Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.
Hi..this is a Linear Feedback Shift Register...which gives you a PSEUDO Random output usually called a PN code. The register has a feedback (the next bit to be loaded into the register) that is an XORED combination of bit number 10, 12, 13 and 15. Thus the output will repeat after [(2^16)-1] bits. So for the period of [(2^16)-1] bit period it is a random generator
Yes, you're right, the circuit is indeed NOT random enough since it is 16-bit long. Unfortunately, ture random number generator is not easy to build. If your task is to build really good RNG, there are some comp. arith. book like Knuth's available.
is there any standard for this ? okie, we use LFSR with some polynomial, lets say. is there any criteria for choosing this polynomial ? or the initial seed value ?
is there any standard for this ? okie, we use LFSR with some polynomial, lets say. is there any criteria for choosing this polynomial ? or the initial seed value ?
The initial seed value can be any number, except for all-zeros or all-ones (depending on your feedback polarity) because that state will lock-up the LFSR.
to understand it better, you cannot make a perfect random number generator. Thats why you could always see Pseudo random projects from the internet. You could make it more random like but introducing some factors, like special gates, shift registers, and some has a clock so that the program will base the seed of the shift register from the clock value.
Hi, If you look at random number papers they usually are also giving a formula for calculating the perfect starting (seed) value. I know of a paper from TI. Not any number is a perfect seed to start. If you use a shift register XOR based generator you should NEVER use 0 as seed since it will not give any change !.
Some random number generates are sensitive to seed value, but an LFSR repeats the exact same sequence no matter what seed value you use (excluding that one lock-up value). The seed simply changes the sequence starting point. However, an LFSR is still very useful because it is easy to build one with an extremely long sequence.
It's true that you can't generate truly random numbers by using only digital logic. You need need to include some fundamentally random factor such as thermal noise. That's the basis behind adsl's suggestion.
To echo47: "an LFSR repeats the exact same sequence no" Not true. LFSR is sensitive to zero and polynomial value that is the same as LFSR. Try something before you make a statement !. The path you select on a LFSR is different depending on the seed value.
ALL pseudo random generators have this problem, no matter what. I do not remember the TI paper but it was very good and explained also the math behind it.
Paul.
Some LINKS for people who know better:
" Since each starting state produces a different (we are considering shifts as different) sequence, there are 2n elements in (f) since there are that many starting states. The sum of two sequences in (f) is again in (f) since the sum will satisfy the same recursion relationship (i.e., the sum corresponds to a different starting state).
We can characterize the elements of (f) in terms of the reciprocal polynomial of f.
"
From: **broken link removed**
hi
the random nos can be generated using LFSRs.various configurations of LFSRs and their advantages very clearly spec in the book DFT-Abromovicci.better u refer this.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
..