On 12/11/2014 3:02 PM, rickman wrote:
> On 12/11/2014 1:40 PM, Phil Hobbs wrote:
>> On 12/11/2014 1:26 PM, rickman wrote:
>>> On 12/11/2014 12:28 PM, Phil Hobbs wrote:
>>>> On 12/11/2014 11:35 AM, rickman wrote:
>>>>> On 12/11/2014 11:04 AM, Phil Hobbs wrote:
>>>>>> On 12/11/2014 10:55 AM, John Larkin wrote:
>>>>>>> On Thu, 11 Dec 2014 09:44:20 -0500, Phil Hobbs
>>>>>>> <hobbs@electrooptical.net> wrote:
>>>>>>>
>>>>>>>> On 12/10/2014 11:26 PM, rickman wrote:
>>>>>>>>> On 12/10/2014 5:37 PM, Phil Hobbs wrote:
>>>>>>>>>>
>>>>>>>>>> Say you're using a 400 MHz, 48-bit DDS to make 10 MHz,
>>>>>>>>>> using a phase increment M = 7036874417767. Being an
>>>>>>>>>> odd number, M is relatively prime to 2**48. Thus the
>>>>>>>>>> sequence of phase accumulator values will not repeat
>>>>>>>>>> for M cycles of the output. This requires 703687.4
>>>>>>>>>> seconds which is more than 8 days, despite an apparent
>>>>>>>>>> periodicity of 100 ns.
>>>>>>>>>>
>>>>>>>>>> The DAC values may repeat more often than this, or
>>>>>>>>>> very nearly repeat (which is what Gerhard was talking
>>>>>>>>>> about on time-nuts) but there is the potential for DAC
>>>>>>>>>> nonlinearities and slewing effects to produce phase
>>>>>>>>>> and amplitude perturbations on time scales of hours to
>>>>>>>>>> days.
>>>>>>>>>>
>>>>>>>>>> Since time and frequency can be measured to absurd
>>>>>>>>>> accuracy, it's quite possible to get easily
>>>>>>>>>> measureable phase errors at surprisingly long time
>>>>>>>>>> scales.
>>>>>>>>>
>>>>>>>>> I really don't get your point. To make a 10 MHz sine
>>>>>>>>> wave from a 400 MHz clock, the phase step will be M/400
>>>>>>>>> where M is the modulus. So your phase increment *can't*
>>>>>>>>> be prime relative to the modulus. Where did you get the
>>>>>>>>> numbers you are working with?
>>>>>>>>>
>>>>>>>>> I have no idea why you are talking about the DAC. If
>>>>>>>>> you are referring to the values fed to the DAC repeating
>>>>>>>>> more often than the phase values, then you are talking
>>>>>>>>> about truncation of the phase values which *is* where
>>>>>>>>> spurs come from.
>>>>>>>>>
>>>>>>>>> So what was your point?
>>>>>>>>>
>>>>>>>> Forget it.
>>>>>>>>
>>>>>>>> Cheers
>>>>>>>>
>>>>>>>> Phil Hobbs
>>>>>>>
>>>>>>> Rickman just wants to argue and insult people. Better to
>>>>>>> ignore him.
>>>>>>>
>>>>>>>
>>>>>> I don't mind arguing, but arguing without plugging in the
>>>>>> numbers to try to understand the example is, *ahem*,
>>>>>> unproductive.
>>>>>
>>>>> So why don't you want to discuss a real example then? You
>>>>> picked three numbers that were not possible, they don't add up...
>>>>> or divide up actually. Do you wish to pick some real numbers?
>>>>> I would prefer to use numbers I can work with on my calculator so
>>>>> a modulus of less than 2^32 would be better.
>>>>
>>>> Get a better calculator. ;)
>>>>
>>>> c:\>mc2 2**48/7036874417767
>>>>
>>>> 39.9999999999965894
>>>>
>>>> That will make 10 MHz with a 48-bit, 400 MHz DDS such as an
>>>> AD9956.
>>>>
>>>> All I need for the argument is that the increment is relatively
>>>> prime to 2**48.
>>>>
>>>> What part don't you like?
>>>
>>> Ok, let's ignore that this does *not* produce 10 MHz. I guess it is
>>> close enough. 10.000000000000900E+00 according to my spread sheet.
>>>
>>>>>>>>>> potential for DAC nonlinearities and slewing effects
>>>>>>>>>> to produce phase and amplitude perturbations on time
>>>>>>>>>> scales of hours to days.
>>>
>>> DAC irregularities are not something that the DDS is about really.
>>
>> The DAC is on-chip. We're talking reality here. You might want to go
>> read a couple of datasheets, e.g. that very nice AD9956 under discussion.
>
> Yes, in a DDS chip the DAC is on most DDS chips, but the fact remains
> that it is not an inherent part of a DDS function. There are digital
> DDS functions which are all digital such as the one discussed previously
> in the Timing Solutions design. Their patent even says it included a
> DDS, but it does not feed a DAC, it feeds a multiplier to do the mixing.
>
> If you want to discuss imperfections of DACs then we can do that, but
> don't blame them on the DDS circuit that feeds it even if it is on the
> same chip. I don't think this is semantics. I don't see any value in
> mixing the three causes of spurs.
>
>
>>> That is an issue of the DAC in question, no? I'm not sure the time
>>> scale would need to be days or whatever.
>>
>> I just calculated that it could be up to a week, with very plausible and
>> typical values. You could make it as long as months if you picked some
>> odd increment closer to 2**47 (200 MHz).
>
> You are talking about the cycle time of the digital pattern. But this
> pattern does not create spurs. All of these numbers are exact.
>
>
>>> I expect they will be more related to the input clock period.
>>
>> You're weaselling. As I said, it's an analogue issue, because the ADC
>> waveform won't repeat exactly (even in the spherical cow universe) for
>> more than a week, so you can get spurs that are only a few microhertz
>> from the carrier.
>
> First, why not dispense with the personal comments, ok? No need to call
> names. If you don't want to discuss this, then let's stop.
>
> When you say the waveform won't repeat "exactly" how does that cause
> spurs? You mean the digital words won't repeat exactly, but that is
> irrelevant. All that matters is that they don't have error.
>
> Any spurs you see at the output of the DAC are due to aliasing from the
> sampling or imperfections from the DAC. A good enough anti-alias filter
> should reduce the anti-alias problem as needed. Of course an anti-alias
> filter can cause other issues you may not like.
>
>
>>> Regardless this comes from the fact that it is producing an analog
>>> signal. You have the same problem in using amplifiers or any other
>>> analog component.
>>
>> No, because there isn't the periodicity issue.
>
> Irrelevant. Amplifiers have distortion just as any analog part does.
> Periodicity is not important to that. Distortion in the amplifiers will
> be related to the signal frequency. Distortion in the DAC will be
> related to the signal frequency as well as the clock frequency since the
> clock shows up in the signal until it passes through the anti-alias filter.
>
>
>>> Is that what you wish to discuss, the DAC caused spurs? I didn't
>>> get this from the referenced discussions, but then they were all
>>> over the map with a lot of imprecise language, so hard to tell.
>>
>> As I said above,
>> >>>>>>>> The DAC values may repeat more often than this, or
>> >>>>>>>> very nearly repeat (which is what Gerhard was talking
>> >>>>>>>> about on time-nuts) but there is the potential for DAC
>> >>>>>>>> nonlinearities and slewing effects to produce phase
>> >>>>>>>> and amplitude perturbations on time scales of hours to
>> >>>>>>>> days.
>> >>>>>>>>
>> >>>>>>>> Since time and frequency can be measured to absurd
>> >>>>>>>> accuracy, it's quite possible to get easily
>> >>>>>>>> measureable phase errors at surprisingly long time
>> >>>>>>>> scales.
>>
>> I'm not sure what you find unclear about that.
>
> Please separate the DAC distortion effects from the digital issues. How
> would the DAC values (I assume you mean the sine values) repeat more
> often than the phase values unless the phase is being truncated? If the
> phase is truncated that *will* produce errors which produces spurs.
>
>
>>> I'll summarize this again for clarity. There are three sources of
>>> errors (spurs) in a DDS. Errors from phase truncation, limited
>>> amplitude resolution and if you are using a DAC to produce an analog
>>> signal, DAC related errors.
>>>
>>> The phase truncation errors are easily avoided by not truncating the
>>> phase which limits the producible frequencies. However, there are
>>> still many ways to produce the frequency of interest. For all DDS,
>>> or digital circuits for that matter, the frequencies must be related
>>> by integer ratios - "rational" ratios. But if you limit your design
>>> to say, modulus of 2^N, you further limit the possible output
>>> frequencies.
>>>
>>> Limited amplitude resolution is also an inherent feature of digital
>>> designs. However, the spurs produced can be made arbitrarily small
>>> by using larger word sizes. No design, analog or digital, is
>>> without spurs. In digital solutions the spurs are easily
>>> controlled.
>>>
>>> DAC related errors are not really digital issues since they are from
>>> the analog portion of the DAC and depend on the details of the
>>> design and implementation. If you need an analog output they are
>>> necessary. But they are analog components and the issues involved are
>>> largely analog. Just as is true with any analog component there are
>>> methods of design that can minimize these errors but they are indeed
>>> inherent and can not be made arbitrarily small.
>>>
>>> In the end - you can eliminate phase generated errors altogether,
>>> amplitude related errors can be reduced to any extent you need while
>>> DAC errors can be minimized, but not to arbitrarily small levels.
>>
>> If you throw away almost everything that makes the DDS attractive in the
>> first place, and reduce it to a divide-by-N counter, you can get rid of
>> all the spurs that aren't harmonics of the output frequency, sure. But
>> then why not use a frequency divider and save $20?
>
> You are over simplifying the DDS. A DDS without spurs is not a divide
> by N counter.
>
> I can't tell if you don't understand, or *what* you don't understand or
> if you are trying to understand what I am posting. You don't seem to
> point to anything I say to show where it is faulty. You just disagree
> and post something that is not really relevant. Are you trying to get
> what I am saying? Should we continue or is this just an argument at
> this point?
>
The distinction between periodic and aperiodic errors is the distinction
between a spur and the noise floor.
You're moving the goal posts, or maybe you really don't understand my
point here, so I agree we should just let it drop.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
hobbs at electrooptical dot net
http://electrooptical.net