Best way to design a sine wave generator with 3 sync outputs

[Kuehnsen]

Actually I got a problem in designing a good sine wave generator!
I got the folllowing needs:

• 3 syncronized outputs!
  Output-Frequency 1 - 10 MHz
  changeabe in amplitude and phase
  low noise outputs

My task is to build an fast digital configurable sine source with this 3 outputs. Every output has a other amplitude. But all running at the same frequency and are absolutly in phase at the first step. Later i want to setup a little phase differency to one channel.

I does the first step by using the AD7008 but this solution got some disadvantages!
First the 3 chanels are only syncronized by the same resonator.
2nd I can only use a 50MHz resonator, so I got a too many noise voltage on the generated sine wave (only 12 samples ca be take from the LUT)
and third I need a resolution of 14bit by the Output DAC to fine setup a amplitude!

So I thinked about to using an other part with a 200MHz resonator and modulate the amplitude of the sine with an DAC on the reference in!

What did you think is there a easier or better solution for my problem, I didn't thinked about?

Thanks for your help.
Kuehnsen[/url]

 

[echo47]

When you say "resonator" do you mean "clock", such as a crystal oscillator?

Sounds like you forgot to put a low-pass filter on the DAC output. The filter should pass everything up to 10MHz, and block 50MHz and above. Your "too many noise voltage" will disappear.

 

[Kuehnsen]

Re: Best way to design a sine wave generator with 3 sync out

When you say "resonator" do you mean "clock", such as a crystal oscillator?

Sounds like you forgot to put a low-pass filter on the DAC output. The filter should pass everything up to 10MHz, and block 50MHz and above. Your "too many noise voltage" will disappear.

Right I mean a crystal ocillator for instance.

I can't use an low-pass-filter cause after this filter the 3 sources are not in phase!

 

[Kuehnsen]

Re: Best way to design a sine wave generator with 3 sync out

With a 20MHz Low-Pass all is well. But how can I syncronize the 3 sine outputs. It should work over a wide temprature range. Maybe a Delay locked loop? But I want to keep at the simplest solution.

 

[Buriedcode]

Re: Best way to design a sine wave generator with 3 sync out

Heres my two cents:

Hmm, I'm no professional at this, but you really do need some form of low pass filter after the DAC, since the output spectrum will have a big spike at the sampling freuency (50Mhz). The phase problem is down to the design of your filter. You want a filter that has a fixed phase response.

Am I right in saying this is the sort of system you want?:

Microcontroller/PC -> DSP -> DAC -> LPF -> 3 different amplfiers with different gains.

Looking at your problem again, what 'exactly' are you trying to do? I mean, if you just want a sine-wave generator, why not go analogue? That way, you could produce 3 very good sinewaves, with a couple of potentiometers to change the frequency (you could using digital pots if you want to control it with a micro controller). And for getting exactly the right frequency, you could use a frequency counter on its output. That way, the oscillator creating the sine waves doesn't have to be accurate (it does have to be stable though), the frequency counter tells you the EXACT frequency of the wave, so you can vary its frequency be hand until you get the right frequency. Or just run the freqncy counter back into a micro controller and let that controll the frequency using 2nd order feedback.

A link here generates a really good sine wave, but its limited to low frequency (<20Khz), but its gives you some ideas:
https://www.edn.com/archives/1997/061997/13di_01.htm

One I've used:
**broken link removed**

If you're going digital (ie: DAC) then you want to have more than 16 samples per period, ideally over 32. You said you were using 50Mhz to achieve a 10Mhz output, but this means that each sinewave period would have 5 samples. This just ain't enough if you want a decent sine wave. Also, the bigger the difference between the sample period (update time of the DAC) and you frequency output the easier it will be to design a filter after the DAC.

Sorry, I'm rambling, but I just spurt out ideas sometimes

Heres how I would do it:
Microcontroller -> PLL chip -> FIR filter (converts square to sine) -> DAC -> LPF -> 3 separate amplifiers with individual gain controls but with same phase response.

Well, hope this helps in some way. As I said, it really depends on how 'clean' you want the sine wave, how complex you can build the circuit, and how much control you want over the 3 outputs.

Buried(in)Code.

 

[echo47]

Re: Best way to design a sine wave generator with 3 sync out

this means that each sinewave period would have 5 samples. This just ain't enough if you want a decent sine wave

As long as you meet Nyquist (greater than 2 samples per sinewave period), and provide a sufficiently steep low-pass filter, you will have a beautiful sinewave.

Or were you referring to a design in one of those links?

 

[Buriedcode]

Re: Best way to design a sine wave generator with 3 sync out

hi again, I'm not being picky but...

As long as you meet Nyquist..

You are right of course, but in order to get a perfect sine wave, you must heavily attuenuate the harmonics. The Nyquist limit is the theorectical limit, assuming you can have a brick wall filter after the DAC.
In a square wave (which is what you'll get with 2 samples) there are only odd harmonics. But it would be a 'nightmare' to design a low pass filter that has a pass band to 10MHz, and a stopband at 30Mhz with attenuation of +40dB. Even a 10-pole filter wouldn't do it.

I'm currently designing a simple digital filter (for baseband digital data for radio) and I'm using 8 samples per bit, thats 16 samples per sine-wave period. This basically creates a sine wave when you put in 0101010. Did a fft on it Vissim, got all harmonics down to -45dB, with the main component being -0.5dB. So a simple 2 pole filter after this would get them down to acceptable levels.

But it all depends on what these 'sine waves' are going to be used for. A few low-level harmonics aren't really visible in the time domain.

Kuehnsen: If you are doing it in digital, try and make most filtering digital. Since you are varying the output from 1-10Mhz, you want to vary the corner frequency of the output filter accordingly. And if you keep the update rate of the DAC the same for any frequency (1-10MHz) you'll only need one simple RC filter at the output.

I'm done

Buried(in)Code

 

[VVV]

Re: Best way to design a sine wave generator with 3 sync out

I think this is how it could be done:

generate two signals in quadrature, one sine, one cosine.
Then sum them in different proportions to obtain the required phase shift. This allows a 0-90 deg phase shift. If more is needed, you can invert one or both of them before the addition; that will give you 0-360 deg phase shift.

Finally, you have three output amplifiers, each with its independent gain to separately adjust the amplitudes.

 

[Kuehnsen]

Re: Best way to design a sine wave generator with 3 sync out

Thanks for your responses.

I did it digitally, actual I use 3 Direct Digital Synthesis Chips from AD like described before. I need 1..10 MHz Accurrate Output and so my recent solution with 50MHz-Clock is very noisy!

I plan to use an Chip with 200MSample's and so I should get a better solution!

I do it that way cause so I got the most control over the 3 generated sine-outputs!

The biggest problem I'm thinking of is the amplitude of the sine. I plan to use chips with 14bit resolution and want to get the full advantage of them.
With a simple RC-Filter the Sine is not clear enough, thatswhy I will try an coupler with the two complimentary outputs of the DAC as balanced input. So I can get an bandpass!

Thanks for your ideas I still thinking over...

Kuehnsen