EMRFD Message Archive 8237

8237 2013-02-07 17:18:36 Nick Kennedy Series & parallel mode crystal oscillators Message Date From Subject The notion of series and parallel mode crystal oscillators has always confused me a bit. When you test a crystal to measure its parameters, you see a series resonant peak and then an anti-resonant null quite a bit higher in frequency, where the holder capacitance resonates with the Lm & Cm combination. I had an idea that that might be the "parallel" mode, but it's not. There's a simple Pierce crystal oscillator as figure 4.25 of EMRFD. The crystal goes from collector to base and a 220p capacitor goes from base to ground. Since collector and base voltages are 180 degrees out of phase, it seems that this crystal plus capacitor network must shift the phase of the driving voltage at the collector by 180 degrees. That's parallel mode? And how much frequency difference are we talking about, in general, to go from one more to the other? I modeled this circuit in LTSpice with some real 8 MHz crystal data. I saw the out of phase voltages but didn't learn much else. I have in my shack the Pierce oscillator as transistor / crystal checker of figure 72 of SSDRA and also a plain old Manhattan Colpitts crystal oscillator. I thought I'd see how one crystal (a 7122 kHz HC49/U for "novice" CW) would act in the two oscillators. In the Pierce oscillator with switchable 150p, 100p, 50p feedback capacitors I got 7123.9, 7124.2 and 7124.7 kHz for an average of 7124.24 kHz. In the Colpitts oscillator I got 7124.3 kHz with a 35p capacitor in series with the crystal already on the board. Shorting that capacitor out I got 7122.4 ... so it's trending toward its nameplate value? In this case, the "load" capacitance is the series combination of the divider capacitors of 270p and 470p or 171p. Probably my closest comparison is with the 150p capacitor in the Pierce and 171p in the Colpitts giving 7123.9 and 7122.4 kHz for a difference of 1.5kHz higher in the Pierce. Looking just at the network (crystal in series with 220p) in LTSpice in the frequency domain, I see that the phase across the 220p capacitor changes fairly quickly from approximately 0 degrees just below series resonance to 180 degrees just above it. Maybe 1kHz or so to make the transition. So, does the series mode (Colpitts) oscillator operate just below series resonance where there's ~0 phase shift and the parallel mode just above it, where there's ~180? For the same specified frequency, would a crystal made for parallel mode have a slightly lower resonant frequency than a series mode crystal, if both are made for the same load capacitance? 73- Nick, WA5BDU [Non-text portions of this message have been removed] There's a good article here; http://www.foxonline.com/ Go to Support then Crystal Oscillator Theory and Quartz Crystal Design Notes. Also here; http://www.qvstech.com/Engineering-Tech-Info.aspx Go to Crystal Application Notes. These are a fairly random pick from my files; most crystal manufacturers will have similar notes I expect. Kerry VK2TIL. Its really simple: Series resonance is the crystal intrinsic series resonance. when a crystal is specified to work as parallel resonance then it also specifies the parallel resonance frequency for a specific load capacitance, usually 10pF to 30pF. This means that if the crystal is specified to have a parallel resonance frequency of 7MHz for a load of 22pF than that's exactly what it would do. If you connect it to a circuit that will show the crystal a much higher capacitance than the one specified for the crystal parallel resonant frequency (as in your example) then the actual parallel frequency will get closer to the series resonant frequency. Victor - 4Z4ME Thanks. From the Fox definition, I'd gather that both Colpitts and Pierce oscillators are "parallel mode" oscillators, since both have capacitors across the crystal's terminals. The fact that one needs a 180 degree phase shift through the network isn't part of the definition. They do say that "Because of the fact that the frequency resulting from the addition of capacitance is higher than the series resonant frequency, it is usually called the parallel frequency, though it is lower than the true parallel frequency." And both oscillators have "load" capacitance equal to the series sum of the feedback capacitors and the tuning capacitor, if any. Right? One of the QVS-Tech papers says that the Pierce oscillator operates within 5 to 40 ppm of the series resonant frequency. Surely they mean the series resonant frequency which includes the effect of the feedback capacitors, not the inherent series resonant frequency of the crystal itself. (?) If the same crystal were operated in a Pierce and a Colpitts oscillator having the same total load capacitance, what would be the operating frequency for each? I'd guess they'd be close but the Pierce would be maybe a couple hundred Hz higher, for a 7 MHz crystal. Maybe I should try it on the bench. 73- Nick, WA5BDU OK, I did give it a try. Same crystal in a Pierce (figure 4.25) and a Colpitts (figure 4.24) oscillator, with capacitors adjusted to give the same loop ("load") capacitance. Two 2N3904s from the same lot. I measured all the capacitors and saw that they were close to the marked value. Since I didn't have a buffer amplifier, I experimented with my 10x probe to my frequency counter and decided that there was too much pull (by listening in my receiver) on the Pierce, so I checked frequencies by zero beating in my receiver. I think I could read within 10Hz or so. With my 7122kHz HC49 crystal I got: Pierce: 7123.885 kHz Colpitts: 7124.125 kHz So they're within 760 Hz but I thought the Pierce would be higher. Does this mean something, or is it about the tolerance of my capacitors, plus strays and other error factors? To check the sensitivity to small changes in C, I added 4.7p in parallel with my 33p "tuning" capacitor on the Colpitts oscillator and the frequency increased by 805 Hz. So It might be too close to call as to whether the Pierce might be higher if the load capacitances were somehow made exactly equal. But I think it's safe to say that with either oscillator type and the same load capacitance, the frequencies will be pretty close. 73- Nick, WA5BDU > If the same crystal were operated in a Pierce and a Colpitts oscillator > having the same total load capacitance, what would be the operating > frequency for each? I'd guess they'd be close but the Pierce would be > maybe a couple hundred Hz higher, for a 7 MHz crystal. Maybe I should try > it on the bench. > > 73- > > Nick, WA5BDU > > > [Non-text portions of this message have been removed]