TCXO (shopping for replacement)

The new counter is great; we have four or five in test stands. But those are ethernet interfaced and run by Python programs. The thing that amazed me about that counter is that, manually run, it won't do some fundamental measurements that the 5370s did.

Python = external pc. No way am I going to run Windows in production.

I'll go with the THS788.

You might not have heard because it's so recent, but there are other OSes that run Python besides Windows. ;)

Cheers

Phil Hobbs

Sure. I run Ubuntu.

All our new production test stands run Linux. We still run some old DOS stuff that I hope will keep going for a few more years.

The coupling mechanism is fairly obvious in that example. Still pretty neat.

linux is worth it just for not having to mess around with licenses and acti vation when you need an extra computer

l.com:

Doom:

to

s by

yeh, mythbusters tried and couldn't make it happen, apparently it takes jus t the right combination of metronomes and wobbly table

After many years of running Linux I would absolutely *hate* to have to go back to Windows. "Once you Slack you'll never go back" as they say. ;-)

Windows. The Web-enabled SCPI Command Interface requires Java.

Neither of which will I run in production.

I'm not even sure about Python. I prefer Delphi. Much better error detection.

Clearly the stiffness of the bolts that are suspending the table, combined with the net eccentric mass of all the metronomes, is somewhere near a resonance. Perhaps they measured that, and set the metronomes to the measured frequency.

Mythbusters almost never understand the science of anything they failed to replicate - or succeeded to, for that matter. For them, it's all about the fireball. A total waste of time for anyone who actually loves science.

Clifford Heath,

Interesting. The burst is long, and has a well-enough known shape that the ADC can sample it enough times to allow you to least squares fit the ADC ob servations to the known shape with just two unknowns - amplitude and start time, where start time is what you want to know.

It would have to have been before 2010 to make it patentable. Granting what we know about what James Arthur and you know about, it would have to have been a high-Q LC circuit (or shorted delay line) that you were kicking - pr obably with a pulse about half as wide as the resonant period of your tank circuit, which you could have got out of another shorted delay line.

True. Not non-existent though, and you do go into for making bespoke electr onics for niche markets, so the implication is that you couldn't make it wo rk

This does suggest that there actually is a market worth exploiting, albeit with a part cheaper than you could make money on.

Not a helpful observation. You don't know much, and don't have to remind us that you are prone to use your imagination in areas where it isn't well-di siciplined by relevant facts

Am 15.09.2017 um 16:38 schrieb John Larkin:

I have been in Co-author Prochazka's lab some years ago. We had an interface to check out: his SPAD and my time stretcher. The thing was to measure the photon flight time between ground and the ISS. The ISS gets a hydrogen maser and a cesium designed for zero gravity (which helps a lot!) as a time base. I have also made the dual mixer down converter that compares the maser and the cesium.

cheers, Gerhard

I wonder what the point was. You are dealing with a very short viewing time, tracking error, doppler, varying atmosphere angles, etc. I don't see how you can make any meaningful measurements under those conditions.

Tell us about your dual mixer down converter and your time stretcher.

another fundamental question:

why bother to measure jitter in the time domain

it is easy to measure phase noise in the frequency domain

this gives you much more information about the signal

and you can always integrate the result to a single jitter "number" if you want to

Because time-domain jitter is what the customer cares about, and the phase noise method imports dubious assumptions, specifically that there are no important phase correlations between components at different modulation frequencies.

White noise has the same PSD as a delta-function, for instance.

Cheers

Phil Hobbs

I'm not sure I understand what you are talking about: "no important phase correlations between components at different modulation frequencies" doesn't ring any bells. Maybe it makes sense to you, but it sure doesn't make any sense to me. Perhaps others may have the same problem.

Phase noise is very important. You can see spikes where power supply harmonics get into the oscillator. You can see the flicker noise frequency knee where it meets the baseline noise. You can see the wideband falloff which tells you the outside limit of the phase noise.

Time domain jitter tells you the distribution of the jitter. This is helpful when you have two or more distributions to deal with.

I think you need both.

I gave an example: white noise has exactly the same ensemble-averaged PSD a s a delta-function. The difference is solely in the phase corelations: all the frequency components of the delta function have phase 0 at t=0.

The usual approach to calculating jitter from phase noise ignores these cor relations, and so will be seriously in error whenever they're important.

Cheers

Phil Hobbs

Still makes no sense. Where are you going to find a delta function in real life.

Many high frequency oscillators integrate the phase noise over a specific band of frequencies. For example, 10khz to 50khz. This gives an optimistic value for jitter, but the pll chains they are in can follow the noise.