Will a DDS generate harmonics?

[neazoi]

Will a DDS generate harmonics?
Lower, higher, how many?

I design a good spectral purity oscillator and I would like to make rought comparisons to a dds http://neazoi.com/singlespanosc/index.htm

 

[FvM]

You have answered the same question before https://www.edaboard.com/threads/258732/
(and got reasonable answers)
Please append follow-up questions to the previous thread.

 

[drkirkby]

Will a DDS generate harmonics?
Lower, higher, how many?

I design an good spectral purity oscillator and I would like to make rought comparisons to a dds http://neazoi.com/singlespanosc/index.htm

IF you have designed an oscillator with good spectral purity, then you would know the answer to your own question.

Dave

 

[neazoi]

Oh I am sorry! you may close this thread

 

[biff44]

Harmonics, spurs, a whole picket fence worth of them! You need to be able to tolerate them to use a DDS.

 

[vfone]

The question looks more like an advertise to your project, which at some point is ok for me. But (as somebody mentioned above also) you have to know the answer from the moment you put in the title of your project a rhetoric question (without the expectation of a reply) regarding DDS.

 

[neazoi]

No it is not, No need to advertize, no adds on my site etc. I usually publish the best projects in local magazines etc.
With the excellent help of the edaboard members I make sure things I make are correct, they drive me in the safe way.
And as the title says "A better alternative to DDS?"
That is a question actually before I make sure

No need to keep this thread any more actually opened by mistake.

 

[volker_muehlhaus]

And as the title says "A better alternative to DDS?"
That is a question actually before I make sure

"Better" depends on your requirements. DDS is great if you need small frequency steps. DDS is not so great with spectral purity close to the carrier.

The mixer topology that you have suggested is ok. It has some advantaged and disadvantages. We had used it with a project many years ago (before DDS and fractional N synthesis) for a wideband synthesizer 100kHz - 200MHz, 10kHz step. In our case, we actually had two VCOs instead of one VCO + one fixed frequency oscillator. By stepping one VCO in 90kHz steps and the other in 100kHz steps, we could get 10kHz steps in the mixed signal without having to go down to 10kHz PFD frequency. But still, phase noise was not so great both in the loop (-85dBc/Hz @ 10Khz) and outside the loop bandwidth.

With your approach, frequency stability and phase noise are detemined by the free running MC1648 VCO. Have you checked if that VCO phase noise performance is acceptable? Have you calculated the expected frequency stability?

 

[neazoi]

"Better" depends on your requirements. DDS is great if you need small frequency steps. DDS is not so great with spectral purity close to the carrier.

The mixer topology that you have suggested is ok. It has some advantaged and disadvantages. We had used it with a project many years ago (before DDS and fractional N synthesis) for a wideband synthesizer 100kHz - 200MHz, 10kHz step. In our case, we actually had two VCOs instead of one VCO + one fixed frequency oscillator. By stepping one VCO in 90kHz steps and the other in 100kHz steps, we could get 10kHz steps in the mixed signal without having to go down to 10kHz PFD frequency. But still, phase noise was not so great both in the loop (-85dBc/Hz @ 10Khz) and outside the loop bandwidth.

With your approach, frequency stability and phase noise are detemined by the free running MC1648 VCO. Have you checked if that VCO phase noise performance is acceptable? Have you calculated the expected frequency stability?

No, at the time I just thought about overall spectral purity. I do not think that phase noise is so significant on HF and lower bands. On microwaves, that is another story. Frequency stability is more important. I hope the FLL will be able to keep this quite stable, the author of the FLL project published on Elektor a few years ago, states a 10Hz accuracy. We will see....

Something that worries me is the fact that I notice fluctuations on the tone of the signal when switching to SSB mode. I have mentioned this again in another thread and I do not know if this is a problem of the varicaps in general or something else.

 

[sky_123]

Phase noise is really significant on HF, because the channels are so closely packed together.
If you are going to efforts to produce a good VCO, then it is a little bit wasted if you do some FLL (I'm guessing this is measuring the frequency with a microcontroller or something?) instead of PLL.

 

[FvM]

I doubt that "FLL" is a commonly used technical term at all. I never heard it before.

But I agree about the performance lacks of this frequency control method. By nature, it will work as a hysteretic controller, not generate an error signal before the frequency measurement is of by one count. Unlike a PLL, it can't be described by a linear model.

 

[volker_muehlhaus]

No, at the time I just thought about overall spectral purity. I do not think that phase noise is so significant on HF and lower bands. On microwaves, that is another story.

Phase noise is extremely important. The amateur radio equipment for HF bands has really tough requirements on phase noise, because this heavily affects received and transmitter performance.

You seem to think of "phase = not relevant", but that is wrong. Phase noise = spectral purity near the carrier, which is most important.

Something that worries me is the fact that I notice fluctuations on the tone of the signal when switching to SSB mode.

That's why low phase noise is important ;-)

 

[neazoi]

Ok here is the FLL, it is usually known as vfo stabilizer. What it does in very brief, it reads the output frequency and compares it with the set (lock) frequency. Then sends the relevant pulses to the vco varicap to correct the vco frequency to match the set one.

The good thing about the mixed oscillator approach is that parts of the circuit can be replaced by lower phase noise ones if desired in the future. It is a modular approach. Filters do not have to be altered is frequencies are kept the same.

I have not found a PLL that could be like 10Hz accuracy or so, on a wide vco bandwidth (48-78MHz), that is why I prefered the "fll". Maybe there exists one?

 

[volker_muehlhaus]

You seem to misunderstand PLL concepts. The PLL will set the frequency to the exact value.

 

[neazoi]

You seem to misunderstand PLL concepts. The PLL will set the frequency to the exact value.

The PLL circuit compares the phase of the input signal with the phase of the signal derived from its output oscillator and adjusts the frequency of its oscillator to keep the phases matched. The signal from the phase detector is used to control the oscillator in a feedback loop. This uses a vco as well, like the FLL. As far as I know the difference is that the FLL locks to the read frequency not the phase.

 

[volker_muehlhaus]

Yes and no. With the PLL, the output signal stability and phase noise (aka spectral purity near the carrier) is determined by the parameters of the reference oscillator, and division ratio. With a good PLL architecture, you can actually improve the phase noise of your VCO by locking it to a clean reference.

As dicussed above, the resulting phase noise is very relevant for your signal quality in practical applications. It looks like your MC1648 is a really bad VCO, and by frequency locking it all the bad phase noise and bad spectral purity will appear in the output signal as well.

 

[neazoi]

Yes and no. With the PLL, the output signal stability and phase noise (aka spectral purity near the carrier) is determined by the parameters of the reference oscillator, and division ratio. With a good PLL architecture, you can actually improve the phase noise of your VCO by locking it to a clean reference.

As dicussed above, the resulting phase noise is very relevant for your signal quality in practical applications. It looks like your MC1648 is a really bad VCO, and by frequency locking it all the bad phase noise and bad spectral purity will appear in the output signal as well.

Do you think something like this will be better?

 

[volker_muehlhaus]

Do you think something like this will be better?

I don't know if this oscillator is better, but you can build better oscillators than the VCO in the MC1648.

But just in general, I think that you concept and oscillator architecture isn't well designed. The signal quality (noise) will be REALLY bad compared to a DDS based solution.

 

[FvM]

I can imagine that the said "FLL" serves a purpose when used with a good (=low phase noise, good short time stable) oscillator to compensate temperature drifts and achieve a convenient tuning. The control loop gain would be designed as a slow PI not to add much phase noise. Associating it with "high spectral purity" as the link in the initial post does, is at least misleading. Their comparision somehow misses the point, I think:

FLL
This is where digital circuits combine with analogue, to stabilize them. A frequency locked loop monitors the output frequency of a VCO and sends the relevant voltage to it, in order to stabilize it, therefore greatly improve stability.

PLL
A phase locked loop is a similar technique to the above that locks to the phase of the VCO to stabilize it. Usually, the range of the oscillator affects the frequency step acuracy, so very wide range accurate PLLs may be difficult to build.

If you want to know about expectable DDS spurs, you should try the Analog Devices tool.

 

[biff44]

FLL, or frequency locked loop, is similar to the PLL, except that frequency is the detector instead of phase. Old timey receivers had FLLs. Some phase noise degeneration circuits can use AC coupled FLL circuitry to clean up phase noise, while a parallel PLL keeps the frequency and phase locked.