PWM modulation on a AM transmitter

[SteenBasse]

Hi here

I'm new in this forum and I hope I post on the right place..

I work on a PWM / PCM project with AM modulation on about 2 mHz. (160 meter band), but I have a problem and hope for help .

As my first step I have a X-tal controled 50% duty cycle square-signal (8V single supply). Before I send my signal to an comparator, it must be convert to a 'nicely symetrical' triangle signal, but here I stop...

I am not a 'OP-AMP' expert, so therefore I ask: Can anybody here give 'a hint' on a good 'set-up / OP-AMP type' for an > 2 mHz. integrator there can do this on the best way.

Thank's and regards Jan from Denmark .

PS: I have try with an 'LM318', but unsuccessfully..

 

[Raven Luni]

The op amp needs to be configured as an integrator. Here is an explanation of how integrators work. The other pages there also explain triangle wave generation and other stuff:

http://www.play-hookey.com/analog/integrators.html

 

[duke37]

Pehaps you mean 2MHz not 2mHz.

PWM produces a signal with a high frequency component, it can be attenuated for say an audio amplifier and does not cause trouble.

In the case of an AM transmitter, any high frequency component will generate a very high bandwidth signal. Spark transmitters were abandoned because of the high bandwith signals they produced. You will probably not be able to produce a legat signal with PWM.

 

[poor mystic]

Hi
As duke37 said, there are high frequency components in the squared waves which mustn't be allowed to get into the transmitter, because they'll jam other signals.
One way to improve matters would indeed be to use op-amps to integrate the signal to a triangle wave, but the best way would be with good filtering. Such filtering would tend to be passive, as in inductors and capacitors, rather than active using op-amps.
In practice, and especially for a one-off design, you'd be best to go to the Amateur Radio people, or 'hams', who probably have a club in your area. They'll be able to help with the filtering. (Filters are something of a 'black art' because the characteristics of the inductors very much depend on the materials and winding techniques, and the whole filter needs to be constructed with quite a lot of care to take account of juxtapositions of components, and probably other things as well.)
~Mark

PS
anyway op-amps aren't fast and reliable enough

PPS look at this: http://en.wikipedia.org/wiki/Bandwidth_(signal_processing). I admit it's a very complex subject, and if you'd like to ask personally send me a PM and we can skype.

 

[SteenBasse]

Hi again and thank's for answers .

First sorry for 'milli Hz' of cause I mean Mega Hz "MHz" .

@ Raven Luni: My problem is not how a Integrator works, my problem is to design it, especially when we get up to 1 MHz and higher.

@ duke37: To generate an AM signal from a PWM signal, it is obviously necessary to have a trailing 'good filter' in the transmitter on the signal frequency, see enclosed picture.

@ poor mystic: In principle, the modulation does not cause unwanted side band, since it is not are 'squares / pulses', but only the actual signal frequency. Therefore, the output filter is based solely on the harmonics of the signal frequency. But let's see how bad things are going and what the 'spectrum analyzer' says in the last end .

Another solution could be to generate (I have come to mind) the pulses on a lower frequency (eg 50 -100 kHz.) And then mix them up to the signal frequency? In this way, the OP AMPs work in a more rewarding area.
Okay, maybe here I must search / ask in a 'hams forum'.

Jan .

Edit:
Here's something to be wiser of (maybe):
http://www.tonnesoftware.com/appnotes/pwm/pwmham.html

 

[poor mystic]

It's true, I jumped the groove.

 

[Raven Luni]

Hi again and thank's for answers .

@ Raven Luni: My problem is not how a Integrator works, my problem is to design it, especially when we get up to 1 MHz and higher.

In that case you're better using a transistor based version.

 

[CDRIVE]

Tittle of this is making any sense to me. In one breath you're talking AM, in another it's PWM. They're not the same. The bottom scope pattern that you posted depict an amplitude (Audio Wave) modulated carrier, while the top pattern is PWM, that's more akin to FM or FSK. Yet the bottom trace shows no evidence of any frequency variation (FM) or FSK. Instead, it's a textbook example of AM.

If it's truly an AM transmitter that you're trying to build then some basic rules of AM transmission are being violated here. Audio should not be injected into the carrier prior to any class C amplification, as you show it. If the modulation is truly audio then faithful reproduction will be impossible, which is why the cleanest AM designs are Plate, Collector, or Drain Modulated.

Also, I don't see the purpose of the oscillator not being sinusoidal to begin with.

 

[SteenBasse]

Hi again .

@ Raven Luni: Use discrete transistors was maybe a better idea? But I'm sure it must be possible to find an OP-AMPs that can do the job at around 2 MHz. Or as I wrote, make the PWM at a lower frequency and mix it up to the signal frequency.

@ CDRIVE: I will try to explain, okay I'm better to write in Danish, so I hope you understand .

The picture of signal curves:
Pulse width signal .. If you change the pulse width while keeping the same 'distance' between the positive edges in the signal, then the frequency has not changed, only the 'energy' in the signal, which is also known if you want to control a motor with PWM. Are you sending the signal through a filter, then you as a result have a varying DC voltage supplied to the engine. It is this relationship, you can also use to produce an AM signal.

You can make an AM signal in 3 basic ways:

1) The good old method right from the radio's infancy, anode / collector modulation: The modulation is applied directly to the PA stage:

Advantage: The individual stage in the transmitter need not be made linear, however, you must 'make room' for the modulation of PA stage.

Disadvantage: In order to modulate 100%, you must have an AF modulator with a power of least 50% of your signal output power. A 100 kW transmitter must have a 50 kW AF modulator (beautiful bottles with water cooling) ..

2) Make the whole transmitter with linear stage:

Advantages: You can modulate the transmitter with a small AF power just after the oscillator and send this 'undistorted' throughout the whole transmitter up to the antenna.

Disadvantage: All your stage in the transmitter must be completely linear, this linearity is not quite easy to achieve and is particularly critical if you send music. No problem on paper but in practice? And if you want an output power of example. 25 watts, then you must make a PA stage of at least 100 watts PEP, so there is room for all necessary sidebands.

3) AM make with PWM:

Advantage: As the entire transmitter has no AM info (all pulses have the same amplitude as a FM transmitter), then it can be made with non-linear stage also the PA stage (running close to the PEP effect). The modulation may as in the linear version (the AF signal to the comperator) may be quite small. 'The secret' is the output filter after the PA stage, this is here the PWM signal is converted into AM, just as the motor control.

Disadvantage: The PWM mechanics must done exactly correctly (risk of phase modulation PM) and must only work within the range 25% to 50% duty cycle, other% will only give 'nonsense' when we talk about AM modulation.

Concludetion: A more simple transmitter (non-linear) and a better use of the PA effect.
I hope I have explained it well enough and a understanding of why I want to do AM with PWM.

If somebody today will make a prof high-power AM transmitter, then it is also is made ​​with PWM.

Greetings Jan .

 

[john monks]

Generating any kind of a triangle wave using an op amp integrator is difficult especially with an LM318 because it is only rated for 50volts/usec. You might consider using a straight square wave generator driving an RC integrator circuit. This signal can be your carrier signal that drives a high speed comparator such as an SP9685. This can drive a buffer transistor that will drive your output stage. The other input of your comparator can be your audio source. This will need to be adjusted to allow 0% duty cycle to close to 50% duty cycle coming out of your comparator. Getting rid of the distortion will be quite a trick because there may be some non-linearities that you need to compensate for. You might start with a RC circuit with 1K resistor and 250pF capacitor.

 

[SteenBasse]

@john monks thank's for answer:

Ass I write in the start:

As my first step I have a X-tal controled 50% duty cycle square-signal (8V single supply). Before I send my signal to an comparator, it must be convert to a 'nicely symetrical' triangle signal, but here I stop...

My problem here is to make 'nice / correct' triangle from my square-signal and you write:

You might consider using a straight square wave generator driving an RC integrator circuit

So as first step the 'RC integrator' is my problem and then come the 'comparator'..

Ok with LM318 is an unfortunate choice as both integrator and comparator?, I went after it could work at high frequencies and what I had in the 'drawer'. SP9685 looks exciting, but it is not, to shoot small birds with cannons? but okay.

So back to the integrator as my next step Jan .

 

[john monks]

Generating a 1.8MHz or 2MHz symmetrical triangle waveform is almost impossible. The human ear can hear very small levels of distortion. That's why I suggested a simple RC device and an SP9685. The only other thing I can think of is a high speed device such as a BFR96 transistor with a small capacitor connected between the collector and the base and driven by your crystal controlled square wave generator to get a triangular waveform. If you don't like an SP9685 what else is there?

 

[SteenBasse]

@ john monks and thank's:

Okay, now I'm in on what you mean by a "RC circuit with 1K resistor and 250pF capacitor". Yes I thought about it enough that triangles are not easy to make symmetrical on this frequencies, such a transistor stage I will try to do.

I love SP9685 and many others , but an operation up against 1 GHz (50 Ohm). I just thought it sounded a little violent, but okay such an 'animal' can of course may be necessary to get an accurate enough comparator.

Jan .