Bit rotation

[SwingeyP]

Hello,

Still working on the AD9850 and now have it working with a fixed frequency.

I want to break the fout variable into 4 8 bit words.

My thinking is something like this, (however it doesn't compile) is this the correct way to go about this?

ad9850_freq = ad9850_mhz + ad9850_khz + hz
fout = ad9850_freq * 34.3597383 '7 makes it too long!
'Ok so now we have the frequency we need to break it down to 4 8 bit words.
For x = 1 To 4
shifted = 0
For i = 0 To 7
shifted = shifted + ShiftRight(fout)
word_array(x) = shifted
Next i
Next x





Cheers - Paul

 

[SwingeyP]

fout is a single. It seems you can't shift a single. fout is only a single because I have to multiply it by 34.3597383

 

[ericgibbs]

hi Paul,
Look at the HW , LW, LB and HB extensions in Oshon

Dim First as LONG

Dim Second as WORD
Dim Third as WORD

Dim Fourth as BYTE
Dim Fifth as BYTE
Dim Sixth as BYTE
Dim Seven as BYTE

Second = First.HW
Third = First.LW

Forth = Second.HB
Fifth = Second.LB
Sixth = Third.HB
Seventh = Third.LB

Do you follow that OK.?
It may work for Single, I cannot confirm that, I dont have FP package

Eric

 

[ericgibbs]

hi,
This is an example of a USB program, note the ASM: used to shift the Bit pattern.
Thats another option you could try.

'chan0
SPICSOn
PORTC.6 = 0
SPISendBits 8, %00000011
SPIReceive ch0hi
SPIReceive ch0lo
PORTC.6 = 1
SPICSOff

ASM:        rrcf ch0hi,F
ASM:        rrcf ch0lo,F
ASM:        movlw 0x0f
ASM:        andwf ch0hi,F

temp0 = (ch0hi * 256) + ch0lo  'one sample result
avg0 = avg0 + temp0  'N samples

Next smpcnt

temp0lw = avg0.LW
temp0hw = avg0.HW

PORTB.2 = 0  'low for 50uSec, high for
PORTB.3 = 1
UsbService
PORTB.3 = 0  'testing only 'scope 15uSec High period

 

[ericgibbs]

hi Paul,
Lets know if you need help with the ASM structure.

BTW: so that I can offer you better help I have ordered from Vlad the full FP upgraded packages, I should be running it by Wednesday

E

 

[SwingeyP]

Hi Eric,

I bought the full package a while ago when it was on 'special offer' I have all the bits & bobs. I must say I love it. It has really helped me a lot. It just needed a better manual which you sent me. (I should read it properly some time). I had just got to .hb .lb etc when you replied.



I see the way that yo have done it and it makes sense. I'll give it a try now. Much simpler than loops and arrays all over the place

Cheers - Paul

 

[SwingeyP]

Ok this is weird.

fout calculates to 243954128 = 1110 10001010 01110001 11010000

110 = 14 = E
10001010 = 138 = 8A
01110001 = 113 = 71
11010000 = 208 = D0

I tried your code and also .3b .4b which I also found in the manual - They both give the same results but I don't understand the results in the watch window.....

ad9850_freq = ad9850_mhz + ad9850_khz + hz
fout = ad9850_freq * 34.3597383  '7 makes it too long!
'Ok so now we have the frequency we need to break it down to 4 8 bit words.
       
word1 = fout.LB
word2 = fout.HB
word3 = fout.3B
word4 = fout.4B


Dim second As Word
Dim third As Word

Dim fourth As Byte
Dim fifth As Byte
Dim sixth As Byte
Dim seventh As Byte

second = fout.HW
third = fout.LW

fourth = second.HB
fifth = second.LB
sixth = third.HB
seventh = third.LB

I would have expected to see the values above NOT those as shown in the Watch Variables window.

BTW I did have that other problem you mentioned with variables not counting properly....

Cheers - Paul

 

[SwingeyP]

try this code as a long perfect as a single weird values. Why?

Dim fout As Single  'try changing this to a long

Dim word1 As Byte
Dim word2 As Byte
Dim word3 As Byte
Dim word4 As Byte


Dim second As Word
Dim third As Word

Dim fourth As Byte
Dim fifth As Byte
Dim sixth As Byte
Dim seventh As Byte

fout = 243954128
loop:
word1 = fout.LB
word2 = fout.HB
word3 = fout.3B
word4 = fout.4B

second = fout.HW
third = fout.LW

fourth = second.HB
fifth = second.LB
sixth = third.HB
seventh = third.LB
Goto loop

This one's a bit of a show stopper :-( If those values as a single are true :-(

 

[ericgibbs]

Hi,
Have now installed and running all latest Oshonsoft packages including FP.
The Scope also now works!!
If you could post your very latest code listing I will try it.

Give me a while to get the hang of the FP etc and I will get back to you

E

 

[ericgibbs]

Tried your code as in post #8.

For a Single, it appears that number is not stored in Binary format, but some other format as you would expect for a FP number.

The program does break in down into Bytes correctly,but they are not what I would expect.

I will keep looking at options.
E

 

[SwingeyP]

Thanks Eric. Just upgraded to 6.97 and seeing the same thing as you. I tried outputting word1 on a port just to see what the actual output was. It does differ as per the watch variables :-(
I guess what I need is a long I can multiply by a single or perhaps some clever way of converting the single to a long.

Almost there so close ......

Thanks for the help and getting the F/P package.

Cheers - Paul

 

[ericgibbs]

hi,
Can you explain why you would want to reduce the SINGLE to BYTES.?

If you could what would be the next step in your program.?
E

 

[SwingeyP]

Hi Eric,

Here's the subroutine. I have commented it for you.

The fout has to be multiplied by 34.35973837 (I miss the last digit off) hence the single for fout. This number is then broken down to 4 words for the AD9850.

convert_freq:

ad9850_mhz = mhz * 1000000
ad9850_khz = khz * 1000

'To set the frequency, the 32 tuning bits are set As:
'(Frequency x 2^32)/DDS Clock Frequency
'Where the DDS has a 125Mhz clock, this resolves To
'Bits = Output Frequency x 34.35973837
'So To set the Output of the DDS To 7.000Mhz, the tuning bits must
'be set To 7000000 x 34.35973837 = 240518168 Or E560418 in Hex.

ad9850_freq = ad9850_mhz + ad9850_khz + hz
fout = ad9850_freq * 34.3597383  '7 makes it too long!
'Ok so now we have the frequency we need to break it down to 4 8 bit words.
       
word1 = fout.LB
word2 = fout.HB
word3 = fout.3B
word4 = fout.4B


'http://designtools.analog.com/dt/dds/ad9850.html
'This util gives each word as for freq 7.1mhz @ 125 clk
'W0 - 80 - 10000000 - 128
'W1 - E  - 00001110 - 14
'W2 - 8A - 10001010 - 138
'W3 - 71 - 01110001 - 113
'W4 - DE - 11011110 - 222
Return

I hope all that makes sense. Word 0 is handled elsewhere and just sets the phase.

Cheers - Paul

 

[ericgibbs]

hi,
OK,
Look at this link, it has a EEE to Binary Convert equation
https://en.wikipedia.org/wiki/Single-precision_floating-point_format

Oshonsoft is a modified 32 bit IEEE 754 format.
There are some Single FP notes in the text file that is in the PICSIM program folder
E

 

[SwingeyP]

Hi Eric.

Wow that looks complicated. Takes me right back to college days and I don't think I really understood it then.
I'll have a go at writing a function but this might take longer than the whole process so far -

I'll see what I can do.

Here's the progress so far with the error, but at least now I know what it is.

Cheers - Paul

 

[ericgibbs]

hi Paul,
I will look at the EEE 745 conversion today, in the meantime if you download the free version of the Z80 IDE from Oshonsoft its got a EEE745 convert utility that may help you understand the format.
E

 

[ericgibbs]

hi
More details....
I will try more sample values later, but it looks as though you may have to contact Vladimir.?
E

Dim word1 As Long
Dim word2 As Long

Dim byte1 As Byte @ 0x60
Dim byte2 As Byte @ 0x61
Dim byte3 As Byte @ 0x62
Dim byte4 As Byte @ 0x63

'test values in decimal and hex from convert utility

fout = 243954128  '= 4d 68 a7 1d
'1 = 3f 80 00 00
'10 = 41 20 00 00
'100 = 42 c8 00 00
'1000= 44 7a 00 00
'10,000= 46 1c 40 00
'100,000= 47 c3 50 00
'1,000,000=49 74 24 00
'10,000,000= 4b 18 96 80
'100,000,000=4c be bc 20

'243954128  = on the Z80 IEEE 754 converter = 4D 68 A7 1D

'243954128  = hex On a calculator = E8A71D0...  NOTE No SIGN Bit!!!
'Adding the SIGN Bit gives =  4D 68 A7 1D
'0100,,1101,,0110,,1000,,1010,,0111,,0001,,1101
'So the above agree

'But
'243954128  =on Watch Window =  9A 68 A7 1D
'As Bit Pattern =1 001,,1010,,0110,,1000,,1010,,0111,,0001,,1101
'as Hex= ..................9A.......68......A7...1D

'1001,,1010,,,,,,,0110,,1000,,1010,,0111,,0001,,1101
'first Byte appears incorrect.???


loop1:
fout = 243954128

word1 = fout.HW
word2 = fout.LW

byte1 = word1.HB
byte2 = word1.LB

byte3 = word2.HB
byte4 = word2.LB

Goto loop1

 

[SwingeyP]

Hi Eric,

There seem to be a couple of issues here.

Firstly, a potential bug in the way the IEEE 754 calculations are done is presenting as you say the first byte incorrectly.
I did a some calculations on paper last night and after a lot of confusion used an online converter to conclude what you have.

Secondly, if it is a bug and it gets fixed, I'll still need some conversion to turnit back into the unsigned number. This looks beyond my ability.
Is there another way I could code the multiply of the single? --> fout = ad9850_freq * 34.3597383 maybe keeping everything as LONGs and working through unit, tens, etc etc?

Cheers - Paul

 

[ericgibbs]

hi,
I have spent some time trying to see how Vlad has modified the IEEE745 standard.

fout = ad9850_freq * 34.3597383

Whats the numeric range ad9850-freq... min max.??

E

 

[SwingeyP]

Well it could be 1 (as in 1 hz) to 30,000,000 (as in 30Mhz).

mhz = 7
khz = 1
hz = 0

I do this to set the initial frequency at 7.1 mhz. However when I expand the project to add a band switch it cold be 1 to 30 as described above.
At present the project is covering the amateur 40m band 7 -7.2 mhz. I start off at 7.1 as it's in the middle of the band.

ad9850_mhz = mhz * 1000000
ad9850_khz = khz * 1000

ad9850_freq = ad9850_mhz + ad9850_khz + hz
fout = ad9850_freq * 34.3597383
I hope this makes sense. I'll post the comlete running code when I get home later today.

Cheers - Paul