How does MAX7400 work?

[Futterama]

Hello forum,

I need a simple digitally variable sine wave output. I found the MAX7400 from Maxim-ic and looked at the datasheet and this example diagram:

**broken link removed**

It looks great, just what I needed.

But one thing I don't understand is the expression "corner frequency".

I just can't see the connection between the filter input and CLK input on MAX7400.

I'm not sure how the above example will work.
If I wanted a 1kHz sine wave output, I should apply a 100kHz square wave to CLK as far as I understand. But what about the input at "Filter IN"?
And which frequency should I change to change the sine wave output frequency?

I hope somebody can help me. Thanks.

Regards,
Futterama

 

[Nigel Goodwin]

It looks pretty simple?.

Like it says, it's just a switched-capacitor low pass filter!.

You provide a squarewave input on the IN pin, and a clock input of 100 times that frequency on the CLK pin. The low pass filter attenuates the harmonics, producing a sinewave at the output.

So, to change the frequency you change BOTH inputs, keeping the difference of 100 times the same (so 2KHz and 200KHz etc.).

'Corner frequency' is the frequency the filter drops sharply off at, giving a 'corner' on a frequency plot.

 

[Futterama]

Well, I don't know anything about low pass filters, I don't know what "harmonics" and "attenuates" is.

Regarding the "corner frequency", the datasheet says:

Varying the rate of the external clock adjusts the filter corner frequency:
fC = fCLK / 100

What does that mean? Is that, if the input frequency is eg. 2kHz, and the CLK is 200kHz, you change the CLK to 210kHz, something I don't understand happens?

 

[Nigel Goodwin]

Well, I don't know anything about low pass filters, I don't know what "harmonics" and "attenuates" is.

Perhaps it's the language difference?, although your English is EXTREMELY good - but "low pass filters", "harmonics" and "attenuates", are VERY basic terms.

"Low pass filter", a frequency sensitive filter, which passes low frequencies, and rejects (actually only 'reduces') unwanted higher ones.

"Harmonics" - multiples of the fundamental frequency, a 1KHz signal will have harmonics at 2Khz, 3KHz, 4KHz etc. The reletive strengths of the harmonics will depend on the exact shape of the wave. A 'perfect' sinewave would have NONE! - but wouldn't happen in practice!. Square waves are high in odd harmonics, 3rd, 5th etc.

"Attenuate" - make smaller, the low-pass filter will make the higher harmonics (unwanted frequencies) much smaller.

Regarding the "corner frequency", the datasheet says:

Varying the rate of the external clock adjusts the filter corner frequency:
fC = fCLK / 100

What does that mean? Is that, if the input frequency is eg. 2kHz, and the CLK is 200kHz, you change the CLK to 210kHz, something I don't understand happens?

It simply alters the frequency of the low-pass filter to be suitable 2.1KHz, if you do this you should alter your input frequency to 2.1KHz.

As the datasheet says, keep the clock frequency 100 times the input frequency, you don't really need to understand anything else?.

 

[audioguru]

I made a similar circuit using a "stepped" digital wave from a CD4018 IC feeding a switched-capacitor lowpass filter IC. The steps reduce lower freguency harmonics so the filter performs "over-sampling".
Its harmonics output is NONE! Yeah, a perfect sine-wave. :lol:

Maybe I cheated a little, because my distortion analyser is built with a switched-capacitor bandpass filter IC, and the tiny amount of distortion from the lowpass filter in the generator is matched and cancelled by the same tiny amount of distortion in the bandpass filter of the analyser.
They both use the same master clock oscillator so their frequency matches perfectly too. :lol:

 

[Futterama]

Nigel, thank you for the explanation, and thank you for your fine words about my english skills 8)

Actually, the other day, I had a visit from a friend, telling me that I don't need a sine wave for my purpose. I wish to send some different frequencies from one PMR (Private Mobile Radio) to another, and my friend told me that I could use square waves. I will be using the microphone input. My friend said that the square waves would not damage the circuits, it would only leave the output stage open for a longer period.

I just thought that I should use sine waves to prevent damaging the PMR.

Any comments on this?

Regards,
Futterama

 

[Nigel Goodwin]

Nigel, thank you for the explanation, and thank you for your fine words about my english skills 8)

Actually, the other day, I had a visit from a friend, telling me that I don't need a sine wave for my purpose. I wish to send some different frequencies from one PMR (Private Mobile Radio) to another, and my friend told me that I could use square waves. I will be using the microphone input. My friend said that the square waves would not damage the circuits, it would only leave the output stage open for a longer period.

I just thought that I should use sine waves to prevent damaging the PMR.

Squarewaves are fine, just keep them at a reasonable modulation level. I don't know if you're a radio amateur?, but a 1750Hz tone burst is used (or at least was used?) to open up repeaters, it was common to use a couple of 555 timers to generate it, or a CMOS 4001 or similar.

 

[Futterama]

Nigel, I am not a radio amateur...

I am not sure what you mean by "reasonable modulation level" - is it the same as the amplitude?

The square wave will be generated by a PIC, and must be "converted" from the PIC's 5V output to microphone level. I found information on this page:

https://www.epanorama.net/circuits/line_to_mic.html

I will measure the peak voltage from the microphone with a scope, before determining the signal amplitude to prevent damage (is that called "overdrive"?).

 

[Nigel Goodwin]

Nigel, I am not a radio amateur...

I am not sure what you mean by "reasonable modulation level" - is it the same as the amplitude?

Yes, near enough, 'modulation level' is effectively how loud it is. It's a term used in transmitters (and other devices as well).

The square wave will be generated by a PIC, and must be "converted" from the PIC's 5V output to microphone level. I found information on this page:

https://www.epanorama.net/circuits/line_to_mic.html

I will measure the peak voltage from the microphone with a scope, before determining the signal amplitude to prevent damage (is that called "overdrive"?).

Yes, all you need is two resistors!.

 

[Futterama]

Nice.

Well, do you know how an amplifier will react to a square wave that is not 50% duty cycle? I am thinking perhaps 10% duty cycle.

 

[Nigel Goodwin]

Nice.

Well, do you know how an amplifier will react to a square wave that is not 50% duty cycle? I am thinking perhaps 10% duty cycle.

It will simply pass it through, just as it would any other signal - but why only 10% duty cycle?.

 

[JimB]

Futterama,

The nomal output from a microphone is just a few millivolts, and this is what the radio will expect on its input.
It is most unlikely that the 5v output from a PIC or 555 would damage the input circuit of the radio, but it will not work very well!

Modulation
A radio transmitter emits a carrier wave, just a sine wave at some frequency, (say 150Mhz). The carrier on its own carries no information, it has to be "modulated".
There are several methods of modulating the carrier:
Switch it on and off, the usual way of sending morse code
Vary the amplitude at the voice frequency, Amplitude Modulation (AM)
Vary the frequency at voice frequency, Frequency Modulation (FM)

My best guess it that your radio will use FM.

In any FM radio system, there is a limit to the level of modulation which can be applied to the carrier, otherwise the signal will spread out and interfer with other signals on adjacent channels.
There are circuits in the radio which limit the amount of modulation (also known as "deviation" in FM systems). There are of course limits to the signal level which the "deviation limiter" can handle without introducing gross distortion to the signal.

So, what Nigel is getting at is, keep the input signal from your PIC/555/W.H.Y. to a level similar to that from the microphone and you will not overdeviate the radio signal and it will sound OK in the receiver.

What are you trying to do by sending this tone through your radio? It may make a difference to our answers.

JimB

Edit, an online dialogue going on as I typed this!

 

[Futterama]

Nigel, ok then.

10% duty was an example. The goal is to use a PMR to send digital information, 8 bit at a time. Therefore 10% as 10bits (8 bit + startbit and stopbit) if all bits are high or low.

The original idea with the sine wave frequencies was to send different commands and data as frequencies like this:

1 kHz tone: Transmission begin
2 kHz tone: Bit value 0
3 kHz tone: Bit value 1
4 kHz tone: Transmission end

But if the circuits can handle digital pulses, I'll just use that :wink:

JimB, thanks for taking time to write, it wasn't for nothing, I sucked it all in :wink:

 

[JimB]

Futterama

A waveform with a 10% duty cycle does not translate to 10bit digital data!

You need a modem, just like you have to connect to the internet via a telephone line.

Your scheme of four tones goes part way to solve the problem, but, dont use harmonically related tones, (eg 1, 2, 3, 4 khz), distortion in the transmitter and receiver will generate harmonics and the decoder will become confused.

Also 4khz is too high an audio frequency, it will be greatly attenuated by the transmitter and receiver filter circuits. Phone transmitters only pass an audio range of 300 to 3000hz.

JimB

 

[audioguru]

A duty cycle that isn't 50% produces even harmonics. Therefore a 1kHz tone will also have a 2kHz tone (and 3k, 4k, 5k etc).
A 50% duty cycle 1kHz square wave won't have 2kHz but will have 3kHz added (and 5k, 7k, 9k etc).
Distortion in the transmitter or receiver circuits would produce 2k, 4k, 6k etc in addition.

 

[Futterama]

A waveform with a 10% duty cycle does not translate to 10bit digital data!

You are right, I don't know what I was thinking...

Regarding "harmonically related tones", again you are right, I didn't think of it.

So, if 4kHz is too high a frequency, 4 good frequencies to use would perhaps be 2000Hz, 2300Hz, 2600Hz and 2900Hz? Or are these in harmonics? There has to be a big gap between the frequencies so the decoder can easily distinct them from each other.

Perhaps I could use 4 duty cycles instead of 4 frequencies, and the decoder (PIC processor) could measure the duty cycle and determine the command or bit from that. How about that?

It seems like I'm getting there now (to the final solution).

I'm really appreciate your help (goes to all of you).

 

[JimB]

The four tones you suggest would be OK, but not optimum for using the available bandwidth.

The idea of using a variable duty cycle (Pulse Width Modulation) is OK but not in this case, the radio is designed the transmit audio frequencies (voice), not DC levels (a variable MS ratio pulse train).

Are you transmitting Asynchronous data from a computer serial port or just a collection of 8 bits?

JimB

 

[Nigel Goodwin]

A waveform with a 10% duty cycle does not translate to 10bit digital data!

You are right, I don't know what I was thinking...

Regarding "harmonically related tones", again you are right, I didn't think of it.

So, if 4kHz is too high a frequency, 4 good frequencies to use would perhaps be 2000Hz, 2300Hz, 2600Hz and 2900Hz? Or are these in harmonics? There has to be a big gap between the frequencies so the decoder can easily distinct them from each other.

Perhaps I could use 4 duty cycles instead of 4 frequencies, and the decoder (PIC processor) could measure the duty cycle and determine the command or bit from that. How about that?

NO! to it all!.

As already suggested above, you are approaching this totally wrong, use a modem for data communication over a voice link - it can be as simple as one tone for zero, and another (non-related) tone for one. There are specific reasons it's done like that!, you can find specific frequencies used from old modem specs.

Also, you are probably breaking your PMR licencing conditions by transmitting data on your voice system, you should check this first of all!.

 

[Futterama]

Well, Nigel, I have read alot about the rules and regulations, and found nothing thats says that I can't send "home composed music" (data tones) through a PMR.

And if all my suggestions are wrong, please direct me in the right direction, give me some hard facts I can use (eg. a link to modem spec or something).

JimB, I'm sending 8bit from a microprocessor (PIC).

I think I'll experient a little, and return when I got more insight in the systems capabilities.

Regards,
Futterama

 

[audioguru]

Some geeks consider the sound of data to be "music to their ears". I don't think the bearurocrats would believe it.