Extreme Low Frequency Project

dannyf

Joined Sep 13, 2015
2,197
Can anyone help me out with ideas or samples?
I would think you want to think about the medium of transmission; and generation of the carrier signal at such a low frequency.

No idea about the 1st but the 2nd can be done easily on a mcu, through DDS.

The driver choice will depend on your transducer (which depends on your answer to the first question). A h-bridge will work; so will a typical DC amplifier (like the types you use for audio).
 

nsaspook

Joined Aug 27, 2009
13,421
In the 70's we were testing a system with aircraft that used a type of Barker code for low frequency low bit-rate secure communications. For effective communication system simple FSK is OK as the base-band method but don't expect a reliable communications channel unless the data is encoded into the proper format for transmission and reception first.

If you needed to send/receive this.

Then you would need to actually send a sequence of bits for synchronizing and to detect/correct errors in transmission for any practical distance at low power levels.
 

DickCappels

Joined Aug 21, 2008
10,227
It is not radio radio when the distance involved is a tiny fraction of a wavelength; it is coupling of electric and magnetic fields.

It is easier to reliably detect changing fields than to detect static fields, with signal strength being directly proportional to the rate of change. If you want to use a low pulse rate, using narrow, fast rise time pulses which you can code by the spacing (frequency) of pulses or pulse groups. This is impulse radio.

Such signals can penetrate concrete, wood, and many other common construction materials, but when it comes up against metalized mylar and steel are a different matter.

I have had some success in the tens of Hz using both electric and magnetic fields over a couple to several meters using relatively small radiators.

One method:

- Transmit fast pulse edges (Tr 5 ns or less) at low frequency.

- Code the timing of the edges in accordance with a Barker Code (post #23 above)

- Detect and low pass the pulse edge energy

- Demodulator the detected & filtered signal with the same Barker code (may by synchronized or asynchronous) in a four quadrant multiplier.

- Integrate the product

- Information ("data") may be overlaid on this "carrier" system by inverting or not inverting the Barker code in the modulator. The result of inverting the modulating Barker code will be the inversion of the polarity of the demodulated signal.

- If the demodulating Barker code is not synchronized with the modulating code, a beat between the two code frequencies will appear in the modulator output. The polarity of the peaks of the beat signal will indicate the whether both the modulating and demodulating Barker codes are of the same polarity or not.

- The shorter the Barker Code the higher the signal gain but the lower the processing gain (ability to ignore noise)

Some related articles:

http://lmgtfy.com/?q=micropower+impuse+radar

user.it.uu.se/~carle/Notes/UWB.pdf

www.vtvt.ece.vt.edu/research/references/uwb/overview/Journal038.pdf
 

Thread Starter

fsoender

Joined Dec 10, 2015
26
Guys, thanks for the replies. I think I'm on to something, the only thing left to know is:

To get higher bit rate, I need to kill the induction in the coil after sending the signal pulse in the H-Bridge as quick as possible.
This so I can send a new pulse, and the charged induction wont interfere from the previous pulse.

What would be the order to "fire" the transistors in the H-Bridge to obtain this sequence?

Would it be to fire 1 & 4 or 2 & 3, then pull 2 & 4 to kill the charge?

Transistor order:

1 3
-Coil-
2 4
 

RichardO

Joined May 4, 2013
2,270
The way you get a coil to change faster is to drive it from a current source, not a voltage. The simplest way to do it is to use a resistor in series with the coil and increase the power supply voltage to make up for the losses in the resistor. This is how it was done with stepper motors before PWM became common.
 

Thread Starter

fsoender

Joined Dec 10, 2015
26
The way you get a coil to change faster is to drive it from a current source, not a voltage. The simplest way to do it is to use a resistor in series with the coil and increase the power supply voltage to make up for the losses in the resistor. This is how it was done with stepper motors before PWM became common.
Sounds good, thanks.
I Seen other papers that one way to do it is to open up transistor 2 and 4 to drain the tank.
 

sailorjoe

Joined Jun 4, 2013
365
If you have an ATMega, you can generate your frequencies directly out of the GPIO pin. Then amplify with any low frequency amplifier, and run that into your coil/solenoid.
 

Thread Starter

fsoender

Joined Dec 10, 2015
26
If you have an ATMega, you can generate your frequencies directly out of the GPIO pin. Then amplify with any low frequency amplifier, and run that into your coil/solenoid.

How is the best way to read the signals sent from this coil, would it be to connect the coil to the ADC on the other end?

What filter would you prefer for this, just a standard low pass filter?
 

Thread Starter

fsoender

Joined Dec 10, 2015
26
If you have an ATMega, you can generate your frequencies directly out of the GPIO pin. Then amplify with any low frequency amplifier, and run that into your coil/solenoid.
I guess I wont be reading square signals on the receiver end, since I fire one half of the bridge then the other, that would create a sin curve?
 

sailorjoe

Joined Jun 4, 2013
365
Fsoender, I was thinking that the coil is used as an antenna, to transmit on one end, and to receive in the other. DickCappels is right in that this isn't really radio, more like electromagnetic coupling between two coils. So sending a square wave at these low frequencies would result in a short positive pulse due to the rising edge of the square wave, followed by a negative pulse due to the falling edge of the square wave. The shape and duration of the pulse depends on the coil inductance and the driver circuit. If the inductance is high enough, you might get something that begins to resemble a square wave. So in this arrangement there's no need for an H-bridge. Full disclosure. I haven't worked at these low frequencies since the seventies. At that time I ran tests on a guidance circuit that started at 0.0001 Hz, and ran up to 100 Hz. And I've never tried to transmit at these frequencies, so please study carefully the information from the other posters. It's good info.
 

Thread Starter

fsoender

Joined Dec 10, 2015
26
Fsoender, I was thinking that the coil is used as an antenna, to transmit on one end, and to receive in the other. DickCappels is right in that this isn't really radio, more like electromagnetic coupling between two coils. So sending a square wave at these low frequencies would result in a short positive pulse due to the rising edge of the square wave, followed by a negative pulse due to the falling edge of the square wave. The shape and duration of the pulse depends on the coil inductance and the driver circuit. If the inductance is high enough, you might get something that begins to resemble a square wave. So in this arrangement there's no need for an H-bridge. Full disclosure. I haven't worked at these low frequencies since the seventies. At that time I ran tests on a guidance circuit that started at 0.0001 Hz, and ran up to 100 Hz. And I've never tried to transmit at these frequencies, so please study carefully the information from the other posters. It's good info.
Sailorjoe,

I understand that it is electromagnetic pulses. I will be using one coil to transmit and another to receive. But they will be transceivers, since I want them both be able to transmit and receive. Im been reading a lot about this, but there is no examples out there to really get some real input.
I've seen a project like this before, and he used a H-bridge and could read various sensor data thru it. I was thinking using a few hundreds of mA, and I should transmit at least 5-6 m in air.

I just have to figure out how to operate the transistors in the bridge the correct way. I think the other guy use one half of the bridge for positive sin and fired the other half for the negative sin.

He then used FSK as transmit protocol.
 

SWer

Joined Dec 20, 2015
17
Here's an idea for both xmit and receive that does not need anything as exotic as synchronous detection - the concept of sending data in this way is nothing more than a half/full duplex modem working at a very low baud rate.
As long as you can reliably transmit a detectable signal such as with an H- bridge controlling the switching between two different and distinct frequencies of a LF transmitter, you can very easily make a receiver which consists of two PLLs, each tuned to one of the two distinct transmit frequencies. The output of each PLL would have an output that serves as a flag to indicate that the PLL is locked - say a 1 for locked, a 0 for not locked. By doing some very simple logic with these two lock-not locked signals a detected one or zero is sent to a serial shift register when the following circuit is triggered.
On the arrival of a one (a start bit )after a minimal delay interval (stop bit period) the receiver turns on the output of its system clock which is 16X the transmitted bit rate. Then following the reception of the start bit, the timing of the receiver counts off 16 clock edges then samples the output of the decoded PLL outputs and shifts that value into the shift register. It continues just as outlined for as many times as needed to send the complete data word. For example, for the familiar protocol of 1 start bit 7 data bits 1 stop bit with parity, the number of shift clocks to the serial shift register would be 9 bit periods. A simple state machine is then used to check the data word, rejecting words with wrong parity, data words that contain unrecognized data, etc. This would be a good job for a simple microcontroller. The receiver would need a xmit circuit if it needed to signal to the transmitter that data needed to be re-transmitted.
The transmitted frequencies are simply fractional-N PLL frequency dividers set to any frequency furnished by a simple xyMHZ crystal clock oscillator. No need for any frequency control as long as the transmitter and receiver both use the same basic crystal oscillators.
 
Last edited:

Thread Starter

fsoender

Joined Dec 10, 2015
26
Here's an idea for both xmit and receive that does not need anything as exotic as synchronous detection - the concept of sending data in this way is nothing more than a half/full duplex modem working at a very low baud rate.
As long as you can reliably transmit a detectable signal such as with an H- bridge controlling the switching between two different and distinct frequencies of a LF transmitter, you can very easily make a receiver which consists of two PLLs, each tuned to one of the two distinct transmit frequencies. The output of each PLL would have an output that serves as a flag to indicate that the PLL is locked - say a 1 for locked, a 0 for not locked. By doing some very simple logic with these two lock-not locked signals a detected one or zero is sent to a serial shift register when the following circuit is triggered.
On the arrival of a one (a start bit )after a minimal delay interval (stop bit period) the receiver turns on the output of its system clock which is 16X the transmitted bit rate. Then following the reception of the start bit, the timing of the receiver counts off 16 clock edges then samples the output of the decoded PLL outputs and shifts that value into the shift register. It continues just as outlined for as many times as needed to send the complete data word. For example, for the familiar protocol of 1 start bit 7 data bits 1 stop bit with parity, the number of shift clocks to the serial shift register would be 9 bit periods. A simple state machine is then used to check the data word, rejecting words with wrong parity, data words that contain unrecognized data, etc. This would be a good job for a simple microcontroller. The receiver would need a xmit circuit if it needed to signal to the transmitter that data needed to be re-transmitted.
The transmitted frequencies are simply fractional-N PLL frequency dividers set to any frequency furnished by a simple xyMHZ crystal clock oscillator. No need for any frequency control as long as the transmitter and receiver both use the same basic crystal oscillators.

That sounds great, have you seen any examples of this on the net?
One thing I'm struggling with is, how is the H-bridge controlled. When you send the signal as you suggest, do you use the whole bridge or only one half of it?

Thanks for your input SWer
 

SWer

Joined Dec 20, 2015
17
That sounds great, have you seen any examples of this on the net?
One thing I'm struggling with is, how is the H-bridge controlled. When you send the signal as you suggest, do you use the whole bridge or only one half of it?

Thanks for your input SWer
 

SWer

Joined Dec 20, 2015
17
I'm sorry to say that I don't know where you could find how to do this on the net. However it is pretty much a classic data transmission protocol that has been used at least since the early days of serial data transmission from computers to say, dumb terminals. Keywords would probably be something like modem operation or UART. It certainly isn't the only way to transmit data in a modem but what I described is a simple way. There used to be integrated chips that contained much of the required circuitry. I can't remember any chip numbers. I would guess that such devices are still available and cheap. Another possibility is to use a microcontroller/microprocessor, some have built-in UARTs. The PLLs are not part of the UART.

I have to add something now that I neglected to describe in my original note:

1. As an example of modem/UART operation, the technique of data transmission protocol has to be taken into account when sampling the input data. Let's say you use something simple like NRZ format. The rising edge of the start bit starts the 16X oscillator/clock. The first sample would be taken at a count of 8 clock periods and then every sample after that would be taken at a count of 16. This puts the actual sampling of the input data at or very near to the middle of the bit period providing some some noise, bit slippage, etc., immunity. The first sample (at a count of 8) is dropped and doesn't get clocked into the shift register. If another data transfer protocol (other than NRZ or NRZI )is used then the position of the first sample (the sample that is ignored) might be sampled at a count of 4 or possibly 12 with every sample after that being at a count of 16. This is usually something the designer would specify, possibly making this delay user selectable if other transmission protocols would be used. Everything works the same after this selectable delay.
This technique, using the PLLs, does incorporate some aspects of synchronous detection, namely a phase comparator comparing the phase difference between a reference frequency and and a signal frequency. The output of the phase comparator is used to drive a VCO to match the frequency of the input signal. However, the operation of the PLL doesn't require external processing of the output I & Q signals from a 2 or 4-quadrant multiplier, it's all self contained. Simple and reliable PLL chips should still be available such as the CD4046B. It's been quite a few years since I used one but I remember that particular chip was easy to use.
 

Thread Starter

fsoender

Joined Dec 10, 2015
26
I'm sorry to say that I don't know where you could find how to do this on the net. However it is pretty much a classic data transmission protocol that has been used at least since the early days of serial data transmission from computers to say, dumb terminals. Keywords would probably be something like modem operation or UART. It certainly isn't the only way to transmit data in a modem but what I described is a simple way. There used to be integrated chips that contained much of the required circuitry. I can't remember any chip numbers. I would guess that such devices are still available and cheap. Another possibility is to use a microcontroller/microprocessor, some have built-in UARTs. The PLLs are not part of the UART.

I have to add something now that I neglected to describe in my original note:

1. As an example of modem/UART operation, the technique of data transmission protocol has to be taken into account when sampling the input data. Let's say you use something simple like NRZ format. The rising edge of the start bit starts the 16X oscillator/clock. The first sample would be taken at a count of 8 clock periods and then every sample after that would be taken at a count of 16. This puts the actual sampling of the input data at or very near to the middle of the bit period providing some some noise, bit slippage, etc., immunity. The first sample (at a count of 8) is dropped and doesn't get clocked into the shift register. If another data transfer protocol (other than NRZ or NRZI )is used then the position of the first sample (the sample that is ignored) might be sampled at a count of 4 or possibly 12 with every sample after that being at a count of 16. This is usually something the designer would specify, possibly making this delay user selectable if other transmission protocols would be used. Everything works the same after this selectable delay.
This technique, using the PLLs, does incorporate some aspects of synchronous detection, namely a phase comparator comparing the phase difference between a reference frequency and and a signal frequency. The output of the phase comparator is used to drive a VCO to match the frequency of the input signal. However, the operation of the PLL doesn't require external processing of the output I & Q signals from a 2 or 4-quadrant multiplier, it's all self contained. Simple and reliable PLL chips should still be available such as the CD4046B. It's been quite a few years since I used one but I remember that particular chip was easy to use.

Thanks for the help, appreciate it!
 

SWer

Joined Dec 20, 2015
17
I forgot to add some info about using an H-bridge.
An H-bridge would just be a current flow controller that switches the polarity of two points on the bridge. In one output condition the +output would simply forward bias a diode that would enable F1 transmission with F2 transmission turned off due to a reverse biased diode. Switching to the output of the bridge would then reverse bias the F1 diode and forward bias the F2 diode. Both F1 and F2 can be turned off by the third state of the H-bridge, both outputs turned off so nothing is transmitted.
 
Top