Who to map modulators +/- output in VHDL?

[ahmad_abdulghany]

According to the figure below,

If I'm going to develop a VHDL code for each of these modulators (i.e. PSK, QPSK, 16-QAM), and as it can be seen for example for the QPSK the mapped output varies between 1 and -1, and for the 16-QAM, it's further stragne as it takes values like +/-3 and so on..
How can this be mapped in a VHDL code, that only have (as far as i know) binary data (i.e. 1's and 0's)

Thanks in advance,
Ahmad,

 

[WM3]

Re: VHDL question

I will give a tip on +-3 issue only: Use the datatype 'real' for such numbers or binary-represent them as vectors.

 

[ahmad_abdulghany]

Re: VHDL question

I will give a tip on +-3 issue only: Use the datatype 'real' for such numbers or binary-represent them as vectors.

I thought in that already, but i think it can't be.. How would it be synthesized? i.e. 'real' datatype?

Anyway, thank you ..

Any more??

 

[Davood Amerion]

VHDL question

VHDL doesnt support analog signals.

 

[ahmad_abdulghany]

Re: VHDL question

I think i'm now very close to the solution,
WM3 was right
But it'll be further difficult than simply using vector, instead, i will represent the analog value (e.g. 3 or -3) interms of binary (and will be transmitted on a BUS)

I didnt' try it yet, but going with it.. I hope i succeed..

Keep on..
Thanks alot,
Ahmad,

 

[Ahmed Ragab]

Re: VHDL question

SA Ahmad,

In all cases your I and Q are buses. Even in QPSK and BPSK you are transmitting +1's and -1's not only a 1 and a 0. You have to perform level mapping as shown in the constellation diagrams you sent earlier.
In QPSK and BPSK a signed vector of 2 bit width would do, but in the 16 QAM you would need to make the I and Q buses 3 bits wide to take values up to +3 or -3.
Good luck ya gameel

Salam.
Ahmed M. Ragab

 

[xstal]

Re: VHDL question

Salam,

Raghab has suggested a good solution for the problem. you can always use signed vector to represent your real values.

Tasnif

 

[ahmad_abdulghany]

Re: VHDL question

SA Ahmad,

In all cases your I and Q are buses. Even in QPSK and BPSK you are transmitting +1's and -1's not only a 1 and a 0. You have to perform level mapping as shown in the constellation diagrams you sent earlier.
In QPSK and BPSK a signed vector of 2 bit width would do, but in the 16 QAM you would need to make the I and Q buses 3 bits wide to take values up to +3 or -3.
Good luck ya gameel

Salam.
Ahmed M. Ragab

Wa alaykom assalaam
Thank you for your replies..

I have some other little questions..

Q1) Will i represent numbers in their two's complement, or sign-magnitude?

Q2) Can I assign using the decimal value and implicitely it understand the corresponding binary value (e.g. I assign value of -3 and it becomes 101 as its two's complement)

Q3) Regarding fractions that may be needed to be used in schems like 8-PSK, i will use a BUS with certain precession. If i used 8-bits will give 256 different levels, and i want them to correspond to values from -1 to 1 with fractions.. each bit or level will correspond to decimal value of 1/16=0.0625 ... i.e. the binary number 00000000≡-0.9375, 100000000≡decimal zero, and 11111111 ≡+1, and in between, values varies between these values with step 0.0625 .. e.g. 00000001≡-0.8750 and so on.. i.e. i have 127 negative values and 128 positive values and the 'zero'.
The question is, if i want to represent a number say=+0.3827 then i will divide it by 0.0625 that gives 6.1232≈+6, then the corresponding binary representation will be 10000110 . This is nither exact or simple.. if this is the correct method but there's an easier way to deal with, so please tell me..

Q4) Finally, as i want to make single block performing all types of modulation and demodulation, then use two MUXes for I and Q BUSes after mapping, to select the required scheme output, so i think i have to fix length of BUS (say 8 bits).. right?


Thank you for helping and being patient as well
Ahmad,

 

[secondlife]

Re: VHDL question

for the first two questions first it is better to go for a 2s complement representation and representing the number in binary format even if they are negative.

 

[ahmad_abdulghany]

Re: VHDL question

for the first two questions first it is better to go for a 2s complement representation and representing the number in binary format even if they are negative.

But, using two's complement will not fit with fractional representation of numbers..right?

 

[nand_gates]

Re: VHDL question

Checkout this may help you!
**broken link removed**

 

[ahmad_abdulghany]

Re: VHDL question

Checkout this may help you!
**broken link removed**

Thank you, but i think the process followed there isn't accurate otherwise, i will use large number of bits to obtain good accuricy guaranteeing my requirements..

 

[Ahmed Ragab]

Re: VHDL question

SA Ahmad,

I must be missing something in your questions, since I find no problem in representing negative fractions in binary form and using 2's complements.

Regarding their representation and the matter of accuracy, you may prefer using this library found at:
**broken link removed**
Check the fixed_pkg.

They are new proposed libraries to be added soon into tools. You will find a function named resize that would take in your number as a decimal fraction and convert it into the vector size you're using with the chosen precision.
Say you need to work with an 8 bit vector that would represent fractions from +1 to -1 then you would choose it to be a signed fixed number "Sfixed(0 downto -7)".
The tool would understand that you have only 1 bit before you're decimal point (which is infact your sign bit) and 7 bits after your decimal point, giving you a minimum resolution of 0.00000005.

You'll have to play a little with your libraries in order to make your tool synthesize them well.

I hope in someway I've made things clearer. If not then don't hesitate to pm me )

Salam,
Ahmed M. Ragab

 

[mmoctar]

Re: VHDL question

I think you can use ROM for mapping your qpsk and qam.., and use it like this
address(rom)≡datainput(map) and dataoutput(rom)≡map output,and you use a mux to select betwen qam and qpsk.....MOC