Delay Circuit

Thread Starter

sorany

Joined Mar 25, 2007
1
hi, i just found out about this site...i hope it is going to be usefull..
i have a project to do does anyone can help me how to build 0.1 second delay circuit. since my project is on producing hermite pulses for ultra wide band UWB..
i would appritiate if any one help me on this.
 

Papabravo

Joined Feb 24, 2006
14,382
This is a new one on me. Could you explain that one more time. I'm just a poor old country boy and I don't understand all the abreviations.
 

Papabravo

Joined Feb 24, 2006
14,382
One problem with the first couple of sites is that access to the papers requires a membership eg. ieee. This makes it hard to offer any suggestions on the use of the technique. Maybe the OP could summarize for us what they are and why he thinks it would be worthwhile to investigate them further. Not in detail, just a thumbnail sketch.
 

Papabravo

Joined Feb 24, 2006
14,382
I read the first three sections and the conclusion. The basic idea is to compress a number of bits of information into a pulse of a particular shape and duration at the transmitter. At the receiver there are a number of correlators which attempt to identify the incoming pulse according to a template, and thus recreate the original codeword.

Returning to the original post, I fail to see the relationship of a delay circuit to either the generation of Hermites or the subsequent recovery by matching the incoming pulses to codeword templates. Perhaps the OP could elaborate.
 

Dave

Joined Nov 17, 2003
6,970
One problem with the first couple of sites is that access to the papers requires a membership eg. ieee. This makes it hard to offer any suggestions on the use of the technique.
True, which is why I suggest we "black-box" the communications aspects and focus on the delay circuit. I await the OP logging back on and shedding some light on this.

Dave
 

Papabravo

Joined Feb 24, 2006
14,382
I am making progress slowly. The generation of hermite waveforms can be done with a discrete time wavefor generator. This can be implemented as an FIR filter with the appropriate hermite function as its impulse response. FIR filters are characterized by "delay" without feedback. In DSP terms the delay is usually related to the sample time which must be very much shorter than the symbol time.

Thus the original question seems to make very little sense if we are talking about analog filters for generating hermite functions. It makes some sense if we are talking about a DSP generation algorithm. The delay however is entirely a function of the DSP clock rate and the length of the calculation loop of the DSP algorithm. You can lengthen it but you probably will have a very hard time shortening it.

I have loacated a reference to a delay element in potential receivers but have not followed up on that piece of reading. I'm waiting for the OP to check in again with some clarifications.
 

Dave

Joined Nov 17, 2003
6,970
Thanks for the information Papabravo.

Do we know what a hermite waveform/pulse looks like, i.e. h[n] Vs k or alternatively H[w] Vs frequency?

FIR filter designs are generally easily implemented using either the Windowing Method or Frequency Sampling Method for simple low/high pass and band pass/stop impulse responses and tools are readily available to deduce the coefficients.

Dave
 

Dave

Joined Nov 17, 2003
6,970
Yes we do. The lowest order one looks like a normal distribution probability density function (pdf) and the higher order ones look like the derivatives of the basic bell shaped curve.
Ok, so by using a bit of initiative and Fourier analysis one should be able to deduce the coefficients using the Frequency Sampling Method. Depending on the high-order ripples of the impulse response a windowing function can be used to remove these ripples beforehand.

Most of this is academic without the input of the OP.

EDIT: Thanks for the link.

EDIT 2: From reading this disertation, maybe the waveform is too complex for a crude mechanism like the FSM?

Dave
 

Papabravo

Joined Feb 24, 2006
14,382
I'd like to discourage the use of PMs for this kind of problem. Let's keep the discussion in the open forum. In your PM you asked if I had access to Matlab and the answer is no. I had a student version when I went back to grad school. Since I got my degree in the spring of 2004 I was compelled to deinstall it. The simulink picture was meaningless to me anyway.

For some insight into the construction of FIR filters whose impulse response is a hermite waveform, I would recommend the hardcover/ebook by Steven W. Smith on Digital Signal Processing.

http://www.dspguide.com/

As I understand the FIR filter, it is a strictly digital signal processing kind of animal. I am by no means convinced that there is any way to construct an analog filter whose impuse response is one of the hermite functions. This would imply that for an N-bit codeword you would need N DSP elements. That will get kind of expensive in chip cost and board space, but I suppose it could be worse. The number of required DSP elements could grow exponentially.

There might be some hope of using really big FPGAs and craming as many DSP cores as you can into it. If none of the above is helpful then I think you are SOL.

Maybe there is somebody on the forum who knows more about this than I do and I will defer to their superior knowledge and experience.
:)
 

Dave

Joined Nov 17, 2003
6,970
The FIR filter is primarily a digital implementation based around onr core idea: given a desired frequency response, how are the discrete coefficients of its impulse response calculated. I have already referenced the Frequency Sampling Method which would allow to deduce the coefficients of an arbitrary frequency response. Now how you go about this for the hermite waveform/pulse is where the skill comes in.

I recommend you read up on the relevant DSP theory, Papabravo has given a good link above, I would also recommend you look at Foundations of Digital Signal Processing: Theory, Algorithms and Hardware Design by Patrick Gaydecki.

Dave
 
Top