I think overtone crystals are used above a particular frequency (maybe 30 MHz...not sure) because the crystal cannot vibrate that fast, and fundamental types below that frequency.
The best thing to do is to ask someone who knows which crystal is suitable for the application such as someone at ICM.
Neil
When I put my Ramsey AM-1 onto crystal
control, I ordered crystals for 1590,
1670, and 1700. They were very expensive,
but worked OK. Each crystal was about 35
dollars. I did not order them all at the
same time, though.
I do not remember what specs I gave. The
crystals were on channel, though, or very
very close. I did not have any trouble with
this. No hetrodynes with other stations on
the same channel at night.
This was more then 5 years ago. I'll have
to look at my notes to see what the company
was. I can't believe it, but all I can
remember is that the company was in Florida.
Actually, I'm not completely sure of that.
either.
For me, It was a lot easier to order a crystal
for the AM BCB than to get a higher frequency
crystal, and build a circuit to divide it down
to the frequency. There was also another company that
made crystals for lower than 1000 kHz, which the
first mentioned company didn't do.
I talked to the other company about getting a
crystal for my CC transmitter on the low end
of the AM BCB. They would make it to spec
for about 75 dollars. I was all ready to go
with this. Then I found out the LPB RC-6A
wasn't working right, and I stopped the
project. It turned out about 5 years would
go by until the LPB RC-6A would be up and
running, which it is now. As you guys know,
I am working on a CC system.
I cannot find the company that makes the
crystals for below 1000 kHz. Maybe they no
longer exist.
I will try to find my notes to see if I can
give you guys any useful info.
Bruce, DRS2
Before the advent of PLL frequency synthesis special order crystals cost about $15 (mid '70s dollars). With the CB craze came the need to reduce price and complexity and ICs and circuits for frequency synthesis were developed where a single crystal can be use as is done in the Ramsey AM-25 and the SSTRAN AMT3000 (though Phil cleverly used off the shelf digital ICs for his design). The cost of individual crystals then increased due to decreased demand.
Today, there is another technology available called Direct Digital Synthesis (DDS). There is a hobby company which sells kits which include a LCD display for about $75. So, for about the cost of two special order crystals this circuit can provide virtually unlimited frequency adjustment with crystal accuracy. I don't have a link handy but a net search should be productive.
Neil
The search resulted in several sits, mostly ham radio guys and their own designs. One apparently is so popular it is completely sold out.
I fully agree with Neil, in these modern days of digital synthesis why mess around with crystals when it may be possible you cannot get the crystal you need, or would have to get some odd ball crystal and count it down or bump up with extra circuitry.
I don't see too many daily commuters still riding horseback these days!
RFB
Neil,
Thank you for your schematic of what you identify as a class E amplifier. Back in the early days of class E, in the 1970s, bipolar transistors were sometimes used instead of the MOSFETs of today. Vce(sat) was used in analysis in place of Rds(on) x Id with MOSFETs. Where BJTs are really different from MOSFETs is that they don't have the parasitic drain-to-source diode that is forward-biased during undershoot at the drain. The undershoot does not occur during true class E operation, but it can occur when the class E amplifier is mistuned. Thus, mistuned "class E" amplifiers can work differntly depending on if a BJT or a MOSFET is used. Strictly speaking, a mistuned "class E" amplifier is not really a class E amplifier at all because the two operating conditions specified by A. and N. Sokal are what define class E operation, and they have to be both met for the amplifier to be class E.
Neil: How do you plan to tune your circuit in actual operation with a loaded monopole? You mentioned using 30 and 50 ohm dummy load resistors for testing; but you will not actually know what the real part of your total antenna system load impedance (from loading coil and ground plane loss) will be. The value of your collector-to-emitter capacitance will have to change depending on your load resistance, in order to get correct class E operation.
There are hams that have built
class E AM transmitters. I think
I saw a schematic for one that
ran 500 watts. I'll have to look
for that.
About this DDS circuit, for generating
RF "anywhere you want" - it sure is
tempting.
I just don't know if I have the eyesight
to build one of those - but it's academic
now, because I'm short on funds.
My CC transmitter really belongs on 620 kHz.
Maybe someday.
This is a really good thread.
Bruce, DRS2
www.classeradio.com same guy who designed the REA mod monitor that used to be advertised here.
Referring to the link you just
mentioned.
Wow!
Bruce, DRS2
Ermi,
I appreciate your comments on BJT vs MOSFET. My impression is that modern Class E designs, according to some reading on the subject, are implemented with FET technology and I wonder if this is due to the fabrication techniques of LSI chips where FETS are easily incorporated. Just a hunch.
Your comments about tuning are worthy of consideration and from my response you will probably see why I was reluctant to disclose this design. I didn't want others to build it and be disappointed in case their antenna system parameters were different than mine.
I did, in fact, know the exact real part of my antenna system Z which is 29 ohms. This was measured by using a current transformer on the feed to the loading coil for current and a pickup at this point for voltage. The transformer was carefully calibrated on the bench by means of dummy load resistors and checked with a Bird Wattmeter. The coil Q was measured by Ben Tongue's method but this is incidental for now.
The design proceeded based on a 29 ohm R load but simulations and bench experiments yielded max efficiency at a load around 99 ohms (89%). My take on a design problem here is that there are not enough degrees of freedom. The limit on input power sets the collector R which affects the output filter component values and there is not freedom to adjust the transfer of Z by the filter and maintain the cutoff frequency desired. Varying the input capacitance to the filter for tuning will disrupt the filter response so I chose not to do so. It seems the filter is, as a Professor of mine explained, "realizable within a constant".
The inclusion of the impedance transformer on the output was originally to transform the load up to the optimum 99 ohms (89%) but the loss due to this transformer was nearly equal to the loss in efficiency due to the mismatch with a 29 ohm load. Even so, the efficiency, DC final input power/RF output power is 86% with the transformer in place.
I did not attempt to measure the loss due to the filter since I don't know how to measure the true RF output with my equipment at the collector due to the harmonics at this point. I also question that if this measurement could be made with RMS values is it meaningful given that the harmonic power will be removed by the filter and/or the loaded antenna system.
Both the simulation and the measurements for a 29 ohm load (transformer in place) showed that the non zero Vce and the Ic were almost mutually exclusive. Here's a link to the simulated Vce and Ic with the voltage yellow and the current green.
The current spike while the voltage is non zero keeps this from being a pure class E operation. I attempted to measure the actual v and i but the i measurement was difficult with my equipment. I placed a .33 ohm resistor between the emitter and ground and scoped the Vcg and the voltage across this resistor to get the current. The waveforms were very nearly those shown in the linked simulation, including the current spike.
I think I mentioned that the operation "approaches Class E" and I won't claim it is Class E. (is approaches class E the same as approaches pregnant?) So the only reliable claim I can make is that this transmitter gives 86% efficiency both with a 29 ohm dummy load and with my antenna system where the power into the antenna system was measured with the current transformer and voltage probe mentioned earlier.
While i did not pursue the waveforms for different loads I did measure the efficiency for loads from 24 to 121 ohms (without the output transformer). The maximum efficiency was at 99 ohms and the lowest in this range was at 24 ohms (80%). This is why I designed the transformer to transform the 29 ohm load up to 99 ohms or thereabouts.
So, as they say, that is how the sausage was made. I will welcome any comments and maybe the design can be improved.
Neil
I've just got to keep reading
your stuff.
Bruce, DRS2
Neil,
Congratulations for being able to measure the real part of your antenna system impedance. Although I have a General Radio 1606B RF Bridge, my own measurements have not been stable enough for me to have sufficient confidence in the results to be sure what the actual load impedance is. Most people will not have any idea at all what their output resistance is when tuning a class E amplifier, particularly if they have severe operating constraints due to very limited space for antenna insatallation.
The proper collector-emitter (drain-source) capacitance is higher for lower load resistance. The class E circuit tends to be fairly forgiving as far as efficiency is concerned when the wrong value is used, provided that the error is not too great. If you have no idea what your load resistane is, you will not know if your capacitance is in the ballpark.
To really know if you are operating class E or not, you need to look at your collector (drain) voltage pulse with an oscilloscope. The Sokals' class E conditions are that, at the instant the transistor switch of the amplifier turns ON, the voltage at the drain or collector is supposed to be zero, and the slope of the pulse is supposed to be zero. Unless these two conditions are met to at least a reasonable extent, by definition, a class E amplifier does not exist. A class E amplifier is not defined only by the circuit configuration (which can be of many different types), but particularly by the operating conditions.
In my own experimental circuit, I have an oscilloscope probe permanently connected to the drain, like a circuit component; correctly terminated when not using the scope. This keeps the circuit configuration the same when measuring or not measuring the drain pulse.
Ermi,
I have tried using a noise bridge to measure antenna Z and it works reasonably well for my low band ham antennas giving believable results and also for dummy loads on the bench but it is useless for the 3 meter antenna systems.
The high signal level and modest frequency involved with this AM system yields stable and repeatable data with my current transformer and voltage probe method. The V and I check I did today as reported in the Antenna Field Test thread to tune the antenna system gave data at resonance similar to that I measured with the antenna indoors. The indoor current runs about 57 mA and the outdoor current was 54 mA. There was nothing that made me suspect the measurements were not good.
Regarding the collector voltage of my transmitter the simulated voltage is seen on the graph I linked and it is not zero when the transistor switches on so this amplifier is not operating pure class E, but based on the efficiency it is close. I tried many values for the series collector L and the parallel to ground C on the input to the filter and there was no indication that there was an improving trend to follow.
It could be that the non zero voltage and the current spike at switching is due to the duty cycle of the driving waveform. This is a square wave with a duty cycle of 45%. Early in the development I varied the duty cycle and chose this value as being in the optimum range but since doing so I have changed the filter components and the collector L and the load resistance. It may be time to revisit the driver signal.
Any thoughts on this?
Neil
Wouldn't it be a bit better if you were working with sine wave out of that driver prior to hitting the final so as to achieve the evading E?
Or will having square wave achieve E be dampened by heavy filtering to shape that final square into a proper sine wave for radio transmission?
Square waves produce a ton of harmonics which require a bit of filtering. So why not filter it out at the driver stage and then there wont be any need for heavy wave shaping at the output which no doubt would rob any increased efficiency factor, resulting in less than expected at the antenna feed point.
RFB
To achieve high efficiency it is necessary that the final transistor have either zero voltage across it with current through it or zero current through it with voltage across it. For this to happen the device has to switch quickly hence the square wave input.
As I view the v and i graph it seems that the switching on is happening just a bit early before the tank has completed the half cycle of ringing and hence my query about the driving waveform duty cycle.
The literature contains information about this duty cycle and I did vary this in the early prototype and found that the efficiency peaked with a duty cycle between 35 and 45%, consistant with the reports of others. Unfortunately, for the final design several component values had been changed and I didn't repeat this test. I will run the simulation again with different duty cycles to try to answer my own question.
Neil
No, that title is not a plug! 😉
"To achieve high efficiency it is necessary that the final transistor have either zero voltage across it with current through it or zero current through it with voltage across it. For this to happen the device has to switch quickly hence the square wave input."
That's what I figured the goal was in your design. Fully understand the class E approach. And makes perfect sense.
Please do let us know those new test results!
RFB