Hi Or just maybe you go into attack mode much to quickly, and thus shut down the conversation before it's ever really started.
Bob On Mar 27, 2010, at 7:52 PM, WarrenS wrote: > > Bob wrote: >> EFC is essentially instantaneous at the "many seconds" level. There's not >> much settling left to do after 20 time constants. > > ALL mostly true, although I have not seen any TCs as long as 50 ms. > EFC inputs on most Osc I've tested, including the 10811, are essentially > instantaneous at <100us TC. (BW > 2 KHz) > > BUT > I find it amazing how a few true statements of basic principal can be used to > counter the facts. > I should of made it more clear, I was NOT talking about TC settling. > The long "resettling" time of hrs I was referring to is of a similar effect > as turning the unit off for a while and then back on. > I suppose that according to your book formula, that should only take 20 > seconds also to come back to within 1e-12. > Maybe the problem is that the Osc does not read the same books, > OR maybe the problem is just that some of us are getting more performance > from them than they were ever designed to give, > and at that level there are a few new unexpected and unknown secondary > effects. > > ws > > ******************* > Hi > > On a properly working crystal oscillator, EFC is essentially instantaneous at > the "many seconds" level. The main delay you see is from the R/C time > constant between the R's in the tuning attenuator and any bypass C's that are > present. Time constants rarely get over 50 ms. Twenty * tau would only be 1 > seconds at the 50 ms case. There's not much settling left to do after 20 time > constants. > > Bob > ****************** > WarrenS wrote: > >> I should of also said; when care is taken, turning the Osc unit over is >> repeatable, >> if just looking at a graph plot where the resolution is several parts in >> e-11. >> >> Pete2 >> >> Addressing your first question only, I'll let the experts address the why. >> >> Yes, I 'think' it does in principal, I would not guess on how long it takes >> though. >> But >> I have not checked that close to be sure. >> I'd never do something that gross during the middle of a test that I was >> trying to see 1e-12 things. >> I do that first thing, long before the high resolution test starts and then >> let things resettle before proceeding. >> Any change in cable position or the Osc moving a bit in its Box or a dozen >> other things could also change >> it that much, including temp gradients in the box, so Don't do that if you >> want 1e-12 results. >> Use the wedge, that I have tested, and is completely reversible and >> repeatable at those levels. >> >> I have noticed that the OSC will EVEN take a long time to RESETTLE (HRS+ ) >> if the EFC voltage is change a lot and then put back. >> Which is why I limit the clamp voltage of my Tbolt to typ + - 100 mv or less >> from its nominal. >> >> ws >> >> ******************** >> >>> Warren, >>> If you turn over an oscillator, is the frequency change >>> completely reversible (to your "under 1e-12 resolution") when it is >>> restored? Thinking aloud, if an hour-glass is turned over twice, the >>> final level will be the same, but the grains will be mixed. A quartz >>> crystal, however, is solid, so hopefully nothing actually moves. >>> Presumably the zero-G axis is with the axis of oscillation at 90 >>> degrees to gravity? >>> >>> Peter (the "other" one :-) >>> >> ***************************> >>> >>>> Another thing I use it for is to test high resolution Freq meters. >>>> Using a calibrated wedge that I can then slide under one edge of the zero-G >>>> Osc box, I can >>>> make small, variable, repeatable, freq changes of under 1e-12 resolution, >>>> something pretty hard to do otherwise. >>>> If I want to make BIG changes like 1e-10, I can rotate the box on any of >>>> its >>>> sides and still use the wedge, >>>> and for a quick check of new equipment, I just turn the box over which then >>>> gives a couple of parts in 1e-9 freq change. >>>> It makes a weird but simple and indispensable variable freq source that is >>>> useful for many things, such as checking the LOOP TC of a TBolt. >>> >>> >> > > _______________________________________________ > time-nuts mailing list -- time-nuts@febo.com > To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts > and follow the instructions there. > _______________________________________________ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.