Distance measurement !! CTMU ??

mglsoft · Joined: 18 Feb 2008 Posts: 48

How accurately can read time CTMU module of a PIC??
I have to measure time of reflection of a signal with maximum precision of an inch.
You can get to do this module?
_________________
MGLSOFT

Mike Walne · Joined: 19 Feb 2004 Posts: 1785 Location: Boston Spa UK

mglsoft · Joined: 18 Feb 2008 Posts: 48

I have read the datasheet for the PIC and the CTMU module section, but can not find answer to this question, so I put my question here.

The distance to be measured is a few inches (2.5 cm or 1 "), and the timing should give me a measurement of the order of picoseconds, to get that accuracy.
The conversion is simple, the speed of sound and the flyback signal give me away.
That's why we need such precision ...
_________________
MGLSOFT

Ttelmah · Joined: 11 Mar 2010 Posts: 19369

You talk about the speed of sound. So you are thinking about using an ultrasonic transducer?. The CTMU can be used for this, and potentially give very good accuracy, _but_ your pulse train will need to be very short, or it'll still be being sent when the reflection occurs at such a short range You are on the 'edge' of different technologies at this sort of distance. The CTMU is normally used to measre capacitance changes,to give a touch sensor ability. This though is not an accurate distance measure, since other factors (what is apporaching, humidity etc.), wiill cause larger changes. If the other object to be detected, is your choice, then inductive sensing would probably be a better choice at such a short range. Have you looked at AN1375?.

Best Wishes

Mike Walne · Joined: 19 Feb 2004 Posts: 1785 Location: Boston Spa UK

asmboy · Joined: 20 Nov 2007 Posts: 2128 Location: albany ny

SherpaDoug · Joined: 07 Sep 2003 Posts: 1640 Location: Cape Cod Mass USA

The PIC counters all work from a clock source. Look at the PIC data sheet to see how fast this clock can be. The fastest PIC clocks I know of are 40MHz.

The counter can count to a resolution of one clock cycle. A 40 MHz clock has a cycle of 1/40M = 25ns. 25 nanoseconds is MUCH slower than the picoseconds you are asking for. A PIC will never measure to such resolution without a lot of special hardware help.

Now you are also asking for 1" at the speed of sound. The speed of sound varies with material, about 1100'/s in air, 6000m/s in steel, etc.. Assuming you are working in air 1" = 1/(1100' x 12) = 76us. 76 microseconds IS something a PIC can measure easily.

Decide what you want, one inch or picoseconds? What material is your sound going through? Let us know when you have some answers.
_________________
The search for better is endless. Instead simply find very good and get the job done.

Ttelmah · Joined: 11 Mar 2010 Posts: 19369

As a comment, the CTMU, can potentially measure times shorter than the PIC clock.
It measures change of voltage, against time.
You measure one element of the equation:
I ⋅ t = C ⋅ V
If (for instance), you pulse an input for time T, with a known capacitance, and I fixed, then you measure V with the ADC, the time 'estimates' can be an order of magnitude shorter than the PIC clock.
I have measured pulse widths of only a very few nSec using this module. However, there is a huge amount of variability, with you needing to calibrate from a known pulse to get any sort of accuracy.

The big problem here is that using a slow traveling signal, like sound, makes measurement relatively easy, but you are back to just how few cycles of the waveform can be generated. One cycle from a 40000Hz transducer, is about 1/3rd inch long in air. This is why it'll be much easier if there is something attachable to the other side that needs detection. Measuring the change in inductance, as a piece of metal approaches, for example, is relatively much simpler.

Best Wishes

mglsoft · Joined: 18 Feb 2008 Posts: 48

The measured distance will be a minimum of 50 meters and maximum 1.8 km.
The accuracy should keep in the entire range.

The CTMU module type has two trigger inputs that perform start and end of a very precise timing.
I understand that measuring time by the CTMU module, is not related to the clock of the PIC.

The intent is to measure the round-trip time of a signal in AM, using it to measure distances very accurately, this signal will be triangulated with two other similar measurements.

I think this answers several questions that respondents have raised here.
_________________
MGLSOFT

asmboy · Joined: 20 Nov 2007 Posts: 2128 Location: albany ny

Mike Walne · Joined: 19 Feb 2004 Posts: 1785 Location: Boston Spa UK

mglsoft · Joined: 18 Feb 2008 Posts: 48

Thanks for answering.

And to think that some people think that the CCS compiler is toy, if they saw the level of responses, probably would not think the same.
If you believe this myself, probably would not come here, not even to look at this forum ...

Well, I called my answers ...

The physical medium is air, since this distance measuring is positioning as an alternative to RTK GPS systems, which in places like where I live, are having problems with solar storms, and will further worsen.
Regarding changes with the weather, sea relative height, etc., I can see that data and make corrections when necessary calculation.
That does not bother me.

The idea is not new, is issued with radio signal AM (amplitude modulation) at a frequency, the signal received at a receiver at a predetermined distance, forward this PING on another frequency and then read it on the device from which it was originally issued signal.
That's where I need to accurately measure the time spent in air emitted PING, to know the relative distance between these two points.
Using the same process on two points arranged in a triangle, should get a substantial improvement in the measurement of distances that may enable me to work quite accurately when you have no proper RTK signal.

They seem possible to put into practice?

This answers all the questions before they left??
_________________
MGLSOFT

Mike Walne · Joined: 19 Feb 2004 Posts: 1785 Location: Boston Spa UK

I've had enough and giving up.

Mike

asmboy · Joined: 20 Nov 2007 Posts: 2128 Location: albany ny

FvM · Joined: 27 Aug 2008 Posts: 2337 Location: Germany

I agree to the comments saying that the discussed problem isn't essentially a PIC timing measurement
problem. IMHO ps resolution delay measurement is the more simple part of the project.

Regarding CTMU features there's a Microchip AN 1375 discussing it in detail. Timing measurement is said to
achieve "below ns" resolution. You should also review the ECN article quoted therein.

There are two basic limitations of CTMU:
- you won't get better than some 100 ps resolution due to jitter and crosstalk of on-chip signals
- absolute range of the measurement is limited to about 1000 units by the 10-bit ADC

If we ignore the obvious other problems for a moment, a CTMU resolution of 0.5 ns coresponds to 5 cm
distance in a RF or optical round-trip measurement. But the maximum range would be only 50 m.