Features of the 7-bit coding for SMS in siemens S45i - GSM-Forum

ntcn · 09-09-2004, 06:21

Has sent СМС with the text ""~!@#$%^&*()_+|`-=\,./?". Has come (on the screen, only correctly displayed): ""!@#$%&*()_+-=./?", actually codes (that was incorrectly defined - the come code of a symbol and an expected code of a symbol with the an expected symbol): $3D instead of $7E ("~"), $00 instead of $40 ("@"), $02 instead of $24 ("$"), $14 instead of $5E ("^"), $11 instead of $5F ("_"), $40 instead of $7C ("|"), $2F instead of $5C ("\").
Hex dump of sent and recv (after decoding from 7-bit codings, latin symbols and numbers in the same message are displayed absolutely correctly in comparison with codes symbols of a computer which have been sent):

Code:

waiting for : 7Eh,21h,40h,23h,24h,25h,5Eh,26h,2Ah,28h,29h,5Fh, 2Bh,7Ch,60h,2Dh,3Dh,5Ch,2Ch,2Eh,2Fh,3Ch,3Eh,3Fh
received    : 3Dh,21h,00h,23h,02h,25h,14h,26h,2Ah,28h,29h,11h, 2Bh,40h,60h,2Dh,3Dh,2Fh,2Ch,2Eh,2Fh,3Ch,3Eh,3Fh

Who knows why so occurs (for example, instead of "@" comes a symbol with a code of $00, but on the screen is displayed correctly, and the symbol ['] has come as ['] (60h), but on the screen was not displayed correctly) ?

dspmobile · 09-09-2004, 08:35

There is a GSM alphabet (3.38) character set - 7bit - that differs from standard ASCII - Latin char. set . It is used in SMS 7bit (160 max for simple SMS or 720 char EMS).

Look at :
http://www.csoft.co.uk/sms/character_sets/gsm.htm

or
http://www.dreamfabric.com/sms/default_alphabet.html

to see exactly how it is.

It is designed to have some accended small latin letters (more common) and also some Capital Greek letters that differ in appearence form Latin. (Full Greek Capital letters alphabet can be formed with Same latin counterparts ).
So it can be used form more languages while sending more characters
in less space - that is paying less.

Other languages that cannot fit in this GSM 7bit set use Unicode char set
that is sending 70 (2 byte) chars in simple SMS and around 320 in EMS.

ntcn · 09-09-2004, 08:58

Quote:

Originally Posted by dspmobile

http://www.dreamfabric.com/sms/default_alphabet.html

Thanks! Otherwise my patch uncorrectly writes some symbols (unicode - correctly, for the 7-bit coding did not know as a correct symbol to receive) in a logfile.

ntcn · 09-09-2004, 22:22

Quote:

Originally Posted by dspmobile

There is a GSM alphabet (3.38) character set - 7bit

In algorithm of the siemens counter a mistake ? In SMS PDU the length of the text can be only 140 symbols, for 7-bit coding have 160 symbols. But the some (^ {} [] ~ |) symbols are coded through esc-sequences and such symbol send as 14 bits. Phone (S45i) at a spelling of such symbols shows, that the symbol = 7 bits. If in the text will be, for example, twenty symbols "[", other symbols in it SMS can be no more than 140 or 120 (actually, it is impossible to trust the counter of symbols in phone) ?

dspmobile · 09-10-2004, 13:51

Although not knowing too much in detail ,
It is correct that the size of a simple SMS has max 140 bytes.
That is why a simple SMS in Unicode (2 byte char.) is then 70 chars .

When using the 7-bit GSM alphabet a special packing of the 160 7-bits occurs to fit in the 140 bytes (8 bits)....

Search the Internet for information on the format of SMS PDU and this packing of 7-bits ....

ntcn · 09-10-2004, 16:07

I wanted to know about symbols with a code 1B (the expanded code for special symbols). This symbol concerns to 7-bits 160 symbols in PDU ? Should concern, but when I write the SMS, that the counter of symbols at S45 incorrectly counts special symbols (for example, at a spelling "tilde" the counter writes that 1 symbol is written, but in fact actually "~" it will be transformed to 1B 3D, 2 symbols). Ok, I think, that it is a trouble of a siemens firmware

More questions are not present, thanks for URL.

rc-flitzer · 09-10-2004, 22:51

@ntcn: You want to tell us, that if you write a char like ~, \ or €, the S45i counts it as one char? Hm, sounds like bad software. The (older) firmware of SL45i counts those chars as each two chars, if we have standard 7bit-code.

I didn't discovered anywhere an input mode that allows 8bit-Code (with SL45i). Has somebody other experience? In my phone I can either type 7bit or Unicode (e.g. Arabian chars).

lalo.lerry · 09-11-2004, 02:31

Quote:

Originally Posted by rc-flitzer

@ntcn: You want to tell us, that if you write a char like ~, \ or ?, the S45i counts it as one char? Hm, sounds like bad software. The (older) firmware of SL45i counts those chars as each two chars, if we have standard 7bit-code.

Yes, unfortunally is exactly what heppends.
Now I know why sometimes I noticed to have sent 2 SMS but counter shown me 160 characters used!

Ntcn, can you fix this in our Left Character Countdown patch?

ntcn · 09-11-2004, 07:53

The existing algorithm of calculation of number of symbols is based on getting the address of structure where saved text of SMS (there the number of symbols begins with 0Bh, for 1 symbol will be 0Ch, for 2 - 0Dh, etc.) and get number of symbols is stored and subtraction 0Bh (sub_F2DB6E, .. , sub_FCB0F8), get the "code page" of the text and if it not unicode - then number of symbols is subtracted from 760 (loc_E9A1F0), if unicode - from 330 (0E9A1EA).. But I cannot find a place where the number of symbols increases on 1 at typing (if to find this function - I can make the separate counter of real number of symbols, instead of the counter of those symbols which on the screen).. The only thing, that I can make - each time by call LCC of a patch _anew_ (on existing symbols) to calculate number of symbols.. But it will be, probably, very long (process of recalculation)..

09-09-2004, 06:21	#1 (permalink)
ntcn Insane Poster Join Date: Feb 2003 Posts: 71 Member: 22759 Status: Offline Thanks Meter: 0	Features of the 7-bit coding for SMS in siemens S45i Has sent СМС with the text ""~!@#$%^&()_+\|`-=\,./?". Has come (on the screen, only correctly displayed): ""!@#$%&()_+-=./?", actually codes (that was incorrectly defined - the come code of a symbol and an expected code of a symbol with the an expected symbol): $3D instead of $7E ("~"), $00 instead of $40 ("@"), $02 instead of $24 ("$"), $14 instead of $5E ("^"), $11 instead of $5F ("_"), $40 instead of $7C ("\|"), $2F instead of $5C ("\"). Hex dump of sent and recv (after decoding from 7-bit codings, latin symbols and numbers in the same message are displayed absolutely correctly in comparison with codes symbols of a computer which have been sent): Code: waiting for : 7Eh,21h,40h,23h,24h,25h,5Eh,26h,2Ah,28h,29h,5Fh, 2Bh,7Ch,60h,2Dh,3Dh,5Ch,2Ch,2Eh,2Fh,3Ch,3Eh,3Fh received : 3Dh,21h,00h,23h,02h,25h,14h,26h,2Ah,28h,29h,11h, 2Bh,40h,60h,2Dh,3Dh,2Fh,2Ch,2Eh,2Fh,3Ch,3Eh,3Fh Who knows why so occurs (for example, instead of "@" comes a symbol with a code of $00, but on the screen is displayed correctly, and the symbol ['] has come as ['] (60h), but on the screen was not displayed correctly) ?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Any Tutorial that explains how to use all features of the Blackberry Tool? [SOLVED]	upa	FuriousGold	7	11-30-2009 00:10
Radio Codes For Cars In The UK	mad ali	CAR Electronics	7	03-15-2009 19:42
Anybody found the routine to read SMS in siemens flash?	kxn	x1x to x45/x50	0	07-30-2003 15:45
Anybody found the routine to read SMS in siemens flash?	kxn	GSM Programming & Reverse Engineering	0	07-30-2003 14:33
What do each of the 7 Nokia Codes do?	Shaocaholica	Phone Unlocking Codes & Maps	2	06-04-2003 22:00

09-09-2004, 08:35	#2 (permalink)
dspmobile Freak Poster Join Date: May 2003 Location: Athens, Greece Posts: 173 Member: 29130 Status: Offline Thanks Meter: 5	There is a GSM alphabet (3.38) character set - 7bit - that differs from standard ASCII - Latin char. set . It is used in SMS 7bit (160 max for simple SMS or 720 char EMS). Look at : http://www.csoft.co.uk/sms/character_sets/gsm.htm or http://www.dreamfabric.com/sms/default_alphabet.html to see exactly how it is. It is designed to have some accended small latin letters (more common) and also some Capital Greek letters that differ in appearence form Latin. (Full Greek Capital letters alphabet can be formed with Same latin counterparts ). So it can be used form more languages while sending more characters in less space - that is paying less. Other languages that cannot fit in this GSM 7bit set use Unicode char set that is sending 70 (2 byte) chars in simple SMS and around 320 in EMS.

09-10-2004, 13:51	#5 (permalink)
dspmobile Freak Poster Join Date: May 2003 Location: Athens, Greece Posts: 173 Member: 29130 Status: Offline Thanks Meter: 5	Although not knowing too much in detail , It is correct that the size of a simple SMS has max 140 bytes. That is why a simple SMS in Unicode (2 byte char.) is then 70 chars . When using the 7-bit GSM alphabet a special packing of the 160 7-bits occurs to fit in the 140 bytes (8 bits).... Search the Internet for information on the format of SMS PDU and this packing of 7-bits ....

09-10-2004, 16:07	#6 (permalink)
ntcn Insane Poster Join Date: Feb 2003 Posts: 71 Member: 22759 Status: Offline Thanks Meter: 0	I wanted to know about symbols with a code 1B (the expanded code for special symbols). This symbol concerns to 7-bits 160 symbols in PDU ? Should concern, but when I write the SMS, that the counter of symbols at S45 incorrectly counts special symbols (for example, at a spelling "tilde" the counter writes that 1 symbol is written, but in fact actually "~" it will be transformed to 1B 3D, 2 symbols). Ok, I think, that it is a trouble of a siemens firmware More questions are not present, thanks for URL.

09-10-2004, 22:51	#7 (permalink)
rc-flitzer No Life Poster Join Date: Mar 2004 Posts: 593 Member: 60520 Status: Offline Thanks Meter: 0	@ntcn: You want to tell us, that if you write a char like ~, \ or €, the S45i counts it as one char? Hm, sounds like bad software. The (older) firmware of SL45i counts those chars as each two chars, if we have standard 7bit-code. I didn't discovered anywhere an input mode that allows 8bit-Code (with SL45i). Has somebody other experience? In my phone I can either type 7bit or Unicode (e.g. Arabian chars).

09-11-2004, 07:53	#9 (permalink)
ntcn Insane Poster Join Date: Feb 2003 Posts: 71 Member: 22759 Status: Offline Thanks Meter: 0	The existing algorithm of calculation of number of symbols is based on getting the address of structure where saved text of SMS (there the number of symbols begins with 0Bh, for 1 symbol will be 0Ch, for 2 - 0Dh, etc.) and get number of symbols is stored and subtraction 0Bh (sub_F2DB6E, .. , sub_FCB0F8), get the "code page" of the text and if it not unicode - then number of symbols is subtracted from 760 (loc_E9A1F0), if unicode - from 330 (0E9A1EA).. But I cannot find a place where the number of symbols increases on 1 at typing (if to find this function - I can make the separate counter of real number of symbols, instead of the counter of those symbols which on the screen).. The only thing, that I can make - each time by call LCC of a patch _anew_ (on existing symbols) to calculate number of symbols.. But it will be, probably, very long (process of recalculation)..