Reader question: Shouldn’t the (effective) bandwidth of the ADC or DAC also be reported in the titles?


Ameya Bhide posted a question in the Q&A section. Ameya wrote:

Hi Bengt,

After being lost in the world of Converters for a while, I always have this fundamental question as follows:-

Industries, Publications always talk about GSamples/sec of an ADC or a DAC and the higher the rate, better the converter is. But very few talk about the Bandwidth in the title of the ADC. Even in publications, I always see the title as ” An XGsamples/sec ADC” where as I think it should be “An X Gsamples/sec Y Mhz BW ADC” since many of these converters do not retain performance up to Nyquist.

Even when looking at the TI ADC’s I need to search the data sheet to find the real bandwidth of the ADC. Shouldn’t the sample rate and BW be always reported together so that one quickly understands what the capabilities of the ADC are?

Could you comment on this?

Thanks,
Ameya

Good question Ameya, and one that touches on a broader topic (scientific reporting practices) that I’ve been planning to bring up for a while. This is a good starting point. Perhaps your question was mostly concerning data sheets? Either way, I’d like to broaden it to include scientific papers as well.

So blog readers, what do you think?

  • Is there an inflation in paper/data-sheet titles?
  • Should we require paper titles to more accurately describe the actual bandwidth of the ADC?
  • Is effective resolution bandwidth (ERBW) the best measure?

If there is an inflation, and a practice to boost performance in the titles, what will happen to the pioneers which put more complete (potentially less impressive) information in the titles? Rejected papers and loss of business? Is the target audience really that gullible? Don’t they read the content of a data sheet or paper? How long can a paper/data-sheet title be 😉 … etc.

Advertisements

4 responses to “Reader question: Shouldn’t the (effective) bandwidth of the ADC or DAC also be reported in the titles?

  1. Hi Bengt,

    Thanks for hitting the nail on the head and bringing up the exact questions that I had. My question is directly towards publications as well as commercial ADC’s. More towards publications for now as I am a PhD student :-). I always been afraid to ask this question since ADC’s as a topic have been studied a lot.

    Thanks,
    Ameya

  2. You’re welcome!
    Personally I believe that “Honesty is King” here. Reading or skimming 1600+ papers, I’ve seen a few variations, and there are few things as off-putting as paper content not living up to the paper title. A smashing title might grab your attention at first, but if reading the paper leaves you with a sour taste in the mouth, what did the authors gain? Now I will remember them as “title-boosters” instead and they could have been better off not tempting me to actually read the content 😉

    That said, it is difficult to capture the complete performance in just the title. That’s why papers also have body text.

  3. One particularly nasty trend I’ve observed is “coarse rounding”, and even truncation. Usually in connection with FOM claims. A fictive example:

    The ADC performance give a certain FOM = 2.4 of some unit. Authors round it to one (!) digit, which gives them a FOM = 2 (that’s a 17% rebate). Then they claim FOM = 2 in the title. Magically, no one ever rounded a FOM = 2.6 to one digit [if lower FOM is better]. But in a similar case I’ve seen truncation applied at least once, and that’s just plain wrong.

    You can see you really got me going here 🙂

    Because of the above, I feel that no reviewer should accept that authors report real-valued (continuous) performance measures at 1-digit precision. Particularly not the parameters for which the scientific competition is the most fierce (i.e. FOM). Can you imagine running 100m at the Olympics and using a stopwatch with only one digit precision? It would be completely ridiculous.

  4. To calm myself down a bit, and to return to your questions, I think that ERBW does not have to be in the title unless the performance degrades much faster over the Nyquist band than normal, in which case it could be mentioned in the title. A certain performance roll-off is normal, and not necessary to mention.

    If the ENOB is several bits lower than nominal resolution N, I think it should be included in the title, or the very least in the abstract.

Comment

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s