IET
Decrease font size
Increase font size
Topic Title: Analogue or digital
Topic Summary: comparing an analogue integrator with a digital solution
Created On: 30 April 2010 04:07 PM
Status: Read Only
Linear : Threading : Single : Branch
Search Topic Search Topic
Topic Tools Topic Tools
View similar topics View similar topics
View topic in raw text format. Print this topic.
 30 April 2010 04:07 PM
User is offline View Users Profile Print this message


Avatar for TCSC.
TCSC

Posts: 173
Joined: 25 April 2007

Apologies if this is a discussed question. I have been away in the wilderness for many years.

If you use an op-amp integrator on the signal from a tacho generator (velocity), you can produce a voltage propotional to how many revolutions the tacho generator has turned (distance). It is quite instructive and fun to do.

However you have to deal with input bias and leakage currents etc. introducting erronous outputs.

Instead it would appear to be much simpler to use a pulse transducer giving a pulse every few degrees of rotation, and simply clock a counter. The output of the counter is then converted from D to A giving a similar analogue display of distance. There would be no problems of errors or drift etc.

It seems to me that for any sort of simple integration of a signal, a digital circuit would be more appropriate than the analogue. So my question is, - Are there any situations where an analogue integrator would be used instead of a digital one? Assuming that resolution or bandwidth isn't a limiting factor eg say you only wanted audio bandwidth and an analogue meter display.
Cheers
 30 April 2010 07:32 PM
User is offline View Users Profile Print this message



jencam

Posts: 608
Joined: 06 May 2007

Originally posted by: TCSCAre there any situations where an analogue integrator would be used instead of a digital one?


Here is an answer from my son.

An analogue integrator is used in situations where a true mathematical integration of a signal is required. The signal being integrated is usually a continuously varying voltage or a train of pulses where the voltage and the duration of the pulses is likely to vary. In the case of the tacho generator, the number of pulses in a given timeframe is of greater importance than the voltage and duration of the pulses, so a pulse counter can be used instead of an integrator. Pulse counters replaced integrators in the vertical synchronisation circuits of some televisions during the early 1980s as they eliminated problems caused by component drift.

It is possible to create a digital integrator using an A/D converter then performing the integration mathematically. The performance of these are subject to quantisation error and Nyquist.
 05 May 2010 09:51 AM
User is offline View Users Profile Print this message



amillar

Posts: 1918
Joined: 28 May 2002

Simplistically: If you want an accurate integration (step size and sampling apart) do it digitally. If you want a simple integrator do it in analogue.

As you say, problems of bias and leakage currents, and also component drifts and non-linearities, make analogue integrators a bit of a nightmare. But practically an integrator will always form part of a larger system, so you tend to choose the technology to match the rest of the system: it is hard to image a digital control system that would exit into analogue just to implement an integrator (although it may make sense the other way around).

The last time I used an analogue integrator* was nearly 20 years ago, to correct the DC offset in an audio system (to avoid the use of coupling capacitors), however this was inside a feedback loop so the exact parameters and stability were not critical.

*Of course a low-pass filter is actually an integrator, but I think we know what we mean by treating them seperately...

-------------------------
Andy Millar CEng MIET CMgr MCMI

http://www.linkedin.com/in/millarandy

"The aim of argument, or of discussion, should not be victory, but progress." Joseph Joubert
 05 May 2010 09:24 PM
User is offline View Users Profile Print this message



jencam

Posts: 608
Joined: 06 May 2007

Originally posted by: amillar
The last time I used an analogue integrator* was nearly 20 years ago, to correct the DC offset in an audio system (to avoid the use of coupling capacitors), however this was inside a feedback loop so the exact parameters and stability were not critical.


Do analogue sample and hold circuits belong in the history books?
 07 May 2010 12:33 PM
User is offline View Users Profile Print this message



amillar

Posts: 1918
Joined: 28 May 2002

As far as I know they are still a key part of many (most?) A-D convertors

-------------------------
Andy Millar CEng MIET CMgr MCMI

http://www.linkedin.com/in/millarandy

"The aim of argument, or of discussion, should not be victory, but progress." Joseph Joubert
 07 May 2010 04:35 PM
User is offline View Users Profile Print this message



saridgway

Posts: 148
Joined: 07 May 2002

Originally posted by: TCSC

Instead it would appear to be much simpler to use a pulse transducer giving a pulse every few degrees of rotation, and simply clock a counter.


Provided the clock is free from noise (not always easy to achieve, as I was reminded when using a similar approach in a 7kW searchlight pan and tilt system a few years ago).

-------------------------
Steve Ridgway MIET
 10 May 2010 10:12 AM
User is offline View Users Profile Print this message


Avatar for TCSC.
TCSC

Posts: 173
Joined: 25 April 2007

Thanks guys. Interesting and useful answers.
Statistics

See Also:



FuseTalk Standard Edition v3.2 - © 1999-2014 FuseTalk Inc. All rights reserved.