conversion and sampling time

Post any questions you may have about our current range of USB data loggers
Post Reply
LarsW
User
User
Posts: 6
Joined: Thu Mar 23, 2006 9:04 am
Location: Lund University

conversion and sampling time

Post by LarsW »

I use ADC-24 and I am have questions about conversion and sampling time:
- Is there a limit on how many channels that can be measured with a certain combination of conversion and sampling time? So that, e.g., 5 channels with 660 ms conversion time and 5 s sampling rate is ok, but using 10 channels with the sample settings will not work (as 660 ms * 10 > 5 s)?
- Are the channels measured sequentially, one after the other?
- If the answer is yes to the above question, then the values are taken at slightly different times? So that one reason for using a short conversion time is to have the different channels measured at almost the same time.
- I guess that for slowly changing signals (essentially DC signals), it is always better to use the longest possible conversion time to reduce noise.

Gerry
PICO STAFF
PICO STAFF
Posts: 1145
Joined: Mon Aug 11, 2014 11:14 am

Re: conversion and sampling time

Post by Gerry »

Hi LarsW,

With our data loggers there is a distinction between sampling interval and conversion time. 'Conversion Time' is hardware related, and is the time taken by the Analogue to Digital conversion process, while 'Sampling Interval' is software related, and is the time in between requests, by PicoLog, for a sample update from the hardware data logger. So, if the Sampling Interval is much shorter than the Conversion Time, PicoLog will be requesting an update too soon, which results in getting a lot of duplicate values from the logger. If the Sampling Interval is too long PicoLog will be missing samples. The ideal would be for both to be exactly the same duration, but in practice this is not possible (as they are not synchronized) so the best ratio is for the Sampling interval to be set slightly shorter than the Conversion Time to minimize the number of duplicate values without missing any values.

Regarding the length of time to perform a conversion, this is purely a restriction due to the physical process of Analogue to Digital Conversion. The more Resolution (detail) that is provided by the conversion process, the longer the process takes to convert the analogue value. That is why in the User Guide on page 11 you will find that as the resolution requested for the converted value by PicoLog increases, so does the conversion time.

Regarding the relative conversion between channels, the Analogue to Digital Converter is multiplexed between the channels, so the channels are converted sequentially. So, you can select shorter conversion times to have all your channels in use converted quicker but, as mentioned in the previous paragraph, you would have to sacrifice some resolution to do so (which may or may not be important to you).

You could reduce noise by using the longest practical conversion time if you were also low pass filtering. However, because you won't be filtering, there are better methods for reducing noise. The real benefit of the slower conversion time, and therefore slower sampling interval will be less redundant data to store and wade through. There are limits on how slow the conversion and sampling should be though, to avoid distorted data due to aliasing effects. It is best to ensure that the conversion time is less than 1/5th of the period for the highest frequency component of interest in the signal. Put another way the conversion time should be less than the fastest rise time of the signal divided by 1.75.

Regards,

Gerry
Gerry
Technical Specialist

Post Reply