PWM AM TX IDEA

Hi,
it looks like the whole concept of generating AM via PWM is wrong. If the TX sends out PWM bursts of carrier and the RX uses a common diode detector, the demodulation is not correct and distortion appears, no matter what the sampling frequency and quantization precision are. The correct process of PWM demodulation requires to convert it to PAM (pulse amplitude modulation) first, see e.g. here.

VBR form Ivan

Ivan wrote:

Ruud wrote:What a relief to see a READABLE schematic!

Haha. Produced by LTspice freeware with Bordodynov's component library, if someone is interested.

VBR from Ivan

Superb design, Ivan! I wasn't well so not able to play with Ltspice. When I stay away from phone, laptop, and homebrew I don't have any pain then. 

With my extensive experience in Ltspice last year I would say it's so accurate only the simulation would do and there's no need to physically build the circuit. I was amazed by its precision and accuracy at RF.
It's a perfect design tool for transmitter and receivers.

Ruud wrote:What a relief to see a READABLE schematic!

Haha. Produced by LTspice freeware with Bordodynov's component library, if someone is interested.

VBR from Ivan

What a relief to see a READABLE schematic!

Hi all, I tried to design a PWM-AM modulator based on the same principle, but without a MCU. It contains three common chips: 4 NAND gates (1x 4011), a binary counter (4024 or 4040, ev. 4060) and a comparator (e.g. LM311, an opamp can be used instead). I used Vcc +12V to get higher speed of CMOS circuits and more output power, but +5V might do as well.

Two NAND gates form a clock generator. In case of using a 4060 its internal gate can be used instead. Also a canned XCO can be used. The crystal has cca 10,2 MHz resonant frequency here, but another one will do. Take the proper outputs of the counter to obtain a carrier in the MW band and a suitable PWM sampling frequency. Here the carrier is 10200 / 16 = 638 KHZ and sampling 10200 / 512 = 20 KHz.

A R-2R ladder together with an RC filter form a sawtooth signal. You can fiddle with the part values to remove the ripple from this signal, as it may adversely affect the quality of the resulting modulation. The sawtooth signal is compared with the input AF signal, PWM is generated on the comparator output. The PWM is logically multiplied (AND function) with the carrier. The result should behave similarly to AM. It can be amplified by an E-class PA (not in the schematic).

This is a raw suggestion, comments are welcome. I have not tested the circuit in practice yet, but the simulation results are promising.

Ivan wrote:

dare4444 wrote:... Given the simplicity of a single Arduino Nano it's okay.

What about the frequency stability of its internal clock generator? AM on MW is quite tolerant, but anyway...

BTW "standard frequencies" in the MW and LW band are divisible by 9 KHz. So 1600 KHz is not standard, 1593 KHz should be used instead.

VBR from Ivan

Oh. The Arduino is run by a 16MHz crystal.

dare4444 wrote:... Given the simplicity of a single Arduino Nano it's okay.

What about the frequency stability of its internal clock generator? AM on MW is quite tolerant, but anyway...

BTW "standard frequencies" in the MW and LW band are divisible by 9 KHz. So 1600 KHz is not standard, 1593 KHz should be used instead.

VBR from Ivan

Ivan wrote:Hi,
here I am.

To be honest, I am not a fan of using digital ICs in analog apps, although it may work and make simple circuits. Strange oscillations and/or increased power consumption may appear near the threshold level. But you are probably right. Try it on your testboard and you will see!

What advantage has using an Arduino to generate a carrier/clock over a simple XCO (or a VFO, if tuning is needed)?

VBR from Ivan

Hi!! You're so right. Arduino would not give standard frequency numbers. For example 16mhz clock / 6 = 2.6666 /2 = 1.3333 Arduino output frequency. 

It only generates 1600, 1000,  and 800 KHz standard frequencies. Given the simplicity of a single Arduino Nano it's okay.

Happy to see you! Thanks for the reply. Did you check the Arduino one? It can generate twelve frequencies in the MW band. Programming takes minute or so and external parts are minimised. 64KHz sampling rate is enough for MW. 

500 to 1600 KHz coverage with 12 different frequencies. 

Frequency 15 - 500 khz 14 - ~530 khz 13 - ~570 khz 12 - ~610 khz 11 - ~670 khz 10 - ~730 khz 9 - 800 khz 8 - ~890 khz 7 - 1000 khz 6 - ~1140 khz 5 - ~1330 khz 4 - 1600 khz 

Hi,
here I am.

To be honest, I am not a fan of using digital ICs in analog apps, although it may work and make simple circuits. Strange oscillations and/or increased power consumption may appear near the threshold level. But you are probably right. Try it on your testboard and you will see!

What advantage has using an Arduino to generate a carrier/clock over a simple XCO (or a VFO, if tuning is needed)?

VBR from Ivan

Ivan, are you there?

R1R2C / (R1+R2)  time needs to be bigger than the pulse period of 1.6MHz (.0065 mS something) 

My values are coming R = 100ohm and C = 22pF for a time constant of. 0011 mS. 

This is a PWM generator. Please advice. Is
 Is it a practical circuit? An arduino can supply 1600KHz carrier or CLK. Q output of cd4013 can drive a 2n7000. The capacitor charges via the positive audio signal as it's DC coupled to LM386. When output of /Q is 0 it discharges the capacitor. When CLK is applied there is constant tug or pull between AF and /Q for the control of capacitor C. The Q and /Q output alternates when CLK is applied. LM386 to Data input is reflected at Q output provided a certain threshold is reached. The charging and discharging of the capacitor with the audio signal and /Q creates a logic 0 or 1 on the Data pin. The CLK keeps the process of charge and discharge repeat itself. The output of Q alternates between 0 and 1, a square wave and its duty cycle  varies in accordance with the audio signal. It leads to PWM modulated RF signal available at Q. Am I correct?

I do not think so. A gate restores the edges and levels of the signal, while a 4066 passes the signal through "as it is".

VBR Ivan

Is CD 4066 a better option than NAND switching?

The first opamp takes in 50KHz square wave. This one is untested. The integrator generates the triangular wave. Class D amplifiers all the way to the final stage should give 90% efficiency. A seperate audio modulator and filter aren't required.

Thanks, Harry! Glad you liked it. No I haven't checked it yet but we can use two nand gates and drive them with 0 and 180 degrees LO square wave for a pushpull 2n7000 output for lower harmonic content. CD4011 is another easy to find nand gate. Could they be replaced by a cascode transistor stage?

Hi Joy,
That is  really interesting idea. Thank you for sharing it.

One question, have you looked at the spectrum analyser? How does it perform across the band for spurious?

BR Harry - SM0VPO