Keyboard Shortcuts
ctrl + shift + ? :
Show all keyboard shortcuts
ctrl + g :
Navigate to a group
ctrl + shift + f :
Find
ctrl + / :
Quick actions
esc to dismiss
Likes
Search
JFET Model
|
That was easy thanks. Chris? On Sun, Dec 22, 2013 at 3:56 PM, <helmutsennewald@...> wrote:
|
|
Hello Chris, I don't know what you expect to learn from substituting a JFET for the tube, but why not try with an actual tube model? You can find symbols for triodes, tetrodes and pentodes in the LTspice installation (look in the "Misc" category), and models in the files section of the group site. You can search for them with your browser by opening: http://groups.yahoo.com/neo/groups/LTspice/files/%20%20Tables%20of%20Contents/all_files.htm Regards, Tony |
|
Thanks, Tony.? I actually got the Clapp oscillator working with a JFET substituted as a the vacuum tube.? I didn't really even need to change the schematic.? I just lowered B+ from 150V to 12V.? Obviously the JFET does not need screen grid bias, so I just left that part out.? I was trying to see how changing voltage changes the resonant frequency.? I guess spice doesn't model that though. Changing B+ has no effect on freq in spice. Regards, On Dec 23, 2013 12:05 PM, <tony@...> wrote:
|
|
John Woodgate
In message
<CANnsUMFQLY+eCBVz8ow+K+HLOREL12FBwcBXhgVO-DuQqOAfzQ@...>, dated Mon, 23 Dec 2013, Chris Maness <chris@...> writes: .? I was trying to see how changing voltage changes the resonantIt ought not to in real life, of course. If it does, it's due to some second or third order effect in the tube, such as a change in effective inter-electrode capacitance. Your tube had 4 or 5 electrodes, too, whereas the JFET has only 3 (or you could say just 2, because drain and source are interchangeable). I guess spice doesn't model that though. Changing B+ has no effect onThe same sort of effect on capacitances might occur in a JFET, and the model might or might not include that. If it includes it, then LTspice will, of course, take it into account. BUT the effect would probably be much smaller that for a tube, so the frequency shift would be correspondingly smaller. -- OOO - Own Opinions Only. With best wishes. See www.jmwa.demon.co.uk Nondum ex silvis sumus John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK |
|
The frequency does change in real life. I am working with the design
toggle quoted message
Show quoted text
limits of late 1950's technology. The radio in question uses an OA2 for a shunt regulator. I thought of replacing it with a high voltage zener, but they can't handle 15mA of shunt current. They also probably have even worse regulating properties than the venerable OA2 Argon discharge tube. A friend of mind sites the miller effect as being the reason for f_o shift. I will have to read up to see if this is indeed the case. Regards, Chris KQ6UP On Mon, Dec 23, 2013 at 1:08 PM, John Woodgate <jmw@...> wrote:
In message |
|
In this case, "Miller Effect" capacitance is just the gate-drain capacitance multiplied by the stage gain. Miller capacitance won't change unless one or the other (gain, or Cdg) changes. Cdg WILL change with drain voltage. I am not sure, but it is pretty likely that the JFET spice model does not include the voltage dependency of Cdg. If so, this is NOT unique failing of LTspice because all of the spice simulators tend to share model definitions. Jim Wagner Oregon Research Electronics From: "Chris Maness" To: LTspice@... Sent: Monday, December 23, 2013 1:25:54 PM Subject: Re: [LTspice] RE: JFET Model The frequency does change in real life. I am working with the design |
|
John Woodgate
In message <CANnsUMFiEqEOs9Si0pke8F-9R34KDT+DGL8kOgkA3oN15_WQ-g@...>, dated Mon, 23 Dec 2013, Chris Maness <chris@...> writes:
A friend of mind sites the miller effect as being the reason for f_o shift. I will have to read up to see if this is indeed the case.Miller effect is feedback via the anode-to-control grid capacitance. It can depend on B+ voltage if the stage gain changes with B+ voltage. I doubt that it does very much in a JFET, operating in 'FET saturation' (pentode region), which is not the same as bipolar 'saturation' (bottomed). And the drain-gate capacitance is much smaller anyway than in some tubes (you need to take external strays into account because of the physical size difference). -- OOO - Own Opinions Only. With best wishes. See www.jmwa.demon.co.uk Nondum ex silvis sumus John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK |
|
So would the change in gain due the change in B+ be the most likely cause of f_o change for a Clapp oscillator with a valve state amplifier?? The reason for the inquiry is at this point mostly academic.? I love learning things as I work my way through fixing an old radio.? If indeed that is the case, then it would be the Miller effect indirectly. Thanks, On Dec 23, 2013 2:29 PM, "John Woodgate" <jmw@...> wrote: In message |
|
John Woodgate
In message
<CANnsUMHruA-cbtCW3yswZaaKDayat7KKSm1wdPE3YZpvtmrAww@...>, dated Mon, 23 Dec 2013, Chris Maness <chris@...> writes: So would the change in gain due the change in B+ be the most likelyOne way to tell is to analyse the circuit in detail, or you could try to find Clapp's original paper: J. K. Clapp, "An inductance-capacitance oscillator of unusual frequency stability", Proc. IRE, vol. 367, pp. 356-358, Mar. 1948. For analysis, look at the size of the a-g1 capacitance compared with the tuning capacitor (the 'capacitance tap' capacitors should be so much larger that their effect is negligible). The consider what internal gain would be needed for a small change in the Cag to affect the frequency by say 1 kHz. If that turns out to give a gain of 10 000, you can be pretty sure that you don't have a Miller effect problem. -- OOO - Own Opinions Only. With best wishes. See www.jmwa.demon.co.uk Nondum ex silvis sumus John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK |
|
You need to be careful about equating a JFET and a vacuum tube directly. Yes, they are very similar, but there are some very important physical differences. Because the frequency changes for one reason with a tube, that does not mean that the change is caused by exactly the same thing in a JFET. As far as the cause being "Miller Effect", it is likely that gain change may be the dominant effect in a tube. But, in a JFET, it is likely to be both. To call one more "Miller Effect" than the other is false. Both are. Because Miller Capacitance depends on BOTH gain and internal inter-electrode capacitance plus any circuit capacitance. Inter-electrode C WILL change significantly with a JFET (and even more in a MOSFET) with electrode voltage. In fact, it will change as the signal voltage swings up and down so that it is time varying at the signal frequency. This makes detailed analysis quite complex. But, suffice it to say, the frequency change with supply voltage is real. Further, the frequency change is due to changes in Miller Capacitance. Finally, change in Miller Capacitance is due to gain change AND due to changes in Cdg.? Jim Wagner Oregon Research Electronics KA7EHK From: "Chris Maness" To: LTspice@... Sent: Monday, December 23, 2013 2:47:35 PM Subject: Re: [LTspice] RE: JFET Model So would the change in gain due the change in B+ be the most likely cause of f_o change for a Clapp oscillator with a valve state amplifier?? The reason for the inquiry is at this point mostly academic.? I love learning things as I work my way through fixing an old radio.? If indeed that is the case, then it would be the Miller effect indirectly. Thanks, |
|
John Woodgate
In message <1932965163.4191125.1387841577938.JavaMail.root@...>,
dated Mon, 23 Dec 2013, Jim Wagner <wagnejam99@...> writes: But, suffice it to say, the frequency change with supply voltage isAgreed, but how big is the effect likely to be? -- OOO - Own Opinions Only. With best wishes. See www.jmwa.demon.co.uk Nondum ex silvis sumus John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK |
|
"I am not sure, but it is pretty likely that the JFET spice model does not include the voltage dependency of Cdg." Actually, I believe the SPICE models for most transistors, including JFETs, are quite
complete ?and include most effects almost anyone would care about. ?(With a few exceptions.) ?I am almost certain this includes capacitance vs. voltage, vs. temperature, etc., which I think is supposed to be modeled rather accurately. ?Of course it depends on the right parameter values being present in the .MODEL statement; but even if those parameters are wrong, or omitted, LTspice still should show the dependency.On the other hand, I would guess that the majority of vacuum tube / valve SPICE models do not. ?Those models are behavioral macro-models, built-up out of things like ideal VCVS and VCCS sources, and might not account for major dependencies on e.g. the supply voltage. Regards, Andy |
|
You are mixing several effects in your question. A Clapp oscillator is a special case of a Colpitts with an added decoupling capacitor between the main LC tank resonator and the grid/base/gate element, depending of course on which type of active device is used in the circuit. Although there are other effects, the main purpose of decoupling with a very small C is to minimize tempco effects on resonant frequency, particularly useful if the LC tank is electrically tuned by the use of a varacap (varacter) for some or all of the tank C. ? Beyond that all solid state junctions, even those of relatively linear devices such as JFETs display some dC/dV effects, and so in general changing applied biases and b+ (+vcc) voltages will cause some
shift in operation frequency for any oscillator circuit. The only real difference is the degree of change, which by careful design and topology can be made quite small. In other cases the effects of "miller" and junction capacitance changes can be used to desired effect to improve circuit performance. ? In the case of JFETs specifically there are devices in the JFET family such as the U310 series which have very low internode capacitances, so low that oscillators with?slightly over 1 Gigaherz center frequencies can be realized. Indeed I have personally designed varacter tuned versions of such an oscillator using miniature solid shield coaxial line as the "LC tank" with extremely low noise and very good temp stability for use in reciever and transmitter circuitry, and used LTspice years later to model the circuit and see how well the predictions of performance matched the actual lab results I had
recorded during bench and temperature testing. LTspice did quite well, although as is always the case some departure from real world results will be expected since some models are ideal. ? Cordially -?RC. On Tuesday, December 24, 2013 2:45 PM, Andy wrote:
?
"I am not sure, but it is pretty likely that the JFET spice model does not include the voltage dependency of Cdg."
Actually, I believe the SPICE models for most transistors, including JFETs, are quite
complete ?and include most effects almost anyone would care about. ?(With a few exceptions.) ?I am almost certain this includes capacitance vs. voltage, vs. temperature, etc., which I think is supposed to be modeled rather accurately. ?Of course it depends on the right parameter values being present in the .MODEL statement; but even if those parameters are wrong, or omitted, LTspice still should show the dependency.On the other hand, I would guess that the majority of vacuum tube / valve SPICE models do not. ?Those models are behavioral macro-models, built-up out of things like ideal VCVS and VCCS sources, and might not account for major dependencies on e.g. the supply voltage. Regards, Andy |
¿ªÔÆÌåÓý?Well, I always say check your model against bench measurements before placing much reliance on any SPICE results.? As for FET capacitance with voltage, that's especially true! I've made plenty of measurements to back up that assertion. If the OP has a specific part he's interested in maybe I can take some measurements so we can test this issue.-- - Win Winfield Hill Rowland Institute at Harvard 100 Edwin Land Blvd Cambridge, MA 02142 617-497-4670 |
|
John Woodgate
In message <52BC2285.9070201@...>, dated Thu, 26 Dec 2013,
Winfield Hill <hill@...> writes: ?Well, I always say check your model against bench measurements beforeAnd vice-versa. I recently traced a difference between simulation and real life to a fault in real life - a 10 ohm resistor where a 100 ohm should have been. -- OOO - Own Opinions Only. With best wishes. See www.jmwa.demon.co.uk Nondum ex silvis sumus John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK |
|
Ron, that is an awesome explanation of what is going on. I actually
toggle quoted message
Show quoted text
was able to fix the problem by swapping out the OA2 regulator. It really settled down. I also had a continuing discussion about how amplitude effects frequency in non SHO (simple harmonic oscillator) with a physics professor friend of mine. A damped LC circuit (RLC) is not a SHO, so if the amplitude of the oscillations change, the frequency should change as well. This is also reflected as a resistance term in the resonance equations for a RLC circuit. For some reason this didn't show up with the FFT of the JFET Clapp I made in LTSpice. Too much fun :D Thanks, Chris Maness On Thu, Dec 26, 2013 at 5:03 AM, John Woodgate <jmw@...> wrote:
In message <52BC2285.9070201@...>, dated Thu, 26 Dec 2013, |
Tony Casey
to navigate to use esc to dismiss