PropScope input bandwidth?
RickB
Posts: 395
Has anyone from Parallax actually measured the worst case -3 dB bandwidth of the PropScope? It's not listed in the specifications.
Rick
Post Edited (RickB) : 12/6/2009 9:40:37 PM GMT
Rick
Post Edited (RickB) : 12/6/2009 9:40:37 PM GMT
Comments
Or are you guys hinting at cunning ways to use it above the Nyquist limit?
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
Ah, it's not so easy as it might look at the first moment!
You have two factors:
* Sampling rate ("digital bandwidth"). It should be 5..10 times higher than the expected signal frequency
* Analog bandwidth, should be below the sampling rate
An example with an 1 MHz signal (sinus):
* analog bandwidth 10MHz (signal will pass unchanged)
* sampling rate 1MHz
-> below Nyquist, your reading will show a much lower frequency or anyhow, complete Smile
* analog 10MHz
* sampling 10MHz
-> good reading
* analog 1MHz
* sampling 10MHz
-> -3dB, Vpp is in fact higher then the readout
* analog 100kHz
* sampling 10MHz
-> zero signal, because signal doesn't pass the analog filter
So, an ideal scope has an adjustable analog bandwidth that is always less than the sampling rate. Of course, that is quite a challenge (and I think that doesn't exist). It's enough when the analog bandwidth is less then the highest sampling rate.
It's always good to know what signal is expected and to start with the highest sampling rate.
Nick
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Never use force, just go for a bigger hammer!
The DIY Digital-Readout for mills, lathes etc.:
YADRO
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Post Edited (Leon) : 12/8/2009 7:47:52 PM GMT
Attached is an app note on digital scope phylosophy as far as band
width and sampling rate. A good article.
The input amp acts as the anti alias filter and it is a designers call as
to how close the the nyqust freq they want to get.
For the prop scope 25Ms div 2.5 = 10Mhz bandwidth.
25Ms div 4 = 6.25 Mhz bandwidth (The conservitve value from the article.)
Attached also from my prop scope is a pic of a 1Mhz square Wave.
This is still looking like a square pulse and would indicate 10 plus Mhz
bandwidth.
Tom
-Phil
Leon
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Amateur radio callsign: G1HSM
Anyway this arrangement is what I would expect for normal ADC setups.
What I was really getting at in my query is that in a scope it might be useful to have an analog bandwidth that is greater than the sampling rate. This would of course allow frequencies higher than the Nyquist limit to be sampled and appear as alias wave forms at a lower frequency.
Provided one is aware of what may be in the input signal so as to be able to tell an alias form the real thing this could be of use.
Variable analog bandwidth would be best.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
My Prop Info&Apps: ·http://www.rayslogic.com/propeller/propeller.htm
The input bandwidth on the PropScope is greater than the sample rate. We were debating whether or not to filtering out everything above the Nyquist cutoff, and after some discussion, including input from the forums, we decided not to. There were several people that expressed an interest in reading the aliased samples to cheat and see higher frequency repetitive waves. The biggest factor was that there is no perfect filter, so if we cut off everything above 12.5 MHz, signals would be attenuated as they approach 12.5 MHz. We still wanted the signal to have minimal distortion at 5 MHz, so we opted away from filtering.
-- David Carrier
P.S. Sorry for taking almost four hours to respond; I didn't even notice the thread 'til now! I'm currently busy writing some documentation on the PropScope hardware that I am sure will make everyone happy.
The data sheet for the differential amp that drives the adc shows that, with the feedback resistors chosen, ignoring circuit strays, the bw would be 24 Mhz max. Stray capacitance and inductance will lower that figure. Aliasing and proper sampling techniques could allow display of repetitive waveforms up to that point before dropping off. But all this is just guessing untill Parallax actually measures it with calibrated test equipment and posts the results.
mikedev: What led you to believe that the Propscope could do what a 2236 can do?
Dave: You posted while I was writing. Are you sure about "The input bandwidth on the PropScope is greater than the sample rate".
With the gain selected for the adc driver, it looks to be about the same as the sample rate, or did I screw up the calculation from the datasheet.
It still needs to be measured to be a legitimate specification.
Post Edited (RickB) : 12/9/2009 1:12:27 AM GMT
Thanks for the updade David!
Jim
That sounds about right. The function generator that we have here only outputs up to 20 MHz, but I did set it to sweep frequencies up to 20 MHz and set my oscilloscope in peak-detect mode, to make a quick network analyzer, and used it to check the analog front end. It stated showing some attenuation, probably a little less than one dB, just before 20 MHz, but it was dropping quickly, and I wouldn't be surprised if it hit 3 dB before the 24 MHz mark. I didn't spend to much time characterizing the bandwidth, because once it passes the sample rate, how far it passes it isn't as important, and we don't have the equipment to find the exact value.
-- David Carrier
Good to see you say that. I'm not crazy then. Can't wait to get my hands on a Propscope.
▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
For me, the past is not over yet.