Post by thread:[PIC]: P 'Two speed' clock support

Hi, I've been looking towards using a PIC for a 'dual speed' operation; running the PIC from a (small) battery during a power failure (sleep then wake up, even at the lower oscillator rate......or maybe run all the time) then resuming full power operation when power is restored. As things go, the oscillator speed for achieving the reduced power mode requirements does not provide enough processing "oomph" when full power is available. Hence, dual-clock operation would be beneficial. I've had a poke around on the mchip site/checked out a few data sheets and I cannot really find too much information or a specific micro that performs this. Mitsubishi etc make micros that specifically support this - is it blatantly obvious how to do this with the PIC and I have missed it, or is it just not possible? I've seen a posting where an RC oscillator is used where additional resistance is switched in and out via an I/O pin (dual speed) but I don't think RC is going to give me the clock accuracy that is required. Any comments would be appreciated. Regards, Dan

I know that it is possible to run a PIC set to RC mode with an external oscillator. I am sure some of our hardware experts can tell you how. Perhaps you can then use real RC for the low speed (assuming you don't need the timing precision in low-power mode) and an external osc module for high speed. One thing to watch out for: when in RC mode there are differences in the oscillator start up timer, and the timing when coming out of sleep. I'd recommend the hardware defaulting to RC, then switching to the XTAL under software control. You are going to have to be very careful about the timing at the transition point to avoid getting a clock cycle that doesn't meet the chip's timing. Bob Ammerman RAm Systems (high performance, high function, low-level software) {Original Message removed}

I believe I've seen oscillators with built in selectable dividers, but I cannot remember where. You can use a seperate oscillator and divider, and use a pic output or two to set the divider when necessary. -Adam D Lloyd wrote: {Quote hidden}> > Hi, > > I've been looking towards using a PIC for a 'dual speed' operation; running the > PIC from a (small) battery during a power failure (sleep then wake up, even at > the lower oscillator rate......or maybe run all the time) then resuming full > power operation when power is restored. As things go, the oscillator speed for > achieving the reduced power mode requirements does not provide enough processing > "oomph" when full power is available. Hence, dual-clock operation would be > beneficial. > > I've had a poke around on the mchip site/checked out a few data sheets and I > cannot really find too much information or a specific micro that performs this. > Mitsubishi etc make micros that specifically support this - is it blatantly > obvious how to do this with the PIC and I have missed it, or is it just not > possible? > > I've seen a posting where an RC oscillator is used where additional resistance > is switched in and out via an I/O pin (dual speed) but I don't think RC is going > to give me the clock accuracy that is required. > > Any comments would be appreciated. > > Regards, > > Dan

part 1 2451 bytes content-type:text/plain; charset=us-ascii Hi, So it would seem that there is no 'on chip' support for such functionality.... The application is cost sensitive so precludes the use of anything too messy/complicated/costly externally. That's not to mention the hazards, as Bob suggested, of dodgy edges. Looks like another route is necessary. Thanks again, Dan |-------------> |(Embedded | |image moved | |to file: | |pic10075.pcx)| | | |-------------> >------------------------------------------------------------------------| |"M. Adam Davis" <spam_OUTadavisTakeThisOuTUBASICS.COM> | |28/06/2000 13:13 | >------------------------------------------------------------------------| Please respond to pic microcontroller discussion list <.....PICLISTKILLspam@spam@MITVMA.MIT.EDU> To: PICLISTKILLspamMITVMA.MIT.EDU cc: (bcc: Dan Lloyd/GBPTD/ABB) Subject: Re: [PIC]: P 'Two speed' clock support Security Level:? Internal I believe I've seen oscillators with built in selectable dividers, but I cannot remember where. You can use a seperate oscillator and divider, and use a pic output or two to set the divider when necessary. -Adam D Lloyd wrote: > > Hi, > > I've been looking towards using a PIC for a 'dual speed' operation; running the > PIC from a (small) battery during a power failure (sleep then wake up, even at > the lower oscillator rate......or maybe run all the time) then resuming full > power operation when power is restored. As things go, the oscillator speed for > achieving the reduced power mode requirements does not provide enough processing > "oomph" when full power is available. Hence, dual-clock operation would be > beneficial. > > I've had a poke around on the mchip site/checked out a few data sheets and I > cannot really find too much information or a specific micro that performs this. > Mitsubishi etc make micros that specifically support this - is it blatantly > obvious how to do this with the PIC and I have missed it, or is it just not > possible? > > I've seen a posting where an RC oscillator is used where additional resistance > is switched in and out via an I/O pin (dual speed) but I don't think RC is going {Quote hidden}> to give me the clock accuracy that is required. > > Any comments would be appreciated. > > Regards, > > Dan part 2 165 bytes content-type:application/octet-stream; (decode) part 3 2 bytes -

Dan Lloyd wrote: >I've been looking towards using a PIC for a 'dual speed' operation; running the >PIC from a (small) battery during a power failure (sleep then wake up, even at >the lower oscillator rate......or maybe run all the time) then resuming full >power operation when power is restored. As things go, the oscillator speed for >achieving the reduced power mode requirements does not provide enough processing >"oomph" when full power is available. Hence, dual-clock operation would be >beneficial. ........... > 1. Checkout the ICS5xx PLL Clock Multiplier chips from http://www.icst.com/ 2. I believe the new Mchp 18Cxxx chips have a configurable PLL clock on board. 3. Also, the 16C73/74/.. chips have a 2nd oscillator on board wired to pins C0/C1, and which can be used with a 32khz xtal. Can be used with sleep mode and any speed xtal on main oscillator. best regards, - Dan Michaels Oricom Technologies http://www.sni.net/~oricom ==========================

How much current would an external clock driver w/ selectable dividers draw as opposed to a "traditional" high-speed crystal w/ pic/sx in sleep mode when not in use? Is there such a thing as a selectable divider regular crystal? Also I've seen it proposed to switch between low speed RC and high speed crystal; how can one do that? Is'nt that a setting burned at program time? I ----- Original Message ----- From: "M. Adam Davis" <.....adavisKILLspam.....UBASICS.COM> > I believe I've seen oscillators with built in selectable dividers, but I cannot > remember where. You can use a seperate oscillator and divider, and use a pic

----- Original Message ----- From: dal wheeler <EraseMEdwheelerspam_OUTTakeThisOuTINSIGHTEK.COM> To: <PICLISTspam_OUTMITVMA.MIT.EDU> Sent: Wednesday, June 28, 2000 2:18 PM Subject: Re: [PIC]: P 'Two speed' clock support > How much current would an external clock driver w/ selectable dividers draw > as opposed to a "traditional" high-speed crystal w/ pic/sx in sleep mode > when not in use? Is there such a thing as a selectable divider regular > crystal? Also I've seen it proposed to switch between low speed RC and high > speed crystal; how can one do that? Is'nt that a setting burned at program > time? > I This can be done by setting the chip to RC mode, but then driving it with an external oscillator (_not_ a crystal!). Again, the trick would be to arrange for 'glitchless' transitions. Bob Ammerman > ----- Original Message ----- > From: "M. Adam Davis" <@spam@adavisKILLspamUBASICS.COM> > > I believe I've seen oscillators with built in selectable dividers, but I > cannot > > remember where. You can use a seperate oscillator and divider, and use a > pic

Hi. I've lost the start of this discussion, which PIC did you intend using, I've an idea that would work on 16C73 or up possibly. Use the RC osc, but have a way to trim the frequency in software. A DAC is probably too expensive but filtering a PWM output might do. Fit a 32KHz xtal on the timer1 osc. Use the crystal to 'calibrate' the main osc. A switched resistor would provide the main frequency switch function, keep the trim down to 10% Never tried it myself, however I've had the luxury of working on mains powered projects so far. Oliver.

Something I havn't tried <grin> but have dreamt up, is using the new flash parts... Seeing as they have self reprogramming, it would be possible to reprogram the config bits at 2100h onwards to change the part from RC to XT. Ok. Now the fun bit. Power a crystal oscillator module from a PIC i/o pin. Drive a crystal oscillator into a resistor do ground, through a schmitt trigger inverter into the top of an RC series combination to ground. Then take the junction from the R and C to the OSC1 pin - with me? The RC is calculated for your initial frequency in RC oscillator mode. The pic would wake from sleep, and because the crystal oscillator is switched off at this point, the R to ground on the input to the schmitt inverter will cause the output to be a logic 1. This provides the necessary +V to run the RC oscillator, into OSC1. When powered up, the PIC switches the Oscillator module on, and then reprograms itself to XT mode. Geddit? This would be intended for timing sensitive applications, where RC is not accurate enough, but when low power standy, possibly interrupt or watchdog wake from sleep is used. Any thoughts? Kevin. ----- Original Message ----- From: Gennette, Bruce <KILLspambruce.gennetteKILLspamTAFE.NSW.EDU.AU> To: <RemoveMEPICLISTTakeThisOuTMITVMA.MIT.EDU> Sent: Monday, July 03, 2000 7:39 AM Subject: Re: [PIC]: P 'Two speed' clock support > I haven't followed this fully myself, but how about an R-C oscillator with > the R being a light dependant resistor. Hook an LED up to an output pin and > you can switch the value of the resistor quite a long way. (need to seal the > pair in a light proof manner) > > Better informed experts than me can calculate the R-C and wether the LED has > to be attenuated with a filter. > > Should give glitch free speed changing as the LED powers up or down. > > Bye. > > {Original Message removed}

Hi, Thanks to everyone offering suggestions to overcome this problem. Unfortunately for many of the suggestions, the application is connected to an electricity meter of which there is no option for a "reset" or "just cycle the power and it will be fine" and so the solution needed to be absolutely bullet-proof. I can think of only the (expensive) switched external oscillator or a multispeed micro being the only real solutions. I'll certainly consider them for other projects, though! Thanks again. Regards, Dan -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

Kevin Blain wrote: > > Something I havn't tried <grin> but have dreamt up, is using the new flash > parts... > > Seeing as they have self reprogramming, it would be possible to reprogram > the config bits at 2100h onwards to change the part from RC to XT. Ok. Now > the fun bit. > You cannot reprogram the CONFIG register from within the PIC. Actually, on writing this I thought hang on, perhaps you can. I fired up the ROMzap PCB and wrote a quick Delphi prog to talk to it and tried to program the config register but no go. This would make sense because you would then be able to easily defeat a code protected device. -- Best regards Tony http://www.picnpoke.com spamBeGonesalesspamBeGonepicnpoke.com -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

>electricity meter, bulletproof Write the clock switching routine such that it also sets OPTION for a short watchdog timeout before switching, and allow the PIC to be reset by the watchdog after changing the clock (both ways). You could use a 74S66F single CMOS chip as clock switch. Just place 120R in series with the other crystal (the 74S66F has ~60R open). Then the Q rule will cause the crystal with the higher Q to run the oscillator. That's it. Or almost. Peter -- http://www.piclist.com hint: To leave the PICList RemoveMEpiclist-unsubscribe-requestTakeThisOuTmitvma.mit.edu>

Though not what was originally requested (2 speed clock with one being crystal, the other being RC), you could possibly do a 2 speed RC by connecting another resistor between the RC input of the PIC and an I/O pin. Make the pin an input for low speed. Program it as a high output (putting the second resistor in parallel with the first) for high speed. Harold ________________________________________________________________ YOU'RE PAYING TOO MUCH FOR THE INTERNET! Juno now offers FREE Internet Access! Try it today - there's no risk! For your FREE software, visit: dl.http://www.juno.com/get/tagj. -- http://www.piclist.com hint: PICList Posts must start with ONE topic: [PIC]: PIC only [EE]: engineering [OT]: off topic [AD]: advertisements

<snip> Though not what was originally requested (2 speed clock with one being crystal, the other being RC), you could possibly do a 2 speed RC by connecting another resistor between the RC input of the PIC and an I/O pin. Make the pin an input for low speed. Program it as a high output (putting the second resistor in parallel with the first) for high speed. <snip> The main problem with using RC arrangements is that I need a clock drift of sub 0.5 seconds per day. I was of the opinion that this was not possible with RC sources (sorry if this formatting is haywire, Lotus Bloats is having a bad day) - certainly not over a wide temperature range? Thanks, though Dan -- http://www.piclist.com hint: PICList Posts must start with ONE topic: [PIC]: PIC only [EE]: engineering [OT]: off topic [AD]: advertisements

D Lloyd <dan.lloydEraseME.....GB.ABB.COM> wrote: >The main problem with using RC arrangements is that I need a clock drift >of sub 0.5 seconds per day. I was of the opinion that this was not >possible with RC sources Wow. 0.5 sec/day is better than 5.8 ppm and that is hard even for a crystal ? You can achieve 0.5 sec/day with a cheap 32 kHz resonator, trimmed manually or electronically at factory, aged (baked), and at relatively constant temperature. A RC oscillator for 5 ppm would probably be made entirely of noble materials to avoid aging, kept at a regulated +/- 0.1 degrees temperature at constant pressure, have temperature compensation (in despite of thermostat) etc. 0.5 sec/day is 0.5/86400 = .0000057870 (5.79 ppm) Peter -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

Harold Hallikainen wrote: > Though not what was originally requested (2 speed clock with one >being crystal, the other being RC), you could possibly do a 2 speed RC by Which reminds me that I did not see the orignal posting. I suggested using a 74S66F switch to change crystals, but for RC, is is possible that a collector-switched bipolar transistor stage will do the trick. This is pure imagination so far, but it may work. The scheme is: PIC + parallel resonant crystal + 2 small capacitors connected as usual. Coupling cap from Xout to base of additional NPN. Emitter of NPN through ~2k7 to GND. Collector of NPN through 10K to PIC IO pin. Collector of NPN through coupling cap to Xin. Base of NPN polarized from collector through ~150k. The Xin coupling capacitor should be very small (the same order of magnitude as the crystal 'pi' capacitor on that side). Now, when the PIC is reset the IO pin is tristate and the NPN collector has no current into it. The NPN arrangement should allow the PIC Xtal oscillator to start as usual in Xtal mode. To switch to RC mode, set the PIC pin to output and HIGH at the same time. Now the NPN stage is biased and the crystal appears as NFB across its input and output. This will hopefully stop the crystal oscillation and then the ac coupled 'gate ring' oscillator formed by the NPN stage its coupling capacitors and the internal PIC oscillator gate should take over at a frequency that is to be determined (but should be obscenely low - tens of kHz probably). Problem: the conducting NPN will draw more current than the PIC ;-) Solution (?) replace NPN stage with a single tristate inverting CMOS gate (4S24?F I think - I've never seen one though they certainly exist) and AC couple its output only this time. Steer the enable pin with the PIC IO pin. Last alternative: Use a 'A' series CMOS inverter gate as external inverter and steer by switching power using PIC IO pin, hoping for sufficient isolation between in and out for the Xtal to oscillate with power off. If anyone will have success with this scheme (which I have no time to try myself) then please report. Peter -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics

part 1 1832 bytes content-type:text/plain; charset=us-ascii Hi, I might have been a bit zealous with the 0.5 sec/day. 1 second per day is more like it, I think. Still not attainable by RC, I think, not over the temp range....... I think I'll have to just keep it simple and go for the on board option. Dan |-------------> |(Embedded | |image moved | |to file: | |pic20241.pcx)| | | |-------------> >------------------------------------------------------------------------| |"Peter L. Peres" <EraseMEplpACTCOM.CO.IL> | |06/07/2000 18:24 | >------------------------------------------------------------------------| Please respond to pic microcontroller discussion list <RemoveMEPICLISTEraseMEEraseMEMITVMA.MIT.EDU> To: RemoveMEPICLISTspam_OUTKILLspamMITVMA.MIT.EDU cc: (bcc: Dan Lloyd/GBPTD/ABB) Subject: Re: [PIC]: P 'Two speed' clock support Security Level:? Internal D Lloyd <RemoveMEdan.lloydTakeThisOuTspamGB.ABB.COM> wrote: >The main problem with using RC arrangements is that I need a clock drift >of sub 0.5 seconds per day. I was of the opinion that this was not >possible with RC sources Wow. 0.5 sec/day is better than 5.8 ppm and that is hard even for a crystal ? You can achieve 0.5 sec/day with a cheap 32 kHz resonator, trimmed manually or electronically at factory, aged (baked), and at relatively constant temperature. A RC oscillator for 5 ppm would probably be made entirely of noble materials to avoid aging, kept at a regulated +/- 0.1 degrees temperature at constant pressure, have temperature compensation (in despite of thermostat) etc. 0.5 sec/day is 0.5/86400 = .0000057870 (5.79 ppm) Peter -- http://www.piclist.com hint: The list server can filter out subtopics (like ads or off topics) for you. See http://www.piclist.com/#topics part 2 165 bytes content-type:application/octet-stream; (decode) part 3 129 bytes -- http://www.piclist.com hint: The PICList is archived three different ways. See http://www.piclist.com/#archives for details.

>1 sec/day which is about 12ppm. By looking in Mouser we see a 32kHz crystal that claims +/- 20 ppm. If you use this one and trim it in factory then you may have a fair chance. Perhaps you can avoid clock switching completely, and run your project all the time at 32kHz. Costs ~= 0.73 in 100's (warning: I do not have the latest Mouser). A TCXO (FOX) that requires no trimming and does 2.5 ppm over temperature and 0.3 ppm over supply will certainly fix your problem, assuming indoor use (restricted temp. range) at a mere $38.85 ea. in 100's ;-) Peter -- http://www.piclist.com hint: To leave the PICList EraseMEpiclist-unsubscribe-requestspamspamBeGonemitvma.mit.edu>