Using DOuble Float in STEP7

thanks all .

after apply these algorithm , i found that there is a limitation in valus .

the maximum value after converting from 64bit to 32bit using FC2 LREAL2DINT is  2147483647 in decimal .

are there any solution of that , because i have valus more than 2147483647.

help please.

Hi

I have created library for S7-300/400 which handles long real (double float)

and a function for PAC energy meter.

Library with the function attached.

The library from Siemens site had a problem - some functions returned different values.

Project as a  .ZIP attached to next post.

Was the format of the double real in the PAC meters different from the IEEE754 standard?

I have a project with CEWE meters that also sends out Wh in 64bit floating format.

But I would like to divide the Wh by 100000 to get MWh with one decimal, I tried to look in the FC0 block that Piotr.M made but I don't really understand the code in NW1

      L     DW#16#3F50624D              // 0.001 H DWord: Sign=0; Exponent = 1013 (-10); Mantissa=0624D

      T     #One_Thausanth[1]           // Load in const 1/1000 H DWord

      L     DW#16#D2F1AA00              // Mantissa (Continued)=0624D

      T     #One_Thausanth[2]           // Load in const 1/1000 L DWord

According to decimal to 64bit real converters I found on internet 0.001 shuold be equal to

3F50624D

D2F1A9FC

Why isnt the last numbers same?

Hi

I did convert it with Excel and it may be not precise enough.

Conversion using excel given me

Exponent:=01111110101 Mantissa:=1 .0000011000100100110111010010111100011010101000000000

Conversion using online tool gives:

Exponent:=01111110101 Mantissa:= 1 .0000011000100100110111010010111100011010100111111100

which differs by 2^-49 before multiplication by Exponent

The constant in FC is changed to proper representation (0.001 = 3F50624DD2F1A9FCHex)

Corrected library attached two posts below.

Thanks to Kicken for spottiong that!!!

I tried the FC0 block, but I found some bugs...

NW0: index 2 in the energy array is transferred 2 times and index 1 is not transferred.

NW3: decimal calculation is wrong.

E.g. if input "I_Energy_Feedback" is 123456789Wh

"Energy_kWh" = 123457kWh

"Energy_decimals" = 123456789 - 123457 = 123333332Wh, I would have expected it to be 789Wh

I don't know how to fix this without getting decimal errors.

NW4: "Energy_kWh" should be converted to "Q_ENERGY_kWh"

NW5: Added to convert "Energy_Decimals" to "Energy_Decimals_Dint"

See attached jpg.

Please correct me if I have misunderstood something.

Hi

I have started simulator and debugged the whole PAC FC block and now it works.

Formula for decimals and copying which You mentioned was wrong.

Also added option to Truncate instead of Round when long real is converted to dint.

Thanks.

Piotr

hi

dint (32bit) capacity is 2147483647

for larger values:

1. use larger unit (for example kWh instead of Wh)

2. if you need decimals you need calculate in long real then have decimal part returned as a second

    variable (one variable shows kWh, another shows fractional part for example from 000 to 999. then

   you display right-justified kWh then . (dot) and left justified decimals with leading zeroes (000 not just 0).

Looking better!

Still some small things:

1. Q_Energy_kWh output is not written anywhere, I suggest to add a MOVE block in NW3 to copy "Energy_kWh_DInt" to the output.

2. Is the code in NW10 only for testing? I don't see the need for it else. Personally I don't like when memory bits are used inside FCs or FBs.

3. If the energy is negative decimals won't work, due to NW6.

Hi

Please find answers below:

1. Q_Energy_kWh output is not written anywhere, I suggest to add a MOVE block in NW3 to copy "Energy_kWh_DInt" to the output.

Corrected - added the MOVE after conversion to dint

2. Is the code in NW10 only for testing? I don't see the need for it else. Personally I don't like when memory bits are used inside FCs or FBs.

Yes. Network removed. I use Markers strictly for testing only (in my programs global variables are in

data block called globla which is divided into structures). this way at the end of commissioning

I can search for any M bits and take them out (any Ms are for test only)

3. If the energy is negative decimals won't work, due to NW6.

Changed the fractions to be always positive. otherwise it would display incorrectly

i.e. when energy is negative kWhs are negative but Whs are positive.

display  xxxxxxx.yyy for positive; -xxxxxx.yyy for negative (it would show -xxxxxx.-yyy when negative given)

in the above example xxxxxxx is linked to kWhs and yyy to wh. the dot is a fixed text and yyy is disoplayed with leading 0s

Attached - corrected project.

Many thanks for sugestions

Piotr