• The forum software that supports hummy.tv has been upgraded to XenForo 2.3!

    Please bear with us as we continue to tweak things, and feel free to post any questions, issues or suggestions in the upgrade thread.

Interesting Items...

what is necessary is to sample at 2x the expected input bandwidth.
Click to expand...
Is that not the same thing? Or are you implying that the sample rate must be 4 times the modulating frequency, as a carrier modulated with 1kHz has a B/W of two kHz, so your implication is that you need a 4kHz sample rate?
 
You are confusing sampling with modulation. Nyquist is about sampling. Fourier is about how a spectrum spreads when a modulation is applied to a carrier frequency. Shannon is about the theoretical limits on how much information can be transmitted in any particular bandwidth. Modulating a carrier is not the same as sampling the modulation signal at a rate equivalent to the carrier frequency.

You could sample a signal at a sampling frequency 2f, and this would be sufficient to resolve the frequency of the signal as long as it was within ±½f of a known base frequency. If the signal is a modulated carrier wave, the modulation can be extracted as long as its maximum frequency is f (this is how direct digital receivers work).
 
Black Hole said:
A 1100Hz sine wave, sampled at 1ms intervals, will look like 100Hz - and a 100Hz sine wave is a solution to those samples - but if you know that the input is in the frequency band 1000-1500Hz, a solution can be found in that band at 1100Hz - so it is not absolutely necessary to sample at at least 2f, what is necessary is to sample at 2x the expected input bandwidth.
Click to expand...

Oscilloscopes that display waveforms purporting to be extremely high frequencies often work this way. The same trick can also be used for looking at eye diagrams when transmission line testing cables for waveform distortion.
 
Yes indeed, such oscilloscopes reconstruct an extremely fine resolution waveform of a repeating signal by taking samples at increasing times through successive cycles rather than displaying the whole of one cycle (allowing time to convert the analogue sample into a digital value before then going to fetch the next sample). This can only work when successive cycles are identical (although a variation in the waveform gets displayed as a smeared trace and is also useful information to the engineer).

Despite the fact that these 'scopes are not digitising the signal at full bandwidth (different use of the term), it still staggers me how short a sample window is required to capture an analogue value to pass on to the ADC, and how precise the timing control is for ramping the sample time through the cycle.
 
I should imagine that a digital signal at a certain bit rate would have a pretty well defined bandwidth and the only problem would be to find the maximum bit rate at which the signal can be encoded and decoded reliably.
 
Mike0001 said:
I should imagine that a digital signal at a certain bit rate would have a pretty well defined bandwidth and the only problem would be to find the maximum bit rate at which the signal can be encoded and decoded reliably.
Click to expand...
That makes no sense to me whatsoever.
 
I'm aware of that, but when you said "has a pretty well defined bandwidth" you presumably meant "requires a pretty well defined bandwidth".
 
When you said what I wrote made no sense whatsoever, you presumably meant that you thought, erroneously, that one word was wrong?
 
Until you supplied further context, I couldn't work out what was wrong. Clearly a digital signal has a very wide and undefined bandwidth. Maybe that particular symbol got corrupted by transmission noise.
 
I wasn't thinking the digital signal would be transmitted unencrypted! I don't know what the bandwidth is when it rides on a phase-shifted carrier sine wave as I am not an expert on this.
 
Last edited by a moderator:
What the heck are you talking about now? By "symbol", I was making an oblique reference to your choice of the symbol "has" in your message stream, as a noise-induced corruption from "requires" (the noise presumably having been introduced between your brain and your typing).

In the proper context of this conversation, symbol refers to the encoding (not encryption) of a number of data bits into one combination of phase and amplitude - thus in QAM-16, one symbol represents 4 data bits.
 
Please identify the disrespect; I can't see where I have been any less respectful of you than you are of me, or where I have been in any way heated. Anybody else agree?
 
Another article in Electronics Weekly, 25th February:

ARM have integrated digital oscilloscope functions onto a processor chip so that they can investigate power rail disturbances within the chip itself. It uses the supply rail to modulate the frequency of a ring oscillator, and measures the phase change. It has a 2k sample buffer and can trade input resolution against timebase: 2.25Gs/s with 10.8mV resolution, down to 400Ms/s with 1.6mV resolution.
 
I'm still waiting for some evidence to back up the assertion of heated disrespect. You don't make that kind of statement and get away with it, particularly not when it's against me!

Is the "disrespect" simply that I happen to disagree with you and you don't like it? Get real.
 
I'm beginning to wonder whether I should join Twitter to see some really good insults thrown about and start watching Jeremy Kyle.
 
You must log in or register to reply here.
Back
Top