- AuthorPosts
- July 5, 2006 at 1:09 am #6643
Well after a while away from the project I decided to begin again in earnest with the setting up of the SSTRAN in an outdoors installation.
I have a 3 meter vertical consisting of a section of 3/4 inch pipe and a section of 1/2 inch pipe soldered together with a reducer coupling. I have this mounted up about 3 feet from the ground and will be putting in the ground / radial system tomorrow.
Well after a while away from the project I decided to begin again in earnest with the setting up of the SSTRAN in an outdoors installation.
I have a 3 meter vertical consisting of a section of 3/4 inch pipe and a section of 1/2 inch pipe soldered together with a reducer coupling. I have this mounted up about 3 feet from the ground and will be putting in the ground / radial system tomorrow.
In terms of making the loading coil, I had a few questions since I wanted to be able to wind up something with the parts and coil forms I have on hand. From the directions given on the SSTRAN site, it would seem that the SSTRAN loading coil is at least 400 uH. Could someone operating near 1600 kHz confirm the loading inductance they ended up with?
What I am trying to understand is why so much inductance would be necessary to tune this antenna. From what I have seen , a 10 foot vertical needs typically 120 uH to 140 uH to bring it to resonance at 1600 kHz. From the mods on the SSTRAN website it appears as though you lower C23 from 820pF to 560pF, and effectively remove the coils L4 to L7 from the circuit by shorting them with S5. This leaves C23 in parallel with trimmer C5.
What I don’t understand is why there is so much capacitance left in the circuit. Is that why the loading coil has to be so large? Wouldnt it be better to simply remove C23 altogether and just use C5 to tune the output? From what I understand, an electrically short vertical acts as a capacitive reactance, so I am imagining it to be in parallel with C23/C5. If you’re going to bother with taps on the loading coil and changing the effective length of the antenna to tune it, why even bother having a variable cap (C5) in the circuit? Or better yet, why not just use it and a fixed antenna length and loading inductance?
Can someone help me understand more in depth about the exact theory behind the SSTRAN tuning procedure as it is written on the SSTRAN web site? I am just trying to make this a little easier if at all possible with the parts I have on hand.
thanks again,
Dave
July 5, 2006 at 2:40 am #13566radio8z
Guest
Total posts : 45366Dave,
I can’t answer all of your questions but here is some information I have that may get you started understanding the situation.
I built an antenna similar to the one on the SSTRAN site. The difference is that I used #14 wire on the same form. To get the required L, I had to “wind back” a second layer from the top of the form. I mounted the antenna and coil over 10 10′ radials layed on top of the earth. I tuned the antenna for best response at 1520 kHz. The inductance at the tap which produced maximum test point reading in the SSTRAN was at an inductance of 250 uH. (measured with the coil removed from the antenna/ground between the tap and the top of the coil with a HP4250A bridge).
Many view the coil function as cancelling the reactance of the radiator. There is another way to look at the interaction between C5/C23 and the antenna and coil. C5/C23 and the coil form a L network and if you include the radiator capacitance, they form a pi network. The antenna C does not appear in parallel with C5/C23 since the coil is between them. L and pi networks are commonly used for impedance transformation in radio transmitters. It is likely, though I have not analysed this for the SSTRAN, that the combination of C5/C23 and the loading coil form a L network to transform the highly reactive impedance of the radiator to one which is acceptable to the transmitter and that adjustment of C5 optimizes this transformation for the transmitter. L4 thru L7 probably serve the same function when using a wire antenna. If this is the case, I see no way to simplify the design, but perhaps you will fine one.
Neil
July 5, 2006 at 1:37 pm #13570Rich
Guest
Total posts : 45366Dave — The calculated base impedance of a ground-mounted, 3 meter copper radiator of 0.625″ OD on 1600 kHz is about 0.106 -j2523 ohms. That would take about a 250 µH coil to resonate it, which calculation tends to support the post of Neil (Radio8Z) that included a real-world measurement.
When this antenna is resonant, its base impedance will be 0.106 +j0 ohms plus the resistance of the coil plus the resistance in the r-f ground connection. The ground path also could be reactive, but ignoring that.
The resistance in the path for r-f earth currents back to the tx system (typically called “ground”) will be the biggest variable here. Even when the antenna itself is resonant, the resistive load that the tx sees will be highly dependent on the “ground” connection.
The tx output circuit needs to be adjustable to deliver its maximum rated power into whatever load it sees, and that is why most Part 15 AM txs provide some means of doing that.
The tx adjustments are not really “tuning” the antenna as in making it resonant. They are only improving the impedance match between the already resonant antenna system, and the r-f output amplifier of the tx.
//July 5, 2006 at 6:36 pm #13573polaris34
Guest
Total posts : 45366Hi Rich and Neil,
Thanks for the response. I can see now what you mean – I had overlooked that the loading coil (not shown on the SSTRAN schematic) would be between the capacitors C5/C23 and the capacitive reactance of the electrically-short vertical antenna.
I guess then what I would like to know is, how does one go about designing their own transmitter/output network?
I always thought part of the benefits of DIY-kits were that they helped you learn theory along the way, but with the SSTRAN, it seems the focus was more on helping the builder get the thing working as-is. I am more of a design enthusiast, having worked mostly with audio and am just getting into the RF side of things. I like to be able to ‘roll my own’, by understanding the underlying principles and theory that go into the decision making process of why something was designed a certain way.
Last night, I made a loading coil from stuff I had on hand. It’s on a polystyrene form about 2 5/8 diameter, with #20 wire. It measures only about 150 uH so I know now from your results that I’ll have to increase the inductance. By the way this gives me an idea. Could I make a variable loading coil by having a ferrite core that I could slide into the center of the coil form, or would this introduce some other undesirable effect? I had also thought about making another coil like this one and having the two of them act together as a variometer.
I know most go with #16 or larger but the only other size I had was #14 and I’m saving that for something that needs it. In any case I know that once you resonate the system, all you have left is a large ground loss and relatively small radiation resistance and loss in the coil, etc. I figured improving the ground system will be the way to maximize efficiency in this type of setup.
I confess I don’t totally understand pi-networks, as in how they transform impedance in the case of inductors and capacitors. I understand transformers I suppose because there you have a voltage ratio. Is there a good reference on how LC pi networks are designed to match impedances?
Dave
[quote=Rich]
The tx output circuit needs to be adjustable to deliver its maximum rated power into whatever load it sees, and that is why most Part 15 AM txs provide some means of doing that.
The tx adjustments are not really “tuning” the antenna as in making it resonant. They are only improving the impedance match between the already resonant antenna system, and the r-f output amplifier of the tx.
//[/quote]
July 5, 2006 at 8:58 pm #13576radio8z
Guest
Total posts : 45366Hi Dave,
I have used an application note from Motorola, AN-267, for design of matching networks. Mine is a printed copy and I don’t know if it is available on the web but if not you could contact Motorola. I haven’t done this for many years, but they used to be very good about sending app notes gratis.
The amateur radio community is a great source for information regarding anything pertaining to radio. The ARRL has many specialized publications such as “The Radio Amateur’s Antenna HandbooK” which you might find useful. Check your library since many in my community have these books for loan.
Do a web search on “antenna transmatch” and you will find many articles which might be of use.
RF is a different world from audio as you are discovering. You mentioned SSTRAN not being instructive in the theory of RF, etc. I cannot speak for them, but I will point out that most consumers of electronic gear just want to plug it in and have it work. For those such as yourself who desire to learn theory and practice, you will have to pursue other means as I have mentioned.
Neil
July 7, 2006 at 5:03 pm #13588polaris34
Guest
Total posts : 45366Hi Neil – thanks for the reference. I was unable to locate that ap note from Motorola, but I did find some online caluculators for deriving C and L values for Pi Matching Networks. Oddly enough, they all spit out L values in the 10 to 20 uH range (not anywhere near the 250uH or so for the loading coils as they seem to be), so I am not sure I am doing this correctly.
What might be helpful would be to draw an equivalent circuit for the whole output section, including the output device (“final”), the network, and the antenna/ground system. What does the antenna look like in terms of an equivalent? I know it is capacitive plus some total R loss and radiation resistance, but how do you find these values?
thanks,
Dave
July 7, 2006 at 10:41 pm #13590radio8z
Guest
Total posts : 45366Hi Dave,
Regarding the Pi matching references you found, the inductance and capacitance will have little to do with the inductance of a loading coil so the values may not be anywhere near 250 uH. There are two things going on here. The pi network functions to provide an acceptable impedance to the transmitter and does so by transforming the impedance at the antenna feedpoint to a suitable value. The second is the loading coil cancels the capacitive reactance at the antenna so the feedpoint impedance is resistive. Unfortunately, at part 15 frequencies and antenna lengths this resistance is a very small number which the pi or other network needs to transform up to a value suitable for the transmitter (usually but not always around 50 ohms).
If you look a few posts back you will see that Rich has modeled a short part 15 antenna and the impedance is Z = .106 – j2523 ohms. The primary way we express a resistive and reactive impedance is in this manner which is known as a complex number.
The resistive term is .106 ohms. The reactive, or imaginary, term is -j2523 ohms. The negative sign means this is a capacitive reactance. This is calculated Xc = 1/(2*pi*frequency*capacitance) ( units are Hz. and Farads ). From Rich’s number and solving for C for 1600 kHz., C = 1/(2*pi*f*Xc) = 1/(2*3.14*1600000*2523) = 39.4 picofarads.
The inductance required to resonate (cancel the Xc) at 1600 kHz. comes from XL= 2*pi*f*L. Solving, L = XL/(2*pi*f), so L = 2523/(2*3.14*1600000) = 251 microhenries.
The equivalent circuit for the antenna is Xc in series with the XL in series with the radiation resistance of the antenna (.106 ohms) in series with the loading coil resistance in series with the ground resistance. In this example, the XC is -2523 which added in series with the XL which is +2523 gives 0 leaving only the .106 resistive term at resonance for the antenna in series with the other resistances I mentioned. Reactances such as XC and XL do not dissipate or radiate power and that is why we use loading coils to cancel these terms.
At resonance, the XL and the XC cancel and the model includes only the resistive terms.
As Rich pointed out, usually the biggest unknown is the ground resistance since this is comprised of many factors. If you build a coil to give around 250 uH. it will resonate the antenna at 1600 kHz. used in this example and you should make the coil resistance as small as practical by using big (#16 or larger) wire. Unfortunately there is an effect called the “skin effect” which increases the effective resistance of a conductor as frequency increases.
What this boils down to is in practice you should build a coil using fairly large wire to cancel the antenna capacitive reactance and bury at least 10 10′ ground radials and mount the transmitter between the coil and the radial ground. It is great to pursue the theory beyond this, but this approach, as reported from many posters here, is about the best you can do for a legal antenna, and going further is probably a lot of effort for diminished return.
Hope this helps.
Neil
July 8, 2006 at 12:40 am #13591radio8z
Guest
Total posts : 45366Dave,
I found a link to the Motorola app note I referenced. It is:
http://www.freescale.com/files/rf_if/doc/app_note/AN267.pdf
Neil
July 8, 2006 at 12:19 pm #13594polaris34
Guest
Total posts : 45366Hi Neil,
Thanks again for your response, and the link to the Ap Note.
If I am understanding your verbal description of the antenna equivalent circuit, it is modeled as a string of series lumped elements, in which you have some total resistance (both “good” and “bad”), capacitive reactance (mostly from the antenna, in the case of a short vertical) and some inductive reactance (mainly from the loading coil). The reactances are then made to cancel at the frequency of interest.
Couldn’t this series of elements then be reduced to one effective series resonant RLC circuit? If so, then I guess I am missing where the idea of the Pi Network comes in. The Series RLC made up by the antenna itself is resonant at the transmitting frequency and is connected then to the final transistor (Q5 in this case) by a large (.1uF) coupling cap, and is in parallel with fixed C23 and variable C5. So I am trying to see where the Pi network is considered to be in this instance, since there is no “L” between the antenna circuit and the variable C that sits on the “transistor side” of the circuit? (I hope this question makes sense – its hard to explain without a diagram.) In other words, it looks to me like you actually have a series tuned RLC circuit (the antenna itself), and then you have a variable capacitance in parallel with that.
Sorry for all the questions – I’m just trying to get a handle on all of this! 🙂
Thanks again,
Dave
[quote=radio8z]Hi Dave,
Regarding the Pi matching references you found, the inductance and capacitance will have little to do with the inductance of a loading coil so the values may not be anywhere near 250 uH. There are two things going on here. The pi network functions to provide an acceptable impedance to the transmitter and does so by transforming the impedance at the antenna feedpoint to a suitable value. The second is the loading coil cancels the capacitive reactance at the antenna so the feedpoint impedance is resistive. Unfortunately, at part 15 frequencies and antenna lengths this resistance is a very small number which the pi or other network needs to transform up to a value suitable for the transmitter (usually but not always around 50 ohms).
If you look a few posts back you will see that Rich has modeled a short part 15 antenna and the impedance is Z = .106 – j2523 ohms. The primary way we express a resistive and reactive impedance is in this manner which is known as a complex number.
The resistive term is .106 ohms. The reactive, or imaginary, term is -j2523 ohms. The negative sign means this is a capacitive reactance. This is calculated Xc = 1/(2*pi*frequency*capacitance) ( units are Hz. and Farads ). From Rich’s number and solving for C for 1600 kHz., C = 1/(2*pi*f*Xc) = 1/(2*3.14*1600000*2523) = 39.4 picofarads.
The inductance required to resonate (cancel the Xc) at 1600 kHz. comes from XL= 2*pi*f*L. Solving, L = XL/(2*pi*f), so L = 2523/(2*3.14*1600000) = 251 microhenries.
The equivalent circuit for the antenna is Xc in series with the XL in series with the radiation resistance of the antenna (.106 ohms) in series with the loading coil resistance in series with the ground resistance. In this example, the XC is -2523 which added in series with the XL which is +2523 gives 0 leaving only the .106 resistive term at resonance for the antenna in series with the other resistances I mentioned. Reactances such as XC and XL do not dissipate or radiate power and that is why we use loading coils to cancel these terms.
At resonance, the XL and the XC cancel and the model includes only the resistive terms.
As Rich pointed out, usually the biggest unknown is the ground resistance since this is comprised of many factors. If you build a coil to give around 250 uH. it will resonate the antenna at 1600 kHz. used in this example and you should make the coil resistance as small as practical by using big (#16 or larger) wire. Unfortunately there is an effect called the “skin effect” which increases the effective resistance of a conductor as frequency increases.
What this boils down to is in practice you should build a coil using fairly large wire to cancel the antenna capacitive reactance and bury at least 10 10′ ground radials and mount the transmitter between the coil and the radial ground. It is great to pursue the theory beyond this, but this approach, as reported from many posters here, is about the best you can do for a legal antenna, and going further is probably a lot of effort for diminished return.
Hope this helps.
Neil
[/quote]July 8, 2006 at 9:05 pm #13596radio8z
Guest
Total posts : 45366Hi Dave,
I agree with your description of the antenna system. I have not analysed the SSTRAN circuit but since the recommendation is to tune the antenna for maximum transmitter output voltage by finding the inductance (selecting the proper tap) and by adjusting the length of the radiator while connected to the transmitter, the adjusted antenna includes any reactance at the output of the transmitter. Perhaps C5 is then used to “tweak” the resonant frequency over a very narrow range for fine tuning, though I still suspect that there is an impedance transformation here.
If you consider only the antenna (including the coil) model at resonance then the only load the transmitter sees is the resistance (radiation, coil, and ground). As the coil tap or antenna length are changed, the reactance added vectorially to the resistance raises the magnitude of the antenna feedpoint impedance. So, at antenna resonance, the impedance is minimum as seen by the transmitter and this should result in a minimum transmitter output voltage if the transmitter output impedance is resistive (we know it is not because of C5/C23). The opposite is what happens…there is a peak in the output voltage as seen at the test point.
Now consider that the antenna and coil are not really resonant at the adjustment peak. If true, then the antenna will appear with reactance in the model and C5/C23 with this reactance will form either a L network with the coil inductance or a tapped C network with the antenna capacitance which might act as an impedance transformer and C5 is really adjusting the impedance match and not the resonant frequency. This is only speculation since I have not written the equations for this type of setup, but if I get inspired I will look into doing so. As you can see in the app note, circuits with multiple reactances can get rather complex to analyse.
What do you think?
Neil
July 8, 2006 at 11:57 pm #13597jbprptco
Guest
Total posts : 45366Neil, David, Rich, One of you should E Mail Phil Boylin the designer and owner of SSTRAN for his explanation of his external antenna circuit. He is deeply knowledgeable about his design and did explain to me once how the external antenna relates to the transmitter. I remember he did tell me that the antenna (coil and radiator) acted as a low pass harmonic filter, hence no pi network at transmitter output. He used to post here on the Forum yet seems to have stopped for quite a while now. He’s been very helpful to me on a number of occasions about his transmitter design and how it works. I don’t own a rangemaster, yet of the 100 mw transmitters that I do own the SSTRAN, when set up and tuned to its indoor antenna, beats the others by far. Since Part15 AM isn’t limited by audio frequency response the SSTRAN is capable of 20 to 20,000 cps and is the best thing I’ve heard on AM when fed by a high quality audio source. Jim B
July 9, 2006 at 2:05 am #13598radio8z
Guest
Total posts : 45366Hi Jim,
Thanks for the suggestion. I am going to leave it to Phil to comment here if he wishes. Perhaps I am going off the deep end and it is much simpler than I am making it appear.
I did not analyse the circuit because my SSTRAN works fine for my yardcasting and there was no need for me to optimize the antenna, but it is fun to kick ideas around.
Neil
July 9, 2006 at 4:36 am #13599radio8z
Guest
Total posts : 45366Dave,
I worked the analysis on the output network we have been discussing. Those not familiar with complex circuit analysis may not follow the math, but the conclusions should be of interest.
The model circuit is a current source (the output transistor) with C1 [corresponds to C5/C23] from the output to ground. From the junction of the current source and C1, a series circuit of L, R, and C2 is connected with the free end of C2 going to ground.
Using X’s and R and complex algebra, the impedance at the junction of the current source and C1 to ground (corresponds to the impedance seen by the output transistor in the transmitter) is:
Z = ( XC1(XL-XC2) – jR)/( R + j(XL – XC1 – XC2) )
When the antenna and loading coil is resonant the Z approximately equals R.
However, off resonance, things get interesting. The impedance becomes complex and is transformed upward from the value of R. This can be useful if the optimum impedance expected by the transmitter is high compared to 30 ohms.
Try these numbers:
R = 30, XL = XC2 = 2513, XC1 = 398. The Z is close to 30 ohms resistive. Now change XL to 2613 and you will see the impedance changes quite a bit. Changing XC1 also changes the impedance, but to a lesser extent.
From this it appears that the network is performing an impedance transformation between the antenna and the output transistor and C5/C23 and the loading coil work together in this regard. The adjustment of C5 and the coil and antenna set these values to maximize the voltage at the transmitter output. Maximizing the voltage out maximizes the current through R which maximizes the radiated power. The sensitivity of the impedance to a change in XC2 explains the observation by several posters that the tuning changes if you approach the antenna since this changes XC2.
Neil
July 9, 2006 at 10:03 pm #13600polaris34
Guest
Total posts : 45366Hi everyone,
I think I might just understand what is going on with this circuit.
I modeled the output stage of the SSTRAN using a free circuit modeling software called B2 Spice. In my model, I used the following values: 39.4pF, 250uH, and 40 ohms. I found that with the parallel capacitance (C23 and C5) left out of the circuit, the antenna is series resonant around 1.6 MHz, but since it is driven by a current source, the voltage across that part of the circuit is at a minimum at that frequency, and the output device is loaded down to whatever the series resistance is. With a better ground system and lower losses, the output stage would be even more loaded down.
I then added the C23/C5 back in. I used a median value of 610pF for the C23/C5 combination. With the caps added back in, as the frequency is swept upward, the impedance first drops, and then shoots up right after that.
From what I can understand, this indicates that the network is both series and parallel resonant. At the higher frequency of resonance, the combined capacitance of C23 and C5 is essentially in series with the antenna capacitance, but then this combined (lower) capacitance acts in parallel with the loading coil to make a parallel resonanct circuit that is resonant at a slightly higher frequency. At this higher resonance, the current through the antenna branch is at a peak.
This would mean two things – the output transistor is loaded by a higher impedance (around 56 ohms), and the current through the antenna is peaked, thus the current through the radiation resistance is peaked and the antenna output should peaked as well. Adjusting any of the reactances (loading coil, antenna, C23/C5) would affect where the peak and dip are, as well as the amplitude of each.
Since I can’t adjust the length of my antenna, I made a split loading coil so that I can adjust the inductance of it by adjusting the spacing of the two halves with respect to one another. As they are wound now, I can get from about 230 to 280 uH which should allow me to tune the antenna in the frequency range of interest (1610 kHz or so).
What do y’all think?
Dave
[quote=radio8z]Dave,
I worked the analysis on the output network we have been discussing. Those not familiar with complex circuit analysis may not follow the math, but the conclusions should be of interest.
The model circuit is a current source (the output transistor) with C1 [corresponds to C5/C23] from the output to ground. From the junction of the current source and C1, a series circuit of L, R, and C2 is connected with the free end of C2 going to ground.
Using X’s and R and complex algebra, the impedance at the junction of the current source and C1 to ground (corresponds to the impedance seen by the output transistor in the transmitter) is:
Z = ( XC1(XL-XC2) – jR)/( R + j(XL – XC1 – XC2) )
When the antenna and loading coil is resonant the Z approximately equals R.
However, off resonance, things get interesting. The impedance becomes complex and is transformed upward from the value of R. This can be useful if the optimum impedance expected by the transmitter is high compared to 30 ohms.
Try these numbers:
R = 30, XL = XC2 = 2513, XC1 = 398. The Z is close to 30 ohms resistive. Now change XL to 2613 and you will see the impedance changes quite a bit. Changing XC1 also changes the impedance, but to a lesser extent.
From this it appears that the network is performing an impedance transformation between the antenna and the output transistor and C5/C23 and the loading coil work together in this regard. The adjustment of C5 and the coil and antenna set these values to maximize the voltage at the transmitter output. Maximizing the voltage out maximizes the current through R which maximizes the radiated power. The sensitivity of the impedance to a change in XC2 explains the observation by several posters that the tuning changes if you approach the antenna since this changes XC2.
Neil
[/quote]
July 10, 2006 at 4:28 am #13603radio8z
Guest
Total posts : 45366Dave,
Nice report on your model. I also modeled it using Electronics Workbench (a GUI for Pspice), but I did it a bit differently. I ran a Bode plot which shows the gain of the network vs. frequency and only saw one peak near 1600 kHz. You reported the impedance which may behave differently from the gain, but we both agree that the voltage will be maximized at the “tuned” frequency.
The equation I derived shows the effect you saw with the possibility of the numerator going to zero and the denominator going to zero at different combinations of the elements in the network. This “dual resonance” shows up in crystals and duplexers among other radio circuits.
Your split coil idea should work since you will have control over XL and C5/C23 for adjustment.
Keep up the good work and let us know how it works in practice.
Neil
- AuthorPosts
- You must be logged in to reply to this topic.