Post by thread:[PIC] Error % charts of PIC USART ?

In the datasheet of 18F2455 we have Table 20-3 Baud Rates For Aynchronous Modes (USART module) (I guess other pics would have the same thing as well) The errors seem to range from 0.16% up to -45.75%. Are all these values valid, or should I be aiming at a certain percentage, I am confused?

That percentage is the actual error if you compare the actual speed of the PICmicro using the mentioned (in the datasheet) crystal and the speed standard. For example, the standard speed is 19200 baud, but with the 0.16% error you will have (in the PIC) 19230 baud. See? Regards, Mauricio Jancic Janso Desarrollos - Microchip Consultants Program Member spam_OUTinfoTakeThisOuTjanso.com.ar http://www.janso.com.ar (54) 11 - 4542 - 3519 > {Original Message removed}

kenasw@btinternet.com wrote : > In the datasheet of 18F2455 we have > Table 20-3 Baud Rates For Aynchronous Modes (USART module) > (I guess other pics would have the same thing as well) > The errors seem to range from 0.16% up to -45.75%. Are all these > values valid, or should I be aiming at a certain percentage, I am > confused? Note that the table are just examples. A lot of them will *never* work. Try to stay (in *your* application) within "a few" % from the nominal speed. Exactly how many "a few" is varies a little depending on who you ask, but 1-2 % would probably be "safe"... Note that there are "special" crystals with "funny" (odd) speeds that makes it easier to be on the spot. Jan-Erik.

----- Original Message ----- From: "Mauricio Jancic" <.....infoKILLspam@spam@janso.com.ar> To: "'Microcontroller discussion list - Public.'" <piclistKILLspammit.edu> Sent: Friday, May 13, 2005 4:37 PM Subject: RE: [PIC] Error % charts of PIC USART ? > That percentage is the actual error if you compare the actual speed of the > PICmicro using the mentioned (in the datasheet) crystal and the speed > standard. > > For example, the standard speed is 19200 baud, but with the 0.16% error you > will have (in the PIC) 19230 baud. > > See? I see now, but I still do not quite understand what I can choose if I want to connect to the PC com port. If I am not mistaken every char is 9 to 10 bits including the stop bit(s), so if the error is -45.75 percent in the table, I would need to multiply this by 9 for the whole char. This will give me an error of 450%. Am I doing something wrong here? > > Regards, > > Mauricio Jancic > Janso Desarrollos - Microchip Consultants Program Member > .....infoKILLspam.....janso.com.ar > http://www.janso.com.ar > (54) 11 - 4542 - 3519 > > > > {Original Message removed}

kenasw@btinternet.com wrote : > I see now, but I still do not quite understand what I can > choose if I want > to connect to the PC com port. If I am not mistaken every char is 9 to > 10 bits including the stop bit(s), so if the error is -45.75 > percent in the > table, I would need to multiply this by 9 for the whole char. > This will > give me an error of 450%. Am I doing something wrong here? Forget about those %-figures that are larger then "a few" % ! No matter what speed the "other end" is using, your PIC shouldn't be off more then "a few" % from that. Try to stay within 1-2 %, and you'd probably be safe... Jan-Erik.

kenasw@btinternet.com wrote: > In the datasheet of 18F2455 we have > Table 20-3 Baud Rates For Aynchronous Modes (USART module) > (I guess other pics would have the same thing as well) > The errors seem to range from 0.16% up to -45.75%. Are all these > values valid, or should I be aiming at a certain percentage, I am > confused? Consider anything over about 2.5% bad. ***************************************************************** Embed Inc, embedded system specialists in Littleton Massachusetts (978) 742-9014, http://www.embedinc.com

----- Original Message ----- From: "Jan-Erik Soderholm" <EraseMEjan-erik.soderholmspam_OUTTakeThisOuTtelia.com> To: <piclistspam_OUTmit.edu> Sent: Friday, May 13, 2005 5:31 PM Subject: RE: [PIC] Error % charts of PIC USART ? {Quote hidden}> @spam@kenaswKILLspambtinternet.com wrote : > > > I see now, but I still do not quite understand what I can > > choose if I want > > to connect to the PC com port. If I am not mistaken every char is 9 to > > 10 bits including the stop bit(s), so if the error is -45.75 > > percent in the > > table, I would need to multiply this by 9 for the whole char. > > This will > > give me an error of 450%. Am I doing something wrong here? > > Forget about those %-figures that are larger then "a few" % ! > > No matter what speed the "other end" is using, your PIC > shouldn't be off more then "a few" % from that. Try to stay > within 1-2 %, and you'd probably be safe... o.k. 2% times 9 = 18% per char. Assuming we are sampling in the middle we have 50% margin, right? so 18% will be o.k. , 3 or 4 percent should still work, anything above this *should* fail. this would make for an interesting experiment ... > > Jan-Erik. > > > > --

> In the datasheet of 18F2455 we have > Table 20-3 Baud Rates For Aynchronous Modes (USART module) > (I guess other pics would have the same thing as well) > The errors seem to range from 0.16% up to -45.75%. Are all these > values valid, or should I be aiming at a certain percentage, I am > confused? Usually if I am designing something using the usart, I choose a crystal frequency that divides well. For the 16-series PICs and PC baud rates, 19.6608 MHz is readily available. For the 18 series with PLL, you will need to choose something 10MHz or less. Cheers, Bob

At 12:14 PM 5/13/2005, Bob Blick wrote: > > In the datasheet of 18F2455 we have > > Table 20-3 Baud Rates For Aynchronous Modes (USART module) > > (I guess other pics would have the same thing as well) > > The errors seem to range from 0.16% up to -45.75%. Are all these > > values valid, or should I be aiming at a certain percentage, I am > > confused? > >Usually if I am designing something using the usart, I choose a crystal >frequency that divides well. For the 16-series PICs and PC baud rates, >19.6608 MHz is readily available. > >For the 18 series with PLL, you will need to choose something 10MHz or less. Yup. When using uarts to talk to other standard devices, it's best to use a baud rate multiple crystal. Then the error is zero, and you don't have to worry.

kenasw@btinternet.com wrote : > > No matter what speed the "other end" is using, your PIC > > shouldn't be off more then "a few" % from that. Try to stay > > within 1-2 %, and you'd probably be safe... > > o.k. 2% times 9 = 18% per char. I'm not sure why you are making it harder then it is... :) I think you are talking about a half-bit time after a full byte, right ? Remember also that the *other end* can be off in the other direction. As have been said before, just stay within 1-2 % and you will probably be safe. Jan-Erik.

At 12:20 PM 5/13/2005, Jan-Erik Soderholm wrote: >KILLspamkenaswKILLspambtinternet.com wrote : > > > > No matter what speed the "other end" is using, your PIC > > > shouldn't be off more then "a few" % from that. Try to stay > > > within 1-2 %, and you'd probably be safe... > > > > o.k. 2% times 9 = 18% per char. 2% is 2%. you'll be 18% of a BIT off at the end of a word.

>> > No matter what speed the "other end" is using, your PIC >> > shouldn't be off more then "a few" % from that. Try to stay >> > within 1-2 %, and you'd probably be safe... >> >> o.k. 2% times 9 = 18% per char. > > I'm not sure why you are making it harder then it is... :) > > I think you are talking about a half-bit time after a full byte, > right ? > Remember also that the *other end* can be off in the > other direction. > > As have been said before, just stay within 1-2 % and > you will probably be safe. Theoretically you are "allowed" a total error from all sources of about 50/N% where N is the total bits used per word. For 8+1+1 = 10 its you can tolerate an all up 5% error. As Jan says, if total error is about 40%of this maximum value things will probably work OK. RM

----- Original Message ----- From: "Jan-Erik Soderholm" <RemoveMEjan-erik.soderholmTakeThisOuTtelia.com> To: <spamBeGonepiclistspamBeGonemit.edu> Sent: Friday, May 13, 2005 6:20 PM Subject: RE: [PIC] Error % charts of PIC USART ? > TakeThisOuTkenaswEraseMEspam_OUTbtinternet.com wrote : > > > > No matter what speed the "other end" is using, your PIC > > > shouldn't be off more then "a few" % from that. Try to stay > > > within 1-2 %, and you'd probably be safe... > > > > o.k. 2% times 9 = 18% per char. > > I'm not sure why you are making it harder then it is... :) I am not making it harder, I will choose what you suggested, I am simply thinking aloud as they say, because I feel it is important to also understand *why* we do what we do. > > I think you are talking about a half-bit time after a full byte, right ? yes > Remember also that the *other end* can be off in the > other direction. right, good point, my guess is that the pc will have a clock which is a multiple of the baud rate so that, as others have said, with such a crystal there will be no error. > > As have been said before, just stay within 1-2 % and > you will probably be safe. > > Jan-Erik. > > > > --

----- Original Message ----- From: "Bob Blick" <RemoveMEbblickTakeThisOuTsonic.net> To: "Microcontroller discussion list - Public." <piclistEraseME.....mit.edu> Sent: Friday, May 13, 2005 6:14 PM Subject: Re: [PIC] Error % charts of PIC USART ? {Quote hidden}> > In the datasheet of 18F2455 we have > > Table 20-3 Baud Rates For Aynchronous Modes (USART module) > > (I guess other pics would have the same thing as well) > > The errors seem to range from 0.16% up to -45.75%. Are all these > > values valid, or should I be aiming at a certain percentage, I am > > confused? > > Usually if I am designing something using the usart, I choose a crystal > frequency that divides well. For the 16-series PICs and PC baud rates, > 19.6608 MHz is readily available. This seems to fit well in the formula of the 18 series, I think > > For the 18 series with PLL, you will need to choose something 10MHz or less. > > Cheers, > > Bob > > > --

> right, good point, my guess is that the pc will have a clock > which is a multiple of the baud rate so that, as others have > said, with such a crystal there will be no error. Are willing to bet your neck on this? If not you'd better assume that the PC designer and you make the same assumptions, so each side can claim half of the error budget (that is, what's left of it after the transceivers, cable effects etc. have taken their part). Wouter van Ooijen -- ------------------------------------------- Van Ooijen Technische Informatica: http://www.voti.nl consultancy, development, PICmicro products docent Hogeschool van Utrecht: http://www.voti.nl/hvu

----- Original Message ----- From: "Wouter van Ooijen" <EraseMEwoutervoti.nl> To: "'Microcontroller discussion list - Public.'" <RemoveMEpiclistEraseMEEraseMEmit.edu> Sent: Friday, May 13, 2005 7:43 PM Subject: RE: [PIC] Error % charts of PIC USART ? > > right, good point, my guess is that the pc will have a clock > > which is a multiple of the baud rate so that, as others have > > said, with such a crystal there will be no error. > > Are willing to bet your neck on this? If not you'd better assume that > the PC designer and you make the same assumptions, so each side can > claim half of the error budget (that is, what's left of it after the > transceivers, cable effects etc. have taken their part). Well, obviously I will start with low error values, then see how far I can push my luck ... > > Wouter van Ooijen > > -- ------------------------------------------- > Van Ooijen Technische Informatica: http://www.voti.nl > consultancy, development, PICmicro products > docent Hogeschool van Utrecht: http://www.voti.nl/hvu > > > --

> > 19.6608 MHz is readily available. > > This seems to fit well in the formula of the 18 series, I think Yes it does, but you can't PLL (ooops, another noun verbalised) that. If you want to run the 18F at high speed with 0% comms error, use 9.8304MHz. Times 4 = 39.3216MHz. For 20MHz devices, no PLL, 19.6608 or 18.432. IOW divisible by 256, not 10 (as in the case of 20MHz, 10MHz, 4MHz etc) PS I really dislike that table because it emphasises comms error rates when it's so easy to attain 0% with better (sensible ?) xtal selection

> > In the datasheet of 18F2455 we have > > Table 20-3 Baud Rates For Aynchronous Modes (USART module) > > (I guess other pics would have the same thing as well) > > The errors seem to range from 0.16% up to -45.75%. Are all these > > values valid, or should I be aiming at a certain percentage, I am > > confused? > > Consider anything over about 2.5% bad. Don't forget to take oscillator accuracy into this; for example, the onboard osc is usually specced at (IIRC) ~1% (for most PICs), ceramic resonators are frequently ~.5%, and crystals are ~20ppm. In the US, Mouser sells in onsie-twosie quantities a ceramic resonator which gives "good" baud rates, 3.68 MHz. Unfortunately, it's surface mount only. Mike H.

So, just curious.... The percentage off related to the baud rate, means it would be hard pressed to communicate with a device running the standard baud rate. What about two PICs talking over serial....at that point if they are both off by the same percentage, then it really shouldnt matter? __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com

> So, just curious.... > > The percentage off related to the baud rate, means it > would be hard pressed to communicate with a device > running the standard baud rate. > > What about two PICs talking over serial....at that > point if they are both off by the same percentage, > then it really shouldnt matter? Correct. In that case, you're simply using a non-standard baud rate, and unless you go through a bridge of some sort that expects a standard baud rate, you'll be fine. You still need to account for oscillator error, though. For example, let's say your particular oscillator technology guarantees +/-1% accuracy. In that case, you might be in trouble, because if one device is at +1% and the other at -1%, the 2% total error may (or may not) cause problems. Most of the wise elders of the list suggest keeping error at or below 1%. I'm not inclined to argue with them. Mike H.