Michael,
I hope you find a nice one. Just watch out for shipping and handling.
The scope I now use is an old Hitachi analog scope, pretty typical of what you will find at auctions in the $20 range. The older scopes tend to have dirty switches and pots which make them unstable. A good squirting with contact cleaner fixes this.
Here's a link to nice, simple test tone generator software.
http://www.world-voices.com/software/nchtone.html
You will need this or something similar to produce test tone waveforms to observe the tx. output on a scope. A nice sine wave makes it easy to view the modulation envelope.
Just for fun, I created an audio file with alternating 0dB and -10dB tones. It is interesting to watch the SSTRAN internal compressor respond to this on the scope.
Neil
Ermi,
First, the AMT3000 is not a "serious" transmitter for long range broadcasting UNLESS it is used with the external base-loaded antenna. Using the internal tuning network and wire antenna will yield a good signal inside a house, but won't go beyond about 200 ft., if you are lucky. So, it is only fair to compare efficiency and range when you are talking about the AMT3000 wired to use the external base-loaded antenna.
Although the Wenzel circuit and AMT3000 circuit are based on the same concept, Wenzel did not invent the circuit, nor did I. Back in the 70's there were "Operational Transconductance Amplifier" (OTA) ICs that were capable of producing very linear AM modulation, but were not particularly suitable for even 100mW RF circuits without a lot of extra components. The innards of these things were similar to the Wenzel and AMT3000 circuit, but much more sophisticated. They used "current mirrors" which are hard to duplicate in discrete circuits. The Wenzel and AMT3000 circuits are similar but are different ways of implementing an OTA using discrete components. The concept works very well.
When the AMT3000 is set up for the external base loaded antenna, Resistor R18 is removed. The AMT3000 circuit matches the low impedance of the series connected loading coil and antenna capacitance by means of an L-net circuit. The Wenzel 1 circuit was puzzling in this respect. The Wenzel 2 circuit corrects Wenzel 1 problems by using the pot core L to series resonate with the antenna. A low-end tap on the coil matches the collector impedance instead of the L-net used in the AMT3000. The AMT3000 advantage is that the small pot core inductor is replaced by a much more efficient 3.5" loading coil.
The Wenzel 2 design was published well after the AMT3000 was put on the market. The AMT3000 was not "based" on the Wenzel 1 design. Rather, the two designs were independent and both based on older technology.
Here are relevant links:
Wenzel 1 and 2:
www.techlib.com/electronics/amxmit.htm
A Carnegie Mellon University EE lab project in Postscript. I have a pdf version if anyone wants it. Just email me at "info at sstran dot com."
www.ece.cmu.edu/~ee321/spring00/lab5.ps
Operational Transconductance Amplifier information:
www.datasheetcatalog.com/datasheets_pdf/C/A/3/0/CA3080.shtml
www.intersil.com/data/an/an6668.pdf
My previous Part15.us post describing the AMT3000 RF output circuit:
www.part15.us/node/883#comment-4258
Just wanted to try to clear things up a bit!
PhilB
Thank you, PhilB, for pointing out to me that the SSTRAN does, indeed, have an impedance transformation circuit for coupling the transmitter to the antenna. As I stated in a previous post, I am not familiar with the SSTRAN circuit, and I was shown the Wenzel 1 circuit, which, I was told, is similar to the SSTRAN.
I am, however, confused about your comments about the priority of the circuits. You said that both of the Wenzels, and the SSTRAN, are based upon an earlier design, namely the OTA. The OTA is a general-purpose device that can be used for many purposes. It is a form of the transconductance multiplier, which is a circuit that was very well-known in the sixties, and probably goes back to the fifties. You seem to be saying that, because Wenzel used the transconductance multiplier for modulation, his design is not original. Unless someone else came up with a radio transmitter circuit similar to Wenzel's I would call his design original. I don't know if his circuit would meet the criteria of patentability, but if he was the first one to come up with a similar circuit, his contribution was at least novel.
You said that the Wenzel 2 design was published after the SSTRAN was introduced to the market. This would explain why the SSTRAN was not based on Wenzel 2. You did not say if the Wenzel 1 design was published before you came up with your circuit. If it was, your design was very likely based on Wenzel 1. Fortunately, you improved Wenzel's first circuit by adding the L network you mentioned, for impedance transformation.
Incidentally, you are the perfect person to answer Neil's question about why he is experiencing distortion when he removes R18.
Ermi,
I hope I didn't mislead you or anyone else with my comment that the SSTRAN circuit is "very similar" to the Wenzel circuit. Since I respect SSTRAN's copyright on their circuit schematic I could not provide a drawing of the actual circuit and I thought this would give you something to study. There are differences which may be significant, and I apologize if I misrepresented this.
If you obtain the data sheet for the Motorola MC1496 IC you will recognize similarities to the Wenzel circuit. I don't know the history of this circuit, and unless there is a patent dispute, it doesn't matter matter to me who "invented" it. What Phil has done is reduce this to practice so many of us can get on the air.
Thank you for your concern and your comment that Phil can answer my question about R18. I conclude that my particular problem is a result of the apparently inappropriate load that my antenna presents to the tx. which is probably quite a bit outside the operating tolerance of this circuit and that my problem is just a result of my "not following directions". My antenna was designed for another application (ham stuff) and since it was already installed I tried to use it with the SSTRAN.
I am in the process of redoing my antenna more in line with the SSTRAN recommendations and I am sure that I can report good results later.
Neil
Just wanted to correct a link I posted previously. I found a pdf version on the web site.
A Carnegie Mellon University EE lab project in Postscript. I have a pdf version if anyone wants it. Just email me at "info at sstran dot com."
http://www.ece.cmu.edu/~ee321/spring00/lab5.pdf
PhilB
Hi Ermi,
I don't think Mr Wenzel invented the 3 transistor balanced modulator circuit. There must have been many versions published over the years. I personally arrived at the circuit by studying the datasheet for the Intersil CA3080 OTA:
http://www.intersil.com/data/an/an6668.pdf
Look at page 2 for the schematic of the OTA and pages 6-8 for AM modulator application. I discarded the idea of using this IC because it would require a linear amp to boost the signal. By implementing a balanced modulator in discrete transistors, you can run the balanced modulator at the 100mW level, so overall component count and cost is reduced.
I did not invent this circuit, nor did Mr. Wenzel. In my case, the value added was thorough spice simulation and recursive changes to refine the circuit.
Who was it that said something like "everything of value has already been invented"? The circuit in the CMU EE lab exercise that I found after "inventing" my version is actually a lot closer to my implementation than either of the Wenzel circuits. This was just a natural convergence to the "best" design by different people. I have no idea where the CMU circuit came from. The fact that it was part of a lab exercise leads me to believe the professor picked it from published information. Here is the link to the lab exercise again:
http://www.ece.cmu.edu/~ee321/spring00/lab5.pdf
Look at the schematic on page 2. Look familiar?
PhilB
I don't think that Wenzel invented the three transistor balanced modulator, either. As I stated in my previous post, the transconductance multipler probably goes back to the 1950s. It was in 1968 that the great Barrie Gilbert perfected the transconductance multiplier by using six matced transistors to eliminate temperature drift.
Strictly speaking, Wenzel's circuit is not a balanced modulator, since there is no carrier suppression. It is a full-carrier AM transmitter and modulator. One transistor is a buffer for the oscillator, which drives the emitter of another transistor, which is the active component of a grounded-base transmitter stage. The third transistor varies the current through the first two transistors to vary their transconductances for modulation.
It was not clear to me why you mentioned the OTA in a previous post. As you said in your last post, however, the OTA was your inspiration for independently developing a circuit similar to Wenzel's.
I never would have guessed this thread would cover so much ground,I LOVE IT!!! Great discussions and resources,a super sharing effort for which you all deserve my gratitude! Kudos,gentlemen. I am impressed!
On another topic,my web site sales have been nil?!
I have had better luck at the hamfests lately,but my finances are still strapped. I was hoping for a better response,alas,I was wrong. When the eagle flies again and I have disposable income,I will do as I said and donate to the great effort put forth here. My failing health is at last now starting to improve so I may return to the working class once again. Its been a long,rough road.....onward and upward!!!
Regards,Lee
http://www.freewebs.com/wilcomlabs/index.htm
The Wenzel Part 15 AM transmitter circuit has been discussed extensively in this thread, and it seems to be a very popular circuit for homebrewing. Links to the Wenzel design are on a couple of the previous posts in this thread. I decided to construct the Wenzel as closely as I could to the original design, while using only parts I had on hand, and then report my observations to the Forum.
My overall impression of the Wenzel is that the circuit is very stable, and there is no problem in getting it to work. The modulation waveform looks very nice (no distortion). However, the design is a lot less efficient than I expected it to be. I expected the efficiency to be similar to what Neil reported for the SSTRAN, around 30%, but I got only 14%, using a 30 pf capacitor in series with a 46 ohm resistor as a dummy antenna at 1.5 MHz. The efficiency became lower as the resistor value in the dummy antenna was reduced. 30 pf is a typical capacitance for a 3 meter rod antenna above ground. 46 ohms is a typical ground resistance. The radiation resistance is orders of magnitude less than the ground resistance, so the efficiency of an actual antenna would be very low.
I built the second Wenzel circuit in the Wenzel link (Wenzel 2), but I at first used the output circuit for the first Wenzel circuit (Wenzel 1), just to see how well the circuit worked. I connected a few feet of wire to the output, not nearly 3 meters long, and I was able to hear the signal anywhere in my house. Of course, the radiated power had to be very low. Then, I increased the output power by connecting an autotransformer and tuning capacitor to the output, according to Wenzel's design. I did not have an 1811 potcore, as specified by Wenzel, so I used an Indiana General F1707-1-Q2 3.5 " OD ferrite toroid with u = 40 and AL = 57. I don't think this toroid is available any more, but other similar toroids are made by other manufacturers. I used a 14/1 turns ratio, as is used in the potcore design. The secondary winding of the transformer is the loading coil, so no external loading coil is used.
The efficiency would have been even lower if the potcore specified for the design had been used. Potcore ferrite, is, in general lossy at RF frequencies. The low-permeability toroid I used has fairly low loss. I measured the Q of the secondary winding to be 265 at 1.5 MHz. Wenzel did not give enough technical details about his recommended potcore. The number, "1811" refers only to the potcocore dimensions, but not the potcore material. The highest frequency material available for the 1811 size is Ferroxcube 3F3, which is specified to 700 kHz max. This material will not give a high Q at the upper end of the AM broadcast band, and the efficiency with the potcore has to be even less than the low efficiency that I obtained. One advantage of the high loss potcore design is that harmonics are suppressed. In my test circuit, I observed what appeared to be a parasitic oscillation. This was surprising, because, a common-base amplifier is supposed to be relatively free of parasitics. This "oscillation" stopped when I removed the 1.5 MHz drive to the final amplifier. After further examination, I found that I did not have a true oscillation, but an enhanced 10th harmonic at exactly 15 MHz. If the lossy potcore specified by Wenzel were used, there would have been no harmonic at 15 MHz.
The principal cause of the low efficiency I observed in my Wenzel circuit is a mismatch between the final stage of the transmitter and the antenna resistance. The optimum load resistance of the final stage is about 1 k ohm. The function of the transformer is to match 270 k ohms, which is the equivalent parallel resistance of the 46 ohm resistor in series with the 30 pF capacitor at 1.5 MHz, to 1 k ohm. The input resistance of the transformer was about a third of what was required, and this was the cause of most of the inefficiency. There is really no way to adjust the impedance matching of the transformer, except by rewinding the transformer to a different turns ratio.
The final stage of the Wenzel circut has a 360 degree conduction angle, so it is definitely not a class C amlifier. It is class A. The voltage waveform at the output is sinusoidal, but the current waveform is roughly a square wave. The maximum theoretical efficiency is 63.7 %.
Since the Wenzel has a common base output amplifier, it receives a large amount of drive power. A small amount of this drive power goes to the output. The input drive power to my test circuit was about 16 mW. According to Part 15.219(a), this drive power should be included in the 100 mW input power allowed. Wenzel included only the DC input to the final stage, not the drive power.
Ermi,
This is a really great report on the Wenzel circuit and it is good to see some bench data. I have read your post once so some I may have missed some of the finer points, but here are some thoughts.
You mentioned that I had reported the efficiency of the SSTRAN to be about 30%. Maybe my post wasn't clear on this but the 31% I reported was for the Ramsey AM-25 with a resistive load of 24 ohms. I measured Pin and Pout for a selection of resistive loads and the maximum efficiency was what I reported. The Ramsey unit uses a class C FET amplifier for the final and I would have expected the efficiency to be higher, but they use a 7 pole LP filter with toroids wound with small gauge wire which might ding the efficiency.
I have attempted to measure the SSTRAN efficiency but have not been able to do so with the confidence I require before reporting the results. This is because I do not know what load would be appropriate. A simple resistor would not work because the output network of the SSTRAN is a L matching network which includes inductance in series with the load resistance to properly transform the impedance of the load back to the collector of the output transistor. PhilB wrote a nice article, which I think was lost in the board "restore", describing this. He stated that the circuit load should be about 800 ohms R at the collector of the final transistor but that would be after the Z transformation from the antenna and coil circuit. I would hope that Phil would repost about this since I am going on memory only here.
You are probably correct about the Wenzel circuit is operating class A and if so the theoretical maximum efficiency would be 50% and that would be under ideal conditions depending on the Q point and load. Your reported 14% efficiency was, I assume, for the Wenzel circuit and not for the SSTRAN.
I have observed by bench testing and by simulation that the SSTRAN waveform can vary quite a bit in quality from textbook perfect to severely distorted as the load is changed. Since the waveform is excellent with my antenna system loading the SSTRAN I did not pursue this.
My experience with RF amps has been with class C and the highest efficiency I have measured is around 70%....far from the theoretical maximum of 100%. Reality and theory depart quite a bit when applied to RF don't they.
Neil
Neil,
My testing was only of the Wenzel, with a modified output transformer. I used a modified output transformer only because I did not have the components for duplicating the original design exactly. I wanted to duplicate the original design as closely as possible. My intent was to evaluate the design, and not to try to improve it.
Sorry that I misunderstood which product you performed your efficiency test on. It would really be nice to have efficiency data on the the most popular transmitters, SSTRAN and Rangemaster. You said on a thread near the end of last year that maybe Part 15 enthusiasts should concentrate on improving transmitter efficency, where some real gains could be made, and not only on antenna efficiency. To know what sort of gains could be made, we need to have data about the transmitters that are used the most.
I am not familiar with the SSTRAN, so I don't know how to design a method for testing its efficiency. From what I have learned on the Forum, there is an impedance matching L network, and also an external loading coil. If the L network is of the capacitive input type, where the output resistance is lower than the input resistance, and the external loading coil tunes out the series-equivalent capacitance of the antenna, supplying a resistor at the output to represent the sum of the loading coil resistance and the ground resistance should be adequate. However, I have no idea if the SSTRAN circuit actually works that way.
Getting data on the Rangemaster has the additional difficulty that it costs a lot, which makes it less likely that the transmitter would be used as a test bed.
Although the Wenzel has a Class A amplifier, since the conduction angle is 360 degrees, it is not a linear amlifier, because the current approximates a square wave while the voltage is a sine wave. This causes the theoretical maximum efficiency to be higher than 50%. If both the current and voltage waveforms were sine waves, the theoretical efficiency would be 50%. Interestingly enough, if the output circuit had enough bandwidth so that both the current ant voltage waveforms were square waves, the maximum theoretical efficiency would be 100%. Actually, in this case, the theoretical efficiency could be slightly higher than 100% because of feedthrough from the driving circuit of the grounded base amplifier.
Ermi,
This causes the theoretical maximum efficiency to be higher than 50%.
Thanks for explaining this. The 50% number assumes a centered Q point and sine waves with a resistive load. I thought perhaps you were thinking of a class B, but this helps.
I believe you have the correct image of the SSTRAN output network. The collector current is supplied though a choke coil. As modified for a base coil antenna, there is a capacitor from collector to ground. From what I understand from Phil's descripton, the system is tuned so that the antenna circuit presents an R + jX load with the jX of the coil providing the inductance for the impedance transforming ell network. This implies that the base coil inductance and the antenna capacitance do not operate exactly at resonance, rather off resonance just enough to provide the proper inductance in the network. This is my understanding and if not correct I hope not to mislead anyone.
I tried my efficiency tests on the SSTRAN before I understood this. Maybe it is time to try it again. I'll dust off the old roller-inductor. I also had good results (compared to actual measurements) with an Electronic Workbench (R) simulation. I'll try that first.
Neil
Now I understand how the SSTRAN output circuit works. For testing efficiency, what is needed is a resistance, R, and an inductive reactance, X, in series to serve as the transmitter load. The resistance and inductance have to be related roughly as R(Rin) = X^2. For the Wenzel, I found that Rin is about 1 k ohm. I think that PhilB said it is about 800 ohms for the SSTRAN. R would represent the sum of the ground resistance and the loading coil resistance. X represents the excess loading coil inductive reactance needed to get the desired impedance transformation. The tuning capacitor would be adjusted for peak output power.
Neil, if you are able to provide (approximately) the needed RL series circuit to serve as a test load , you will be able to measure the efficiency of the SSTRAN. An actual measurement would be a lot better than a simulation.
I rewired the output of my Wenzel circuit to incorporate the SSTRAN test circuit I proposed in my last post. I did not mention in my post that the test circuit requires a DC blocking capacitor. I used a .01 uF ceramic. The SSTRAN connected to an actual antenna does not require such a capacitor because the antenna is a capacitor.
I first used a 46 ohm load resistor in series with a 22 uH inductor.
As was stated in my previous post, the 22 uH inductor represents antenna loading coil inductance in excess of what is required to tune the antenna to resonance. This excess inductance forms the inductor of a capacitor-input L network for transforming the load resistance (46 ohms, in this case) to the optimum load resistance at the collector of the final transmitter stage. My efficiency measurement was only 27%. This is nearly twice as good as the efficiency I obtained with my version of the original Wenzel circuit, but this is considerably less than 63.7%, which is the theoretical maximum efficiency.
(Neil pointed out that, since the Wenzel is a class A amplifier, its theoretical maximum efficiency should be only 50%. The Wenzel final stage, however, is overdriven, causing the current waveform to be a square wave, and this causes the theoretical maximum efficiency to increase.)
The amplitude of the current waveform is constant with loading (nominally 20 mA p-p.) This is because the final stage of the Wenzel transmitter is approximately a current source. However, the amplitude of the voltage waveform varies directly with the load resistance seen by the collector. When I obtained only 27% efficency, the amplitude of the voltage waveform at the collector was only about 60% of the optimum value, which is nominally 20 V p-p, since the nominal DC voltage across the output stage is 10 VDC. I found that increasing the test inductance increased efficiency. The best efficiency I was able to get was 41%, which I obtained by using a 30 uH inductor. I was not able to get the full 20 V p-p at the collector output because clipping of the collector waveform started at around 18 V p-p. If the full 20 V p-p range were available, the efficiency would have been higher.
In my test of the original version of the Wenzel, I observed something that looked like parasitic oscillation, but was actually an enhancement of the 10th harmonic of the transmitter at exactly 15 MHz. In this circuit, I observed something similar: an enhancement of the 34th harmonic at exactly 51 MHz. If such enhanced harmonics occur, they have to be suppressed, because the short Part 15 AM antenna probably works a lot better at the harmonic frequency than at the fundamental.
My efficiency calculations are based only on the DC input, and do not include the appreciable drive power to the emitter of the common-base final stage. In a common-grid tube amplifier, about 10% of the drive power feeds through to the output. In the Wenzel, it is much less than that. This is because transistors have a lot more transconductance than tubes. Nevertheless, please be aware that the way Section 15.219(a) reads, drive power is included in the 100 mW input power that is allowed. The rule excludes filament power, but nothing else. There has been a lot of discussion about Section 15.219(b) lately, but Section 15.219(a) has to be considered, also.
This thread was originally initiated by Wilcom Labs, because he was interested in developing a tube circuit for Part 15 AM. Since his original post, he has said that he has dropped this project, but then he said that it may be on again.
In my first post in this thread, I said that tubes should be used only if there is a compelling technical reason for using them. The only such technical reason that I can think of is increased transmitter efficiency. I am not at all sure, but maybe tubes will more readily provide higher efficiency than solid state devices. To find out, it is necessary to test some tube designs.
Wenzel's article briefly states that tubes might be used to improve Part 15 AM transmitter designs, but he does not elaborate. He mentioned the interesting fact that 100 mW applied to a 3 meter antenna develops a couple of hundred volts rms across the antenna. (Short antennas are, indeed, "hot.") Wenzel suggests that, since tubes are high-voltage devices, it might be easier to supply the needed high voltage if they were used. He did not, however, propose any tube circuit that can provide high voltage to the antenna.
The tests I reported in my last post indicate that the SSTRAN can have more than 40% efficiency. Using the SSTRAN as the baseline, about 3 more dBs are available if efficiency is improved. This is not a big improvement, but it is worth pursuing. There is no efficiency data available on the Rangemaster, which has claimed in its advertising that its stated 1-2 mile range is due to its superior circuit design, which supplies more power to the antenna than other designs. If the Rangemaster, or any other transmitter that is currently available, has 70% or more efficiency, further efficiency improvements will make very little difference, and there is no need for a tube design to improve efficiency.
Wilcom Labs: Please pursue your tube design, if you can, and report your results. We have learned a lot about short antennas, but there is a lack of publicly-available data about transmitter design. This is why I studied the Wenzel circuit and reported on it. If efficiency is your goal, I suggest:
1. Use a triode, not a pentode. Don't even think about screen modulation. Don't use a tuned circuit at the input of the final stage, or you might make a TPTG oscillator.
2. Use plate modulation.
3. Make the conduction angle really small. Typically 140 degrees is used, with 90 degrees for frequency multipliers. Get the conduction angle below 90 degrees, if you can.
4. You only need the tube for the final stage. The drive circuit can be solid state. You may be able to make a better drive circuit with transistors than with more tubes.