Ultra low frequency VCO
in [Design]
|
Prev: AofE 3rd Edition
Next: Labelling Prototypes
From: Phil Hobbs on 12 Jul 2010 11:50 John Larkin wrote: > On Mon, 12 Jul 2010 10:40:00 -0400, Phil Hobbs > <pcdhSpamMeSenseless(a)electrooptical.net> wrote: > >> Jim Thompson wrote: >>> On Fri, 09 Jul 2010 14:08:28 -0400, Phil Hobbs >>> <pcdhSpamMeSenseless(a)electrooptical.net> wrote: >>> >>>> whit3rd wrote: >>>>> On Jul 8, 12:29 pm, Phil Hobbs >>>>> <pcdhSpamMeSensel...(a)electrooptical.net> wrote: >>>>> >>>>>> I don't know that -100 dBc/Hz is that hard at 60 Hz. I bet you could do >>>>>> that by running a bog standard multivibrator at 1024*1024*60 Hz and >>>>>> dividing down. You'd need a sine shaper, but the phase noise goes down >>>>>> by N**2 >>>>> Eh? I'd think it's N**0.5 (the multivibrator has cumulative but >>>>> random errors). >>>> The time jitter of the edges stays the same, but the resulting phase >>>> error goes down by a factor of N due to the division. Phase is like >>>> amplitude, so you have to square it to get the noise power--hence N**2. >>>> >>>> Cheers >>>> >>>> Phil Hobbs >>> Hey Phil! How come no comment on conservation of charge and energy? >>> You have a dog in this show ?:-) Weenie! >>> >>> ...Jim Thompson >> I'm mainly here to talk about electronics. One-upmanship also tends to >> intimidate the newbies, which I really don't want to do. I try not to >> dispense Bad Info myself, and try to help other people's >> misunderstandings when I can. Otherwise I just read with interest and >> learn stuff. >> >> Whit3rd seems to be talking about the phase correlations rather than the >> instantaneous phase noise. Both multivibrators and LC resonators obey >> equations with full locality, i.e. neither one has any memory at all. >> >> For instance, if you have a 1 MHz resonator with a Q of a million, it >> takes a second or so to get its phase to change when you put PM on the >> drive waveform. OTOH, if you change the resonant frequency suddenly, >> e.g. by putting 100V on a Y5V tank capacitor, the resonant frequency >> changes immediately--much faster than 1/Q cycles. >> >> Because of the switching action, multivibrators intermodulate the >> switching element's noise at all frequencies, which makes their jitter >> much worse; also the effective Q of a multivibrator is less than 1, >> which means that there isn't any significant filtering action from the >> resonator. (That's frequency-domain way of thinking about what Whit3rd >> is talking about in the time domain--the conservation of energy issue is >> easier to think about if there's a natural bandwidth limit to the >> sqrt(t) behaviour.) The physical origin of the phase modulation doesn't >> change the way it varies with division ratio, though. >> >> Cheers >> >> Phil Hobbs > > One interesting and often overlooked part is the coaxial ceramic > resonator. It's essentially a shorted transmission line formed in a > block or tube of hi-K ceramic, usually by silver or copper plating it. > They are usually treated by the RF boys as resonators or inductors, > but they really act like time-domain transmission lines. TCs are in > the single-digit PPMs and Qs in the hundreds or thousands. Dielectric > constants are in the hundreds or thousands, so they are very short for > their delay/frequency. > > Remarkable parts. I use them to make instant-start/instant-stop > oscillators in the 600 MHz range. As a VCO, they will have very low > phase noise, somewhere between an LC and a quartz crystal. > > John > I confess I'm one of the ones who overlooked them...where do you get them, and do they come in Y5V? Cheers Phil Hobbs -- Dr Philip C D Hobbs Principal ElectroOptical Innovations 55 Orchard Rd Briarcliff Manor NY 10510 845-480-2058 hobbs at electrooptical dot net http://electrooptical.net
From: John Larkin on 12 Jul 2010 11:52 On Mon, 12 Jul 2010 11:43:29 -0400, Phil Hobbs <pcdhSpamMeSenseless(a)electrooptical.net> wrote: >Jim Thompson wrote: >> On Mon, 12 Jul 2010 10:40:00 -0400, Phil Hobbs >> <pcdhSpamMeSenseless(a)electrooptical.net> wrote: >> >>> Jim Thompson wrote: >>>> On Fri, 09 Jul 2010 14:08:28 -0400, Phil Hobbs >>>> <pcdhSpamMeSenseless(a)electrooptical.net> wrote: >>>> >>>>> whit3rd wrote: >>>>>> On Jul 8, 12:29 pm, Phil Hobbs >>>>>> <pcdhSpamMeSensel...(a)electrooptical.net> wrote: >>>>>> >>>>>>> I don't know that -100 dBc/Hz is that hard at 60 Hz. I bet you could do >>>>>>> that by running a bog standard multivibrator at 1024*1024*60 Hz and >>>>>>> dividing down. You'd need a sine shaper, but the phase noise goes down >>>>>>> by N**2 >>>>>> Eh? I'd think it's N**0.5 (the multivibrator has cumulative but >>>>>> random errors). >>>>> The time jitter of the edges stays the same, but the resulting phase >>>>> error goes down by a factor of N due to the division. Phase is like >>>>> amplitude, so you have to square it to get the noise power--hence N**2. >>>>> >>>>> Cheers >>>>> >>>>> Phil Hobbs >>>> Hey Phil! How come no comment on conservation of charge and energy? >>>> You have a dog in this show ?:-) Weenie! >>>> >>>> ...Jim Thompson >>> I'm mainly here to talk about electronics. One-upmanship also tends to >>> intimidate the newbies, which I really don't want to do. I try not to >>> dispense Bad Info myself, and try to help other people's >>> misunderstandings when I can. Otherwise I just read with interest and >>> learn stuff. >> >> There's no one-up-man-ship involved. Larkin won't (or can't, because >> he doesn't really understand it) show where the extra charge came >> from. You (or Win) could put a stop to Larkin's nonsense. Larkin >> displays me as a fool, and the newbies don't know any better, so >> they'll never ever learn the correct solution unless someone >> (politically :) respected steps in. > >I don't know about that. It isn't that difficult to calculate a circuit >with two caps, an inductor, and an elf who opens and closes a switch at >the right moments. It does help to know elementary differential equations. > >I haven't actually followed the original discussion closely enough to >know who made the first technical error. The larger error IMO is to >keep getting into these tiresome p***ing contests, which I decline to >do. If what you want is merely to have the correct solution posted, >post it and let's move on to some electronics. > >Cheers > >Phil Hobbs I don't think any specific problem has been clearly stated, such that it can be analyzed. My comment, that seems to have ruffled feathers, is that one shouldn't assume as a working tool that charge, coulombs stored in various capacitors in a circuit, is conserved. Sometimes it is, sometimes it isn't, sometimes the concept is silly. The argument did make me go back and review some basics, which is good. Messing with all this digital and software and opamp stuff can make the old EE101 math get rusty. John
From: John Larkin on 12 Jul 2010 12:07 On Mon, 12 Jul 2010 11:50:36 -0400, Phil Hobbs <pcdhSpamMeSenseless(a)electrooptical.net> wrote: >John Larkin wrote: >> On Mon, 12 Jul 2010 10:40:00 -0400, Phil Hobbs >> <pcdhSpamMeSenseless(a)electrooptical.net> wrote: >> >>> Jim Thompson wrote: >>>> On Fri, 09 Jul 2010 14:08:28 -0400, Phil Hobbs >>>> <pcdhSpamMeSenseless(a)electrooptical.net> wrote: >>>> >>>>> whit3rd wrote: >>>>>> On Jul 8, 12:29 pm, Phil Hobbs >>>>>> <pcdhSpamMeSensel...(a)electrooptical.net> wrote: >>>>>> >>>>>>> I don't know that -100 dBc/Hz is that hard at 60 Hz. I bet you could do >>>>>>> that by running a bog standard multivibrator at 1024*1024*60 Hz and >>>>>>> dividing down. You'd need a sine shaper, but the phase noise goes down >>>>>>> by N**2 >>>>>> Eh? I'd think it's N**0.5 (the multivibrator has cumulative but >>>>>> random errors). >>>>> The time jitter of the edges stays the same, but the resulting phase >>>>> error goes down by a factor of N due to the division. Phase is like >>>>> amplitude, so you have to square it to get the noise power--hence N**2. >>>>> >>>>> Cheers >>>>> >>>>> Phil Hobbs >>>> Hey Phil! How come no comment on conservation of charge and energy? >>>> You have a dog in this show ?:-) Weenie! >>>> >>>> ...Jim Thompson >>> I'm mainly here to talk about electronics. One-upmanship also tends to >>> intimidate the newbies, which I really don't want to do. I try not to >>> dispense Bad Info myself, and try to help other people's >>> misunderstandings when I can. Otherwise I just read with interest and >>> learn stuff. >>> >>> Whit3rd seems to be talking about the phase correlations rather than the >>> instantaneous phase noise. Both multivibrators and LC resonators obey >>> equations with full locality, i.e. neither one has any memory at all. >>> >>> For instance, if you have a 1 MHz resonator with a Q of a million, it >>> takes a second or so to get its phase to change when you put PM on the >>> drive waveform. OTOH, if you change the resonant frequency suddenly, >>> e.g. by putting 100V on a Y5V tank capacitor, the resonant frequency >>> changes immediately--much faster than 1/Q cycles. >>> >>> Because of the switching action, multivibrators intermodulate the >>> switching element's noise at all frequencies, which makes their jitter >>> much worse; also the effective Q of a multivibrator is less than 1, >>> which means that there isn't any significant filtering action from the >>> resonator. (That's frequency-domain way of thinking about what Whit3rd >>> is talking about in the time domain--the conservation of energy issue is >>> easier to think about if there's a natural bandwidth limit to the >>> sqrt(t) behaviour.) The physical origin of the phase modulation doesn't >>> change the way it varies with division ratio, though. >>> >>> Cheers >>> >>> Phil Hobbs >> >> One interesting and often overlooked part is the coaxial ceramic >> resonator. It's essentially a shorted transmission line formed in a >> block or tube of hi-K ceramic, usually by silver or copper plating it. >> They are usually treated by the RF boys as resonators or inductors, >> but they really act like time-domain transmission lines. TCs are in >> the single-digit PPMs and Qs in the hundreds or thousands. Dielectric >> constants are in the hundreds or thousands, so they are very short for >> their delay/frequency. >> >> Remarkable parts. I use them to make instant-start/instant-stop >> oscillators in the 600 MHz range. As a VCO, they will have very low >> phase noise, somewhere between an LC and a quartz crystal. >> >> John >> > >I confess I'm one of the ones who overlooked them...where do you get >them, and do they come in Y5V? > Don't know. We buy ours from Skyworks/Trans-tech, and they have Er values from 10 to 90. TCs are very good somehow. Lots of people make these things and the little, high-frequency ones are cheap in quantity. John
From: George Herold on 12 Jul 2010 12:15 On Jul 12, 11:09 am, Phil Hobbs <pcdhSpamMeSensel...(a)electrooptical.net> wrote: > j wrote: > > The point is that a lot of this jiber-jaber is pointless. Without the > > OP giving a better definition of the problem its a guess at best > > which measurement technique is required. > > > He never did state the basis for his phase noise number, nor did he > > have an offset frequency. > > > The challenge in making 100 dBc or better measurements is a function > > of the offset frequency and bandwidth. Center frequency isnt the > > issue here. > > You may not be interested, but perhaps other folks are. And how big an > offset frequency can he have on a 60 Hz carrier, anyway? > > Cheers > > Phil Hobbs > > -- > Dr Philip C D Hobbs > Principal > ElectroOptical Innovations > 55 Orchard Rd > Briarcliff Manor NY 10510 > 845-480-2058 > hobbs at electrooptical dot nethttp://electrooptical.net Yes! I've enjoyed the discussion. Say could someone explain the the 100 dBc of phase noise spec. I've been thinking of this a one part in 10^5 of jitter in the period. So for instance a 1 Hz signal the jitter is less than 10 micro seconds aand for a 1 MHz signal a jitter of 10 pico seconds. Is that right? George H.
From: John Larkin on 12 Jul 2010 12:22
On Mon, 12 Jul 2010 09:15:37 -0700 (PDT), George Herold <gherold(a)teachspin.com> wrote: >On Jul 12, 11:09�am, Phil Hobbs ><pcdhSpamMeSensel...(a)electrooptical.net> wrote: >> j wrote: >> > The point is that a lot of this jiber-jaber is pointless. �Without the >> > OP giving a better definition of the problem it�s a guess at best >> > which measurement technique is required. >> >> > He never did state the basis for his phase noise number, nor did he >> > have an offset frequency. >> >> > The challenge in making �100 dBc or better measurements is a function >> > of the offset frequency and bandwidth. �Center frequency isn�t the >> > issue here. >> >> You may not be interested, but perhaps other folks are. �And how big an >> offset frequency can he have on a 60 Hz carrier, anyway? >> >> Cheers >> >> Phil Hobbs >> >> -- >> Dr Philip C D Hobbs >> Principal >> ElectroOptical Innovations >> 55 Orchard Rd >> Briarcliff Manor NY 10510 >> 845-480-2058 >> hobbs at electrooptical dot nethttp://electrooptical.net > >Yes! I've enjoyed the discussion. Say could someone explain the the >100 dBc of phase noise spec. I've been thinking of this a one part in >10^5 of jitter in the period. So for instance a 1 Hz signal the >jitter is less than 10 micro seconds aand for a 1 MHz signal a jitter >of 10 pico seconds. > >Is that right? > >George H. The usual oscillator phase noise spec would be " -100 dBc/Hz " at some offset frequency from the carrier. It's often given as a curve. It is possible to convert the curve to an RMS jitter spec... I have a program around somewhere that some s.e.d. guy posted. John |