Broadband transmission line transformer design ?

[G0HZU_JMR]

Anyway.... the bottom line is that the waveform on the left transistor collector can't possibly look like a full/classic AM waveform. The circuit can only push a current waveform into that transistor that looks like half of a classic AM waveform. eg the 'green' current waveform in my earlier post. Because the load is a pure resistance it can't fill in the top half of the waveform 'above' 24V when that transistor is off. i.e. there is no energy storage/release mechanism so no flywheel effect. So this will be the waveform that gets fed back via the 27pF cap rather than a miniature version of the classic AM envelope on the other collector. However, I can only describe what I see happening, not what was in Charles' head wrt the true purpose of the 22R resistor and the 27pF cap.

[G0HZU_JMR]

I guess the other thing to add for Al, is that if you want to stick with this modulator and feed it to a different antenna type then you ideally need to tune/match that antenna such that it presents a load of about 1200 ohms at resonance to the modulator if you want to get the best from it in terms of RF/modulation swing without bad distortion. The MiniVNA spec says it measures impedance in the range 1-1000 Ohms so this was why I suggested that you test to see how well it can measure various impedances near the spec boundaries. I've not used a MiniVNA but I hope/expect it could easily measure 1200 ohms at 1.4MHz with minimal uncertainty.

[G8HQP Dave]

The waveform does not look like full AM simply because it has the baseband present too. Take it through a highpass filter and you will see AM. No rectification, no top half to 'fill in'. Quite different from, say, a Class C output which does need the flywheel effect of the tuned circuit to fill in the other half of the waveform.

The 27pF provides such a high pass filter so it is just AM which is sent back to the base to cause FM trouble.

It was your mention of 'rectified' which I queried because there is no rectification, no need for a flywheel etc.

[Radio Wrangler]

It's active in all four quadrants of the baseband and carrier signals, through having a DC offset in the baseband to make AM rather than DSB.

Baseband to output should be linearly related. Carrier to output will show limiting.

David

[G0HZU_JMR]

I guess it's easy to take things too literally (and that applies to both of us) but my aim was to quickly/loosely describe the waveform in the time domain arriving at the 22R resistor.

For everyone's benefit, probably the best thing I can do is post up a scope plot with the 22R resistor included and they can decide if my description of what the waveform looks like in the time domain was reasonable. Note that I've turned down the RF frequency from 1.6MHz to to about 20kHz to make the trace easier to decipher wrt the green trace. Because there is 'just' a 22R resistor in the collector the voltage waveform corresponds faithfully to the green current trace in my earlier sketch with 100% modulation. Flip the green trace upside down and you will see a similar waveform shape in the scope plot for 100% modulation. The top of the scope plot corresponds to 24V DC and corresponds to the 'zero current' area on the green trace.

with 24mA pkpk current going into the 22R resistor there should be just under 0.6V pkpk seen here at the collector and the scope is 0.2V/div because I used a x10 probe.

I guess it's possible dissect my words and conclude that I think a resistor is a rectifier (!) but I just described the waveform visually in the time domain as if it was a rectified version of an AM waveform. But that doesn't mean I think the 22R resistor is a rectifier... Maybe I should have just posted up a scope plot.

[Radio Wrangler]

Yup, that scope picture is just what I'd expect. There is a current source which conveys the baseband plus a DC offset (to eventually give the AM a carrier) and it is switched by something approximating a square wave at the carrier frequency between two collectors. look at any one collector and you see the modulated current source chopped at the crystal oscillator frequency. Look at the loop and you'll see the current shaped into an RF sine and the DC/LF components removed (shunted by the LF Z of the loop). Look at the spare collector and you see the current converted into a voltage. A baseband sinewave witha DC offset, chopped at the crystal osc freq.

David

[G0HZU_JMR]

Yes, that describes it well.

The constant risk with writing stuff down on a forum is that every written word can be scrutinised in slow time and misinterpretation (and the subsequent defence of the misinterpretation) starts racking up the post count

Much better to be in the same room with pencils and pads etc. I think I should have posted the scope waveform rather than cause confusion with my attempt at a written description

[Radio Wrangler]

Having the job of designing an AM transmitter here in the 21st century brought it home to me how much modern devices were optimised for more complex (pun alert) modulation schemes. Finding something without a 0.7v gate protection diode in the -ve direction is difficult and you want to play class-C and beyond non-linear modes for efficiency.

It definitely felt like swimming against the current.

David

Oh and the modulator needed to be a feedback loop job with a very linear detector to meet the limits on modulation harmonics.

[G0HZU_JMR]

I wonder what people do to find crystals for this thing? I guess 1.4MHz might be popular and Al is moving to this frequency. I'm not back at work until Tuesday and I doubt I'll find anything in the graveyard cupboards there. I'm due to place an order to Farnell here at home tomorrow for a load of stuff but they don't have anything. I had a look on ebay but had no luck with various keyword searches.

Now I've got the 22R inline I tried fitting the 27pF and 56pF caps but unsurprisingly I saw no benefit. With the antenna fitted the 56pF causes a need for a very subtle retune in frequency but that's about it. This thing is pretty good already in terms of modulation distortion when the 22R in the current source emitter is removed and I can't see any instability or bad spectral spreading in my version.

In his notes Charles says these extra caps need to have short leg connections so maybe this is a clue. As already suggested by David, they could just be an insurance/fix against instability as many people will lay this out using various construction methods and some may use fairly fast transistors. But mine is ugly built on a ground plane and is tame when looking up to 1GHz on my analyser.

[Argus25]

Originally Posted by G8HQP Dave

Your best bet is to measure the inductance of the loop or calculate it. Then calculate the radiation resistance (proportional to loop area) and add the conductor resistance (calculate or measure) to get the total resistance. Then at a particular frequency you can calculate the series impedance. From that you can design a matcher.

I mentioned on other posts about the original circuit that:

"The radiation resistance of the loop antenna is proportional to the square of its cross sectional area and the number off loop turns squared. Also proportional to the fourth power of its operating frequency"..not the Area directly.

This means as noted that its proportional to the 4th power of the linear dimension, like the 4th power of the frequency, so lets say you halved the linear size of the loop and halved the operating frequency, your radiation resistance has dropped by a factor of (16 x 16), awful isn't it ?

But Astral Highway, I think I can help you with a different tactic on the whole problem of how to design with broadband transformers for this AM transmitter application. Really, the best move is to separate the properties of the amplifier and antenna, rather than some matching network arrangement for a specific output stage to some specific antenna, its very very limiting for your future experiments.

Simply settle on some impedance of available transmission line (coax) I used 50 Ohms, but 75 is actually a bit easier for this application.

It is then it is dead easy to design the broadband transformer to match any amplifier output to 75R by experiment. Simply put in a 75R dummy load on the output. Then adjust the turn's ratio of your transformer so that when at maximum power or peak envelope power of your output transistor/s is such that their collector voltage is not near clipping and still in the linear region, but has as much amplitude as possible.

If say the volume of the core material in your transformer is too small (the inductance not high enough in conjunction with the primary turns) there will be waveform distortion that gets worse as the operating frequency drops, so, either use a bigger sized core, or increase the number of turns on the primary and secondary. Its all very easy. Type 77 material from CCI is good from about 1 to 30 MHz. And the RF400 sized core is fine for a few watts @ 1500kHz.

One way you can get better linearity of the carrier sine wave if there is a tad of distortion is to tune the primary with a capacitor which helps a little, usually the Q is low and it doesn't make it too frequency selective (You'll notice that 1100pF tuning cap on the output transformer for the class AB amplifier circuit I posted on the other thread).

Then, when you build your loops, make them so they have a 75R input impedance. As noted before the requires a tap in at 1 turn, then a series capacitor to neutralize the inductance of this turn and an input matching auto-transformer. Its quite easy to get the input impedance to 50R and even easier to 75R, at its operating frequency. You can measure it easily with a signal generator and series resistance.

With this arrangement you can also send the output of your amplifier down a 75R coax cable to your antenna, and you will have a generalized rather than specific system that you can build other antennas and amplifiers to drive.

(Also the other tip, it is better to do the amplitude modulation at low energy levels using a linear modulator and then feed the modulated carrier to the power output amplifier, rather than modulating the output stage). If I was using your circuit, I would dispense with the collector choke, feed the collector into the primary of an RF400 broadband transformer, then feed its split secondary into the bases of a pair of transistors in class AB (like the circuit I posted) and then via another broadband transformer into a 75R load.

[Argus25]

Originally Posted by G0HZU_JMR

I wonder what people do to find crystals for this thing? I guess 1.4MHz might be popular and Al is moving to this frequency.

A while back I found some very nice 1480kHz crystals on ebay, so they do crop up from time to time. I did quite a few experiments with them to find the better method to make a reliable and relatively distortion free L/O. In the end I settled on a circuit, that I attached on another post using the MPF102. It out-performed any single transistor oscillator I could make for low distortion.

[G8HQP Dave]

Originally Posted by Argus25

I mentioned on other posts about the original circuit that:

"The radiation resistance of the loop antenna is proportional to the square of its cross sectional area and the number off loop turns squared. Also proportional to the fourth power of its operating frequency"..not the Area directly.

Yes, you are correct. I should have checked before typing.

Going back to the 'rectification', the issue is that AM plus baseband with a square carrier looks exactly the same as rectified AM of twice the amplitude. With a sinewave carrier they would look different, so it would be more obvious that no rectification is taking place.

The other half of the diffamp has a 22R resistor in the collector. When modulation is applied this can't swing over the full modulation cycle because the load is resistive. So it will produce a tiny rectified version of the modulated RF envelope here but this should have low distortion in the rectified envelope.

It was the "can't swing over the full modulation cycle because the load is resistive" which concerned me. This implied the need for a flywheel, when there is no such need. It can swing over the full modulation cycle precisely because the load is linear. As I said, all you need to recover AM is a low pass filter to remove the baseband.

[G0HZU_JMR]

It was the "can't swing over the full modulation cycle because the load is resistive" which concerned me. This implied the need for a flywheel, when there is no such need. It can swing over the full modulation cycle precisely because the load is linear. As I said, all you need to recover AM is a low pass filter to remove the baseband.

Ok, to clear this up finally, all I meant to imply was that the waveform at the 22R collector resistor can't go/swing higher than 24V (to make it look on the scope like a classic AM envelope) if there is just a resistor there. So the voltage waveform will remain with the same 'shape' (i.e. shape that looks like rectified AM) as the green 'current' waveform. We can now agree that it doesn't mean I think a resistor has rectified something.

If I described a cloud in the sky and said it was looked like an elephant it doesn't mean I think there must be an elephant in the sky for example. The mistake I made was to try and describe the shape of a complex waveform with words (that could cause confusion) rather than just post the scope waveform itself. If I'd done that I don't think there would have been an issue

[G0HZU_JMR]

A while back I found some very nice 1480kHz crystals on ebay, so they do crop up from time to time.

Ok thanks, I'll try various searches for that frequency. Whatever crystal frequency I eventually find in the 1.4-1.6MHz range will mean tuning the antenna to suit it. It's much easier/nicer to do it the other way round with a sig gen

[G0HZU_JMR]

But Astral Highway, I think I can help you with a different tactic on the whole problem of how to design with broadband transformers for this AM transmitter application. Really, the best move is to separate the properties of the amplifier and antenna, rather than some matching network arrangement for a specific output stage to some specific antenna, its very very limiting for your future experiments.

Yes, if he decides to put an amp between the modulator and the antenna then this would definitely make the system more versatile from an experimentation point of view and might make life easier wrt the MiniVNA.
One niggle with the original design is that it runs from 24V and maybe it could be modified to suit 12V operation if it only has to operate into an amp rather than the original 1200R loop.

[G8HQP Dave]

You said

So this will be the waveform that gets fed back via the 27pF cap rather than a miniature version of the classic AM envelope on the other collector.

This is the waveform which appears at the collector. As the 27pF cap is too small to send much baseband current back, the feedback to the base will be full AM not apparently rectified AM. The point you appeared to be making in your original post about this was that something different happens at the two LTP collectors. I questioned that because the same thing happens at both collectors: a current arrives consisting of DC, baseband and modulated RF. The only difference is that the output collector presents different impedances for DC, baseband and RF so there will be different amplitudes of DC, baseband and RF voltages there.

That doesn't mean I think a resistor has rectified something.

I never thought that you did. You appeared to think that the LTP rectified something, and subsequent statements seemed at first to confirm this. As Radio Wrangler puts it

It's active in all four quadrants of the baseband and carrier signals, through having a DC offset in the baseband to make AM rather than DSB.

So we can agree that the feedback to the LTP input base is AM, which carries the risk of creating some FM in the oscillator?

[G0HZU_JMR]

Argus, can I ask what kind of range have you achieved with a fairly modest PA and an antenna designed for some degree of far field radiation efficiency?

By range I mean effective range for a reasonable S/N ratio rather than ultimate detectable range. In reality, it probably doesn't really matter much if the transmitter leaks out over a larger range than necessary, it's hardly going to be classed as a pirate station

However, it would be interesting to know what typical range performance you actually get with this approach.

it's hardly going to be classed as a pirate station

Maybe not, still illegal in the UK though, with possibly draconian measures taken (£5000 fine, six months prison and all things electrical confiscated). I think it is about time the powers that be (here in the UK) did something like the USA where they have a 100mW power (DC input) limit and antenna height/length restrictions, one can dream!

In essence, if you want an AM (MW) transmitter for your lovely radios make sure it is only just powerful enough for your needs.

[G0HZU_JMR]

This is the waveform which appears at the collector. As the 27pF cap is too small to send much baseband current back, the feedback to the base will be full AM not apparently rectified AM.

Yes, I just (correctly?) described the waveform shape at the collector as looking like rectified AM.

The 27pF cap will feed some of this back to the diff amp input

In my quote above I didn't quantify what 'some' actually means because I didn't think I had to. In terms of how much this (or other factors) might cause FM in the crystal oscillator I can't say for now because I simply don't know. I plan to build the oscillator when I can find a crystal

[G0HZU_JMR]

In essence, if you want an AM (MW) transmitter for your lovely radios make sure it is only just powerful enough for your needs.

I think that's why I like Charles' near field antenna. The energy stored (rather than radiated) in the near field will produce a rapidly decaying bubble of useable range.

If one is fortunate enough to live in a huge sprawling house with a shed at the bottom of a long garden then it's not going to give a big enough bubble but I think the circuit could be souped up a bit to give a bit more range. But it's not going to work across something like a converted farmhouse with various outbuildings