ad: rfparts

Attempting to build a 7 MHz receiver.

Discussion in 'Homebrew and Kit Projects' started by MOOSFET, Mar 6, 2013.

Thread Status:
Not open for further replies.
  1. MOOSFET

    MOOSFET QRZ Member

    I'm not a ham, but the idea has always interested me, and so I've decided to test my interest by building a radio receiver.

    I'm not an electronics noob. It's been my hobby since I was a kid, and I've done a lot of things in digital electronics, including building a Z80 computer, and playing around with some microcontrollers. I've also built an EEG machine, which is basically the extent of my analog experience. I haven't done much with radio in the past (other than generally figuring out that it's hard) but it certainly shouldn't be something I can't do. I also once read a Radio Shack book about radio transceivers, which explained the whole super-heterodyne process (which is, interestingly, information that I've found useful for a lot of things despite not actually using it for radio yet).

    I decided to build a 40 meter receiver because the frequencies are within range of my 20 MHz oscilloscope (thus avoiding questions like "is it oscillating or not?"), and because a few 40 meter receiver schematics were the first ones I happened to find on the internet. At first I tried following a schematic (a simple regenerative receiver) as closely as possible, but it didn't work (and I'm tempted to include "obviously"), and at the time I figured it was because I didn't have the exact toroids listed. So then I decided that if I could simply make an LC tank oscillate, then I'd be able to construct a suitable inductor and capacitor pair, and move on from there. That started my week-long mission to build an oscillator.

    I have an air variable capacitor to use a tuner. I first spent a day or two trying to build an oscillator using it and an inductor as a tank circuit, with the intent of measuring the frequency in order to determine the value of the capacitor, but with no luck since I couldn't get it to oscillate.. Then I read on this forum about a "bridge circuit" and so I simply connected the output of a 4 MHz crystal oscillator to the variable capacitor, and subsequently decided it is variable between 15 and 450 pF, which is what I thought it was, but I never wrote it down and the web site where I bought it no longer exists.

    I then decided to connect another capacitor in parallel to the variable capacitor, and used Wolfram Alpha to calculate what size that capacitor should be, and what inductor should be used, to make it tune between 7.0 and 7.3 MHz. I also solved for some larger ranges with 100 kHz and 300 kHz buffer spaces on each side of the band. I then tried again to build an oscillator, but again with no luck.

    Then I decided the problem might be the solderless breadboard I was using. Too lazy to solder the whole circuit together, I instead simply soldered the tank circuit together. This made it work, and my oscilloscope confirmed I was near 7 MHz.

    About this time I decided I should also play with one of my 74HC4046 phase-locked loop chips which I bought probably ten years ago and never even tried to use. I used an 82C54 programmable interval timer for the two frequency dividers, along with an AT89S52 microcontroller connected to a PC via an FT245RL to program the clock divisors. Getting that to work was actually rather easy, which made me wonder if I shouldn't skip the idea of building a simple receiver and just build a super-heterodyne receiver.

    So then I decided to play with the LM1496 balanced modulator chips I'd bought at the same time and similarly ignored for the last ten years. I simply wired the thing up following one of the examples in the datasheet (or so I thought) and connected my oscillator (the first one, not the PLL) to it, along with a piece of wire for an antenna (being too lazy to build a pre-amp and just playing around) and connected the output to an LM386 audio amplifier. In order to have a signal strong enough to receive, I used my MP3 player to modulate the PLL signal, figuring that since the two circuits were right next to each other, it should be strong enough to be received, and without an antenna, it shouldn't radiate enough to bother anyone. To my complete surprise, this actually seemed to work. Even more to my surprise, I could receive other signals as well. Tuning the capacitor to the correct place, I could hear several streams of morse code at once, and tuning it elsewhere, I heard some Spanish radio, and somewhere else was my MP3 transmitter, though it was the worst signal of them all due to my modulation strategy being terrible. Strangely, the signals were receivable even after disconnecting the piece of wire I was calling an antenna. This was particularly strange with the Spanish radio station, as I can't imagine it being anywhere nearby.

    The next day, however, it didn't seem to work at first, until eventually I powered on my PLL circuit as well, which is on a separate solderless breadboard, at which point it worked again. So I tried tuning the PLL to different frequencies, but it seemed to make no difference what frequency it was tuned to -- only turning the variable capacitor changed the tuning of the received signal. Then I noticed that I forgot two very important resistors from the LM1496 balanced oscillator IC, such that the chip technically wasn't connected to a power source, which just deepened the mystery of how this radio worked. So I connected the resistors, but then it no longer worked. ...but, for the moment, I decided to forget about it.

    Instead I decided I needed to change the capacitor and inductor attached to the tuning capacitor. It was difficult to adjust the oscillator to receive a particular signal, and it would drift a lot. I considered attaching a large knob to it, something to make my movements more precise and remove the need for my hand to be anywhere near it when adjusting it, but I also decided I needed to remove the excess 300 kHz I'd designed into it both above and below the band. So I removed the capacitor and inductor, and attached a new larger capacitor and a new smaller inductor.

    Once again, I could no longer get it to oscillate. I spent the last two or three days trying to solve this situation, by replacing the inductor and capacitor again (although with the same values), and twice attempting to build the entire oscillator as a soldered-together mess of components that isn't attached to the solderless breadboard. I also spent some time filtering even more 60 Hz from my power supply, and I also tried powering the circuit with batteries. ...but, no matter what, it just doesn't want to oscillate anymore.

    I'm beginning to suspect that the problem is that some combinations of capacitor and inductor work better than others. Obviously, for any given target frequency, there's going to be a range of capacitance values and corresponding inductor values. Do they all oscillate equally well? I've been unable to turn up any information on this topic on the internet, despite searching on several occasions, as I've suspected some values work better than others since the beginning of this adventure.

    So, trying to figure this out for myself... I can't imagine why, with ideal components, the particular values of inductor and capacitor should matter. With nothing to dissipate the energy, it should oscillate forever. So the problem must be in non-ideal aspects, and all of the stray inductance and capacitance just add to the tank circuit, and so the problem must be entirely with energy lost due to unwanted resistance. However, I've already been making the inductors with very thick wire, and the variable capacitor is rather thick itself. (This might, however, explain why soldering the two together originally made it work, since doing so got rid of numerous resistances due to connections on the solderless breadboard.)

    I have thought of one other thing... Since P = I^2 * R, it would seem that reducing the current flowing in the tank would be far more important than reducing the resistance of the inductor windings. ...and, I think, that means that a tank circuit using a larger inductor and a smaller capacitor will work better, because a larger inductor's current will increase more slowly while a smaller capacitor will discharge more quickly, and the two together mean that the current is lower even though the frequency is the same. ...and, given that the I is squared while the R is not, using a larger value inductor should always work better, even if doing so requires using thinner wire. So I decided I'd try larger inductors and smaller capacitors and see if that helps.

    So I calculated some new capacitor and inductor values, but this time assuming I'd put the capacitor in series with the variable capacitor, but the results (1 pF in series with the 450 pF variable capacitor) made me suspect the tuning range would be incredibly lopsided, and so I spent some hours trying to find some application / web site to graph it for me before giving up and writing a Perl script to output the graph in postscript. It was far worse than I imagined, even after picking less extreme values (22 pF in series) which actually result in a far larger tuning range (thus making my original problem even worse). So it seems there's no point in trying a capacitor in series with the tuning capacitor.

    So I looked online, but I don't see anyone selling the things. Not even the cheap little plastic ones that come in those Radio Shack kits. I'm really not sure what exactly the things would be called. Been searching for "variable capacitor" and the like. The last ones I bought were from Ocean State Electronics, but they appear to have been out of business for years.

    So, I have questions, obviously:

    1. Any advice on building an RF oscillator? Obviously I've read web sites and books, but for the most part, all I find are schematics that don't work, schematics without labeled component values, and a lot of "amplifiers oscillate and oscillators amplify" almost as if to say "yeah, no one really knows why these things work," which doesn't really make sense since how they work is rather simple, at least as I understand it. ...and that does make me wonder if there just isn't something to it that isn't well documented, like this idea I have about the current flow being more detrimental than the circuit resistance.

    2. What about this idea I have about the current flow? Do you think it matters? Is it an idea you've heard before? ...or am I way off and you can actually point me in the right direction?

    3. Any ideas on how to use a 450 pF variable capacitor as if it were much smaller, like 20 pF or 100 pF, but in a way that doesn't result in a painfully exponential tuner?

    4. ...or should I just forget about that, and instead try to build one? I do have quite a lot of double-sided copper clad board that I'll probably never use for anything. However, the required tolerances seem annoying (even if I make it huge) and that stuff is a PITA to cut, so I don't think I like this idea. I may see what I can build from cardboard and aluminum foil.

    5 ...or should I just rip some plates out of my air variable capacitor to reduce the capacitance? I really hate to do this, as I've loved the things ever since I first saw one when I was like 10 years old. Something about the things just seems so awesome. However, if I'm right with my inductor/capacitor trade-off idea, then to have the things be of such a large capacitance would actually be quite pointless and so I might as well turn them into something useful. ...but I really hate to do this without being certain that they're useless in their current form.

    6. ...or should I forget about tunable capacitors and build a tunable inductor instead? It's certainly fewer moving parts, but I imagine difficulty with respect to getting linear performance out of it, e.g. I might glue an inductor core to a screw, thus allowing me a lot of turns to move the core a small distance, but if, as it turns, it wobbles closer and further from one side of the coil, the results may suck a lot. I'll have to search the internet for ideas on this one.

    Also, I'm curious just how easy I should expect reception of some 7 MHz activity to be. I live in New Paris, Ohio, if checking a map will help to figure this out. So far I've just been using a ten foot wire as an antenna. I know from my experience with CB radio that using such an antenna wouldn't pick up any transmitter that was further than a mile away, but I've also heard that the ham bands aren't nearly as noisy, and I once had a CB that could tune a little bit of 28 MHz, and IIRC there is a huge difference in the noise floor between the two bands. There's also that morse code and spanish radio I heard, though honestly I have no clue what frequencies those actually were since I have no clue why that circuit worked at all, and so I don't know if I should expect 7 MHz to be so easy to receive that I don't even have to know what I'm doing, or if that was some other band and 7 MHz will actually be very difficult to receive without building a 1/2 wavelength dipole outside (which I don't believe my yard is even large enough for, nevermind that I just don't want to buy the materials).
     
  2. KO6WB

    KO6WB Premium Subscriber QRZ Page

    Wow, this is exhausting to read. I'm taking a ten minute break:rolleyes:!
    Okay, first off oscillators are not that hard to make provided you use the correct components.
    The inductor you're using is insufficient for the task. It's too low in inductance and the capacitance is too much for a small tuning range of 7 to 7.3 MHz.
    Okay, capacitors can be reduced in value by placing a value in series with it. This divides the capacitance and it will always be lower then the lowest value.
    So, when you put a 1pfd capacitor in series with a 425pfd capacitor your total capacitance is less then 1pfd. On the 40 meter band that's much too small for any good tuned circuit.
    It's also much too small to get a reliable feedback loop to make an oscillator work. To lower the capacitance of a 10-450pfd variable the series capacitance could be something like another fixed value capacitor. for this example and to make it easy use a 450pfd capacitor in series with the variable capacitor. Simply put the capacitance will divide by the ratios of the two capacitors in series. So, a 450pfd in series with another 450pfd capacitor will equal 225pfd. With the variable capacitor at it's lowest value the total capacitance is now just a bit lower then the lowest capacitance of a single part alone. So at the minimum setting of the variable capacitor your total capacitance is now about 8pfd or so.
    Now, you would have a 8-225pfd variable capacitor. A much more manageable value for a 40 meter receiver.
    Let's focus on the inductor. First of all for your first attempts use an air wound or coil form capacitor and stay away from the torodial cores until you are at ease with the other approaches.
    Next, a regenerative will help you get satisfaction in a more rapid manner then jumping to a super-hetrodyne as a first receiver.
    So, if you follow the directions in this posting you will end up with a working example of a very good receiver. You can add things to it as you progress. If you want to make a super-hetrodyne then this little receiver can be the intermediate frequency (IF) amplifier section. More about this later. For now look at this website and follow the direction to make the coil and have the capacitor set to the value needed.
    It's here at; http://hevirred.multiply.com/journal/item/42/40-meter-Regenerative-Receiver.
    This is the schematic of it.
    [​IMG]

    If you notice the "Course tuning" capacitor value is just 40pfd. This should give you some idea of where things were going wrong with your previous attempts. Too much capacitance followed by way too little.
    Now, the fine tuning in this example is by varying the voltage to a simple diode that is reverse biased. It has a 10pfd capacitor in series with it so the capacitance of the fine tuning is going to always be much less then 10pfd.
    This is an interesting approach. The greater the reverse bias on the diode the smaller it's normal capacitance will be. Remember in a perfect world a diode will not have capacitance but in a real application there are two opposing areas with an insulator between. This is a capacitor. Not a very good one but in this case it's value is enough to work very well for this purpose.
    Looking over the schematic you'll find the builder did the same thing on the "regen calibrate" portion of the circuit. That's another voltage variable capacitor using a diode in reverse bias.
    All of these parts are easily found at Digikey and other online electronic parts retailers.
    The 2.5 mH choke needed is; http://www.digikey.com/product-detail/en/6302-RC/M10164-ND/1949388.
    All the parts on this receiver are available at Digikey except for the specified variable capacitor for tuning. In this case eBay is your friend. This will do just fine; http://www.ebay.com/itm/Variable-Ca...428?pt=LH_DefaultDomain_0&hash=item3ccf1acf1c. With this capacitor the tuning range is going to be much greater then just the 7-7.3MHz you want. That's why there's a fine tuning control. This capacitor can also be just a small trimmer that you set and never need to touch it again unless readjustment is needed. You also need a marker of some sort to find where the 40 meter band is located on the tuning. A signal generator or a crystal oscillator that is already working in another piece of equipment located in or very near the 40 meter band will give you something to look for. If you tune in WWV then you're either on 5MHz or 10MHz. Hard to tell which one because the sound exactly the same. With the 40pfd capacitor as the tuning it is possible to tune either of them or perhaps even both. The 5MHz signal at the higher capacitance and the 10MHz signal at the low capacitance range of the 40pfd capacitor. You just have to play with it.
    Try this and see what'll happen.
    Let us know how you're doing and BTW the 40 meter amateur band is much noisier then the CB or 10 meter bands ever are. There is usually activity there all the time. An antenna of 35 feet of wire will do fine for listening. At night the upper part of 40 meters has shortwave broadcast stations.
    Hope this helps
    73
    Gary
     
    Last edited: Mar 6, 2013
  3. KL7AJ

    KL7AJ Ham Member QRZ Page


    I highly recommend you build a regenerative receiver for 7 MHZ. Lots of activity, and low enough frequency so things aren't too touchy. There are gobs of great circuits available online.
     
  4. W1VT

    W1VT Ham Member QRZ Page

    http://www.amazon.com/Communication...602937&sr=8-1&keywords=communication+circuits

    Communication Circuits: Analysis and Design [Hardcover]
    Kenneth K. Clarke (Author), Donald T. Hess (Author)

    Here is a book that explains how oscillators really work--it is an expensive
    textbook for engineers, but you may be able get your hands on one via an inter-library loan.

    It is a hard sell, but getting your math up to speed is often necessary to understand some
    of the wonderful books that have been published.

    Zack Lau W1VT
     
    Last edited: Mar 6, 2013
  5. KI6J

    KI6J Ham Member QRZ Page

    I've built a few versions of this and every one worked on the first try.
    http://www.amqrp.org/kits/tin_ear/TinEar Manual.pdf

    A loaded dipole, like the "shorty 40" will do a much better job receiving than a random length wire. There are lots of signals to receive on and near 40m.
     
  6. MOOSFET

    MOOSFET QRZ Member

    Yes, it's evident my communication skills suck. I came here to ask a question about oscillators, but I titled my post "attempting to build a 7 MHz receiver" and then wrote a huge block of text about my failure to build a receiver. I have no idea how I expected anyone to get "help me with my oscillator" out of that. I didn't even post a schematic of what I'm trying to do. I may create a new thread later for my oscillator problem and try to use my brain this time by giving it a proper title, useful description, and a schematic. I learned to use Spice yesterday (it's rather easy to use and quite useful), and so I'm working on testing out various ideas with it, so I'm going to see how far I get with that before bothering people again.

    Yes. That's why I didn't even try it. The only reason it came up was because I was trying to find values that make the tuning range of my variable capacitor appropriate for the band. For example, when I first tried to use capacitors in parallel, I created these two formulas:

    7000000 = 1 / (2 * pi * sqrt(l * (0.000000000450 + c)))
    7300000 = 1 / (2 * pi * sqrt(l * (0.000000000015 + c)))

    The "l" being whatever inductor I would need, and the "c" being whatever capacitor I would need to put in parallel with my 450 pF variable capacitor, in order that the minimum and maximum capacitance of my variable capacitor would result in frequencies at the top and bottom of the 7 MHz band. I then submitted the equations to Wolfram Alpha which kindly told me that I'd need a capacitor of 5.0 nF and an inductor of 96 nH. This works quite well. Using this calculator, if I plug in 95 nH and 5.45 nF (5 nF + 450 pF) I get 6.958 MHz, and if I plug in 5.015 nF instead, I get 7.2535 MHz, and so both ends of the tuning correspond to both ends of the band. The frequencies are a bit off, but the range is 295.5 kHz, and so just tweaking the inductor a bit can put it in the right place. (In fact, just using 95 nH instead of 96nH seems to work, and makes sense since, when rounding the results from Wolfram Alpha, I probably should have avoided rounding both in the same direction.) Also, if I solve for the middle position on my variable capacitor, which is 232.5 pF I get 7.1012 MHz which is very close to the midpoint of the frequency range (which is 7.10575) which indicates that the tuning range is not exponential, but rather very flat.

    However, since I was suspecting that my oscillator was failing because I need a larger inductor and smaller capacitor, I instead decided to look for a solution that results in a smaller capacitance.

    So I decided to use capacitors in series instead. I used the same method to calculate the ideal capacitor to use. This resulted in 1.5 pF, which I agree would never work. Also, since it's so un-matched against the 450 pF range of the variable capacitor, I was sure the tuning range would be incredibly exponential, and after graphing it, I saw that 99% of the band lies in 1% of the range of the variable capacitor, which is of course unusable. So I tried widening the band that I was calculating for, in order to get a less extreme capacitance that I should connect in series, but even widening it to the range of 6.0 to 9.3 MHz (which is a range so wide that the ham band is only 10% of the range) only gets the series capacitance up to 22 pF. While that may work electrically, it still leaves the tuning so exponential that the ham band lies in only 1% of the range of the variable capacitor, and so even if it would oscillate, it would be unusable.

    So I was hoping that this is a common problem and that there's some particular circuit that people use (obviously more complex than simply a single capacitor in series) that solves the problem better. ...but I'm probably hoping for too much. People probably just do the logical thing and get a smaller variable capacitor.

    I also assumed as much, but I'm looking for information about what exactly determines what will and won't work, rather than simply what experience or gut feelings tell people. For example, any pair of inductor and capacitor that resonates at 7 MHz leaves either the inductor or the capacitor seeming too small to me. Capacitances less than 1 uF seem so small they almost might as well not even exist, and inductances less than 1 uH similary seem so small that I wouldn't naturally expect them to be useful. Yet obviously I'm wrong because people make oscillators that oscillate at frequencies above 5 kHz all the time. Similarly, I might look at 5 nF and 90 nH and decide the two are out of balance simply because 5 is so much less than 90, but they don't measure the same thing, and indeed, the difference in magnitude is only because of how large humans decided that a farad and a henry should be.

    So I'm not simply looking for something that will work better, but rather, I'm looking to know why something will work better.

    I did create a simple LC tank in Spice, which verifies for me that larger inductances with smaller capacitances oscillate more easily when there's some small resistance involved. (Obviously, without the resistance, the oscillation continues forever regardless of the capacitance and inductance chosen.) Not content with such a simple conclusion, I've started trying to simulate an whole oscillator (with active components to keep the oscillation going, rather than allowing it to decay), but so far Spice is doing a wonderful job of simulating the fact that I'm too much of a moron to build an oscillator. I just learned to use Spice yesterday, but clearly I should have learned to use it a long time ago. When a circuit doesn't even work in simulation, I know it isn't because of poor connections or the stray capacitance of my solderless breadboard, which allows me to sooner come to the conclusion that I need a better circuit.

    I'll post more about what exactly I'm trying to do later, along with a schematic to make it more apparent why it isn't working.

    I imagine so. I'm trying to solve this oscillator problem as a step towards eventually building a regenerative, for as long as I can't get my tank circuit to oscillate in an oscillator circuit, I figure it has no chance of working in a regenerative circuit. As I understand it, a regenerative works as an oscillator with feedback just on the edge of failing to oscillate, and so if my circuit can't actually oscillate when it has sufficient feedback, then it isn't going to work.

    Again, I'll post a schematic later. I probably should have started with one, as I don't know how I was expecting anyone to help without knowing exactly what it is I'm trying to do.

    I'm actually not looking for more receiver schematics, though I do appreciate the attempt to help. For some reason I can't actually do anything that I don't entirely understand, because I can't convince myself to do things that I don't understand the necessity of. It should probably be considered a learning disability.

    For example, ages ago, when I was just beginning to learn electronics, I recall trying to build an amplifier, except by leaving out a lot of biasing resistors and bypass capacitors since I didn't yet understand why they were necessary. Instead I just looked at a schematic, saw all of these extra parts, whereas the description of what I was doing would just focus on the transistor being the amplifier, and at most mention the purpose of the resistor attached to the collector. So, naturally, none of my amplifiers worked. The simple thing to do would have been to simply follow an amplifier schematic and therefore end up with a working amplifier, but I wasn't actually trying to build an amplifier. I was trying to learn how to build an amplifier. ...and memorizing a schematic isn't really the same thing, and so I wasn't going to add those resistors and capacitors until I saw why they were a necessary part of making the circuit work.

    So, when I look at other people's schematics, depending on the schematic in question, I either see a circuit where every part is essential because I know why they've done what they've done, or (as is the case with the schematic you've posted) I see a lot of components connected in unfamiliar ways which don't make any sense to me. For example, just the portion between the antenna and the transformer primary makes no sense to me, with the antenna being attached via the emitter of the transistor, and the base connected directly to the positive supply. I'm only left to guess that maybe the other side of the transformer is used as an inductor in an oscillator and that somehow this allows the antenna signal to alter the way that works, but even with that assumption I still don't see how it accomplishes even that. So, even if I built the circuit and it worked, I couldn't claim to have "built" a radio, but rather, merely "assembled" one.

    (Which, again, just highlights my terrible communication skills, since the majority of what I posted does seem to ask for people to post some schematics, and only at the end do I start asking for what I actually want.)

    Another issue with random schematics is that they usually use parts I don't have. For example, the MPF102 I see in both of the schematics posted so far. In neither schematic does there appear to be any biasing on the gate, which means that the 2N7000 MOSFETs I have aren't going to be drop-in replacements. However, I have seen the part used in enough schematics that I was thinking it's probably useful enough that I should have some, but neither DigiKey nor Mouser stock the part, and Jameco wants $1 a piece for the things, and I'm not paying that much for them when, if I actually knew what I was doing, I should be able to build a radio with the parts I already have.

    In short (I type too much), I'm less interested in building a radio, and more interested in designing one. That said, I'm not entirely crazy -- I am looking at other people's designs because I realize that things are done the way they're done for various reasons, but until I understand those reasons, my "learning disability" prevents me from applying the knowledge of others.

    Yes, but again, I'm looking to understand why some values will work better than others.

    For example, imagine we're trying to find a useful mass for a paperweight. Obviously 100 kg would make a terrible paper weight, as would 100 mg. While people could simply tell me that I should stick to values around 100 to 500 g or so, and while that would certainly be useful advice, it isn't what I'm looking for. What I want to know is what makes larger or smaller values fail as paper weights. What I want to know is that I'm trying to find a balance between what sort of mass a human can easily lift and move around, and what sort of mass is sufficient to hold down a sheet of paper against the wind. ...and, not just that, but rather some information about how strong winds are, and how the surface area of paper captures the energy of the wind and how the weight needs to be sufficient that the amount of energy captured isn't enough to lift the weight, and similar information about what sort of mass a human can lift, and how the larger the mass gets, the more inconvenient the use of the paper weight will be. I want to not only choose an appropriate value, but to also understand why that value is appropriate.

    So far, all I've found on the internet is that you choose an inductor and capacitor to resonate at the frequency you want, but I've read nothing about any other constraints, and so as I would understand it, any value of capacitance should work assuming I choose an appropriate inductance, and vice versa, but I believe there are more constraints than that.

    ...and the MPF102, but again, I don't really want to just follow someone else's schematic.

    I disagree. Are there any popular online stores for these things? Like I said, I'd even settle for the little plastic ones. I'm also still not convinced that I shouldn't just try to build one. The capacitance I need is small enough that I could simply use one large pair of plates, and I could easily take care of the issue of keeping the plates separated by just putting a piece of paper between them. I'll probably just end up doing this. It'll be junk, but it's likely that once I get something to work, I'll immediately start working on something new using digital tuning, so it isn't like the quality matters that much as long as it does what it needs to do. I'm also not opposed to pulling plates out of the variable capacitors I have, but I do want to be certain that I'm not going to wish I hadn't sometime later when I discover that my problem actually had nothing to do with my capacitance/inductance ratio being too high.

    Do you mean "noisy" as in that there are a lot of stations? Because when I was talking about noise, what I meant was that on the CB band, during the day, sometimes I'd hear nothing but static with the signal strength meter indicating a very strong signal, like 80% of its full scale, and that was with a proper, large, base station antenna. At night the noise level would go down, but it was still large enough that despite all of the transmitters being (at least) four watts, receiving anything more than a few miles away was difficult.

    Judging from what I read on the internet, I think the ham bands are different. The frequencies may be nearly all in constant use, but what you'll hear is people transmitting intelligible signals rather than just noise.

    So my question is, even though I live about 50 miles from the nearest major city, and 5 miles from even a small city (I live in a small "village"), should I expect to have any signals nearby that are easily received with just a piece of wire as an antenna? Note that I'm not trying to receive anyone in particular as I don't know anyone with a transmitter. I'm curious to what degree I should expect that by random chance there is anyone nearby enough transmitting enough power somewhere on that band that I'll be able to detect anything at all. In other words, if I turn on my receiver and hear nothing at all, does it mean the receiver doesn't work, or does it mean I need to invest in a better antenna.

    I ask because, even though antennas are just wire, I don't presently own any that is of an appropriate gauge and of sufficient length, nor do I have any 50 ohm coax. While I could buy some, my "fun" budget is extremely limited due to disability, and so it is unwise for me to spend any money on anything I am not certain I actually need to make this work, since if I'm wrong, that just makes it more difficult to buy what I actually do need. So I'm trying my best to make do with what I already have, and so if it's possible to make do with a piece of wire, then that's what I'm going to do for now. Once I have a working radio, then I'll may take interest in receiving more distant signals and look into buying materials for a proper antenna, but for now I'd rather not if I don't have to.

    I did look at aprs.fi to get some idea of the local ham situation. I don't know what to make of it, though, as I don't know how to expect APRS activity to correlate to 7 MHz activity, nor do I know how to expect a piece of wire to work as an antenna for signals which, if they exist at all, probably come from Dayton or Cincinatti, which are ~50 miles away, or if I'm lucky, Richmond which is only ~5 miles away. Like I was saying, with a CB radio I'd be in "no way in hell" territory despite the near-certainty of activity in Richmond, only saved by the fact that occasionally a trucker drives through town and transmits something from less than a mile away.

    Not going to happen with my library as it's mostly entertainment. Kids go there to play on the computers and borrow movies, and occasionally read a work of fiction. They no longer even have the slim pickings they had when I was a kid. Now it's all stuff like this, and really very little of even that. I haven't had a library card in ten years simply because I got tired of going there and never finding anything I want.

    I'm not opposed to books, but before I spend $100 on one, I'm going to have to be quite convinced it's useful.

    I found the link for the one book I did read, Basic Communications Electronics. I gave it away years ago, but it was quite good at explaining the super-heterodyne process, though, because of my "learning disability" I was quite convinced the authors were full of it -- that maybe the process worked but not because multiplying 1000 Hz by 1 MHz gives you 0.999 MHz + 1.001 MHz, but rather just something similar. However, I eventually wrote some computer programs and figured out it isn't just an approximation, but rather, it's simply a mathematical fact. Ten years later and I'm still amazed that reality works like that.

    That's actually a much easier sell than a $100 book. I've seen far too many books that are utterly worthless. It's quite common with computer books to find that no only does the author barely understand the subject himself, but the book is also full of mistakes. In particular, I remember reading one of the "Forrest Mims" series of books from Radio Shack, which are largely the source of my belief that I shouldn't expect anyone's schematics to make sense or even work, where I built the circuit only to have the transistor immediately begin to smoke. After looking closer at the schematic, I couldn't see how it could possibly do anything else. I have a rather strong dislike for all things from Radio Shack. In particular those electronics kits they used to sell, with the components attached to springs you'd connect together with wires. While the projects always worked (which amazes me given how I've had terrible luck with schematics from any other source) the book did virtually nothing to teach electronics. Only a vague paragraph at best even attempted to teach anything, the other two or three that accompanied each circuit just explained how to use it. I had at least five of those kits when I was a kid and never really even learned how to use a transistor as a switch, let alone in any linear fashion. The books would have to explain the difference between voltage and current for one thing, and that might require math, and a lot of text, both of which they stayed far away from. You can't learn anything from mindlessly following other's designs.

    Anyway, I'm quite good with math, so feel free to throw as much at me as you like. Indeed, with Wolfram Alpha, one almost doesn't even need to be good at math. I was once trying to find the proper scaling factor for a certain window function I was using in a Fourier transformation and, looking for an easy answer, I just typed in average of sin(x) for 0 < x < pi and to my amazement it gave me an answer, despite the difficulty of the query and that I didn't even try to look up how I should state the problem for it to understand it. Apparently computing has advanced to the point that it doesn't matter that I slept through half a term of pre-calc before dropping out of high school as I can just get the internet to solve anything I don't know how to do myself.
     
  7. KO6WB

    KO6WB Premium Subscriber QRZ Page

    Ahhh, yes, what you are running into is the "in theory it should work" and the "in the real world it won't" type of conflict. If you think about it the coil used in the regenerative receiver gives you the values for a stand-alone circuit that will oscillate.
    Much more relationship of components needs to be realized to make successful circuits from scratch.
    The Q of the components is another avenue to explore. If your inductor or capacitor values are incorrect in this area the circuit could very well be loaded down and cannot break into oscillation.
    Another thing that needs to realized is the resonance of the circuit. This occurs when Xc=Xl. This is Xc which is capacitive reactance and Xl is the inductive reactance. When they equal the remaining component is the resistance of the parts and the reactance becomes zero (in a perfect circuit).
    So, there is a manner in which components can and should be selected to offer the proper operation with the best characteristics. So, what is needed to obtain and maintain oscillations is an amplifier that can obtain gain of at least 1 at the frequency of operation. You need regenerative feedback to obtain and maintain the oscillation. Lastly, you need a frequency determining component to keep the oscillator from just wandering all over within it's area of gain.
    Now, it appears you have overly simplified the frequency determining components by only finding the simple resonance.
    In this manner; ωL = 1/ωC so the frequency at which this occurs is:
    xx.jpg
    In your manner of obtaining the values this is where you seem to have stopped.
    So, this can be a problem in it's own regard. The other is in understanding the actual values being used.
    Capacitance in the ufd range is usually applied more to the realm of audio oscillators. Where as the values for the high frequency range are defined in pfd ranges.
    The inductance of a Henry is again in the lower frequency ranges of audio while mH being from audio to lower HF regions. The next step is to uH which is the HF range for most applications.
    The values are in the following scale;

    [TABLE="width: 600, align: center"]
    [TR]
    [TD]milli[/TD]
    [TD]m[/TD]
    [TD]10[SUP]-3[/SUP] = 0.001[/TD]
    [/TR]
    [TR]
    [TD]micro[/TD]
    [TD]µ[/TD]
    [TD]10[SUP]-6[/SUP] = 0.000,001[/TD]
    [/TR]
    [TR]
    [TD]nano[/TD]
    [TD]n[/TD]
    [TD]10[SUP]-9[/SUP] = 0.000,000,001[/TD]
    [/TR]
    [TR]
    [TD]pico[/TD]
    [TD]p[/TD]
    [TD]10[SUP]-12[/SUP] = 0,000,000,000,001[/TD]
    [/TR]
    [/TABLE]

    So the reference to the use of a nH in a resonant circuit isn't usually considered as a valid value until you get in to the VHF/UHF range of frequencies.
    So, on a fundamental level of "in theory" you are having a difficult time determining why your circuit won't work.
    Raise the nH values into the uH values and find the inductance in this range. For your capacitance depart from the nFD range and lower it to the pFD range.
    The values you have in the nH and nFD are so low in reactance and Q as to be useless at the 7MHz range. It'll look more like a short circuit to the use in the 7MHz range. If you short circuit the input circuit to an amplifier it cannot possibly do any amplification. No amplification = no oscillation. Remember, you need a gain of at least 1 to oscillate.
    From reading your posting, it seems you are not at the level of being able to make the circuit function on the basic level.
    The super-hetrodyne process is not a multiplication process. It's a mixing process where the components of the mixing are a product of the sum and difference of the two signals.
    You have a desire to build and that's great. You should look at examples that already work and using the math to figure out why it works the way it does.
    Practical application and understanding of the manner in which things do and do not function takes a bit of time.
    The fundamentals are a necessary area for the understanding of the practical applications.
    Return to the fundamentals and review what is or is not a proper application of theory and real world applications.
    You will find that the math will also take you to the point where you will discover the use of the components you are applying to the one circuit you wish to build, will cause it not to work.
    From there you can find the model to do exactly what you wish to do.
    You have overly simplified the circuit. Electronics is much more complex.
    Keep on working on it.
    73
    Gary
     
    Last edited: Mar 7, 2013
  8. G0HZU

    G0HZU QRZ Member

    Wow that was a marathon to read :)

    I think I understand your frustration here. I think that going online or wading through books with application notes for 'proven' oscillator circuits is just going to increase your frustration because the background info about the design will probably not be included :(
    Sadly, one of the disappointing aspects about amateur radio kit design is that the circuit design is rarely discussed in any detail beyond a basic description. Here in the UK we have a monthly magazine from the RSGB called RadCOM and it has some really good stuff about building simple but very effective homebrew circuits. But there isn't really much in the way of theory or design strategy in the articles which is a real shame. I guess there isn't the space available in the magazine to justify this as the theory probably has limited appeal compared to just providing the construction information alone. So it has to be limited in terms of page space :(

    However, one good way to help understand oscillator theory is through simple modelling on a PC.
    I was introduced to VCO design/modelling techniques nearly 25 years ago using an old DOS PC with some very basic linear analysis tools. I learned more in one hour about oscillator circuit optimisation (analysed in open loop) than I could extract from reading lots of theory books.

    These days the design tools are much more powerful. Eg tools like AWR MWOffice can analyse a basic oscillator in open loop and closed loop and give a simulation of the gain margin, phase response, phase noise, harmonics and output power etc. It will even model the noise contribution from tuning varactor diodes etc.

    But there are also plenty of free tools available that can help with the basic design and understanding of oscillators. eg it is very easy to analyse an oscillator in open loop using a freebie linear simulator.

    This will reveal basic info like gain margin, phase slope (affects phase noise and stability) and also the most likely frequency of oscillation. Are you familiar with these techniques?


    As you begin to look beyond basic simulations towards high performance designs and maybe begin to be concerned about VFO drift etc then there is no substitute for experience so this is a good time to revisit the theory books and those 'proven' designs used by hams for decades. i.e. once you have a feel for the basic theory (from playing with a simulator) then maybe the theory books and application notes will make more sense and you can fill in any gaps in the info yourself :)
     
  9. G0HZU

    G0HZU QRZ Member

    I think a mixer IS a multiplier? If you multiply two sine waves you get sum and difference frequencies as the result

    I think the following trig identities apply for the multiplication process (but I am a bit rusty on the equations)


    sinA.sinB = (cos(A-B) - cos(A+B))/2

    cosAcosB = (cos(A-B) + cos(A+B))/2

    sinA.cosB = (sin(A-B) + sin(A+B))/2

    etc
     
  10. MOOSFET

    MOOSFET QRZ Member

    I assume you think I mean multiplication as some type of amplification.

    What I mean is that mixing is when the instantaneous voltage of one input signal is multiplied (mathematically) by the instantaneous voltage of the other signal, and the output is the result of this multiplication (except that the circuit probably creates a different DC offset). The multiplication of two frequencies, A and B, has the effect of creating a new signal that is the sum of two different frequencies given by A+B and A-B.

    In equation form, it looks like this: 2 * cos(2 * pi * a * t) * cos(2 * pi * b * t) = cos(2 * pi * (a + b) * t) + cos(2 * pi * (a - b) * t)

    To simplify that equation, realize that to make a cos() function output a waveform at a specified frequency, you simply use "cos(2 * pi * frequency * time)" which solves to the instantaneous voltage of a sinusoidal waveform of the specified frequency at the specified time, with the frequency units being hertz and time units being seconds. So all the cos() functions are doing is telling us what the voltage of each signal is going to be for any point in time. So it simplifies to just "2 * v(A) * v(B) = v(A+B) + v(A-B)" which describes what happens in a balanced mixer.

    I have "found" a copy of this book. It's almost useful.

    For example, at the beginning of the chapter on sinusoidal oscillators, it says "In order to sustain sinusoidal oscillations, a network must have a pair of complex conjugate poles in the right-half complex plane when power is applied at t = 0." That sure sounds like it might be useful information, but unfortunately, it explains nothing about these complex conjugate poles. I know what complex numbers are, and figure poles are some sort of solution to some sort of equation. I briefly considered looking up the definition on the internet, but soon realized that nowhere does the book mention what equations are being solved, or which part of the circuit is being analyzed, or anything, about where these complex conjugate poles, if they exist in the circuit, come from. Then I did look it up on the internet, and found that these poles are the roots of a quadratic equation. That's fine, I know about quadratic equations and roots. However, that still leaves me with the problem that the book doesn't mention what equation we're finding the roots of, or where it comes from. So I read further, where I see a block diagram of a feedback loop, which has an amplifier (just a triangle) and two squares. I have no idea what the squares are supposed to represent. They have "H{1}(P)" and "H{2}(P)" written in them, with {x} representing a subscript. It then uses these in an equation it doesn't bother to explain the source of, to calculate something... Ok, so now I can calculate something, using an equation I don't know the purpose of, and using numbers which came from squares in my circuit which I have no idea what they represent. Awesome!

    So I thought, maybe the author is expanding on a previous concept introduced earlier. So I go to the beginning of the book, and find this near the beginning of the first chapter:

    (I'll represent variable subscripts by putting the subscripts in {} brackets, and exponents with the ^ symbol.)

    "Our key assumption is that the emitter current and the base-emitter voltage of the transistors are related by Eq. (1.1-1):

    i{E} = I{ES}e^(v{BE}q/kT)

    v{BE} = (kT)/q * ln(i{E}/I{ES})

    where k = 1.38 x 10^-23 J/°K is Boltzmann's constant, q = 1.6 x 10^-19 C is the electronic charge, and I{ES} is the emitter saturation current."

    ...and if you think the way I typed those equations is difficult to read, rest assured the representation in the book is no better. In particular, the exponent to which e is raised is shrunk to fit in superscript, meaning that "v{BE}q/kT" looks nearly like "vBEq/kT" which looks more like gibberish than math.

    Like I said before, I've got nothing against math itself, but this book reminds me too much of my experience with trying to learn how to perform a fast Fourier transformation, which ultimately ended with me figuring out how to do it myself in far less time than I spent trying to understand other people's explanations of the algorithm. You can't just toss out equations like that and expect anyone to have a clue what you're talking about without bothering to explain what all of the terms in the equation are and at least giving a vague idea of why the equation does what it does with those terms.

    That said, I was impressed that the final result of the above equation was a table that shows the relation between the base-emitter voltage and the emitter current in a transistor, simply because that makes it the first text I've ever seen (other than transistor datasheets) to acknowledge that the voltage drop over a base-emitter junction isn't always exactly 0.7 volts. However, that just makes me wonder what the point of that math was. I can look that sort of thing up in a transistor datasheet, so why do I want to spend an hour per page trying to understand this textbook? It's like it's throwing around math just to show off that the author knows how to do this sort of thing, which might actually explain why there's no attempt at explaining any of it. It isn't educational, it's just a demonstration.

    Then I recalled that the beginning of the first chapter did sort of state that it was a demonstration of what was to come. So I skipped ahead to the second chapter, but again I'm left with the impression that the book is just a paper someone wrote to demonstrate that they know how to do these things, rather than a text written to teach someone about these things. It continues to put formulas everywhere without explaining where any of them came from or what any of them do. The author is also a huge fan of subscripts, which drive me insane. Must every resistance in a circuit be represented by the variable R? There are other letters, but he prefers to pretend they aren't available. Unless they're greek letters, in which case he'll use as many as possible. It's like trying to read a math article on Wikipedia, where the authors attempt to find the most obscure and non-obvious manner in which to explain even basic arithmetic.

    I suppose one might assume I need to become better at math, but it isn't like I don't understand addition, subtraction, multiplication, exponents, roots, and everything else in these equations. What I'm not understanding is the equations themselves, because the book makes no effort at all to explain them. It just tosses them out there as if they're something I should already be familiar with, but if that were the case, then why would I be reading this book?

    I do appear to have a gap in my understanding...

    The mention of "complex conjugate poles" in that book made me start to think that perhaps I'd do well to start thinking about the circuits in terms of phase angles rather than simply voltage amplitudes. Then something about all of this made me start to question whether the best way to create oscillations is to amplify the signal on the tank circuit and then use positive feedback to strengthen it. It just seemed like the feedback should be slightly advanced in phase if it is to increase the strength of the oscillations. That feels like what I do when I swing a pendulum at least.

    So I created an AC voltage source in Spice, and fed it into a tank circuit via a resistor, with the frequency of the AC voltage matching the resonant frequency of the tank circuit. Doing so, I found that it does appear that the oscillation of the tank circuit is the same phase as the AC voltage source. Then I tried coupling the source to the tank with a small capacitor. Doing that, I found the oscillation in the tank would build up, then die down to nothing, then build up again repeatedly. I guess my AC frequency and the tank resonance weren't exactly matched. I looked closer and found that when the tank oscillations increase in amplitude, the voltage source lagged the oscillation by 90°, but when the oscillations decrease in amplitude, the voltage source phase is advanced relative to the oscillation phase. This was the opposite of what I was expecting. So I thought about this for a while, and realized it made sense, since it is when the AC voltage source is passing the zero axis on its positive transition, that is when the voltage is changing at the highest rate, and therefore, that is when the most current is flowing through the capacitor, since current flows through a capacitor only when there is a change in voltage. So then I tried coupling with an inductor instead, and found that the voltage source led the oscillation by 90°.

    So it would seem that in any case, the feedback into the tank circuit needs to be such that the tank receives positive current when the voltage is positive, and negative current when the voltage is negative. This was a new revelation as I generally don't pay much attention to current, but it seems I need to start thinking about it when I try to design circuits.

    Anyway, since it seemed that passing a voltage through a capacitor advances the phase of the signal by 90°, I looked again at the amplifier in my oscillator. First the signal from the tank passes through a capacitor, advancing it 90°. Then it passes through an NPN-based inverting amplifier, advancing it another 180°, to 270° total. Then another capacitor couples it to the next amplifier, which advances it another 90° because of the capacitor and another 180° because of the second amplifier, bringing it to 540°, or removing 360° from that, 180° total. Then there's another capacitor before the signal is fed back into the tank circuit, making the feedback loop 270° rather than the desired 360°. That might explain why it doesn't work.

    (and, like I said before, I probably should have posted a schematic to begin with, as someone could have easily figured that out)

    All this time I was using two transistors because, since each amplifier is inverting, I thought I needed two in order to create positive feedback. However, it seems I only need one. When the input from the tank is rising across the zero point, the voltage at the base of the transistor is at its peak because of the capacitor that links the two together, and so the voltage at the collector is at its lowest point, which means that the current passing through the capacitor back to the tank circuit is also at its minimum (zero, not negative), and about to begin going positive, which is exactly what I need at the tank circuit since it's voltage is about to go positive and I need to feed it current at the same polarity as its voltage.

    Perhaps I shouldn't have paid so little attention to the AC chapters in that college-level electronics textbook I once had. I vaguely remember reading something like all of this in it at one time. I just kind of ignored it all since I wasn't interested in 60 Hz circuits and it didn't occur to me that the same concepts would apply to radio. Unfortunately I no longer have the book.

    I'll probably take some time to look over the AC chapters of All About Circuits, though I don't know how far I'll get as their backwards current notation makes me crazy and I need to keep my brain calm in order to learn anything. I should probably look for a different resource.

    Any other suggestions for books and or web sites? Obviously I'll search for both on my own. I'm just interested if anyone knows of any they've found in the past that they thought were particularly well-written.
     
  11. G0HZU

    G0HZU QRZ Member

    Not sure what happened to my last couple of posts (deleted?)

    I don't think you are correct here and I think MOOSFET is correct :) The mixing in a typical superhet receiver IS a multiplication process despite the fact the output terms appear to us as sum and difference terms. Why else is a product detector ( a form of mixer) called a 'product' detector (product implies multiplication)

    Also a mixer on a schematic has the X symbol embedded in it to give you a clue that it involves multiplication of the LO and RF signals.
    I think the following trig identities apply for the multiplication process (but I am a bit rusty on the equations)

    sinA.sinB = (cos(A-B) - cos(A+B))/2
    cosA.cosB = (cos(A-B) + cos(A+B))/2
    sinA.cosB = (sin(A-B) + sin(A+B))/2

    Looking at the output on the right in each case you do get sum and difference products falling out of the maths and I think this is what initially surprised MOOSFET.
    But MOOSFET analysed it using maths and found out that mixing IS a multiplication process. If you still don't agree with MOOSFET then try inserting the above equations into excel and graphing the results.

    You will see the maths (multiplication) gives sum and difference terms (sine waves) appearing in the excel graphs. So if you mix 28MHz and 38MHz you will get 10MHz and 66MHz as the main mixer output terms for the simple maths above. In a real world (imperfect) mixer there are lots more terms as we all know but this requires more maths to model.

    Getting back to oscillator design, the best way (IMO) to understand the basics of how a typical circuit works is to model it in open loop on a linear simulator. Just trying to get it to oscillate in closed loop on a SPICE simulator isn't going to give enough insight.

    You also need to be aware of what aspects of the design contribute to high phase noise and poor stability.
    MOOSFET, have you analysed your circuits in open loop?
     
  12. MOOSFET

    MOOSFET QRZ Member

    Strange... They're there now, but I don't recall seeing them there before.

    I had a similar boost to my understanding when I first got an oscilloscope. Being able to see what is going on makes understanding it a lot easier.

    I think that some of that has to do with the tendency of texts to oversimplify things. Either they're telling you something that's sort of correct but not exactly correct, and so you think you know something but you really don't, or they're telling you something that's exactly correct, but by then you've learned that they like to approximate things, and so you trust your own idea of how something works over the factual information you're given in the text, and so once again you think you know something but you really don't. The result is that the only way to really understand what's going on is to see for yourself what actually happens, then you know what is correct and what isn't.

    Not too much. I just learned to use the Spice circuit simulator a few days ago, and so still have a lot to learn about it, but in general it doesn't seem difficult to use. I did try an open loop analysis just an hour ago of the two transistor amplifier I was using. According to that, at 7 MHz it has a gain of 39 and a phase shift of 0.01 degrees. That should work from the point of view of everything I've ever been told about what is required for oscillation, but when I build it, it's very difficult to get it to work.

    I suspect that much of the problem is that I'm trying to use the same point in the tank circuit as both input and output. I've never actually seen an oscillator schematic that does this, and there may be a very good reason for that, as it certainly seems like a bad idea even to me. However, I'm not convinced that I really understand it at this point, and so I continue to try to make it work.

    I should probably add a schematic already...

    [​IMG]

    No component values as I've essentially tried them all, not only in real life but also by writing a script to run thousands of spice simulations with different values, attempting to find what works best. I've also tried variations on the circuit, like not using a resistor on the emitters of the transistors, and instead biasing them with a 10k potentiometer so that the emitter is 0 volts. Also, I usually put capacitors between the emitters and -5 volts when I remember, which is only about half the time. (...and so I forgot to put them in the schematic.) Once you add some real-world resistance, like 0.01 ohms of resistance in the tank circuit, it generally doesn't seem to work very well. I imagine there is a large problem with getting the circuit to sense more of the tank circuit's state than it senses of its own output, while at the same time providing more energy to the tank circuit than it removes from it. While that seems like an impossible obstacle to overcome, I'm working under the assumption that the portion of the amplifier's output which isn't correct, due to it mostly amplifying its own output rather than the tank circuit's output, will be shorted to ground by the tank circuit, thereby making it possible for the amplifier to provide additional energy without it mattering so much that the additional input energy isn't exactly correct.

    The most similar oscillator schematic I've seen is this one, in that the tank circuit uses just one capacitor and one inductor, but even it doesn't attempt to use the same point in the tank circuit as both input and output.

    Also, after reading some of the AC portion of All About Circuits, I came to realize that the phase shift through a capacitor isn't going to be exactly 90° any more than it is 0°, but instead it will depend on the impedance of the signal source, the capacitor, and the destination, as the three create a voltage divider and the phase shift depends on how all of that math works out. I think I understand how to calculate that now, but I'll have to test myself later, using the simulator to check my results. Anyway, since my capacitors offer little to essentially no impedance to 7 MHz depending on which ones I choose to use, it works out that the phase shift through the capacitors is actually much closer to 0° than to 90°.

    So you can't do this open loop simulation in Spice? What I found searching for "open loop" was that it's just an oscillator with the input connected to an AC voltage source instead of to the oscillator's tank circuit, in order to analyze how the circuit works over a range of input frequencies, which is something that can be done in Spice.

    ...and that's probably something that would be very useful which I know nothing about, and not just for my design but in general.

    Every reference I've read about oscillators has said "just use an amplifier that provides positive feedback and a gain greater than unity" and then promptly told me to use a Colpitts or Hartley design and pray that it works, rather than explain how to design an oscillator from scratch in such a way that I don't have to pray that it will work. Unfortunately, it seems there are far more people writing about the topic than actually understand it in great detail. It all leaves me with the impression that the standard way to design an oscillator is to toss together some components and see if they do anything.
     
  13. KO6WB

    KO6WB Premium Subscriber QRZ Page

    Anybody can come up with a complex math model to explain the simple. In the case of 28MHz and 38MHz the simple sum is 66MHz anf the difference is 10MHz. Plus and minus, simple, accurate and appropriate for the actions taking place.
    There is no multiplication or, in the case of difference, a division that can explain the mixing process more accurately. It may explain it in exactly the same terms at the expense of making the simple more difficult.
    As for the making of an oscillator the impedance of the tuned circuit becomes important. A parallel tuned circuit have a very high impedance when at resonance. A series circuit will be a low impedance at resonance.
    Simple basics. Having looked at the circuit diagram there is no way that one will work.
    Look at a oscillator using two transistors in this arrangement and note how the feedback network is in series not parallel.

    How many degrees of phase shift must the feedback circuit (the box in this schematic) introduce to the signal in order for this two-stage common-emitter amplifier circuit to oscillate?
    [​IMG]
    [TABLE="width: 100%"]
    [TR]
    [TD][TABLE="align: center"]
    [TR]
    [TD="align: center"][/TD]
    [/TR]
    [/TABLE]
    [/TD]
    [/TR]
    [/TABLE]

    [FONT=Verdana, Arial, Tahoma, sans-serif]Why is this amount of phase shift different from that of a single-transistor oscillator?[/FONT]

    [FONT=Verdana, Arial, Tahoma, sans-serif]The zero degree shift is the answer but not in the way you have implemented it.
    This is where things are going wrong. Yes capacitors and inductors do introduce a shift of their own characteristics but when you combine them
    what will happen? The lead and lag cancel. So, what happens to phase shift?
    Look at it in a different manner and use the tuned circuit in a different configuration.
    Hope this helps
    73
    Gary
    [/FONT]
     
  14. AF6LJ

    AF6LJ Ham Member QRZ Page

    If your cap and inductor are in series that should make that circuit oscillate.
     
  15. G0HZU

    G0HZU QRZ Member

    Yes but do you realise that BOTH the sum and difference terms are formed by the multiplication and they both appear in the RHS of the simple trig identity for the sinA*sinB identity where A is the RF and B is the LO? The same simple multiplication produces BOTH the sum and difference terms in the result as cos(A+B) and cos(A-B).

    To properly (accurately) understand a mixer/multiplier it is best to embrace these simple trig identities I posted up earlier for multiplying two sine waves because you can use the relevant identities to quickly predict the phase shift of the sum and difference terms in a mixer to show how IQ image cancelling mixers can be produced with a few basic building blocks.
    It's a lot harder to explain how an image cancelling mixer works without the maths because how else do you explain to the student how you get different phases for the sum and difference terms for the I and Q channels? Are they meant to just accept the word of the teacher that they give this phase shift? Such mixers are part of the future of ham radio as they get used in some SDR type receivers so it is worth exploring the maths/theory.

    The trig identities I posted up can be found in any schoolkid's maths book if they took maths at a reasonable level :)

    Yes you can use SPICE but in my experience the most efficient tool is a linear simulator aimed at RF design because you can plot things like group delay, loaded Q and return loss easily when doing the open loop analysis and you can define ports easier. Also, you have to be careful with how you reference the phase on a SPICE simulator because you can easily reference it incorrectly and get confused as to how much phase there really is around the loop.

    The models are usually easier to manage in RF simulators compared to SPICE but I guess there will be some SPICE users that will have addressed all the above issues in one way or another :)

    Also, SPICE can be really useful when modelling the circuit in closed loop so if you are happy with using SPICE then stick with it :)

    The basic design goals beyond getting the phase around the loop correct are to try to get high loaded Q. One of the most efficient ways to monitor loaded Q (QL) is to measure the group delay because you can measure this easily with a VNA and also plot it easily on a linear simulator to compare the two.

    Group delay is rate of change of phase wrt frequency and you want LOTS of phase slope around the zero point if you want low phase noise as the zero point is more tightly defined with a steeply changing phase wrt frequency. Also, you need to keep the zero phase point close to the gain peak of the feedback network for best performance.

    So you need to preserve this QL/phase slope/group delay when you attach your amplifier. So you have to pay attention to loading effects. You also have to try to avoid too much loop gain (causes undesirable heavy saturation)

    That's why open loop analysis can be very useful. You can also analyse for negative resistance if you find this concept easier to work with :)

    They say that in order to design an oscillator you first have to design an amplifier.
    It's a good idea to choose a decent and proven amplifer design and I would recomment you choose either a JFET source follower or a common base BJT amplifer for your initial oscillator design analysis. These amplifers are fairly easy to work with and understand.

    So I would suggest you throw the dual stage CE amplifer circuits in the bin. Both are pretty dire :0) The second one poses a question that can't be reliably answered because the phase shift through the two stage amplifier isn't certain. It might be close to zero degrees but because it looks like it will be affected by Miller effect (high gain in both stages with bypassed emitter resistors and unknown gain) the phase shift could be quite different to zero degrees and this will depend on stage gain, frequency and also the component values (eg the coupling caps can introduce phase shifts). So I would see it as a trick question :)
     
  16. G0HZU

    G0HZU QRZ Member

    [​IMG]

    Probably the most relevant amplifier topology to investigate is the common drain JFET amplifier. At 7MHz it has very high input impedance and fairly low small signal output impedance (ballpark 80 ohms)

    It will also show very low phase shift at 7MHz so you just need to design a tapped tank circuit between source and gate that gives 0 degree phase shift around the loop at 7MHz.

    For example you can tap into the tank inductor (Hartley) or you can tap into the capacitor (Colpitts) when designing the low impedance tap point at the source.
    This type of circuit is easy to analyse for small signal response in open loop (break the loop at the node of the low Z tap point and the JFET source). But you do have to be wary of large signal performance once oscillation equilibrium is reached. This is because you do not want to have sufficient RF voltage at the gate to forward bias the PN junction here.
    If this happens then it is bad for the JFET.

    So a degree of caution is needed for large signal operation (espeicially if you are trying for a very high loaded Q) but even if you design it for way too much loop gain you can still get it to oscillate.

    There are methods to negate the forward bias problem at the gate (eg a clamping diode and/or adding a second tap for the gate connection) but for now just try getting it to oscillate and try to master the art of tapping the tank and maintaining high loaded Q and simulating and proving where it will oscillate.
    Note that most real world designs will include an RF choke in series with the source biasing resistor but it isn't essential to have this if you just want to make a basic oscillator.

    I can show you how to design the network if you like but I think it is better for you to have a go first. You can google for capacitive tap impedance matching programs to get an idea how to design the tapped tank circuit. I wrote my own in BASIC many years ago (and then chaged to QuickBASIC and then DOS when I bought my first PC) I think I got the design equations from an old ARRL handbook from the 1980s.
     
  17. MOOSFET

    MOOSFET QRZ Member

    What software would you recommend then? I only use Spice because I've heard of it before and obtaining a copy for Linux wasn't difficult. I've only used it for a few days so it isn't as if switching to something else at this point is much more difficult than continuing to use Spice. However, I tried searching for other software and failed to come up with any names, so I must not be searching for the right thing.

    I tend to get lost when people start talking about Q. I mostly just try to ignore it. Every time I read about a definition of it, it always involves someone taking some quality they like about a component and dividing it by another quality they don't like, and calling the result a "quality factor" and then shortening that to just Q, since apparently in electronics we're just not happy unless everything is represented by some single uppercase letter. ...and what I tend to get from that is "Q is how much the author of this book likes a component" and thus it never seems worth paying much attention to. Consequently, when they later talk about a circuit having high Q, all I get out of that is "they like this circuit more than that circuit" which doesn't seem very helpful.

    I guess what mostly puts me off of it is its unit-less nature. If you want to talk about the resistive losses in an inductor vs. the inductance, then talking about henries per ohm seems like a useful idea, but to just refer to it as "Q" seems unnecessarily non-specific. Then to sum up an entire circuit in terms of Q seems only to ignore the issues of which components you care about the Q of and which you don't, and in what proportions, and indeed, how the Q of an inductor even compares to the Q of a capacitor since, if anyone had bothered to give the Q values units, we'd realize we're comparing entirely different things. ...or perhaps you aren't talking about a Q that is derived from the components' Q, but some other Q that is calculated in some other way. It's kind of hard to tell when the single letter Q is used to represent any aspect of anything that makes it better or worse than any aspect of anything else. It's just hopelessly vague.

    That seems like it'd be easy enough to do with Spice. Here's an example from an open-loop analysis:

    Code:
    Index   frequency       vm(10)          vp(10)         
    0       6.500000e+06    3.516660e-06    1.480847e+00
    1       6.600000e+06    3.927143e-06    1.470275e+00
    2       6.700000e+06    4.435478e-06    1.457168e+00
    3       6.800000e+06    5.081246e-06    1.440491e+00
    4       6.900000e+06    5.928361e-06    1.418558e+00
    5       7.000000e+06    7.086725e-06    1.388447e+00
    6       7.100000e+06    8.761107e-06    1.344618e+00
    7       7.200000e+06    1.137447e-05    1.275279e+00
    8       7.300000e+06    1.591737e-05    1.150897e+00
    9       7.400000e+06    2.486435e-05    8.802113e-01
    10      7.500000e+06    3.830202e-05    1.926599e-01
    11      7.600000e+06    3.037544e-05    -6.78407e-01
    12      7.700000e+06    1.899311e-05    -1.06221e+00
    13      7.800000e+06    1.324781e-05    -1.22417e+00
    The plain-text output isn't much fun, particularly with its insistence of always using scientific notation, but I wrote a Perl script to make graphs of the data that comes out of it, so it works.

    Well, I'm as much interested in knowing why what I'm trying to do won't work as I am in finding something that will work. It's just the way I learn things. When I find something that does something other than what I expect it to, then it means there's something I don't understand about it, and so I stick with it until I understand why. It makes learning a slower process, but I feel like I understand things better in the end. That said, I will have to play with some other designs as well, in order to see how they behave in the simulator so that I can then compare what I see with them to what I see with my current design.

    ...and this is what confuses me the most about electronics. All of this is stuff that people have figured out long ago and yet it seems to be absent from any text I've read. Indeed, I've been wondering all week exactly how I'd go about figuring the impedance of the input and output of a transistor amplifier as it seems like it'd be useful to know, but I've never read anything that even hinted that the subject exists. There's all sorts about the impedance of passive components, but toss in an active component and suddenly the subject no longer exists.

    I'd really love to find a book that explains topics like this, as it certainly seems like there's enough I don't know to fill a few books, but I'm starting to feel like there's a law of nature that requires that any sufficiently advanced topic be either obfuscated to the point that it is incomprehensible, simplified to the point that it is useless, or simply ignored entirely.

    I guess I will need to get some JFETs then. I've never actually used them before. At the moment all I have are 2N3904 / 2N3906 BJTs, and 2N7000 MOSFETs. I'm thinking of getting these since they're cheap, supposedly designed for RF and, judging simply from the schematic symbol, the source and drain are interchangeable which seems to be what most schematics I see want, given the gate is always attached in the middle instead of near the source.

    Oh, certainly. I haven't had much time to experiment in the last few days, so unfortunately I've barely even begun to look at open loop simulation. I'll come back to the forum again when I feel like I've run out of things to learn from that and/or I've run into another dead end.
     
  18. G0HZU

    G0HZU QRZ Member

    There's lots of questions to answer so I think I'll first try and remove your confusion over Q.

    UNLOADED Q

    If you consider that reactive components store energy (eg inductors or capacitors store energy by their very nature) then unloaded Q tells you the ratio of this energy stored compared to the energy wasted (dissipated) in its real world (unwanted internal) resistance in that component.

    So if you wind yourself a 3uH inductor using a Micrometals T50-6 powdered iron core (about 25 turns?) and look at the impedance on a VNA at 7MHz you might see 132 ohms inductive reactance in series with 0.6 ohm resistance at 7MHz on the VNA display with the marker set at 7MHz.
    From this you can work out the Qu (unloaded Q) of the component.

    It is simply the ratio of series reactance (that stores energy) to series resistance (that wastes it away). In this case Qu
    =132/0.55 = 240

    So the unloaded Q (Qu) of the component is 240. From this figure and the 3uH inductance you can reverse engineer the losses in the component at 7MHz by simply working out the reactance using X = 2*Pi*F*L

    Then get the series resistance from the Equation for Qu = X/Rs

    Rs = 132/240 = 0.55 ohms. (at 7MHz)

    So Qu at a given frequency is quite a useful thing to know :)

    Here's a set of typical Q curves for some T50-6 inductors. You can see how Qu changes with frequency.

    [​IMG]


    That's about it really...
    Loaded Q is a bit more complicated in terms of definition but are you happy with the above definition for Qu?
     
  19. G0HZU

    G0HZU QRZ Member

    I can't edit typos in my posts* but this bit below:

    ...should say 0.55 ohms on the VNA instead of 0.6 ohm.


    LOADED Q

    Loaded Q relates to the selectivity of a tuned circuit and is a figure of merit based on Fc/BW where Fc is the centre frequency and BW is the 3dB bandwidth.

    So in the case of an oscillator where you want high loaded Q (high QL = high selectivity and qives high rate of change of phase wrt frequency for low phase noise) then you ideally need to design for this selectivity using components that have high UNLOADED Qu because if the ratio of unloaded Q (Qu) to loaded Q (QL) gets too small then the circuit will get very lossy.

    So you need to keep the Qu/QL ratio reasonably high or make sure you have enough loop gain to overcome the losses with adequate margin.

    In a typical oscillator at 7MHz then you might use an inductor with Qu = 240 as in my previous post.

    But a typical feedback network might be designed for a LOADED Q (QL) of 24 giving a Qu/QL ratio of 10. This lowish ratio will give some measurable insertion loss in the LC network.


    *Also my posts can take several days to appear as I've been put back on moderator approval and I can't send or receive PMs anymore. I think I've upset somebody at some point LOL.

    So getting help from me is going to be a slow process...
     
  20. G0HZU

    G0HZU QRZ Member

    You mentioned books to read and there are some classic RF design books eg by Bowick and by Hayward but I'm going to recommend this one below as a semi technical reference.


    This is my copy from when I was a student at Uni and I guess the fact that you can see that I read this book until it fell apart tells you why I rate it so highly.

    However, it's not all good news as there are a lot of typo errors in the schematics and there is only very basic design theory and very simple maths. The section on small signal class A amplifiers has a lot of design errors/mistakes when calculating impedance (or at least the revision that I have does). So I'd avoid that part of the book.

    The reason I recommend it is because it contains a LOT of hands on experience of RF design and pretty much every page has practical tips and advice. Also there are lots of classic RF circuits in there also the technical descriptions are really only aimed at technician level.

    I skimmed through the section on VFO design and it is a very good introduction with lots of good info. But you will need to read something more advanced alongside this book.

    i.e. this is an excellent 'hands on' book to have alongside the more technical books by Bowick or Hayward or Rohde but don't trust the worked examples as there are lots of obvious typos when it got transferred to print :(

    [​IMG]
     
Thread Status:
Not open for further replies.

Share This Page

ad: w5yi