John Larkin's LC oscillator
in [Basics]
|
Prev: Twin T circuit wanted
Next: And blocking oscillators
From: Jim Thompson on 16 Jun 2010 23:05 On Wed, 16 Jun 2010 20:02:05 -0700, John Larkin <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: >On Wed, 16 Jun 2010 19:15:30 -0700, >"JosephKK"<quiettechblue(a)yahoo.com> wrote: > >>On Tue, 15 Jun 2010 20:45:37 -0700, John Larkin >><jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: >> >>>On Tue, 15 Jun 2010 20:36:11 -0700, >>>"JosephKK"<quiettechblue(a)yahoo.com> wrote: >>> >>>>On Mon, 14 Jun 2010 22:18:22 -0700, John Larkin >>>><jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: >>>> >>>>>On Mon, 14 Jun 2010 22:06:53 -0700, >>>>>"JosephKK"<quiettechblue(a)yahoo.com> wrote: >>>>> >>>>>>On Mon, 14 Jun 2010 11:15:00 -0700, Jim Thompson >>>>>><To-Email-Use-The-Envelope-Icon(a)On-My-Web-Site.com> wrote: >>>>>> >>>>>>>On Mon, 14 Jun 2010 10:50:58 -0700 (PDT), dagmargoodboat(a)yahoo.com >>>>>>>wrote: >>>>>>> >>>><snip> >>>>>>>Class-A implies _linear_, does it not? Or do we have a Larkin >>>>>>>definition for today ?:-) >>>>>>> >>>>>> >>>>>>The definition i have always heard is that it is conducting through the >>>>>>whole cycle. Nothing about linearity. And that is one fougly non-linear >>>>>>circuit. >>>>> >>>>>Only if you want it to be. It can be designed to be a nice smooth >>>>>class A oscillator with precise automatic gain control that servoes >>>>>oscillation amplitude to almost exactly 2*Vcc, with a low TC. Not bad >>>>>for one transistor. >>>>> >>>>>John >>>> >>>>So post that version instead. >>> >>>I did, several days ago. >>> >>>ftp://jjlarkin.lmi.net/LC_YDx.gif >> >>No way sport, that is running deep class C. > >It is not. Simulate it and look at the whole cycle. Pay attention to >the directions of the various currents. > >The pic above is zoomed up on the small interval, about 50 us, when >the collector swings below the base. That's when the AGC happens, as >charge is pulled out of the base capacitor and transistor bias, and >transconductance, are servoed down. In fact, the drive into the tank >is sustained over the whole rest of the cycle. The emitter current is >a nice smooth curve and it doesn't cut off at the opposite peak of the >cycle. > >One reason the amplitude is so predictable and stable is that it's NOT >running "class C." > >Try it. Note especially the waveform at the base. > >>> >>>And I described it long before that. >>> >>>ftp://jjlarkin.lmi.net/Ships_Bell.JPG >>> >>>I can't help it if people keep over-driving it, and then complaining >>>that it's over-driven. >>> >>>John >> >>The output is reasonable like a proper TC oscillator, but it is running >>class C. > >No. The gain mechanism is active through almost the whole cycle. But >the naming of "class A" versus "class B" is a ham-radio sort of thing. >It doesn't explain what the circuit is actually doing. The circuit >doesn't care what letter you call it. I don't either. > >John > Single point declaration: WRONG! Young bucks proceed at your own peril. ...Jim Thompson -- | James E.Thompson, CTO | mens | | Analog Innovations, Inc. | 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 | The only thing bipartisan in this country is hypocrisy
From: Phil Hobbs on 17 Jun 2010 00:21 John Larkin wrote: > On Wed, 16 Jun 2010 19:15:30 -0700, > "JosephKK"<quiettechblue(a)yahoo.com> wrote: > >> On Tue, 15 Jun 2010 20:45:37 -0700, John Larkin >> <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: >> >>> On Tue, 15 Jun 2010 20:36:11 -0700, >>> "JosephKK"<quiettechblue(a)yahoo.com> wrote: >>> >>>> On Mon, 14 Jun 2010 22:18:22 -0700, John Larkin >>>> <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: >>>> >>>>> On Mon, 14 Jun 2010 22:06:53 -0700, >>>>> "JosephKK"<quiettechblue(a)yahoo.com> wrote: >>>>> >>>>>> On Mon, 14 Jun 2010 11:15:00 -0700, Jim Thompson >>>>>> <To-Email-Use-The-Envelope-Icon(a)On-My-Web-Site.com> wrote: >>>>>> >>>>>>> On Mon, 14 Jun 2010 10:50:58 -0700 (PDT), dagmargoodboat(a)yahoo.com >>>>>>> wrote: >>>>>>> >>>> <snip> >>>>>>> Class-A implies _linear_, does it not? Or do we have a Larkin >>>>>>> definition for today ?:-) >>>>>>> >>>>>> The definition i have always heard is that it is conducting through the >>>>>> whole cycle. Nothing about linearity. And that is one fougly non-linear >>>>>> circuit. >>>>> Only if you want it to be. It can be designed to be a nice smooth >>>>> class A oscillator with precise automatic gain control that servoes >>>>> oscillation amplitude to almost exactly 2*Vcc, with a low TC. Not bad >>>>> for one transistor. >>>>> >>>>> John >>>> So post that version instead. >>> I did, several days ago. >>> >>> ftp://jjlarkin.lmi.net/LC_YDx.gif >> No way sport, that is running deep class C. > > It is not. Simulate it and look at the whole cycle. Pay attention to > the directions of the various currents. > > The pic above is zoomed up on the small interval, about 50 us, when > the collector swings below the base. That's when the AGC happens, as > charge is pulled out of the base capacitor and transistor bias, and > transconductance, are servoed down. In fact, the drive into the tank > is sustained over the whole rest of the cycle. The emitter current is > a nice smooth curve and it doesn't cut off at the opposite peak of the > cycle. > > One reason the amplitude is so predictable and stable is that it's NOT > running "class C." > > Try it. Note especially the waveform at the base. > >>> And I described it long before that. >>> >>> ftp://jjlarkin.lmi.net/Ships_Bell.JPG >>> >>> I can't help it if people keep over-driving it, and then complaining >>> that it's over-driven. >>> >>> John >> The output is reasonable like a proper TC oscillator, but it is running >> class C. > > No. The gain mechanism is active through almost the whole cycle. But > the naming of "class A" versus "class B" is a ham-radio sort of thing. > It doesn't explain what the circuit is actually doing. The circuit > doesn't care what letter you call it. I don't either. > > John > > It's a cute circuit, but letting the BJT saturate causes horrible close-in phase noise. Cutoff is a much better behaved self-limiting mechanism than saturation, if you can't use separate AGC. 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 17 Jun 2010 00:40 On Thu, 17 Jun 2010 00:21:18 -0400, Phil Hobbs <pcdhSpamMeSenseless(a)electrooptical.net> wrote: >John Larkin wrote: >> On Wed, 16 Jun 2010 19:15:30 -0700, >> "JosephKK"<quiettechblue(a)yahoo.com> wrote: >> >>> On Tue, 15 Jun 2010 20:45:37 -0700, John Larkin >>> <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: >>> >>>> On Tue, 15 Jun 2010 20:36:11 -0700, >>>> "JosephKK"<quiettechblue(a)yahoo.com> wrote: >>>> >>>>> On Mon, 14 Jun 2010 22:18:22 -0700, John Larkin >>>>> <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: >>>>> >>>>>> On Mon, 14 Jun 2010 22:06:53 -0700, >>>>>> "JosephKK"<quiettechblue(a)yahoo.com> wrote: >>>>>> >>>>>>> On Mon, 14 Jun 2010 11:15:00 -0700, Jim Thompson >>>>>>> <To-Email-Use-The-Envelope-Icon(a)On-My-Web-Site.com> wrote: >>>>>>> >>>>>>>> On Mon, 14 Jun 2010 10:50:58 -0700 (PDT), dagmargoodboat(a)yahoo.com >>>>>>>> wrote: >>>>>>>> >>>>> <snip> >>>>>>>> Class-A implies _linear_, does it not? Or do we have a Larkin >>>>>>>> definition for today ?:-) >>>>>>>> >>>>>>> The definition i have always heard is that it is conducting through the >>>>>>> whole cycle. Nothing about linearity. And that is one fougly non-linear >>>>>>> circuit. >>>>>> Only if you want it to be. It can be designed to be a nice smooth >>>>>> class A oscillator with precise automatic gain control that servoes >>>>>> oscillation amplitude to almost exactly 2*Vcc, with a low TC. Not bad >>>>>> for one transistor. >>>>>> >>>>>> John >>>>> So post that version instead. >>>> I did, several days ago. >>>> >>>> ftp://jjlarkin.lmi.net/LC_YDx.gif >>> No way sport, that is running deep class C. >> >> It is not. Simulate it and look at the whole cycle. Pay attention to >> the directions of the various currents. >> >> The pic above is zoomed up on the small interval, about 50 us, when >> the collector swings below the base. That's when the AGC happens, as >> charge is pulled out of the base capacitor and transistor bias, and >> transconductance, are servoed down. In fact, the drive into the tank >> is sustained over the whole rest of the cycle. The emitter current is >> a nice smooth curve and it doesn't cut off at the opposite peak of the >> cycle. >> >> One reason the amplitude is so predictable and stable is that it's NOT >> running "class C." >> >> Try it. Note especially the waveform at the base. >> >>>> And I described it long before that. >>>> >>>> ftp://jjlarkin.lmi.net/Ships_Bell.JPG >>>> >>>> I can't help it if people keep over-driving it, and then complaining >>>> that it's over-driven. >>>> >>>> John >>> The output is reasonable like a proper TC oscillator, but it is running >>> class C. >> >> No. The gain mechanism is active through almost the whole cycle. But >> the naming of "class A" versus "class B" is a ham-radio sort of thing. >> It doesn't explain what the circuit is actually doing. The circuit >> doesn't care what letter you call it. I don't either. >> >> John >> >> > >It's a cute circuit, but letting the BJT saturate causes horrible >close-in phase noise. Cutoff is a much better behaved self-limiting >mechanism than saturation, if you can't use separate AGC. The original requirement was for very high amplitude stability, to excite the LVDT-like inclinometer thing. The signal was synchronously detected, so phase noise, or a little distortion, didn't matter. This was done about 35 years ago, when it was harder to do this sort of thing. One transistor did it, with amplitude stable to a very small fraction of a per cent. Even today, with a bucket of opamps, it would be non-trivial to do an externally AGCd oscillator with this sort of longterm, low-TC amplitude stability. How could you use cutoff to get this sort of amplitude stability? Actually, this sort of does use cutoff, in that increasing collector swing backs off the base voltage and reduces Gm. James' addition of a Baker schottky is even better, because that limiting happens a bit above saturation type voltages. I didn't have schottkies when I did this one. John
From: Phil Hobbs on 17 Jun 2010 09:44 John Larkin wrote: > On Thu, 17 Jun 2010 00:21:18 -0400, Phil Hobbs > <pcdhSpamMeSenseless(a)electrooptical.net> wrote: > >> John Larkin wrote: >>> On Wed, 16 Jun 2010 19:15:30 -0700, >>> "JosephKK"<quiettechblue(a)yahoo.com> wrote: >>> >>>> On Tue, 15 Jun 2010 20:45:37 -0700, John Larkin >>>> <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: >>>> >>>>> On Tue, 15 Jun 2010 20:36:11 -0700, >>>>> "JosephKK"<quiettechblue(a)yahoo.com> wrote: >>>>> >>>>>> On Mon, 14 Jun 2010 22:18:22 -0700, John Larkin >>>>>> <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: >>>>>> >>>>>>> On Mon, 14 Jun 2010 22:06:53 -0700, >>>>>>> "JosephKK"<quiettechblue(a)yahoo.com> wrote: >>>>>>> >>>>>>>> On Mon, 14 Jun 2010 11:15:00 -0700, Jim Thompson >>>>>>>> <To-Email-Use-The-Envelope-Icon(a)On-My-Web-Site.com> wrote: >>>>>>>> >>>>>>>>> On Mon, 14 Jun 2010 10:50:58 -0700 (PDT), dagmargoodboat(a)yahoo.com >>>>>>>>> wrote: >>>>>>>>> >>>>>> <snip> >>>>>>>>> Class-A implies _linear_, does it not? Or do we have a Larkin >>>>>>>>> definition for today ?:-) >>>>>>>>> >>>>>>>> The definition i have always heard is that it is conducting through the >>>>>>>> whole cycle. Nothing about linearity. And that is one fougly non-linear >>>>>>>> circuit. >>>>>>> Only if you want it to be. It can be designed to be a nice smooth >>>>>>> class A oscillator with precise automatic gain control that servoes >>>>>>> oscillation amplitude to almost exactly 2*Vcc, with a low TC. Not bad >>>>>>> for one transistor. >>>>>>> >>>>>>> John >>>>>> So post that version instead. >>>>> I did, several days ago. >>>>> >>>>> ftp://jjlarkin.lmi.net/LC_YDx.gif >>>> No way sport, that is running deep class C. >>> It is not. Simulate it and look at the whole cycle. Pay attention to >>> the directions of the various currents. >>> >>> The pic above is zoomed up on the small interval, about 50 us, when >>> the collector swings below the base. That's when the AGC happens, as >>> charge is pulled out of the base capacitor and transistor bias, and >>> transconductance, are servoed down. In fact, the drive into the tank >>> is sustained over the whole rest of the cycle. The emitter current is >>> a nice smooth curve and it doesn't cut off at the opposite peak of the >>> cycle. >>> >>> One reason the amplitude is so predictable and stable is that it's NOT >>> running "class C." >>> >>> Try it. Note especially the waveform at the base. >>> >>>>> And I described it long before that. >>>>> >>>>> ftp://jjlarkin.lmi.net/Ships_Bell.JPG >>>>> >>>>> I can't help it if people keep over-driving it, and then complaining >>>>> that it's over-driven. >>>>> >>>>> John >>>> The output is reasonable like a proper TC oscillator, but it is running >>>> class C. >>> No. The gain mechanism is active through almost the whole cycle. But >>> the naming of "class A" versus "class B" is a ham-radio sort of thing. >>> It doesn't explain what the circuit is actually doing. The circuit >>> doesn't care what letter you call it. I don't either. >>> >>> John >>> >>> >> It's a cute circuit, but letting the BJT saturate causes horrible >> close-in phase noise. Cutoff is a much better behaved self-limiting >> mechanism than saturation, if you can't use separate AGC. > > The original requirement was for very high amplitude stability, to > excite the LVDT-like inclinometer thing. The signal was synchronously > detected, so phase noise, or a little distortion, didn't matter. This > was done about 35 years ago, when it was harder to do this sort of > thing. One transistor did it, with amplitude stable to a very small > fraction of a per cent. Even today, with a bucket of opamps, it would > be non-trivial to do an externally AGCd oscillator with this sort of > longterm, low-TC amplitude stability. > > How could you use cutoff to get this sort of amplitude stability? > Actually, this sort of does use cutoff, in that increasing collector > swing backs off the base voltage and reduces Gm. James' addition of a > Baker schottky is even better, because that limiting happens a bit > above saturation type voltages. I didn't have schottkies when I did > this one. > > John > > For the requirements, that was a good choice--I was making a more general point about self-limiting oscillators. Saturation is certainly more stable with temperature than cutoff, and has that switching action that you're using. I don't know if a Schottky would be an improvement from that point of view--you still have the 2 mV/K slope. Widlar did a cute temperature-compensated breakpoint amp using saturation, by driving several feedback networks from the summing junction via emitter followers with collector resistors. When the BJT saturated, the beta dropped and the extra resistors suddenly became part of the feedback network. (National AN4, Figure 8, http://www.national.com/an/AN/AN-4.pdf.) 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 17 Jun 2010 11:34
On Thu, 17 Jun 2010 09:44:44 -0400, Phil Hobbs <pcdhSpamMeSenseless(a)electrooptical.net> wrote: >John Larkin wrote: >> On Thu, 17 Jun 2010 00:21:18 -0400, Phil Hobbs >> <pcdhSpamMeSenseless(a)electrooptical.net> wrote: >> >>> John Larkin wrote: >>>> On Wed, 16 Jun 2010 19:15:30 -0700, >>>> "JosephKK"<quiettechblue(a)yahoo.com> wrote: >>>> >>>>> On Tue, 15 Jun 2010 20:45:37 -0700, John Larkin >>>>> <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: >>>>> >>>>>> On Tue, 15 Jun 2010 20:36:11 -0700, >>>>>> "JosephKK"<quiettechblue(a)yahoo.com> wrote: >>>>>> >>>>>>> On Mon, 14 Jun 2010 22:18:22 -0700, John Larkin >>>>>>> <jjlarkin(a)highNOTlandTHIStechnologyPART.com> wrote: >>>>>>> >>>>>>>> On Mon, 14 Jun 2010 22:06:53 -0700, >>>>>>>> "JosephKK"<quiettechblue(a)yahoo.com> wrote: >>>>>>>> >>>>>>>>> On Mon, 14 Jun 2010 11:15:00 -0700, Jim Thompson >>>>>>>>> <To-Email-Use-The-Envelope-Icon(a)On-My-Web-Site.com> wrote: >>>>>>>>> >>>>>>>>>> On Mon, 14 Jun 2010 10:50:58 -0700 (PDT), dagmargoodboat(a)yahoo.com >>>>>>>>>> wrote: >>>>>>>>>> >>>>>>> <snip> >>>>>>>>>> Class-A implies _linear_, does it not? Or do we have a Larkin >>>>>>>>>> definition for today ?:-) >>>>>>>>>> >>>>>>>>> The definition i have always heard is that it is conducting through the >>>>>>>>> whole cycle. Nothing about linearity. And that is one fougly non-linear >>>>>>>>> circuit. >>>>>>>> Only if you want it to be. It can be designed to be a nice smooth >>>>>>>> class A oscillator with precise automatic gain control that servoes >>>>>>>> oscillation amplitude to almost exactly 2*Vcc, with a low TC. Not bad >>>>>>>> for one transistor. >>>>>>>> >>>>>>>> John >>>>>>> So post that version instead. >>>>>> I did, several days ago. >>>>>> >>>>>> ftp://jjlarkin.lmi.net/LC_YDx.gif >>>>> No way sport, that is running deep class C. >>>> It is not. Simulate it and look at the whole cycle. Pay attention to >>>> the directions of the various currents. >>>> >>>> The pic above is zoomed up on the small interval, about 50 us, when >>>> the collector swings below the base. That's when the AGC happens, as >>>> charge is pulled out of the base capacitor and transistor bias, and >>>> transconductance, are servoed down. In fact, the drive into the tank >>>> is sustained over the whole rest of the cycle. The emitter current is >>>> a nice smooth curve and it doesn't cut off at the opposite peak of the >>>> cycle. >>>> >>>> One reason the amplitude is so predictable and stable is that it's NOT >>>> running "class C." >>>> >>>> Try it. Note especially the waveform at the base. >>>> >>>>>> And I described it long before that. >>>>>> >>>>>> ftp://jjlarkin.lmi.net/Ships_Bell.JPG >>>>>> >>>>>> I can't help it if people keep over-driving it, and then complaining >>>>>> that it's over-driven. >>>>>> >>>>>> John >>>>> The output is reasonable like a proper TC oscillator, but it is running >>>>> class C. >>>> No. The gain mechanism is active through almost the whole cycle. But >>>> the naming of "class A" versus "class B" is a ham-radio sort of thing. >>>> It doesn't explain what the circuit is actually doing. The circuit >>>> doesn't care what letter you call it. I don't either. >>>> >>>> John >>>> >>>> >>> It's a cute circuit, but letting the BJT saturate causes horrible >>> close-in phase noise. Cutoff is a much better behaved self-limiting >>> mechanism than saturation, if you can't use separate AGC. >> >> The original requirement was for very high amplitude stability, to >> excite the LVDT-like inclinometer thing. The signal was synchronously >> detected, so phase noise, or a little distortion, didn't matter. This >> was done about 35 years ago, when it was harder to do this sort of >> thing. One transistor did it, with amplitude stable to a very small >> fraction of a per cent. Even today, with a bucket of opamps, it would >> be non-trivial to do an externally AGCd oscillator with this sort of >> longterm, low-TC amplitude stability. >> >> How could you use cutoff to get this sort of amplitude stability? >> Actually, this sort of does use cutoff, in that increasing collector >> swing backs off the base voltage and reduces Gm. James' addition of a >> Baker schottky is even better, because that limiting happens a bit >> above saturation type voltages. I didn't have schottkies when I did >> this one. >> >> John >> >> > >For the requirements, that was a good choice--I was making a more >general point about self-limiting oscillators. > >Saturation is certainly more stable with temperature than cutoff, and >has that switching action that you're using. I don't know if a Schottky >would be an improvement from that point of view--you still have the 2 >mV/K slope. The base voltage needs to be +1 jd for the thing to oscillate at equilibrium. The clamp device, c-b junction or schottky, has about the same tc, but in the opposite direction. In the c-b case (with a low secondary swing, not some awful class C thrashing, and a big base cap, so it's not a blocking oscillator) the collector swing that pulls current out of the base capacitor touches zero, near where the transistor saturates. In fact, the tc of the c-b junction is canceling the tc of the e-b junction. Not perfectly, because they are effectively used at different currents. Because of reverse beta, some of the tank energy can be clamped through the emitter, which is a bit negative at that instant. But even that current flow happens when the c-b junction is forward biased. Schottky: suppose we need +0.6 on the base for stable oscillation. The collector will dip down to, say, 0.3, the schottky will conduct and pull charge out of the base cap, and it will stabilize there. The tc's almost cancel. If you assume the amplitude limiting is just brute clamping of the swing, the tc's still cancel. No saturation, since the c-b junction doesn't get forward biased. This assumes you've chosen a suitable base resistor, not jamming too much current into the base, and the transistor has reasonable beta. Actually, the base AGC thing must be happening. With small feedback voltage into the emitter, a sim shows that the transistor is on throughout the cycle, "class A." So Ic ~= beta * Ib, on average. Average Ic would be large if the current through the base resistor were all actually going into the base. But it's not: the base cap is being discharged at the negative swing of the collector, stealing base current and reducing transconductance, and that is exquisitely sensitive to p-p amplitude. Some tank energy does of course get lost to the emitter... both mechanisms are at work. Fortunately, both stabilize the amplitude. In my sim, if you short the tank, the supply current goes up about 30x. (Hmmm, Rb could be bigger!) Output swing is 10.11 volts p-p with a 5-volt supply. Several people have modeled this circuit with small base caps and low transformer ratios and lots of base bias current. That works differently. John |