EEVblog Electronics Community Forum
Electronics => Beginners => Topic started by: metrologist on August 29, 2019, 07:26:46 pm
-
Is it incorrect to specify sample rate in terms of Hz?
Sampling Techniques: Real-time / Equivalent Time / Statistical Sampling
Continuous Sample Rate: 100 MHz
Effective Sample Rate: 10 GHz
Or is MS/s and GS/s a more correct way to specify this? Can any supporting discussion around that also be provided?
Context: https://boonton.com/Portals/0/Products/Real%20Time%20Power%20Sensors/RTP5000/RTP_8PG_2019.pdf
Thank you.
-
1Hz = 1/s so technically correct :)
Also common in audio, but S/s is more common in scopes to avoid confusion with bandwidth.
BTW, what the fuck is one megasiemens per second? :-DD
-
BTW, what the fuck is one megasiemens per second? :-DD
Never heard it before but it doesn't sound to comfortable :horse:
-
There's no convention for samples/second in the SI system, but both S (siemens) and s (seconds) are already reserved.
My favourite would be to write it in full: Gsamples/second. That way, there's no ambiguity, and that's what communication is about :)
-
Yes, I meant Sa/s
But there was a question on if there could be two samples per Hz due to rising/falling edge sampling.
-
Yes, I meant Sa/s
But there was a question on if there could be two samples per Hz due to rising/falling edge sampling.
I see no ambiquity. If the sampling was done on the rising and falling edge of a 1MHz clock, then the sample rate would be 2MHz, or 2M samples per second.
-
The ambiguity is whether the 100 MHz sample rate specified in the datasheet could mean 200 MSa/s. I play more in O-Scope land and generally see samples per second. There is the following mention in the user guide:
The RTP5000 Real-Time Power Sensors have exceptional trigger stability of less than 100 ps trigger jitter
regardless of the trigger source which yields much greater waveform detail because a stable trigger point
yields a stable waveform. Using dedicated trigger circuitry rather than software-based triggering provides
precise timestamping of relative trigger-to-sample delay. This precision permits the use of random
interleaved sampling (RIS) for repetitive waveforms with resulting effective sampling rate of 10 GS/s which
permits accurate, direct measurement of fast timing events without requiring interpolation between
samples.
I appreciate the dialog. Thanks
-
1Hz = 1/s so technically correct :)
Also common in audio, but S/s is more common in scopes to avoid confusion with bandwidth.
BTW, what the fuck is one megasiemens per second? :-DD
I sort off get the joke . A Siemen is a measurement of electric conductance, electric susceptance, and electric admittance.The reciprocal of resistance, reactance, and impedance.
Also called a mho.
-
The ambiguity is whether the 100 MHz sample rate specified in the datasheet could mean 200 MSa/s.
There is no ambiguity, because 100 MHz sample rate equals to 100 MSa/s and equals to 100 M cycles, which is equals to 100 MHz.
200 MSa/s is 200 MHz sample rate and means 200 M cycles per seconds, which means 200 MHz.
-
The ambiguity is whether the 100 MHz sample rate specified in the datasheet could mean 200 MSa/s.
There is no ambiguity, because 100 MHz sample rate equals to 100 MSa/s and equals to 100 M cycles, which is equals to 100 MHz.
200 MSa/s is 200 MHz sample rate and means 200 M cycles per seconds, which means 200 MHz.
What if a sample is taken on both the rising and falling edge? I guess your 50 MHz sampling clock would produce a 100 MHz sample rate, which would still be 100 MSa/s. It was more of a semantic question of whether sample rates are properly specified in Hz.
-
When the re-naming of cps = cycles/second to Hz was first proposed, a conservative German scientist objected “One might as well name the liter/minute the Falstaff.”
-
Funnily enough, there is becquerel - the count of decays per second in a radioactive material.
-
BTW, what the fuck is one megasiemens per second? :-DD
I sort off get the joke . A Siemen is a measurement of electric conductance, electric susceptance, and electric admittance.The reciprocal of resistance, reactance, and impedance.
Also called a mho.
i dont get the joke, but i'm not too serious either. if unitless (count or cycles) per sec is Hz, any unit per sec, is the rate of change, such as m/s, kmph, dT/dt (temperature), V/us. hence anything that change conductivity with delta 1Meg in one sec is one megasiemens per second. i'm not sure if such material exists though.
-
I often see Msps or Gsps in oscilloscope specs. Not a standard, but conveys the meaning.
-
What if a sample is taken on both the rising and falling edge?
You only know if you are on a falling or rising edge when sampling more often than the signal changes (Nyquist-Shannon) and what causes the sample to be taken is the elapse of time, without regard for the signal.
Triggering requires either software with enough samples to be taken or separate hardware.
-
What if a sample is taken on both the rising and falling edge? I guess your 50 MHz sampling clock would produce a 100 MHz sample rate
if sample is taken on both rising and falling edge, then sample rate Sa/s and sample frequency Hz will be still equals.
For example if you use clock source 50 MHz and use rise and fall edges to produce samples, then you will get 100 MSa/s = 100 MHz = 100 M cycles per second sample rate. Nothing confusing :)
Sample rate means how many samples per second is produced. If you put strobe for each sample, you will get square wave with frequency which is exactly equals to sample count per second. And it doesn't matter what is source oscillator frequency.
-
What if a sample is taken on both the rising and falling edge? I guess your 50 MHz sampling clock would produce a 100 MHz sample rate
if sample is taken on both rising and falling edge, then sample rate Sa/s and sample frequency Hz will be still equals.
For example if you use clock source 50 MHz and use rise and fall edges to produce samples, then you will get 100 MSa/s = 100 MHz = 100 M cycles per second sample rate. Nothing confusing :)
Sample rate means how many samples per second is produced. If you put strobe for each sample, you will get square wave with frequency which is exactly equals to sample count per second. And it doesn't matter what is source oscillator frequency.
I agree with radiolistener. No ambiguity exists unless the reader chooses to add unnecessary confusion to a simple topic. Sampling at 100 MHz means sampling at "100 Million per second" since the "Hz" unit is literally "per second". Whether or not a DDR clock is used internally is moot. I am absolutely certain that my oscilloscope, which can sample at 16 GHz real time, does not have any 16 GHz clock internally. It actually samples on multiple phases of a slower clock.
In other words: Hz, MHz, GHz, can specify rate of anything, not just a clock signal.