Post by thread:RS232 possible using internal oscillator?

Bruce Cannon wrote: > > Hello: > Reading the specs for 12Cxx, looks like calibrated osc value can vary nearly > +-500kHz. Has anyone used these parts (with internal osc) for serial comm, > and if so, how? You will be, as long as you re-synch on every start bit on RX. On TX, you can also get away with it if your inter-character spacing is long enough. -- Friendly Regards /"\ \ / Tjaart van der Walt X ASCII RIBBON CAMPAIGN spam_OUTtjaartTakeThisOuTcellpt.co.za / \ AGAINST HTML MAIL |--------------------------------------------------| | GSM Technology for Positioning and Telematics | | Cellpoint Systems SA http://www.cellpt.com | | http://www.wasp.co.za/~tjaart/index.html | | WGS84 -26.0124 +28.1129 Voice +27 (0)11 2545100 | |--------------------------------------------------|

<x-flowed>Bruce Cannon wrote: > Reading the specs for 12Cxx, looks like calibrated osc value can vary nearly > +-500kHz. Has anyone used these parts (with internal osc) for serial comm, > and if so, how? I use the 12C509 @ 2400 baud & the 12C672 @ 9600 baud with no problems. The graph for '672 shows 50-100KHz variation near 25 Celcius not 500KHz. With a regulated 5V supply there should be no problems. Even 100KHz off would be about 2.5% error - still only 25% over a 8N1 byte length. If you sample right in the middle of each bit that still leaves a good margin. Our serial routines are bit-bashed and I used the simulator to tweak the basic bit timing. Regards... ___________________________________ David Duffy Audio Visual Devices Ph: +61 7 38210362 Fax +61 7 38210281 .....AVDKILLspam@spam@uq.net.au Unit 8, 9-11 Trade Street Cleveland, Qld, 4163, Australia. ___________________________________ </x-flowed>

I am using the 12C509 with internal osc. and bit banging RS232 at 9600 baud (8N1). It works fine as long as you plug in correct osc. calibration. It is in one of our products and have seen no problems yet from about 100 units. Bruce Cannon <bcannon@JPS.N To: PICLISTKILLspamMITVMA.MIT.EDU ET> cc: Sent by: pic Subject: RS232 possible using inte rnal microcontrolle oscillator? r discussion list <PICLIST@MITVM A.MIT.EDU> 08/12/99 11:48 PM Please respond to pic microcontrolle r discussion list Hello: Reading the specs for 12Cxx, looks like calibrated osc value can vary nearly +-500kHz. Has anyone used these parts (with internal osc) for serial comm, and if so, how? Bruce Cannon Style Management Systems 1228 Ceres ST Crockett CA 94525 (510) 787-6870 http://www.jps.net/bcannon Remember: electronics is changing your world...for good!

Hello all: I never do a good enough job of explaining myself to start with when I ask questions on this list! Sorry. I should have mentioned that I've already done serial comm with the parts, from 2400 to 9600, in my shop. I noticed some occasional comm problems and dug deeper into the data sheets, where I saw that (in the 508A-519 parts at least), the range for the calibrated osc varies from 3.55 to 4.31MHz. Which I think is another way of saying you could be 450kHz off. And so while it's clear you could tweak by trial and error at both ends to get reliable transmissions based on specific parts for playing around or small volume, it seems it might be tougher to mass produce a cheap high volume product which talks to a PC at, say, 4800bps. (The PC isn't going to tweak, it's just going to choke). What I really meant to ask about was about anyone's experience with a high volume design which works. Thanks for the input, Bruce Cannon Style Management Systems 1228 Ceres ST Crockett CA 94525 (510) 787-6870 http://www.jps.net/bcannon Remember: electronics is changing your world...for good! {Quote hidden}> I use the 12C509 @ 2400 baud & the 12C672 @ 9600 baud with no > problems. > The graph for '672 shows 50-100KHz variation near 25 Celcius > not 500KHz. > With a regulated 5V supply there should be no problems. Even > 100KHz off > would be about 2.5% error - still only 25% over a 8N1 byte > length. If you > sample > right in the middle of each bit that still leaves a good > margin. Our serial > routines > are bit-bashed and I used the simulator to tweak the basic bit timing. > Regards... > ___________________________________ > David Duffy Audio Visual Devices > Ph: +61 7 38210362 Fax +61 7 38210281 > .....AVDKILLspam.....uq.net.au Unit 8, 9-11 Trade Street > Cleveland, Qld, 4163, Australia. > ___________________________________ >

> And so while it's clear you could tweak by trial and error at both ends to > get reliable transmissions based on specific parts for playing around or > small volume, it seems it might be tougher to mass produce a cheap high > volume product which talks to a PC at, say, 4800bps. (The PC isn't going to > tweak, it's just going to choke). What I really meant to ask about was > about anyone's experience with a high volume design which works. You could, however, tweak on the PIC end. The bit time is going to be the same as the narrowest width observed during a char. If you can get the PC to send you a couple of chars, you could sniff out the proper speed values. You'd need to measure and find the width of the bit-time from looking at bytes with a timer (or timing loop) It's workable, provided the host sends reasonable chars, not nulls. Actually, $FF would be easy, since all you'd ever see would be the start bit, or maybe $FE so you can check the last zero and fine-tune.

By high volume you probably mean thousands. For me it's hundreds. Anyway we are shipping units in hundreds using the internal osc of a 12C509 for short strings of RS232 at 9600 baud. I will qualify this and say if I had it to do all over again I would add a $0.40 crystal and two $0.05 caps to guarantee a more accurate osc frequency. We do not do any tweaking with the present OTP's. Just be sure you don't overwrite the built in osc cal values. (Maybe you can't overwrite them I don't remember now) Bruce Cannon <bcannon@JPS.N To: EraseMEPICLISTspam_OUTTakeThisOuTMITVMA.MIT.EDU ET> cc: Sent by: pic Subject: Re: RS232 possible using internal microcontrolle oscillator? r discussion list <PICLIST@MITVM A.MIT.EDU> 08/13/99 01:11 PM Please respond to pic microcontrolle r discussion list Hello all: I never do a good enough job of explaining myself to start with when I ask questions on this list! Sorry. I should have mentioned that I've already done serial comm with the parts, from 2400 to 9600, in my shop. I noticed some occasional comm problems and dug deeper into the data sheets, where I saw that (in the 508A-519 parts at least), the range for the calibrated osc varies from 3.55 to 4.31MHz. Which I think is another way of saying you could be 450kHz off. And so while it's clear you could tweak by trial and error at both ends to get reliable transmissions based on specific parts for playing around or small volume, it seems it might be tougher to mass produce a cheap high volume product which talks to a PC at, say, 4800bps. (The PC isn't going to tweak, it's just going to choke). What I really meant to ask about was about anyone's experience with a high volume design which works. Thanks for the input, Bruce Cannon Style Management Systems 1228 Ceres ST Crockett CA 94525 (510) 787-6870 http://www.jps.net/bcannon Remember: electronics is changing your world...for good! {Quote hidden}> I use the 12C509 @ 2400 baud & the 12C672 @ 9600 baud with no > problems. > The graph for '672 shows 50-100KHz variation near 25 Celcius > not 500KHz. > With a regulated 5V supply there should be no problems. Even > 100KHz off > would be about 2.5% error - still only 25% over a 8N1 byte > length. If you > sample > right in the middle of each bit that still leaves a good > margin. Our serial > routines > are bit-bashed and I used the simulator to tweak the basic bit timing. > Regards... > ___________________________________ > David Duffy Audio Visual Devices > Ph: +61 7 38210362 Fax +61 7 38210281 > AVDspam_OUTuq.net.au Unit 8, 9-11 Trade Street > Cleveland, Qld, 4163, Australia. > ___________________________________ >

>What I really meant to ask about was > about anyone's experience with a high volume design which works. If I had to do it, I would do "autobaud". That requires the PC to send at least one known character to the device to measure the bit time. An odd-code ASCII char works best, since the start bit is always zero, you can time the start bit. Hayes modems, of course, use 0x41h ("A" of the first "AT command ) to autobaud. Measure a bit time in cycles and remember it. Then the oscillator frequency doesn't matter as long as its stable and reasonably fast. If you have to be unidirectional, of course, this won't work. But if you can, you get auto baudrate setting in the bargain. Barry. ------------ Barry King, KA1NLH Engineering Manager NRG Systems "Measuring the Wind's Energy" Hinesburg, Vermont, USA http://www.nrgsystems.com

Bruce Cannon wrote: >Hello: >Reading the specs for 12Cxx, looks like calibrated osc value can vary nearly >+-500kHz. Has anyone used these parts (with internal osc) for serial comm, >and if so, how? > Certainly! We've had good results using '508 devices with 9600 baud comms that rely on the internal oscillator for timing. Our measurements show that it's pretty good at normal temps and ~5V Vdd. To reduce timing errors though, keep transmissions short (stick to 8N1, and preferrably 8N2 if you can afford the time). With 8 data and 1 start and even 5% timing error, you're only looking at 45% total error. By receiving or sending only a few bytes at a time, the receivers have a chance to reset and errors don't accumulate. A 256bit block would likely be impossible to send reliably. Jim Nagy Elm Electronics ICs for Experimenters http://www.elmelectronics.com/

At 20:48 12/08/99 -0700, you wrote: You have had some good replies so far. One point I would make is that there are two different internal OSCCAL schemes used by PICs and one gives much better resolution than the other. Now this is from memory so please check. The 12C508/9 has the poorer scheme. The 12C508A and 509A, 12CE518 and 519, 12C67x have the better scheme. I am not sure about the 16C505. I think it is reasonable to surmise the parts with greater OSCCAL resolution are more accurate than those with the poorer scheme. Maybe this is worth further comment from piclisters? Jim {Quote hidden}>Hello: >Reading the specs for 12Cxx, looks like calibrated osc value can vary nearly >+-500kHz. Has anyone used these parts (with internal osc) for serial comm, >and if so, how? > >Bruce Cannon >Style Management Systems >1228 Ceres ST Crockett CA 94525 >(510) 787-6870 >http://www.jps.net/bcannon > >Remember: electronics is changing your world...for good! > Regards, Jim Robertson NEWFOUND ELECTRONICS ________________________________________ Email: KILLspamnewfoundKILLspampipeline.com.au http://www.new-elect.com MPLAB compatible PIC programmers. ________________________________________