FRECUENCY MEASUREMENT

[davicente]

Hi all,

I have to measure frecuency inside a 1Ghz pulsed signal, about 500ns wide. I think it is too narrow to try a traditional PLL, but maybe mixing with a DDS until i obtain a continious DC level (ripple should be specified depending on resolution:10Khz) and then reading their registers to obtain the frecuency
Any other ideas, papers, other examples????

Thanks in advance

DAVID

 

[nbucska]

Hi davicente:

<nbucska@pcperipherals DOT com> subj: eng-tips
read FAQ240-1032

 

[VisiGoth]

You have a resolution limit around the reciprocal of the pulse width. For you that sounds like 2 MHz. The CRB limit is SNR and time aperture limited. There should be good formula's around for this, I refer to Steve Kay's book, but any good eference will have it.

Do you want to do this with analog HW? how about an IFM? This is an autocorrelation method, the

http://WWW should

help out. A delay coil, splitter, mixer and scope should do a rough one. Do you have a digitial scope? What is the range of the possible freqencies? that is important for some methods.

If you want 10 kHz and you can ensue that the frequency does not change, you will have to integrate a lot of pulses. I would think you Alan vaarince would inhibit you from reaching 10 kHz accuracy, but 10 kHz resolution is easy, just make up the LSB's!

A gated frequency couter would do this all for you.

 

[davicente]

Hi again,
I can't use an external meter. I need to make a littel module to be integrated into a biger system.
Frecuency is 1GHz and shouldn't vary more than those 10Khz, but you must be prepared to mesurea wider range, some MHz i was thinking about.
Pulse width is from 350ns to 50us.
The most important thing is the 10Khz accuracy in the frequency measurement.
AS digitalizing at 1Ghz is expensive and need a lot of space, i was thinking in a HW module(DDS to mix with the carrier to be measured as DDS can control frequency in quite less than 10 Khz)
Mixing down to digitalize in a lower frecuency and then calculate frecuenccy with fft would be another option??
as pulse width will be random between limits commented above, number of samples will vary and accuracy in fft calculation could make accuracy poor than 10KHz??
I've been looking for measurement methods in the web but i haven't found nothing interesting
thanks again

DAVID

 

[VisiGoth]

Your ideas sounds good except for the DDS. You want a pure clean LO on your mixer. Since you will have a fixed LO it may be easiest to quadrature split the lO by 90- degrees and use 2 A/D. You would make a zero IF receiver. Then digitize with a pair of audio A/D's unless you really really need 1 MHz range. I assume you have a trigger mechanism so you will not have to have a free running detector. You still have an aperture limit problem on determining 10 kHz on a narrow pulse. For instance, what if your pulse was only one or two samples wide? Just what frequency would you attribute that to? If you have no side information on the amplitude, then you will have to do a frequency estimate. If you have a 250 ns pulse, how much of that will get filtered in your mixer and A/D Nyquist filter? Can you show us your CRB limit on the frequency estimator vs PW and SNR? You might be suprised on how it relates to 10 kHz.

 

[davicente]

Hi VisiGoth
What i am thinking is to make the DDs to sweep its frecuency in order to mix it with the fix frecuency I have to measure.In theory, when both frequncies are pretty the same, I will obtain from the mixer something near a DC voltage, when this happens, DDS should stop sweeping and a I can read DDs frequency register.
Pulse width will be 350ns as minimum and 60us as maximum
with this architecyute i will not need to digitalize

I hope this will hekp you to understand what i have in mind

Thanks

DAVID

 

[VisiGoth]

If you are sweeping your DDS and need to search across your +/- 1 MHz range you are actually compounding your problem by atempting to do this inside of a 350 ns PW. You can look up the equation for narrow band FM modulation to find out what will happen.

Instead of a heterodyne chirp as you suggested, you could autodyne (as mentioned in my post earlier). That is, use the pulse itself as the LO, then you will always mix down to DC. Now the trick is to control the "phase" of the DC as a function of the RF frequency. Now there is really no DC phase as such, but you can build two mixes, offset by 90 degrees. If you put a delay line in series with the RF path, you will get a phase shift as a function of frequency that is different from the LO path (which does not have that delay line). You wil get a sine wave out of one mixer and a cos out of the other. The phase of the sin and cos will depend on the frequency times the length of the cable. phase = w * tau where tau is the time delay of the cable. Looking up IFM should give you a beter descriptioin.

 

[nbucska]

If the frequencyis critical, why can't you use a fixed
oscillator and gate the output on and off? You may need
a shielded assermbly etc.



<nbucska@pcperipherals DOT com> subj: eng-tips
read FAQ240-1032