! Habib is a dribbler ?>:-} ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| Phoenix, Arizona  85048    Skype: Contacts Only  |             | 
| 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.

you're talking about correlated system issues. goes without saying those components have to be dealt with. you don't want them setting your overall system spec. again, those contributions can be calculated.

but ultimately timing systems specs starts with a master clock phase noise profile. in the case of downconverter all sources have to be considered.

then it becomes obvious that spectral profiles can be shaped to maintain jitter, (time domain), specs.